Three Essays on Development Economics in China
by
Nancy Qian
B.A., University of Texas at Austin (2000)
M.A. Massachusetts Institute of Technology (2002)
Submitted to the Department of Economics
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
at the
MASSACHUSETTS INSTITUTE OF
June 2005
© Nancy Qian 2005
The author hereby grants to Massachusetts Institute of Technologypermission to
reproduce and
to distribute copies of this thesis document in whole or in part.
Signature of Author..
.... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
..... .
Department of Economics
/f\
May 2005
c,I
Certifiedby............
.. .........................................
Esther Duflo
Professor of Economics
Thesis Supervisor
Certifiedby............
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.. .. ..
...
.
.
Abhijit Banerjee
Ford International Professor of Economics
Thesis Supervisor
Certifiedby....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
I .
.
.
.
.
.
.
.
.
Joshua Angrist
Professor of Economics
Thesis Supervisor
Accepted
by.......
....
.......................
...................................
Peter Temin
Chairperson, Department Committee on Graduate Students
AHOCIiiv =,;
Three Essays on Development Economics in China
by
Nancy Qian
Submitted to the Department of Economics
on May 2005, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
Abstract
This dissertation is a collection of three independent essays in empirical development economics
using data from China. In the first two chapters, I examine the determinants of choices within
the household. In the first chapter, I estimate the causal effects of total income, relative female
and relative male income on sex imbalance. The second chapter studies the effects of relaxations
in the One Child Policy on sex ratios and family size and then exploits the exogenous variation
in family size caused by the relaxations to estimate the causal effect of family size on school
enrollment. The third chapter is a descriptive study of income inequality for top income earners
in China during 1986-2002 and the potential redistributive effectiveness of progressive income
taxation.
Thesis Supervisor: Esther Duflo
Title: Professor of Economics
Thesis Supervisor: Abhijit Banerjee
Title: Ford International Professor of Economics
Thesis Supervisor: Joshua Angrist
Title: Professor of Economics
2
Three Essays on Development Economics in China
by
Nancy Qian
Submitted to the Department of Economics
on May 2005, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
Abstract
This dissertation is a collection of three independent essays in empirical development economics
using data from China. In the first two chapters, I examine the determinants of choices within
the household. In the first chapter, I estimate the causal effects of total income, relative female
and relative male income on sex imbalance. The second chapter studies the effects of relaxations
in the One Child Policy on sex ratios and family size and then exploits the exogenous variation
in family size caused by the relaxations to estimate the causal effect of family size on school
enrollment. The third chapter is a descriptive study of income inequality for top income earners
in China during 1986-2002 and the potential redistributive effectiveness of progressive income
taxation.
3
Acknowledgements
I thank my thesis advisors Esther Duflo, Abhijit Banerjee and Joshua Angrist for
much more than words can express.
I thank Daron Acemoglu, David Autor, Dick Eckaus, Sendhil Mullainathan
and
Thomas Piketty for sharing their numerous insights with me. I also thank my
fellow students Karna Basu, Matilde Bombardini, Thomas Chaney, Shawn Cole,
Antara I)utta, Rema Hanna, Geraint Jones, Bill Kerr, Jin Li, Byron Lutz, Daniel
Paravisini, Ver6nica Rappaport, Tali Regev, Ruben Segura-Cayuela, Henry Tang,
Petia Topalova and Ding Wu for making learning fun; and especially Ivan FernandezVal for being the ideal office-mate in every respect.
I thank Alfred Norman, Daniel Slesnick, Jacklin Chou and Kenneth Fortson for
encouraging me to attend graduate school in economics.
I acknowledge financial support from the National Science Foundation Graduate Research Fellowship, the Social Science Research Council Fellowship for Development
and Risk and the MIT George P. Schultz Fund.
I am particularly
grateful to Ashley Lester for his intellectual input, continuing
support and endless patience.
I dedicate this thesis to my family who have enriched my life, but especially my
parents, Shie Qian and Jun Lou.
5
I
Contents
1 Missing Women and the Price of Tea in China: The Effect of Relative Female
Income on Sex Imbalance
13
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . ............
. . . . . . . . . . ...................
13
1.2
1.3
Background
. . . . . . . . . . . . . . . . . . . . . . . . . . . ............
. . . . . . . . . . ..................
17
1.2.1
Previc
)usWorks
................................ ..........
1.2.2
Agriciultural Reforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.2.3
Tea aind Orchard
Conceptual
Production
. 17
. . . . . . . . . . . . . . . . . . . . . . . . . 21
F'ramework ..................................
.
24
1.3.1
Decisi on Rule ..................................
24
1.3.2
House!hold Utility ................................
25
.......................................
28
1.4
The Data
1.5
Empirical Sti rategy...................................
30
1.5.1
Identi fication ..................................
30
1.5.2
Basic Results ..................................
32
1.5.3
Differ, ences-in-Differences
1.5.4
Robus
1.5.5
Two Stage Least Squares
......................................
stness
..................................
.
.........
......................................
1.6 Results on E,ducation Attainment ......................................
1.7
Conclusion
1.8
Appendix-
. 34
.
36
. 37
.......................................
robustness of Linear Specification ....................
2 Quantity-Quality: The Positive Effect of Family Size on School Enrollment
7
34
40
47
in China
49
2.1 Introduction .......................
...
49
2.2
Background ........................
...
53
2.2.1
...
53
2.2.2 Rural Education ................
...
54
2.2.3
...
55
2.3 Data ............................
...
56
2.4
Empirical Framework .................
...
57
2.4.1 Identification ...................
...
57
2.4.2 The Effect of the 1-Son-2-ChildRelaxation . .
...
59
2.4.3
...
63
...
66
2.5
Family Planning Policies .............
Conceptual Framework ............
The Effect of Family Size on School Enrollment
Conclusion
........................
3 Income Inequality and Progressive Income Taxation in China and India,
1986-2010
3.1 Introduction
............
73
............
..
. . . . ..
. . . 73
...
..
. . . 76
...
. . . . ..
. . . 77
...
3.2 Data and Methodology ..................
. . . . ..
3.3
Top Income Shares in China and India, 1986-2001 . . .
..
3.4
Progressive Income Taxation in China and India, 1986-2010 ....
8
. . . 79
.
This dissertation is a collection of three independent essays in empirical development economics using data from China. In the first two chapters, I examine the determinants of choices
within the household. In the first chapter, I estimate the effects of total income, relative female and relative male income on sex imbalance.
The second chapter studies the effects of
relaxations in the One Child Policy on sex ratios and family size and exploits the exogenous
variation in family size to estimate the effect of family size on school enrollment. The third
chapter is a descriptive study of income inequality for top income earners in during 1986-2002
and the potential redistributive effectivenessof progressive income taxation.
The first chapter is entitled "Missing Women and the Price of Tea in China: The Effect of
Relative Female Income on Sex Imbalance". This essay evaluates the effects of an increase in
relative adult female income, and increase in relative male income and an increase in overall
income on sex imbalance for cohorts born during 1962-1990 in rural China. For these cohorts,
the fraction of males in the surviving population increased from 50% to 54%. The first question
this study answers is whether economics factor into parents' desire for girls relative to boys. It
also addresses a long standing debate between economists who claim that richer households demand relatively more girls than boys and evolutionary biologists who hypothesize the opposite,
that boys are luxury goods relative to boys. Furthermore, the study addresses the questions
of whether it is necessary to increase female income as well (or instead) of increasing overall
household earnings and whether this is due to the correlation between productivity increases for
mothers and daughters or to an increase in the woman's bargaining power within the household.
The main difficulty in empirically estimating the observed association between sex ratios and
economic conditions cannot be interpreted as causal since it reflects omitted variables such as
sex preference. This paper exploits regional variation in sex-specific labor input across crops
and exogenous increases in agricultural income and relative female and male income caused by
post-Mao reforms in China which increased the value of planting cash crops relative to staple
crops. In particular, it uses the increase in relative value of tea and the fact that women tea is
picked by women to estimate the effect of an increase in relative female income on sex ratios; and
the increase in relative value of orchards and the fact that men have an advantage in producing
orchards to estimate the effect of an increase in relative male income on sex ratios. I am also
9
able to estimate the effect of an increase in total household income without changing the relative
shares of male and female income by estimating the effect of planting sex-neutral cash crops on
sex ratios. The results show that increasing income alone has no effect on sex ratios whereas
increasing relative female income increases survival rates for girls and increasing relative male
income decreases survival rates for girls. Moreover, the findings show that increasing mothers'
incomes increase education attainment for all children while increasing fathers' incomes decrease
education attainment for girls and have no effect on boys' education attainment.
The results
can be best explained by a model of the household where mothers value education more than
fathers and an increase in relative female income increases the weight of the mother's preferences within the household, and thereby increase female survival rates and education spending
on both children. The policy implications are clear. Making households richer will not alleviate
sex imbalance. Policy makers should target women either through subsidies or by increasing
their employment opportunities. My calculations show that a 20% increase in rural household
income that went entirely to the adult females of the household would have balanced China's
sex ratios.
The second chapter is entitled "Quantity-Quality:
The Positive Effect of Family Size on
School Enrollment". This essay studies the effect of having an additional sibling on the school
enrollment of the first born child. Policy makers in many developing countries view restricting
population growth as a helpful measure in increasing average human capital. Their belief is
consistent with the observed negative correlation between quantity and quality of children across
countries and across households within countries. However, because parents simultaneously
choose the quantity and quality of their children, the observed correlation between family size
and child outcomes cannot be interpreted as causal. This study uses the exogenous increase
in family size caused by a relaxation in the One Child Policy to estimate the effect of family
size on school enrollment. Specifically, it uses the relaxation which allows a rural household to
have a second child if the first is a girl. This relaxation began in 1982 and enforcement varied
across regions. I use this time and regional variation to first show that the "l-son-2-child"
rule increased family size for first born girls. And contrary to previous findings, the results
show that this exogenous increase in family size to find that an additional sibling increased
school enrollment of the first child by 10-20%. Furthermore, I am able to use the relaxation
10
to examine the causal impact of the One Child Policy on sex ratios, which until now has not
been carefully studied.
I show that the One Child Policy decreased the fraction of girls in
the surviving population by up to 10 percentage-points in some regions and the relaxation
was only partially successful in mitigating the problem. Although more research is needed to
understand the effect of quantity-quality outside of the one child context, these findings cast
doubt on the hypothesis that quality is monotonically decreasing with quantity and the idea
that restrictive family planning policies will necessarily help to achieve higher average human
capital investment for developing countries.
The third chapter is entitled "IncomeInequality and Progressive Income Taxation in China
and India, 1986-2010". It is coauthored with Thomas Piketty. This paper evaluates the
prospects for income tax reform in China during the coming decade (with a comparison to
India), and argues that such reforms should rank high on the policy agenda in these two countries.
Due to high average income growth and sharply rising top income shares during the
1990s, progressive income taxation is about to raise non-trivial tax revenues in China and India
and to become an important political object. According to our projections, the income tax
should raise at least 3% of Chinese GDP in 2010 (versus less than 1% in 2000 and 0.1% in
1990), in spite of the 20% nominal rise in the exemption threshold that took effect in 2004.
The fact that progressive income taxation is becoming an important policy tool has important
consequences for China's ability to finance social spendings and to keep under control the rise
in income inequality associated to globalization and growth. Due to faster income growth and
to a higher fraction of wage earners in the labor force, the prospects for income tax development
look better in China than in India. This potential is however limited by the fact that Chinese
top wage-earners are currently severely under-taxed relatively to top non-wage income earners.
11
I
Chapter 1
Missing Women and the Price of
Tea in China: The Effect of Relative
Female Income on Sex Imbalance
1.1
Introduction
Mlany Asian populations are characterized by highly imbalanced sex ratios. For example, only
48.4% of the populations of India and China are female in comparison with 50.1% in western
Europe.
Amartya Sen (1990, 1992) coined the expression "missing women" to refer to the
observed female "deficit" in comparing sex ratios of developing countries with sex ratios of
rich countries. An estimated 30-70 million women are "missing" from India and China alone.
This phenomenon is almost certainly due to behavioral factors that reflect a preference for
male children (selective abortion, infanticide and/or neglect).1 In the long run, male-biased sex
'There are two recent studies arguing that sex imbalance is caused by biological factors unrelated to economic
conditions. Norberg (2004) finds that parents living with a spouse or opposite-sex partner were 14% more likely
to have a male child in the U.S. However, since there is no evidence of increased cohabitation during this period
in China and divorce rates were rising, her findings would suggest that the observed sex imbalance under-reports
sex selection.
In a recent study of the impact of hepatitis B on sex ratios, Oster (2004) argues that 75-85% of the observed
sex imbalance in China is explained by the effect of hepatitis infection of pregnant mothers on miscarriage of
female fetuses. However, Figure 1 shows that sex imbalance increases for cohorts born 1976 and after. (By
comparing sex ratios by age from the 1982 Population Census with the 1990 Population Census, Figure 1 shows
that in 1990, there are more males for every age under 14. In other words, the increasing sex imbalance observed
13
ratios can benefit women by increasing their price in the marriage market (Angrist, 2002). A
more immediate concern, however, is that to select the sex of the child, parents without access
to pre-natal gender revealing technologies must resort to infanticide or other forms of neglect
which ultimately lead to the death of a child.
Economists, beginning with Becker (1981) and Rosenzweig and Schultz (1982), have long
argued that sex imbalance responds to economic conditions. The negative cross-country correlation between income and sex imbalance corroborates this argument.
However, the sex
imbalance within China is increasing, not decreasing, despite rapid economic growth. Figure 1
plots sex ratios by age from the 1982 and 1990 China Population Censuses and the 1990 U.S.
Population Census. It shows that in 1982, sex ratios by age in China were very similar to that
of the U.S. But in 1990, there are more males for every age under 14. In other words, sex ratios increased for cohorts born 1976 and after, which coincides with the beginning of post-Mao
market reforms that led to an increase in GDP per capita and an increase in the gender wage
gap. 2 This is consistent with the alternative explanation most recently posited by Foster and
Rosenzweig (2001) which argues that sex ratios respond to sex-specific economic conditions.
For example, parents may wish to avoid having female children when marriage requires a large
dowry. Or, the demand for girls relative to boys may increase when female productivity increases. The main empirical challenge in establishing the link between sex ratios and economic
conditions (including sex-specific economic conditions) is that both sex ratios and economic
variables reflect omitted variables such as sex preference.
The principal contribution of this paper is to develop and implement a strategy that captures
the causal effect of economic conditions on sex ratios in China using exogenous variation in
regional incomes over time. In particular, I exploit the variation in intensity of adult female
and male labor input across crops and the exogenous variation in agricultural income caused by
in the 1990 census data is a cohort effect rather than an age effect).Since there is no evidence of an increase
in hepatitis B infection rates during this period (if anything, infection rates should have decreased due to the
introduction of a vaccine), and because infection rates are likely to be correlated with socioeconomic variables
which may affect sex imbalance directly,.it is likely that her results overestimate the true effect of hepatitis B.
2Accurate
estimates of rural incomes during the early reform period are prevented by both the lack of data and
the fact that rural workers did not receive wages. However, there is a general consensus between conventional
wisdom and studies done using retrospective data that the gender wage gap began increasing with the onset
of market reforms. This is consistent with the fact that before the reform, compensation for workers were set
according to education, experience and skill. There was no differentiation, at least officially, between sexes (Cai
et. al., 2004, Rozelle et. al. 2002).
14
two post-Mao reforms (1978-1980). The identification strategy of using exogenous price changes
in sex-specific agricultural products to identify effects of changes in female-to-male wage ratio
is similar to Schultz's (1985) study on Swedish fertility rates in the late 19th century. He used
changing world grain prices to instrument for changes in the female-to-male wage ratio.
In
China, women are more productive in picking tea and men are more productive in orchard
production.
ence, an increase in the relative value of tea increases the total income and
the relative female income in tea producing households while an increase in relative value of
orchards increases the total income and the relative adult male income in orchard producing
households.
This study uses a differences-in-differences framework to compare sex ratios for cohorts
born before and after the reforms, between counties that plant and counties that do not plant
sex-specific crops that experienced a value increase due to the reform. First, I estimate the
effect of an increase in adult female income holding adult male income constant on sex ratios
by estimating the effect of an increase in relative tea value on sex ratios. Second, I estimate
the effect of an increase in adult male income while holding adult female income constant on
sex ratios by estimating the effect of an increase in relative value of orchards on sex ratios.
Third, I investigate the effect of an increase in total household income without changing the
proportion of female and male incomes by estimating the effect of an increase in the relative
value of sex-neutral cash crops on sex ratios. Finally, by repeating the experiments above for
education attainment, I am able to estimate the effects of increasing total and relative incomes
on education investment for boys and girls.
Setting the study in China during the period of 1962-1990has the advantages that migration
was strictly controlled and little technological change occurred in tea or general agricultural
production and sex-revealing technologies were unavailable to China's rural population for most
cohorts in the study (Diao et. al., 2000; Zeng, 1993). The implementation of the One Child
Policy largely controls for family size. To ensure that the results are not confounded by family
size controls, I repeat the study on a sample containing only ethnic minorities (non-Han) who
have never been subjected to family planning policies. The results between the non-Han sample
and the whole sample are very similar. 3 An additional benefit of this study is that by measuring
3
Rural areas in China received relaxations to the One Child Policy beginning in 1982. Using relaxation data
15
the effect of sex-specific wages on sex imbalance and education investment, it study can speak
to concerns regarding the impact of China's increasing gender wage gap.
The results show that an increase in relative adult female income has an immediate and
positive effect on the survival rate of girls. In the early 1980s, in rural China, increasing adult
female income by US$7.70 (10% of average rural household income) while holding adult male
income constant increased the fraction of surviving girls by 1 percentage-point.
4
Conversely,
increasing adult male income while holding adult female income constant decreased survival
rates for girls. Increasing total household income without changing the relative incomes of men
and women had no effect on survival rates. I also find that increasing adult female income while
holding adult male income constant increased education investment for both boys and girls,
whereas increasing adult male income while holding adult female income constant decreased
education investment in girls and had no effect for boys. Increasing total household income
without changing the relative shares of male and female income had no effect of education
attainment for anyone. The results imply that the post-reform increase in gender wage can
partly explain the increase in sex imbalance and the decrease in rural education enrollment
during the 1980s.5 Furthermore, they shed light on the mechanisms underlying the empirical
results. While the effects on survival can be explained by either a model of the household with
intra-household bargaining or by a unitary model of the household where parents view children
as a form of investment, the results on education are not consistent with a model where children
are viewed as pure investment goods and where households are unitary unless the returns to
education for girls are negatively correlated with male income and the returns to education
for both boys and girls are positively correlated with female income. Therefore, the results for
survival and education investment together suggest that at least part of the effect is due to
changes in the bargaining power of the woman in the household.
In addition to being of general scientific interest, the results point to the possibility of nonfrom the China Health and Nutrional Survey, I find that tea counties and non-tea counties are equally likely to
receive the relaxation. The sample with policy enforcement data amongst provinces which plant tea is not large
enough to be used in statistical analysis.
4
This estimate is calculated using the empirical results of this study, data on crop composition from the 1997
Agricultural Census and net income data reported by Etherington and Forster (1992). The estimate assumes
that the elasticity of demand for girls relative to boys with respect to relative female income is constant.
5
Both male and female school enrollment decreased in China during the early reform period (Hannum and
Park, Mimeo).
16
coercive policies that can affect sex ratios. In particular, the results presented here suggest that
factors that increase the economic value of women will also increase the probability that female
infants are carried to term and female children live to adulthood.
The paper is organized as follows. Section 2 describes the literature and policy background.
Section 3 presents the conceptual framework. Section 4 describes the data. Section 5 discusses
the empirical strategy and results for sex ratios. Section 6 discusses the results for education.
Section 7 offers conclusions.
1.2
1.2.1
Background
Previous Works
Since Becker (1981) first argued that sex preference reflects underlying economic conditions,
many studies have shown that sex ratios are often correlated with household income. However,
the nature of this relationship is anything but settled. Becker (1981) theorizes that increased
income leads to an increase in the relative demand of girls. This is consistent with Burgess and
Zhuang's (2001) study using micro-level data from two provinces in China which shows that
boy-preference occurs more in poor households. On the other hand, the Trivers-Willard (1973)
hypothesis claims that higher status individuals have more male children (1973). In support of
the latter view.,Edlund (1999) shows that in India, poor states have more girls and rich states
have more boys. To add further controversy, Gu and Roy (1995) show that for China, the
poorest and richest regions have the least sex imbalance. And Li (2002) found no correlation
between sex ratios and factors such as household income, parents' education and the amount
of monetary fine associated with the One Child Policy for children born during 1982-1987.
Beginning with Ben Porath's studies of female labor supply (1967, 1973, 1975), studies have
also shown that relative female income and/or education matters. For India, Rosenzweig and
Schultz (1982) showed that female children receive a larger share of household resources relative
to male children in communities where women's expected labor market employment is relatively
high. Studies by Clark (2000) and Das Gupta (1987) in India; and Thomas et. al. (1991) in
Brazil show that increased wages and/or education for adult women are positively correlated
with health and education outcomes for girls. For China, Burgess and Zhuang (2001) show that
17
boy-preference is stronger in areas with fewer non-farm employment opportunities.
If men are
more valuable for farm labor, their results suggest that boy-preference is positively correlated
with the value of adult male labor. In order to estimate the causal effect of sex-specific economic
incentives on survival rates, Foster and Rosenzweig (2001) exploit sex-specific, regional and
time variation in returns to human capital caused by the practice of patrilocal exogamy and
productivity increases during the Green Revolution in India. 6
The empirical findings outlined above can all be explained by the unitary model of the
household in which households maximize one utility function (e.g. parents have identical preferences or one member of the household dictates his preferences). Because parents maximize the
potential earnings of their children in the unitary framework, increased adult female outcomes
will increase relative investment in girls if the former reflects an increase in the relative returns
to having (investing in) a girl. However, another reason why female income may differentially
affect girls and boys is that household decisions are made as a result of a bargaining process
in which the income of each family member can affect their bargaining power. For Ghana,
Thomas (1994) shows that allocation of resources for girls relative to boys is strongly correlated
to the education status of mothers relative to fathers. Duflo (2002) directly tests the unitary
hypothesis by comparing the effect of pension payments to grandmothers to the effect of payments to grandfathers on health outcomes for girls in South Africa. She shows that contrary
to unitary model predictions, pension payments to grandmothers, which cannot be interpreted
as an increase in female productivity, benefit girls while payments to grandfathers do not. For
China, Park and Rukumnuaykit (2004) find that household composition has differential effects
on fathers' and mothers' nutrient consumption. They argue that this is inconsistent with the
unitary model.
1.2.2
Agricultural Reforms
Pre-1978 Chinese agriculture was characterized by an intense focus on grain production, allocative inefficiency, lack of incentives for farmers and low rural incomes (Sicular, 1988a; Lin,
1988). Agricultural policies aimed at subsidizing urban industrial populations with cheap food
centered around production planning.
6
After agriculture was unified in 1953 (tong gou tong
Patrilocal exogamy is the practice for married couples to reside with families of husbands.
18
xzao), planning included mandatory targets for crop cultivation, areas sown, levels of input
applications and planting techniques by crop. Amongst these targets, sown area was the most
important, in part, because it was easier to enforce (Sicular, 1988a).
Central planning divided crops into three categories. Category 1 included crops necessary
for national welfare: grains, all oil crops and cotton. Procurement prices for grain during this
period were generally 20%-30% lower than market prices (Perkins, 1966) and market trade of
these products was strictly prohibited (Sicluar, 1988a). Category 2 included up to 39 products,
including: livestock, eggs, fish, hemp, silkworm cocoons, sugar crops, medicinal herbs and tea
(Sicular, 1988b).7 Category 3 included all other agricultural items (mostly minor local items);
these were not under quota or procurement price regulation.
Under the unified system, the central government set procurement quotas for crops of categories 1 and 2 that filtered down to the farm or collective levels. Quota production was
purchased by the state at very low prices. These quotas were set so that farmers were supposed
to retain enough food to meet their own needs. But in reality, farmers were left with little
remaining surplus (Perkins, 1966). Non-grain producers produced grain and staples for their
own consumption and sold all cash crop output to the state at suppressed prices. Farmers had
very little incentive to produce more than their quota.
After the Great Famine (1959-1961), the government re-emphasized grain production by
increasing procurement prices for grain relative to other crops. The state resorted to commercial
and production planning to carry out the objectives of grain production (yi liang wei gang) and
self-sufficiency (zi i geng sheng). The government increased production by enforcing mandatory
sown area targets for crops and promoted self-sufficiency by purchasing but not selling grain and
oils in rural areas. Mandatory sown area targets often required cultivation on land unsuitable
for grain. Grain production grew at substantial cost of other production. Production declined
for crops which competed with grain for land. Living standards declined significantly in areas
suitable for commercial crops (Lardy, 1983).
Post-Mao era reforms focused on increasing rural income, increasing deliveries of farm products to the state, and diversifying the composition of agricultural production by adjusting rel7
The number of crops in each category changed over time. And the number of crops reported in for each
category for a given year may vary across sources.
19
ative prices and profitability. Two sets of policies addressed this aims. The first set of policies
gradually reduced planning targets and reverted to earlier policies of using procurement price
as an instrument for controlling production (Sicular, 1988a). In 1978 and 1979, quota and
above quota prices were increased by approximately 20%-30% for grain and certain cash crops.
By 1980, prices had increased for all crops.
Although category 1 crops benefited from the
price increases, emphasis was placed on cash crops from category 2. The second set of policies,
named the Household Production Responsibility System (HPRS), devolved responsibility from
the collective, work brigade, or work team to households (Johnson, 1996; Lin, 1988). The HPRS
was first enacted in 1980 and spread through rural China during the early 1980s, devolving all
production decisions and quota responsibilities to individual households. The HPRS allowed
households to take full advantage of the increase in procurement prices by partially shifting
production away from grain to cash crops when profitable.
Together, the two reforms contributed to diversification of agricultural production, greater
regional specialization, and less extensive grain cultivation (Sicluar, 1988a). There was an immediate and significant increase in the output of cash crops (Johnson, 1996; Sicular 1988a).
However, although the value of all crops increased, continued emphasis on rural-urban subsidization of grain and other category 1 products caused the relative value of category 1 products
to decrease. 8 I will compute the income from each crop directly in the next section, but the
increase in the relative value of category 2 crops is also reflected in the disproportionate growth
in output of category 2 crops in comparison with category 1 crops. Figures 2A and 2B show
that although output for category
crops increased, there is no change in the rate of increase.
Figures 2C and 2D show that the rate of increase for suburban vegetables and orchard fruits,
both category 2 crops, accelerated after the reform. Similar increases can be observed for tea,
another category 2 crop, in Figure 3.
In a second round of reforms designed to reduce the fiscal burden of grain subsidies, the
state increased urban grain retail prices and stopped guarantees of unlimited procurement of
category 1 products at favorable prices. On average, contract procurement prices for grain were
35% lower than market prices (Sicular, 1988a). This change, combined with the de-regulation
8The central government complained that staple crop targets were under-fulfilled while production of economic
crops greatly exceed plans (Sicular, 1988a).
20
of other crops, further decreased the relative-profitability of category 1 products.
Complete substitution away from producing grains was prevented by the state's continued
enforcement of household level grain production quotas and its suppression of intra-rural grain
trade.
As late as 1997, virtually every agricultural household planted staple crops (Eckaus,
1999). Using the 1997 Agricultural Census, Diao et. al. (2000) show that on average, 80% of
sown area is devoted to grain and that self-sufficiency in grain was still an important part of
Chinese agriculture.
One possible cause of the magnitude and speed of the response of the Chinese agricultural
sector is the low labor productivity in the agricultural sector resulting from migration and
other labor controls. Calculations for the marginal productivity of labor in Chinese agricultural
production vary greatly. However, most studies agree that the high population-to-land
ratio
and labor market and migration controls result in low marginal productivity in rural areas
during this period. Households living in areas with the appropriate natural conditions can then
easily expand into cash crop production in response of new economic opportunities.
This is
consistent with the fact that agricultural households very rarely hired labor from outside the
family. In 1997, 1 per 10000 rural households hired a worker from outside of the immediate
family (Diao e. al., 2000). Since migration and labor market controls were more strict in the
1980s, it is most likely that the households studied in this paper hired even fewer non-family
members. Plentiful cheap adult labor would also reduce demand for child labor.
1.2.3
Tea and Orchard Production
This section discusses male and female labor intensities in tea and orchard production and how
the production of each reacted to post-Mao reforms. I will also directly estimate the income from
each crop and show that: the reforms increased income from category 2 cash crops (including
tea and orchards) relative to income from category 1 staple crops; and income from tea did
not exceed income from other category 2 cash crops. The latter fact addresses the possibility
that the effect of income on sex ratio is not linear. An increase in income from tea (orchards)
translates into an increase in total household income as well as an increase in relative female
(male) income. On the other hand, sex neutral cash crops only affect total household income.
To discern whether sex ratios are responding total income or relative female (male) income, I
21
estimate the effect of sex-neutral cash crops on sex ratios. However, if the income effect on
sex ratio is non-linear such that there exists some threshold income which must be met before
income will affect sex ratio, this strategy will only work if income from tea does not exceed
income from sex neutral cash crops.
Across Asia, tea is mainly picked by women. Labor input data by sex and crop is not available to examine sex specialization directly; however, in a study of South Indian tea plantations,
Luke and Munshi (2004) show that 95% of workers are female. The most commonly cited
reasons for why adult women have an absolute advantage in picking tea over adult men and
children is that tea picking favors small and agile fingers. In general, the value of the tea leaves
increase with the tenderness (youth) of the leaf. Adult women have a particular advantage over
children, who are considered more careless, in picking green tea leaves, which is worthless if
broken. 9 In addition, tea bushes are on average 2.5 feet (0.76 meters) tall, which disadvantages
taller adult males. For China, the specialization caused by women's physical advantage might
have been increased by strictly enforced household grain quotas that forced every household
to plant grain. In households that wished to produce tea after the reform, men continued to
produce grain while women switched to tea production. It follows that for tea planting households, an increase in tea value increased both the total household income and the relative value
of adult female labor. Moreover, monitoring of tea picking is made difficult by the fact that tea
picking is a very delicate task and that the quality and value of tea leaves vary greatly with
the tenderness of the leaf. This resulted in almost no hired labor. Hence, the relative value of
female labor increased in households that could produce tea despite the availability of cheap
outside labor.
In contrast, height and strength yields a comparative advantage for men in orchard producing areas. 10 For orchard producing households, an increase in the value of orchard fruits
increased both total household income and the relative value of adult male labor.
9
Breakage causes tea leaves to oxidize and blacken.
l°Adult men have a comparative advantage in orchard production during both sowing and picking periods.
Sowing orchard trees is strength intensive as it requires digging holes approximately 3 feet (0.91 meters) deep.
The strength requirement is re-enforced by the fact that Chinese soil is composed of 85% rock. The height of
apple trees and orange trees range between 16-40 feet (4.9-12.2 meters) and 20-30 feet (6.1-9.1 meters). The
height of the trees mean that adult males have advantages both in pruning and picking over adult females and
children. Orchard trees that are most commonly observed in orchards today are either genetically modified
(stunted) to be short or kept short by constant pruning.
22
The presence of child labor cannot be ruled out in any agricultural production.
However,
adult labor surplus resulting from land shortages and labor market controls leaves little demand
for child labor. In section 4 of this paper, I will establish that the identification strategy is robust
to the possibility that children and adult males (females) contribute to tea (orchard) production.
The main effect of post-Mao reforms for tea production was to increase picking. Considered
a priority crop, tea production was collectivized in the 1950s. Procurement and retail were
completely nationalized by 1958. During the Cultural Revolution, the government pursued an
aggressive expansion of tea fields. However, since farmers had little incentive to produce and
tea picking is more difficult to enforce than sowing, most of the sown fields were left wild and
untended until the post-Mao era, when the HPRS disaggregated 500 state tea farms into over
90,000 household level tea production units. Tea bushes were restored by extensive tending
and pruning (Forster and Etherington, 1992). The procurement price for tea, which was largely
unchanged between 1958-1978, doubled between 1979 and 1984. Figure 3A shows the increase
in procurement; price and yield for tea. Since there was no change in sown area during this
period, the yield increase reflects an increase in picking, which, in turn, reflects an increase in
the value of female labor.
Data for agricultural income by crop is not available during this period. Crop composition
for the average household in tea planting counties from the 1997 Agricultural Census and data
on net income by crop from tea planting households in 1982 (Etherington and Forster, 1992)
suggest that in tea producing counties, tea comprises of 1-4% of total household net income.
To examine the change in value of crops over time, I calculate the approximate gross income
by crop using data on output per standard labor day by year by crop and procurement price
by year by crop. 11 Figure 4A shows the national annual gross income from category 1 crops
and tea. After 1979, income from tea increased at a faster rate than income from grains. I
will exploit this increase to estimate the effect of an increase in relative adult female income on
sex ratios. Figure 4B shows that the calculated income from orchard production increased at a
faster rate than income from category 1 crops. I will exploit this increase to estimate the effect
of an increase in relative male income on sex ratios.
1
Data on output per standard labor day by year by crop is reported by the National Bureau of Statistics of
China. To the best of my knowledge, labor supply does not vary across years.
23
Amongst category 2 crops, the government maintained more control on tea than other
crops. Tea was viewed as a political symbol by the central government from the early 1950s.
In 1984, tea was one of the nine crops to remain under designated procurement price. The
central government continued to maintain a retail monopoly on tea up to the early 1990s. Until
the late 1980s, China exported tea at subsidized prices. Part of the subsidy was achieved by
suppressingprocurement prices of tea (Etherington and Forster, 1992). Consequently, although
price for tea grew significantly after 1979, tea was not as profitable as many other cash crops.
Figure 4C shows that the gross income from tea experienced similar increases to other category
2 cash crops immediately after the reform. By 1983, the rate of increase was less than income
from other category 2 crops although the income from tea continued to increase.
1.3
Conceptual Framework
This section presents a simple model of sex imbalance.
I use this framework to show that
adult income affects the desirability of daughters relative to sons through two mechanisms:
first by changing the consumption value of having a girl relative to having a boy; and second
by changing the investment value of having a girl relative to having a boy. Moreover, it shows
that if households are not unitary (e.g. parents do not have identical preferences), a change in
adult income can also affect the relative desirability of girls by changing the bargaining power
of each parent within the household. The model generates empirically testable predictions for
the unitary case.
1.3.1
Decision Rule
For most cohorts in this study, family size was constrained by China's family planning policies.
Thus, I make the simplifying assumption that all households have exactly one child. The only
decision which faces parents is the sex of their child. Because parents do not have access to
prenatal sex revealing technology, parents select the sex of their child by deciding to keep or
kill a child once she is born. Conditional on having a girl, parents for each household i compare
the maximum utility that they can derive from a girl and the maximum utility they can derive
from a boy, and will choose to keep a girl if VH-VH
24
> ci, where V" is the household's indirect
utility in the state of the world where it has a child of sex s, s E {g, b}, and ei is the cost of sex
selection for household i.
The probability of having a girl can be written as:
Pr(S =g) = Pr (e < V -
b )
= F(V H -Vb)
(1.1)
An increase in the probability of keeping a girl will be reflected in the population as an
increase in the fraction of girls.
Let yp, p
if (vf-)
ayp
{m, f} denote parents' (mother's and father's) incomes. Given that F'(.) > 0,
> 0, then the probability of keeping a girl is increasing in parental income.
Henceforth, denote ryp-
1.3.2
a
- V.
Household Utility
The utility of parent p is uP(c), where p E {m, f} and s, s
{g, b}, indicates the state of
the world (sex of the child). c is each parent's consumption bundle.
I normalize the price
of consumption to equal 1. In each state s, parents pool their income and maximize the
weighted sum of the mother's and father's utilities, usm(c), usf(c),subject to a household budget
constraint comprised of the incomes of the father, mother and a child of sex s, y, Ym and Ys
Credit markets are assumed to be perfect such that parents can borrow against the child's adult
income. For convenience, I represent parents' consumption and investment decisions in a one
period model. The indirect utility function in state s, Vs(y), is the maximand of the following
household utility function.
VH = mxu'(c) + (1-y)uf(c)
C
s.t. c = Yf +Ya +Ys
The investment value of a child is characterized by the inclusion of his/her income in the
budget constraint.
mrnother'sand fther's
The weight, z, which characterizes bargaining power, is a function of the
income ratio. Hence, the mother's bargaining power is increasing in her
income and decreasing in the father's income. Note that the unitary model is simply the special
case of the bargaining model where parents have identical utility functions, u
25
= uf
Assume that the productivity of a child is positively correlated with the productivity of
parents such that a child's income is a function of his/her parents' incomes, Ys
Ys(Yf, Ym).
Furthermore, assume that the correlation is stronger between a child and a parent of the same
sex such that
Oyg > y and
Yb
0
Ym
a&yf
Yf
>
0
Yb
aYm
When parents decide whether they wish to keep or kill a girl, they solve for the maximum
utilities they can achieve in the two states of the world where they have a girl or a boy. For each
state s of the world, s E {g, b}, parents solve the Lagrangian for household utility maximization
Ls = maxuu
(c) + (1 - ) uf(c)
-
As [c- (yf + yi + ys)]
The effect of a parent's income on the probability of having a girl is
ryp
-
a/
OP-
[(
UM
M
b
-
/
U
Uf'
u-ub)I
F
Y
Ybl
+ A ayp -Ab- y-
+ 9 - Ab
(1.2)
It follows from the first order conditions that A is the bargaining weighted sum of the
mother's and father's marginal utilities from income in the state of the world where the household has a child of sex s A9 - Ab is the relative "pure income effect" of having a girl as opposed
to having a boy. Holding other variables constant, the effect of a parent's income on the probability of having a girl is increasing in the relative pure income effect. This means that if a
daughter complements income more than a son, Ag > Ab, an increase in income will increase
the desirability of daughters relative to the desirability of sons. In other words, an increase in
parents' income will increase the probability of having a girl if girls are luxury goods relative
to boys. Henceforth, I call this the relative "consumption value" from having girls.
The terms in the second brackets characterize the relative "investment value" from having
a daughter. Holding other variables constant, the relative desirability of a girl will increase if a
girl's income increases more with the parent's income than a boy's income,
9 >
Dyp
0
yb
Yp'
The terms um - um and u f - uf are the mother's and father's utilities from having a girl
relative to having a boy. As long as parents do not have the same relative "sex preferences",
Ug
-
b
Uf f -9
f ,and bargaining power depends on income, - 4 0, an increase in parental
b
OYP
26
income will also affect the probability of having a girl by affecting the bargaining power of each
parent. Otherwise, equation (1.2) reduces to the unitary case.
In the general case, if parents view children as only a form of consumption, children's
income will not be included in the budget constraint and the terms,
Yg,
aYp
yb will drop out
Dyp
of equation (1.2). Similarly, if parents view children as only a form of consumption in the
unitary case, equation (1.2) reduces to Ag- Ab, the pure income effect. Since the pure income
effect is identical across all sources of income, the effects of mothers' and fathers' income on the
relative desirability is also identical in this case,
ryf.
ym =
Therefore, the joint hypotheses
that households are unitary and parents view children as only a form of consumption can, in
principle, be tested by comparing the effect of an increase in adult female income and the effect
of an increase in adult male income on population sex ratios.
The difference between the effects of the mother's income and the father's income for the
general case can be written as
Y- Yf
y
+A (
[u
&g
_/
> 0, since ay. >
b
-
)-
Ab (yb
'9 Yg >
Dyf' &pym 0 yf
b
(1.3)
_ Yb
&'Jf <a Yb
Dym &yf
Equation (1.3) shows that changes in the mother's income and the father's income will have
different effects on the probability of having a girl because they affect each parent's bargaining
power differently and because the correlation between each parent's income and a child's income
is different for boys and girls.
If households are unitary and parents view children as a form of investment, equation (1.3)
reduces to the bracketed terms. The difference in mothers' and fathers' income effect on the
relative desirability of girls is the difference in the correlation of the mother's and father's
incomes with the relative investment value of a daughter. It follows that mothers' and fathers'
incomes will only have different effects on investments in education or other factors that affect
child productivity if they have different effects on the returns to education (or other factors).
Therefore, if returns to education can be controlled for, the joint hypotheses that households
27
are unitary and parents view children as a form of investment can be rejected if the effect
of increasing relative adult female income on education attainment differ from the effect of
increasing relative adult male income.
1.4
The Data
The analysis of sex ratios uses the 1% sample of the 1997 Chinese Agricultural Census, the 1%
sample of the 1990 China Population Census and GIS geography data from the Michigan China
Data Center matched at the county level.1 2 The sample includes 1,621 counties in China's 15
southern provinces, south of the Yellow River (Huang He) where any tea is planted.
13
Map
1 show that these counties are dispersed throughout southern China. The 1990 census data
contain 52 variables, amongst which are data on sex, year of birth, education attainment, sector
and type of occupation, and relationship to the head of household. Because of the different
family planning policies and market reforms experienced by urban areas and rural areas, I limit
the analysis to rural households. The individual and household level data are aggregated to
the county level to match the agricultural census data. The number of individuals in each
county-birth year cell is retained so that the regression analysis are all population weighted.
Reliable data for procurement prices and output are not available for this period at the
county level. For the sake of scope, accuracy and consistency between areas, this study uses
county level agricultural data on the sown area from the 1% sample of the 1997 China Agricultural Census. Agricultural land is allocated by the village to farmers based on the number
of members per household and quality of land. Land is usually allocated for 15 year terms
(Burgess, 2004). There is no market for buying or selling land.
Using 1997 agricultural data to proxy for agricultural conditions in the early 1980s introduces measurement error. It is also possible that the counties that which tea in 1997 are the
counties which had stronger girl preference prior to the reform. In this case, comparing sex
ratios in tea counties that plant tea in 1997 to tea counties that do not plant tea in 1997
12
This section describes the 0.1% sample of the 1990 Population Census. The analysis of education uses data
from the 0.1% sample of the 2000 Chinese Population Census, which is described in Appendix Table A3. The
organization of the two censuses are similar.
l 3 Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hunan, Hubei, Guandong, Guangxi, Sichuan,
Guizhou and Shanxi.
28
will confound the effect of planting tea with the effect of underlying girl-preferences. However,
as discussed earlier, the government emphasis on tea planting during the Cultural Revolution
meant that the main determinant of whether a region had tea fields was geographic suitability
rather than sex preferences preferences. Specifically, tea grows best on warm and humid hill-
tops. The population density of the Chinese countryside and even distribution of hills through
out southern China means counties that plant some tea should not be very different from their
neighboring counties that plant no tea (in other respects).
To assess whether counties that do not plant tea are good control groups for counties that
plant tea, I loo)k for systematic differences between the treatment and control groups. While
I will exploit differences over time in both types of counties, any differential evolution is more
likely to be due to the relative income effect if the counties are otherwise similar. The average
demographic characteristics and education attainment shown in Table 1 Panel A are very similar
between counties that plant some tea and counties that plant no tea. The difference in ethnic
composition will be controlled for in the regression analysis. The descriptive statistics for sector
of employment in Panel B show that in both types of counties, 94% of the population is involved
in agriculture. Panel C shows that households in tea counties farm less total land on average,
devote more land to rice, garden production and less land to orchards. On average, agricultural
households have very little farmable land, 4.06-4.85 mu (0.20-0.32 hectares) per household.
Households in counties that plant tea have only 0.15 mu (0.02 hectares) of land for tea.
For a visual representation of the similarity in agricultural production between tea producing
counties and non-tea producing counties, refer the Maps B-lE, which show agricultural density
and production by crop. The black colored counties are counties which produce some tea. The
gray shaded counties are counties which produce some garden vegetables (Map 2A), orchard
fruits (Map 2)
and fish (Map 2C). Map 2D shows counties which produce some tea and
counties where the average farmable land per household exceeds the median of 4 mu (0.27
hectares). These maps show that tea producing counties are not geographically distant from
counties that produce other cash crops.
29
1.5
1.5.1
Empirical Strategy
Identification
The main problem in identifying the effect of increased relative female-to-male earnings on
child outcomes is that both may be in part related to omitted household and community
characteristics.
For example, in communities with no male-bias, adult women will earn more
and parents will view female and male children as equally desirable. In communities with strong
male-bias, where adult women earn less and parents strongly prefer boys over girls, we will find a
positive correlation between adult female income and girl survival rates. However, since female
earnings and girls' survival rates are jointly determined by sex preference, the correlation would
not reflect the effect of female income from the effect of sex preference on girls' survival rates.
This problem is addressed by exploiting the increase in relative value of tea caused by post-Mao
policies during 1978-1980. The exogenous variation in relative adult female earnings allows me
to estimate the causal effect of an increase in relative adult female earnings on relative survival
rates of girls.
First, I estimate the effect of the agricultural reforms on girl survival rates in tea planting
regions. The identification strategy uses the fact that the rise in adult female income varied
across region and time of birth. Substantial variation in amount of tea sown existed across regions. Therefore, the number of surviving female children should have increased in tea planting
regions for cohorts born close to and/or after the reform, and the increase should have been
larger for regions that planted more tea. 1 4 I use a differences-in-differences estimator to control
for systematic differences both across regions and across cohorts. Only the combination of these
two variations is treated as exogenous. In other words, I compare relative survival rates between
counties which plant tea and counties which do not plant tea, for cohorts born before and after
the reform. Comparing sex ratios within counties for cohorts born before and after the reform
differencesout time-invariant community characteristics. Comparing tea planting communities
to non-tea planting communities differencesout changes that are not due to planting tea. Thus,
14
The exact timing of the response in sex ratios to the reform depends on the nature of sex selection. If sex
selection is conducted by infanticide, the reform should only affect sex ratios of cohorts who were born after the
reform. However, if sex selection is conducted by neglecting young girls, the reform can also affect sex ratios of
children who were born a few years before the reform.
30
the causal effect of planting tea can be identified as long as tea planting areas did not experience
changes which were systematically different from non-tea planting areas.
Figure 5A plots the fraction of males of each birth year cohort for tea planting counties and
counties which do not plant tea. It shows that prior to the reform, tea counties had higher
fractions of males and after the reform, tea counties had lower fractions of males. The fact that
the change in relative sex ratios between tea and non-tea counties occurred for cohorts born
immediately after the reform suggests lends credibility to the identification strategy.
The date of birth and whether an individual is born in a tea planting region jointly determine
whether he/she was exposed to the relative adult female income shock. In other words, tea is
a proxy for female earnings. The validity of the identification strategy does not rely on the
assumption that only women pick tea. If men or children picked tea, the proxy for relative
female income will exceed actual relative female income. Hence, the strategy will underestimate
the true effect of relative female income on sex ratio. If there are any unobserved time-invariant
cultural reasons that both cause women to pick tea and affect the relative desirability of female
children, the effect will be differenced out by comparing cohorts born before and after the
reform. The identification strategy is only in question if there is some time varying difference
which coincides with the reform.
For example, if the attitudes which drive sex preference
changes in tea planting counties at the time of the reform, the estimate of the effect of planting
tea ,will capture both the relative female income effect and the effect of the attitude change.
Or, if the reason for women to pick tea was changed by the HPRS, the pre-reform cohort will
be an inadequate control group. While I can not resolve the former problem, the latter concern
is addressed by instrumenting for tea planting with time invariant geographic data.
Second, I use the increase in value of orchard fruits relative to other crops to investigate
the effect of an increase in relative male income on sex ratios. Third, I investigate whether
the increase in tea value affects relative survival rates because of the increase in relative female
income rather than an increase in total household income. I estimate the effect of the reform on
girls' survival in regions that plant any cash crops (including tea and orchards) that experienced
equal or more value increase than tea.
The identification strategy is based on the increase in the value of category 2 crops relative
to category 1 crops, for which prices continued to be suppressed, and category 3 crops, which
31
were never regulated. Therefore, the effect of category 1 and category 3 crops on sex ratios
should not change after the reform. I estimate the effect of category 1 and category 3 crops
on sex ratios. Figure 5B shows that indeed the effect of category 1 and 3 crops were identical
before and after the reform.
1.5.2
Basic Results
To see that the effect of tea and orchards on sex ratios is due to the post-Mao agricultural
reforms and not due to other changes in these regions, I check that the effect of tea and orchard
on sex ratios increased in magnitude at the time of the reform. The unrestricted effect of tea
planted for each birth cohort can be written as
1990
sexic-= a + E (teaix dl)/l + yi+ Oc+ tic
(1.4)
1=1963
The fraction of males in county i, cohort c is a function of: the interaction term between
teal, the amount of tea planted for each county i, and dl, a variable which indicates if a cohort
is born in year ; %, county fixed effects; and Oc, cohort fixed effects. The dummy variable for
the 1962 cohort and all of its interactions are dropped.
fi is the effect of planting tea on the fraction of males for cohort 1. If the effect of tea on sex
ratios was due to the reform, fi should be zero until approximately the time of the reform, after
which, it should become negative. The estimates for the coefficients in vector p1, reported in
Table 2 column (1), are statistically significant for cohorts born after 1979. Figure 6A, the plot
of the estimates of l, clearly shows the link between the increase in tea value and the decrease
in the fraction of males. The estimates oscillate around 0 until 1979, after which, they steadily
decrease. To test the joint significance of the effect of planting tea for cohorts born before the
reform and for cohorts born after the reform, I estimate the F-statistic for each cohort. They
are 3.59 and 2.05, both statistically different from 0.
In a similar regression, I estimate the effect of orchard planted in each county i on the
fraction of males in county i, cohort c.
1990
sexic= o + E (orchardix dl)61+ ?i + Oc+ ic
1=1963
32
(1.5)
The coefficients in vector 5l are plotted in Figure 6B. The plot shows that the effect of
planting orchards on the fraction of males becomes positive after 1979. The estimates, reported
in Table 2 column (2), are statistically insignificant. However, the F-statistics for the interactions for the pre-reform cohort and the post reform cohort are 0.82 and 1.75. This means that
while being born in an orchard planting county before the reform has no effect on sex ratios,
the effect of being born in an orchard planting county after the reform is jointly significantly
different from 0.
Figure 6C plots the coefficients from a similar regression estimating the effect of all category
2 cash crops on fraction of males. The plot shows that the effect of cash crops on sex ratio
experienced no change after the reform. Table 2 column (3) presents the estimates.
The F-
statistics for the pre-reform cohort and the post reform cohort are 1.32 and 1.37. Neither are
statistically different from 0.
Because the relatively few counties produce tea or orchards while all counties produce grains,
the reference group in equations (1.4) and (1.5) are counties that produce grains. Consequently,
controlling for the amount of orchards planted should not affect the unrestricted estimates of
the effect of tea from equation (1.4). To check that the unrestricted estimates are unchanged
by including controls for orchards and cash crops, I estimate the following equation.
1990
1990
1=1963
1990
1=1963
sexic =-- E (teai x dl)/1 + E (orchardix dl)31 +
(1.6)
(cashcropx dl)pl+ Hanice+ + i + ?c+ Sic
1=1963
Teai is a continuous variable for the amount of tea planted in each county i. The dummy
variable indicating that a cohort is born in 1962 and all its interactions are dropped.
The
estimated coefficients for the vectors /1, 61 and Pi are reported in Table 3. The similarity
between these estimates and the unrestricted estimates from equation (1.4) and (1.5) can be
seen in Figure ID, which plots the coefficients for tea and orchards. The figure shows clearly
that before the reform, sex ratios were very similar between tea and orchard regions, whereas
after the reform, planting orchards increased the fraction of males while planting tea decreased
the fraction of males. However, the estimates for tea are no longer statistically significant.
33
1.5.3
Differences-in-Differences
To summarize the effect on sex ratios, I estimate the following equation where the fraction
of males in county i birth cohort c is a function of the interaction term of a dummy variable
for whether a county plants tea and a dummy variable for whether a cohort is born after the
reform, controlling for the amount of orchards and all category 2 cash crops planted, fraction
of Han, county fixed effects, and a dummy variable for being born after the reform.
sexi
= a~+ (tea x post)l + (orchardi x postc) 2
(1.7)
+(cashcropi x postc)3 3 + Hanjic + /i + postc2y+ sic
The differences-in-differences estimator,
1,
is the difference in the fraction of males between
cohorts born before and after-reforms between tea planting counties and counties which do not
plant tea. orchardi and cashcropi are continuous variables for the amount of orchards planted
in county i. All standard errors are clustered at the county level. The estimates in Table 4
columns (3) and (4) show that planting tea decreased the fraction of males by 0.7 percentage
points, whereas planting orchards increased the fraction of males by 0.9 percentage points. Both
estimates are statistically significant at the 1% levels. However the estimate for the effect of all
cash crops, 3, is not significantly different from zero. Because the absolute increase in income
from tea does not exceed that of other cash crops (Figure 4C), I conclude that increase total
household income has no effect on sex ratios.
1.5.4
Robustness
Migration
If migration patterns differed significantlybetween tea and non-tea areas, and between orchard
and non-orchard areas, the OLS estimates could be capturing the effects of migration rather
than of income changes. Cohorts born after the reform are 11 years of age or younger in the
1990 Census. Hence, migration would bias the estimates if households with boys are more likely
to migrate out of tea areas and households with girls are more likely to migrate out of orchard
areas. Migration controls, however, made migration of entire households impossible. Another
possible cause for bias is if amongst pre-reform cohorts, females were more likely to migrate out
34
of tea areas and males were more likely to migrate out of orchard areas. However, because strict
migration controls suppressed long term migration from rural areas throughout the period of
the study, migration is unlikely to be a serious issue.
To address this problem, I estimate the upper and lower bounds of the absolute value of the
effect of planting tea and orchards on sex ratios by estimating equation (1.7) in a sample where
migrants are assumed to be women in tea counties and men in orchard counties. To construct
the inferred populations, the fraction of urban residents in each province that report they are
not born in that city and the population of the province are used to calculate the maximum
possible number of rural-urban migrants per province. The population of each county is then
used to calculate the fraction of provincial population in each county. I then add the multiple
of this fraction and the maximum number of migrants for that province back into each county.
Since the post reform cohort is less than 10 years of age and migration of children is not likely,
I assume that the new additions are all born prior to the reform. To estimate the lower bound
of the effect of tea, the new additions to the pre-reform cohorts in tea counties are assumed to
be female. To estimate the upper bound of the effect of tea, the new additions are assumed
to be male. Similarly, for the lower bound of the effect of orchard, all the added inferred
migrants in orchard counties are assumed to be male. To estimate the upper bound, all the
inferred migrants are assumed to be female. The estimated bounds are very similar to the OLS
estimates on the reported population, ruling out the possibility that the results are driven by
migration.
Cohort Trends
Cohort fixed effects control for variation across cohorts that do not also vary across counties.
They cannot control for county-varying cohort trends which may have occurred over the 29
years of this study. I address this issue by including linear cohort trends at the county level. In
order to make the estimates comparable to the 2SLS estimates in the next section, I restrict the
sample to only counties for which there is geography data and estimate the same specification as
the second stage of the 2SLS. This specification does not explicitly control for orchards because
planting orchards can be endogenous for the same reasons as those discussed in the next section
35
for tea. I estimate
sexic = a + (teai x postc)l + (cashcropi x postc)/32
(1.8)
+Hanic( + ,i x trendc,+ 4 i + postc? + sic
Teal is a dummy variable indicating whether a county plants any tea. h x trendc is the
interaction between county specific fixed effects with a linear time trend.
Columns (1) and
(2) of Table 5 shows estimates without and with the the county-level cohort trend. The point
estimates are similar and both statistically significant at the 5% level. Thus, the OLS estimates
are robust to changes across counties over cohorts.
1.5.5
Two Stage Least Squares
Two problems motivate the use of instrumental variables. First, using 1997 agricultural data
to proxy for agricultural conditions in previous years will introduce measurement error which
may bias the estimate downwards. Second, the OLS estimate will suffer from omitted variable
bias if families which prefer girls relative to boys switched to planting tea after the reform. In
this case, the OLS estimate will overestimate the true effect of an increase in the value female
labor because it will confound the aforementioned effect with the sex-preferences of households
which switched to planting tea after the reform. I address both problems by instrumenting for
the tea planting with the average slope of each county.
Tea grows in very particular conditions: on warm and semi-humid hilltops, shielded from
wind and heavy rain. Hilliness is a valid instrument for tea planting if it does not have any direct
effects on differential investment decisions and is also not correlated with any other covariates
in equation (1.10). Map 2 shows the slope variation in China, where darker areas are steeper.
Map 3 overlays the map of counties which plant tea onto the slope map. The predictive power
of slope for tea planting can be seen by comparing the tea planting counties with the steep
regions in Map 2. I use the GIS data pictured in Map 2 to calculate the average slope for each
county and estimate the following first stage equation, where both the amount of tea planted
and slope is time-invariant. Note that since orchards is also an endogenous regressor, the 2SLS
36
specification does not separately control it. The first stage equation is
teal x post
= (slopei x post,)A + (cashcrop x post,)p
(1.9)
+Hanic( + a + 4i + postc7+ eic
The predicted residuals are used to estimate the following second stage regression.
sexi
=
(teai x postc)i3+ (cashcrop x postc)
(1.10)
+Hanic~+ a + bi + postjY + Eic
Column (3) of Table 5 shows the first stage estimate from equation (1.9). The estimate for
the correlation between hilliness and planting tea, A, is statistically significant at the 5% level.
Column (4) shows the two stage least square estimate from equation (1.10). The estimate is
larger than the OLS estimate and statistically significant. Column (5) shows the two stage
least squares estimate controlling for county-level cohort trends.
The estimate is similar in
magnitude to the OLS estimate but no longer statistically significant. It is important to note
that the estimates with and without trends are not statistically different from each other. The
estimate without trends is larger in magnitude but also less precisely estimated.
The 2SLS
estimate in column (5) shows that conditional on county-level cohort time trends, the OLS
estimate is not biased. Furthermore the OLS and 2SLS estimates in columns (2) and (5) are
almost numerically identical to the initial OLS estimate in column (1). These results give
confidence to the robustness of the initial OLS estimates of the effect of tea and orchards.
1.6
Results on Education Attainment
The main results of the effect of relative adult earnings on sex ratios rejected the hypothesis
that households are unitary and parents view children only as a form consumption. However,
since increasing adult agricultural earnings also increase the earnings potential of children, these
results do not distinguish the hypothesis that households are unitary and increasing mothers income increases the survival rates of girls by increasing the relative investment value of girls from
the alternative hypothesis that increasing female income may increase the survival rates of girls
37
through increasing female bargaining power. To gain further insight in the household decision
making process, I investigate the effect of adult income changes on education attainment.
Recall that in the unitary model where parents view children as a form of investment, the
decision to invest in a child's education depends solely on the returns to education.
Hence,
increasing mother's income and increasing father's income will only have different effects on
education investment for children if they have different effects on returns to education. Similarly,
increasing mother's income and increasing father's income will only have different effects on the
relative education investment for girls if they have different effects on the relative returns to
education for girls. Because there is no income data from this period, I cannot explicitly
control for returns to education. However, returns to education are presumably low for manual
agricultural labor. Under the assumption that returns to education are the same for planting
tea and for planting orchards, I can test the hypothesis that households are unitary and parents
view children as a form of investment by estimating the effect of relative female income and
relative male income on education attainment.
This analysis uses county-birth-cohort level data from a 0.05% sample of the 2000 Population
Census. 15 In order to isolate the sample to children who had completed their education, I
restrict the sample to cohorts born between 1962 and 1982. Individuals in the sample should
not be affected by the Cultural Revolution since disruptions to schools were generally isolated
to urban areas. 16 I use cohorts which had not yet reached public preschool age at the beginning
of the reforms (born before 1976) as the pre-reform control. 17
The empirical strategy is the same as before. I estimate the following equation to examine
the effect of planting tea, orchards and all category 2 cash crops on education attainment for
the all individuals. I then repeat the estimation for the sample of girls, the sample of boys and
the difference in education between boys and girls.
eduyrsic = (tea * postc)/l + (orchardi* postc)/3
2 +
(1.11)
(cashcrop * postc)/ 3 + Hanic( + at+ i + postc"y+ Sic
15
Descriptive statistics are in Appendix Table A3.
16I repeat the experiment on the sample of cohorts born after 1967, who did not begin primary school until
after 1974, when schools were re-opened. The results are similar and statistically significant.
17Children enter public preschools at age 4 or 5 in China. Public nursery schools, targeted at children age 1-4,
are not available to most rural populations.
38
eduyrsic is the average years of education attainment for individuals born in county i, birth
year c. The estimates in column (1) of Table 6 show that planting tea increased overall, female and male education attainment by 0.2, 0.25 and 0.15 years. On the other hand, planting
orchards decreased female education attainment by 0.23 years and has no effect on male educa-
tion attainment. These estimates are statistically significant at the 1% level. Planting orchards
had no effect on male education attainment. The estimates in Column (4) show that planting
tea decreased the male-female difference in education attainment whereas planting orchards
increased the difference. The latter is statistically significant at the 1% level. The estimates for
all category 2 cash crops are close to zero and statistically insignificant.
I re-estimate equation (1.11) with continuous variables for the amount of tea and orchards
planted in each county i. Columns (5)-(8) of Table 6 show that the estimates have the same
signs as the estimates with the dummy variables in columns (1)-(4).
The estimates show
that one additional mu of tea planted increases female education attainment by 0.38 years
and male education attainment by 0.5 years, whereas one additional mu of orchards decreases
female education attainment by 0.12 years and has no effect on male education attainment.
Note that the effect of income from tea increases male education attainment more than for
female education attainment and that cash crops in general have no effect on female education
attainment but decreases male education attainment.
To observe the timing of the effect of tea on education attainment, I estimate the effect of
planting tea by birth year.
1982
eduyrsic --
1982
(tea x dl)l + E (orchardi x dl)Jl +
1=1963
1982
(1.12)
1=1963
(cashcrop x dl)pl + (Hanic + ce+ 'i
+ +£ic
1=1963
The dummy for the 1962 cohort and all its interactions are dropped. The estimated coefficients for each cohort 1 in vectors /1, Jz and P are shown in Appendix Table A4. I plot the
three year moving averages of the estimates for female education attainment in Figure 7. It
shows that female education attainment was similar between tea and orchard areas until 1976,
after which it increased in the former and decreased in the latter.
By showing that the income effect for education is not equal across different sources of
39
income, the results for education, like the main results for sex ratios, reject the hypothesis that
households are unitary and parents view children as only a form of consumption.
Moreover,
if returns to education are not affected by the reform, these results cannot be explained in
the context of a unitary model where children are a form of investment. The findings that an
increase in adult female income increases education attainment for all children while an increase
in adult male income decreases girls' education attainment and has no affect on boys can only
be explained by the unitary model (where parents view children as a form of investment) if an
increase in tea value increases returns to education of both boys and girls while an increase in
orchard value decreases returns to education of girls and does not affect that of boys.
The lack of income data prevents a direct analysis of the returns to education. However,
there are reasons to think that the returns to education are not differentially affected by the
reforms. First, evidence from India shows that returns to education for tea workers is close to
zero (Luke and Munshi, 2004). Second, there was no technological change in tea or orchard
production that would have changed the relative productivity of girls (Foster and Rosenzweig,
2001). Overall, the unitary model where parents view children purely as a form of investment
can only explain these results in the context of unlikely scenarios, the results can be easily
explained with a model where mothers value education more than fathers and increasing the
mother's income will increase the investment in education for all children because it increases
her bargaining power within the household.
1.7
Conclusion
This paper addresses the long standing question of whether economic conditions factor into
parents' demand for girls relative to boys. Methodologically, it resolves the problem of joint
determination in estimating the effect of changes in adult income on survival rate of girls by
exploiting changes in total household income and sex-specific incomes caused by post-Mao
reforms in rural China during the early 1980s. I find that increasing total household income
without changing the relative incomes of men and women had no effect on survival rates of
girls or education attainment.
Increasing female income while holding male income constant
had a large and immediate positive effect on the survival rates of girls and increased education
40
for all children.
Conversely, increasing male income while holding female income constant
immediately decreased survival rates and education attainment
of girls. The results reject
the joint hypothesis that households are unitary and parents view children as only a form of
consumption.
Furthermore, the unitary hypothesis where parents view children as a form of
investment can be rejected unless implausible assumptions are made about returns to education.
The empirical findings give a clear and affirmative answer to the question of whether sex
imbalance responds to economic incentives in the short run. In addition, increasing total household income without changing the relative shares of female and male income will have no effect
on survival rates. In association with the increased gender wage gap, these results can help
explain the increased sex imbalance and the observed decrease in rural education enrollment
in post-reform China. Policy makers who aim to decrease excess female mortality or to increase education investment should create policies that increase proportional adult earnings of
women. For example, an effective method of immediately decreasing excess female mortality is
to increase the relative income of adult women. The results indicate that for rural China in the
early 1980s, increasing female wages by US$15 (20% of household income) without changing
male wages would have balanced sex ratios.
41
Bibliography
[1] Angrist, Joshua and Krueger, Alan "Empirical strategies." The Handbook of Labor Economics, Vol. III, Chapter 23, North Holland, 1999.
[2] Banister, J. China's Changing Population, Stanford, California: Stanford University Press,
1987.
[3] Becker, Gary A Treatise on the Family, Cambridge, Massachusetts: Harvard University
Press, 1981.
[4] Ben-Porath, Yoram "The Production of Human Capital and the Life Cycle of Earnings."
The Journal of Political Economy, 75(4) Part 1, 1967, pp. 352-365.
[5] Ben-Porath, Yoram "Economic Analysis of Fertility in Israel: Point and Counterpoint."
The Journal of Political Economy, 81(2) Part 2, 1973, pp. S202-S233.
[6] Ben-Porath, Yoram and Welch, Finis "Do sex preferences really matter?"
The Quarterly
Journal of Economics, 90(2), 1976, pp. 285-307.
[7] Bourguignon, Francois, Martin Browning, Pierre-Andre Chiappori and Valerie Lechene
"Incomes and Outcomes: A Structural Model of Intra-household Allocation." Journal of
Political Economy, 1993.
[8] Browning, M. and Chiappori, P. A. "Efficient intra-household allocations: A general characterization and empirical tests." Econometrica, 66(6), 1998, pp. 1231-1278.
[9] Burgess, Robin "Mao's legacy: Access to land and hunger in modern China." London
School of Economics Working Paper, 2004.
42
[:L0]Burgess, Robin and Zhuang, Juzhong "Modernisation and son preference." London School
of Economics Working Paper, 2001.
[11] Cai, Fang, Albert Park and Yaohui Zhao "The Chinese Labor Market." University of
Michigan Working Paper, 2004.
[12] Clark, S. Son preference and sex composition of children: Evidence from India." Demography, 31(1), 2000, pp. 21-32.
[13] Coale, A. J. and Banister, J. "Five decades of missing females in China." Demography,
31(3), 1994, pp. 459-79.
[14] Croll E., I). Davin, et. al. China's One Child Family Policy, London: Macmillan, 1985.
[15] Das Gupta, Monica "Selective discrimination against female children in rural Punjab,
India." Population and Development Review, 13(1), 1987, pp. 77-100.
[1.6]Diao, Xinshen, Yi Zhang and Agapi Somwaru "Farmland holdings, crop planting structure
and input usage: An analysis of China's agricultural census." Trade and Macroeconomics
Discussion Paper, No. 62, 2000.
[17] Duflo, Esther "Grandmothers and granddaughters:
Old age pension and intra-household
allocation in South Africa." World Bank Economic Review, 17 (1),2003, pp. 1-25.
[18] Eckaus, Richard "Report #50: A study of the effects of natural fertility, weather and
productive inputs in Chinese agriculture." MIT Joint Program on the Science and Policy
of Global Change, 1999.
[19] Edlund, L. "Son preference, sex ratios and marriage patters." Journal of Political Economy,
107(6), 1999, pp. 1275-1304.
[20] Foster, Andrew D. and Rosenzweig, Mark R. "Missing women, the marriage market, and
economic growth." Brown University Working Paper, 2001.
[21] Foster, Andrew D. and Rosenzweig, Mark R. "Technical change and human capital returns
and investmnent: Evidence from the Green Revolution." American Economic Review, 86(4),
1996, pp. 931-953.
43
[22] Garg, Ashish and Morduch, J. "Sibling rivalry and the gender gap: Evidence from child
health outcomes in Ghana." Journal of Population Economics, 11(4), 1998, pp. 471-493.
[23] Gu, B. and Roy, K. "Sex ratios at birth in China with reference to other areas in East
Asia: What we know." Asia-pacific Population Journal, 10(3), pp. 17-42.
[24] Hannum, Emily and Park, Albert "Educating china's rural children in the 21st Century."
Mimeo.
[25] Jensen, Robert "Fertility preferences and female disadvantage in the developing world:
Equal treatment, unequal outcomes?" Harvard Kennedy School Working Paper, mimeo.
[26] Johnson, Gale "China's rural and agricultural reforms: successes and failures." Chinese
Economics Research Centre at the University of Adelaide, 96(12), 1996.
[27] Lardy, Nicholas R. Agriculture in China's Modern Economic Development. New York, New
York: Cambridge University Press, 1983.
[28] Li, Guanghui "Effects of the One-Child policy on the Number and Sex Composition of
Children in China," Mimeo, 2002.
[29] Lin, Justin (1988) "The household responsibility system in China's agricultural reform:
A theoretical and empirical study." Economic Development and Cultural Change, Vol. 36,
pp. s199-s224.
[30] Luke, Nancy and Munshi, Kaivan "Women as agents of change: Female incomes and
household decisions in South India." BREAD Working Paper, 2004.
[31] Oster, Emily "Hepatitis B and the case of the Missing Women." Harvard Working Paper,
2004.
[32] Park, Albert and Rukumnuaykit, Pungpond "Eat drink man woman: Testing for gender
bias in China using individual nutrient intake data." University of Michigan Working Paper,
2004.
[33] Perkins, Dwight H. Market Control and Planning in Communist China. Cambridge, Massachusetts: Harvard University Press, 1966.
44
[34] Rosenzweig, Mark R. and Schultz, T. P. "Child mortality and fertility in Colombia: Individual and community effects." Health Policy and Education, Vol. 2, pp. 305-348.
[35] Rosenzweig, Mark. R. and Schultz, T. P. "Market opportunities, genetic endowments, and
intrafamily resource distribution:
Child survival in rural India." The American Economic
Review, 72(4), 1982, pp. 803-815.
[361 Rozelle, Scott et. al. "Gender Wage Gaps in Post-Reform Rural China." World Bank
Working Paper, 2002.
[37] Schultz, T. P. "Demand for children in low income countries." Handbook of Population and
Family Economics, New York: Elsevier Science, 1997, pp. 349-430.
[38] Schultz, T. P. "An economic model of family planning and fertility." The Journal of Political Economy, 77(2), 1969, pp. 163-180.
[39] Schultz, T.P. "Changing World Prices, Women's Wages, and the Fertility Transition: Sweden, 1860-1910." The Journal of Political Economy, 93(6), 1985, pp. 1126-1154.
[40] Schultz, T. P. and Zeng, Yi "Fertility of rural China: Effects of local family planning and
health programs." Journal of Population Economics, 8(4), 1995, pp. 329-350.
[41] Sen, Amartya "More than 100 million women are missing." New York Review of Books,
1990.
[42] Sen, Amartya "Missing women." British Medical Journal, 304, 1992, pp. 587-588.
[43] Sicular, Terry "Plan and market in China's agricultural commerce." The Journal of Political Economy, 96(2), 1988a, pp. 283-307.
[44] Sicular, Terry "Agricultural planning and pricing in the post-Mao period." The China
Quarterly, No. 116, 1988b, pp. 671-705.
[45] Strauss, John and Thomas, Duncan "Health, nutrition and economicdevelopment." Journal of Economic Literature, 36(2), 1998, pp. 766-817.
[46] Thomas, Duncan "Like father, like son; like mother, like daughter: Parental resources and
child height." Journal of Human Resources, 29(4), 1994, pp. 950-988.
45
[47] Thomas, D., Strauss, T. and M. Henriques "How does mother's education affect child
height?" The Journal of Human Resources, 26(2), 1991, pp. 183-211.
[48] Trivers, R. L. and Willard, D. E. "Natural selection of parental ability to vary the sex ratio
of offspring." Science, 1973, pp. 179, 190-192.
[49] White, T. "Birth planning between plan and market: The impact of reform on China's
One-ChildPolicy." China's Economic Dilemmas in the 1990's: ... Studies in Contemporary
China, London: Sharpe, 1992, pp. 252-269.
[50] Zeng, Y., Tu, P. et. al. "Causes and implications of the recent increase in the reported sex
ratio at birth in China." Population and Development Review, 19(2), 1993, pp. 283-302.
46
1.8
Appendix - Robustness of Linear Specification
The empirical analysis of sex imbalance uses the fraction of males in the existing population
as the dependent variable. To check the robustness of the additivity implied by the linear
specification, I repeat the experiment using the log of male-to-female ratios as the dependent
variable. Using log odds restricts the sample to county-birth year cells where there are both
males and females. I estimate equations (1.4), (1.5) and (1.6) using the new dependent variable.
The estimates are shown in Table A1 and plotted in Figures (A)-(A4).
The effects of tea,
orchards and category 2 cash crops are statistically significant and very similar to the linear
estimates.
I estimate the differences-in-differences effect using equation (1.7) with the new
dependent variable. The estimates are shown in Table (A2). They are statistically significant
at the 5% level. The estimates in column (2) show that planting tea decreases the relative
proportion of boys by 2.9% and planting orchards increase the relative proportion of boys by
2.7%7.This translates to a 0.6 percentage-point decrease in the fraction of boys from planting
tea and a 0.5 percentage-point increase in the fraction of boys from planting orchards. These
estimates are very similar to the linear specification estimates reported in Table 3.
47
Chapter 2
Quantity-Quality: The Positive
Effect of Family Size on School
Enrollment in China
2.1
Introduction
The trade-off between quantity and quality of children has been a question of long standing
interest in labor economics. Understanding this tradeoff is especially relevant to developing
countries today as policy makers in these countries attempt to curb population growth as a
way of increasing average human capital investment. Both China and India, the world's two
most populous countries, have experimented with different family planning policies to limit
family size. Examining the trade-off between quantity and quality of children is of first-order
importance for evaluating the effects of past policies as well as for constructing effective future
ones.
Empirical findings on the quantity-quality trade-off are conflicted. On one hand, the effect
of family size on education has been found to be negative by Rosenzweig and Wolpin (1980) in
India; by Goux and Maurin (2004) in France; by Conley (2004), Berhman et. al. (1989) and
Stafford (1987) in the U.S. On the other hand, studies by Lee (2003) in Korea, Kessler (1991)
and Guo and VanWey (1991) in the U.S. have found that family size has no effect on education
49
while Gomes (1984) found that family size was positively correlated with education attainment
for first born children in Kenya. Adding to the controversy, in studies using data from Norway
and'the U.K., Black et. al. (2004) and Iacavou (2004) use dummy variables for family size
instead of the traditional continuous variable and found that while family size and education
outcomes are negatively correlated for children from households with two or more children,
children from one-child families perform worse than children from two-child and three-child
families. Black et. al. (2004) estimates that only-children, on average, attain 0.21 years less of
schooling than other children.
The main empirical challenge in estimating the effect of family size on child outcomes is
caused by two sources of endogeneity. The first source arises from parental heterogeneity. For
example, if parents who value education more also prefer to have fewer children, the correlation
between quantity and quality will be driven by parental preferences rather than by family
size. To address this problem of joint determination, past studies have exploited the exogenous
variation in family size caused by multiple births or the sex composition of the first two children
(Rosenzweig and Wolpin, 1980; Conley, 2004; Lee, 2003). However, both instruments fail
the exclusion restriction since they affect child outcomes other than family size. In a study
of Indonesia, Duflo (1998) found that twin births of younger siblings were correlated with
higher mortality rates of the first born child. She argued that short birth spacing may be a
channel through which an increased number of children lower their average quality; the strain
on resources is especially problematic if the household is credit constrained.
The sibling-sex
instrument is equally problematic: Dahl and Moretti (2004) and Ananat and Michaels (2004)
find that sibling sex composition directly affect divorce rates. Using sibling sex composition
has the additional limitation that it requires the sex of children to be randomly assigned, and
consequently, it cannot be used in a country with sex-selection such as China (Qian, 2004).1
The second source of endogeneity arises from heterogeneity in the quality of the first child.
For example, if parents are more likely to have a second child when the first child is of high
quality, the OLS estimate of the family size effect will be biased upwards.
Instead of looking for exogenous variation in family size, Guo and VanWey (1991) and Black et. al. (2004)
attempt to control for the unobserved differences across households by controlling for household fixed effects
in panel data. However, fixed effects estimates are biased if unobserved household-level heterogeneity are time
varying.
50
This principal contribution of this paper is to address the two sources of endogeneity by
exploiting regional and time variation in relaxations of China's One Child Policy. Specifically,
it uses the relaxation that allows families to have a second child if the first child is a girl to
instrument for the family size of first born children born before the relaxation was announced.
Three facts are exploited: first, an individual is only affected by the relaxation if she is born
in a relaxed area; second, amongst first born children born in relaxed areas, only girls are
affected; and third, amongst first born girls born in relaxed areas, the relaxation only affected
girls whose family size were constrained by the initial One Child Policy (born in 1976 or after). 2
The instrument for family size is the triple interaction term of an individual's sex, date of birth
and region of birth. The interaction between whether a girl was born in a relaxed area and
whether she was born in 1976 or after estimates the effect of the relaxation of family size. The
additional comparison with boys controls for region specific changes in school provision that
affected boys and girls similarly.
There are two main benefits in setting this study in China. First, family planning policies
provide a unique source of exogenous variation in family size. Second, this study can evaluate
the effects of the One Child Policy, one of the most restrictive and large scale family planning
policies ever undertaken.
While demographers and sociologists have conducted descriptive
studies of the policy's impact on fertility, the lack of local enforcement data has prevented an
examination of the causal effect of the One Child Policy on child outcomes.
The empirical findings show that the One Child Policy decreased the fraction of girls amongst
first born children in the surviving population by up to 10 percentage-points in certain regions.
This increase in sex imbalance mostly reflects an increase in excess female mortality since prenatal sex revealing technologies which enable selective abortion were not introduced to the
population in this study until the end of this period (Zeng et. al., 1993). The results show that
consistent with official reports, the 1-Son-2-Child relaxation was implemented in communities
that experienced larger increases in boy-biased sex selection after the One Child Policy. The
relaxation immediately decreased the level of sex selection although sex ratios did not return to
their initial pre-One Child Policy levels. The difference in levels of sex selection between relaxed
2
The One Child Policy began in 1978-1980. However, prior to that were policies which encouraged birth
spacing of at least four years between children. I show that the One Child Policy was actually binding for the
family size of cohorts born in 1976 and after.
51
and un-relaxed regions suggests that parents who kept girls in relaxed regions are on average
different from parents who kept girls in un-relaxed regions. Comparisons between these two
types of regions will therefore suffer from selection bias. In particular, if parents of girls born
in regions with the relaxation value education for girls more than parents of girls born in other
regions, the two stage least squares estimate using the triple interaction term to instrument for
family size will confound the family size effect with parental preferences and be biased upwards.
To address this, I will estimate a lower-bound of the absolute value of the effect of family size
on school enrollment.
The first stage results show that the relaxation increased family size for first born girls
and had no effect on the family size of first born boys. The two stage least squares estimates
show that an additional sibling increases school enrollment of the first born child by 18-20%
on average. This finding can be explained by a model where there are fixed costs in education
or a model where other children are complements in each child's production function.
The
plausibility of the latter hypothesis is strengthened by the finding that the family size effect
varies by the age gap between the two children.
Although more research is needed to understand the effect of quantity on quality outside the
one-child context, the empirical findings of this paper suggest that there is a strong only-child
disadvantage and consequently reject the hypothesis that quality is monotonically decreasing
with quantity.
The results cast doubt on the idea that restricting family size will necessar-
ily help to increase average human capital investment in developing countries. Policy makers
should weigh the benefits of implementing restrictive family planning policies against the substantial "costs" related to sex selection and the long run consequences of the resulting sex
imbalance.
In addition, policy makers who wish to restrict family size to one child should
consider implementing programs which increase interaction between children.
The paper is organized as follows. Section 2 discusses family planning policies, education
in rural China and the conceptual framework. Section 3 describes the data. Section 4 presents
the empirical results. Section 5 offers interpretation of the results and concluding remarks.
52
2.2
2.2.1
Background
Family Planning Policies
In the 1970s, after two decades of explicitly encouraging population growth, policy makers
in China enacted a series of measures to curb population growth.
The policies applied to
individuals of HElanethnicity, who comprise 92% of China's population. Beginning around 1972,
the policy "Later [age], longer [the spacing of births], fewer [number of children]" gave economic
incentives to parents to space the birth of their children at least four years apart.
The One
Child Policy was formally announced in 1979. Actual implementation began in certain regions
as early as 1978 and enforcement hardened across the country until the policy was firmly in
place in 1980 (Croll et. al., 1985; Banister, 1987). Past studies generally consider the One
Child Policy to have only affected the family size of cohorts born after 1978-1980. However,
this paper will show that because of the previous four year birth spacing rule, the One Child
Policy affected cohorts born in 1976 and after.
Policy tightened gradually and second births became forbidden except under extenuating
circumstances. Local cadres were given economic incentives to suppress fertility rates. In the
early 1980s, parts of the country were swept by campaigns of forced abortion and sterilization
and reports of female infanticide became widespread (Greenlaugh, 1986; Banister, 1987).
Local governments began issuing permits for a second child as early as 1982. However,
permits for a second child were not made widespread until the Central Party Committee issued
"Document 7" on April 13, 1984. The two main purposes of the document were to: 1) curb
female infanticide, forced abortion and forced sterilization; and 2) devolve responsibility from
the central government to the local and provincial government so that local conditions can
be better addressed. It asked cadres to deal with each case individually and move away from
inflexible, uniform enforcement. The document allowed for second births for rural couples with
"practical" difficulties, and strictly prohibited coercive methods (Greenlaugh, 1986). The main
relaxation following Document 7 is called the "-son-2-child"
rule. It allows rural couples to
have a second child if the first child was a girl (Greenlaugh, 1986).3 The explicit purpose of
3
Practical difficulties included households where a parent or first born child was handicapped,
was engaged in a dangerous industry (e.g. mining).
53
or if a parent
this relaxation was to decrease female infanticide of the first born child.
White (1992) found that 5% of rural households were allotted second child permits in 1982.
These permits were generally granted to regions with extremely high levels of infanticide. After
Document 7, the permits expanded to 10% of the rural population in 1984, 20% in 1985 and
50% by 1986.
Document 7 made provincial governments responsible for both maintaining low fertility
rates and decreasing infanticide. While the exact process of granting permits is unclear, I use
county level data on family planning policy to show in the next section that the probability
for a county to obtain the 1-son-2-child relaxation is positively correlated with the rate of prerelaxation sex selection, and both are positively correlated with distance from the provincial
capital. These facts most likely reflect that in order to maintain low aggregate fertility rates
and decrease female infanticide, provincial governments granted relaxations to regions that were
distant to the administrative capital, and where female infanticide was more prevalent.
The
higher prevalence of sex selection in rural areas can be due to both more boy-preference in
distant rural areas and the fact that geographic distance increases the provincial government's
difficulty of preventing infanticide. 4 Issues of identification that arise from the correlation of
obtaining a relaxation and sex selection will be addressed explicitly in section 4.
2.2.2 Rural Education
Inequality in education provision greatly increased during the 1980s both across provinces and
across counties within a province. Inequality between school finance increased as changes in the
fiscal system reduced subsidies from rich regions to poor regions. The system of "eating from
separate pots" (fen zou chi fang) devolved expenditure responsibilities from the central and
provincial governments onto local governments in order to give the latter stronger incentives
to generate revenue. The ratio of the per capita schooling expenditure in the highest spending
province to the lowest spending province doubled in one decade.
Many rural schools were closed; rural enrollment rates dropped dramatically and did not
recover until the mid to late 1990s (Hannum and Park, mimeo). Using spending data from
4
Levels of income between counties with some relaxation and counties with no relaxation are comparable in
the CHNS data. This is consistent with the findings of Qian's (2004) study of rural China, where she finds that
sex selection was driven by the female-to-male income ratio and not by total household income.
54
Gansu, Hannurn and Park (mimeo) found that per capita school expenditure was positively
correlated with income and significant variation in school quality existed across counties. They
found little variation within counties, suggesting that studies examining education outcomes
should focus on variation at the county level.
Hannumrn(1992) show that difference in school provision between rich and poor areas are
much greater for middle school and high school than primary school. This is consistent with the
CHNS data used in this study, where primary school enrollment remained stable while middle
school and high school dropout rates increased for poor areas (Hannum and Park, mimeo).
The CHNS data show that counties with some relaxation and counties with no relaxation
have similar geographic access to schooling in 1989. However, the data does not reveal quality
of schooling or the changes in school availability during the early 1980s. Because relaxed areas
tend to be more rural, it is likely that the quality of schools declined in relaxed areas during
the same time that the 1-son-2-child relaxation took effect. To control for this, I will compare
outcomes for girls to boys within counties. The strategy is robust as long as the changes in
school quality and the economic conditions that determine school quality in relaxed areas have
the same impact on boys and girls.
2.2.3
Conceptual Famework
There are two models in the economics and sociology literature that predict an interaction
between the quantity and quality of children. The quantity-quality model, known in sociology
as the "resource dilution" model, dates backs to Becker (1960), Becker and Lewis (1973) and
Becker and Tomes (1979). They theorized that when income increases, parents who prefer
that children within a household have equal quality will want to increase the average quality
of their children.
Their model predicts that quality monotonically decreases with quantity.
An alternative model is the "confluence model", which to date, has not been explored in the
economics literature. Psychologists Zajonc and Gregory (1985) argue that children benefit from
interacting with adults and teaching younger children. The quantity and quality of children
are inversely related because increasing the number of children decreases the adult-to-children
ratio within a household. At the same time, children from one-child families and the youngest
child from a multi-child family are worse off because they cannot take advantage of the learning
55
which comes from teaching younger children. This model, therefore, predicts an inverse "U"
shape for the relationship between quantity and quality of children. This is consistent with
findings from Iacavou's (2004) study of children in the U.K. She finds that although general
family size is negatively correlated with measures of school performance, first born children from
one-child families perform worse than first born children from two-child families. Moreover, the
only-child effect decreases for children who interact more with other children outside of school.
2.3
Data
This paper matches data from the 0.1% 1990 Population Census with data from the 1989
China Health and Nutritional Survey (CHNS) at the county level. The 1990 Population Census
contains 52 variables including birth year, region of residence, whether an individual currently
lives in his/her region of birth, sex and relationship to the head of the household. The data
allows children to be linked to parents. Thus, family size and birth order of children within a
household can be calculated. Because the identification is partially derived from the region of
birth, the sample is restricted to individuals who reported living in their birth place in 1990. The
CHNS uses a random cluster process to draw a sample of approximately 3,800 households with
a total of 16,000 individuals in eight provinces that vary substantially in geography, economic
development, public resources, and health indicators. Most importantly, the survey provides
detailed village and township level information on family policy enforcement.
Since ethnic
minorities were exempt from all family planning policies, I restrict the analysis to four provinces
which are mostly composed of individuals of Han ethnicity. The matched dataset contains 21
counties in four provinces. 5 These provinces exclude rich coastal provinces or poor interior
provinces.
For the analysis of family size and education, the sample is restricted to first born children in
cohorts born during 1972-1981. This has two main advantages. First, all children in the sample
have access to public schooling in 1990. Second, including children born after the relaxation
may induce bias in the 2SLS estimate. After the relaxation, parents who prefer larger families
may choose to keep girls. This means that the 2SLS estimate will show that girls with larger
5
Liaoning, Jiangsu, Shandong and Henan.
56
family size are better off. But the estimate will be partially driven by parental preferences.
Exclusion of first born children born after 1981 removes this possibility.
The descriptive statistics in Table 1 Panel A show that counties with no relaxation are
very similar along demographic characteristics to counties with some relaxation. Each has 52%
boys on average and is mainly composed of ethnic Hans. Children in relaxed counties have on
average one more sibling than children from counties without the relaxation. Approximately
65% of children are enrolled in school.
The data shows that counties with some relaxation are almost four times as far from the
provincial capital as counties with no relaxation. Distance to school is similar between the two
types of counties.
Panel B of Table 1 describes the data for first born children from one-child families and
from families with two or more children. 47% of children in multi-child families are boys while
60% of one-child families are boys. Children without siblings are on average enrolled in school
12% more than children with at least one sibling.
2.4
Empirical Framework
2.4.1
Identification
Sex, date and region of birth jointly determine an individual's exposure to the 1-Son-2-Child
relaxation.
The relaxation allowed parents to have a second child only if the first born child
was a girl. Therefore, family size should be positively correlated with being a girl. The One
Child Policy, introduced around 1980, followed family planning polices which encouraged birth
spacing of at least four years. Consequently, the relaxation should only affect girls born 1976
or after.
The interaction between whether a girl was born in a relaxed area and whether she was born
1976 or after estimates the effect of the relaxation on family size. The additional comparison
with boys controls for education provision changes that affected both boys and girls similarly.
The instrument for family size is the triple interaction of an individual's sex, date and region
of birth.
Only the combination of the three is exogenous. The exclusion restriction for the
instrument is that it must be correlated with family size and have no direct effect on school
57
enrollment or other right hand side variables.
Like simple differences-in-differences estimators, cohort-invariant differences across regions
are differenced out by the comparison across cohorts.
Changes across cohorts which affect
different regions similarly are differenced out by the comparison across regions. The triple
difference adds the advantage that cohort varying differences that affect boys and girls similarly
across regions are also differenced out by the comparison between girls and boys within each
cohort and region. The exclusion restriction is only violated if a change with differential impacts
on relaxed and un-relaxed regions and on boys and girls occurs at the same time the relaxation
took effect. In other words, the 2SLS estimate will be biased only if there is a sex-specific,
region-specific change for the treated cohort.
I find in the next section that consistent with official reports, the extent of the relaxation is
strongly correlated with the extent of sex selection for One Child Policy cohorts (1976-1982).
The determinants of sex-selection may also affect education investment differentially for boys
and girls. For example, Qian (2004) shows that increasing male-to-female earnings increase
boy-biased sex-selection. She also shows that increasing male-to-female earnings has no effect
on education investment for boys but decreases education investment for girls. This means
that sex-selection is correlated with lower education investment for girls relative to boys. This
will not bias the estimates as long as the correlation is time invariant, in which case it will be
differenced out by the before and after comparison. 6
A potential source of bias introduced by the One Child Policy is the selection of parents who
choose to keep girls. Parents who choose to keep girls born during 1976-1982 in relaxed counties
may have different preferences from parents who keep girls in counties without the relaxation.
For example, if parents who decide to keep girls in relaxed counties also value education more
than parents who keep girls in non-relaxed counties, the 2SLS estimate will overestimate the
true effect of family size on school enrollment.
To address the problem of sample selection, I construct an alternative sample where the
"extra" boys from relaxed counties in the actual sample are taken out and replaced with girls
so that for each cohort, the sex ratio is equivalent between counties with some relaxation and
6
The CHNS does not have accurate data on individual income within the household since much of rural
production is conducted at the household level and income cannot be accurately assigned to individual members.
Consequently, I cannot directly examine the role of relative earnings in this study.
58
counties without any relaxation. In order to estimate the lower bound of the absolute value of
the effect of family size on school enrollment, I remove only boys who are not enrolled in school
and add girls who are not in enrolled in school. This increases the average enrollment rate
for boys born 1976-1982 in counties with the relaxation, and decreases average enrollment rate
for girls in counties with the relaxation. 2SLS using this "stacked" sample will underestimate
the true effect of family size on school enrollment.
Thus, using the actual sample and the
constructed sample, I will be able to estimate the upper and lower bounds of the absolute value
of the family size effect.
2.4.2
The Effect of the 1-Son-2-Child Relaxation
Effect on Family Size
One benefit of this policy experiment is that it is possible to check whether the policy was
enforced correctly by estimating the effect of the policy on family size for boys and girls separately. If the policy was correctly enforced, it should increase the number of siblings for girls
born 1976 and after and have no effect on boys. The following equation separately estimates
the effect of the relaxation on family size for boys and girls born during 1962-1981.
1981
sibsitc=
E
(relax x dil)/l + yt + a + &c+ Vitc
(2.1)
1=1973
The number of siblings for individual i, born in county c, birth year t, is a function of: the
interaction term of relaxc, the extent of relaxation in county c and d,
whether the individual was born in year ;
a dummy indicating
t, birth year fixed effects and
c, county fixed
effects. The reference group is comprised of individuals born during 1962-1972. It and all of its
interaction terms are dropped. For all regressions, standard errors are clustered at the county
level.
/3j is the effect of being born in a relaxed county on family size for an individual born in year
I. The estimates for girls and boys are shown in Table 2, columns (1) and (2). The estimates
for girls are statistically significant at the 1% level for the affected cohorts (born 1976-1981).
The estimates for boys are statistically insignificant. The coefficients are plotted in Figure 2A.
It shows that family size for boys and girls were similar for cohorts born 1973-1976, after which
59
the family size for girls increased and the family size for boys remained the same.
This difference in the effect of the relaxation on family size between boys and girls can be
written as the interaction between sex, date of birth and region of birth.
1981
1981
(relaxc x girlitc x dil)Bi +
sibsitc =
E
(relax x dil)Sl
(2.2)
1=1973
1=1973
1981
+
E3
(girlit x dil)(j + (relax, x girlit,)A+ girlitI
1=1973
+ C + 7t + / + it
The number of siblings for individual i, born in county c, birth year t, is a function of:
the triple interaction term of relaxc, the extent of relaxation in county c, girlitc, a variable
indicating whether a child is a girl and di, a dummy indicating whether the individual was
born in year ; the interaction term of relaxc and dil; the interaction term between girlitc,
and di; the interaction term between relaxc and girlitc; girlitc; -t, birth year fixed effects;
and %bc,county fixed effects. As before, the reference group of cohorts born 1962-1972 and
all its interactions are dropped.
l is the difference in the effect of being born in a relaxed
areas on family size between girls and boys. The estimates should be zero for cohorts who
were not affected by the One Child Policy and relaxation (1973-1976) and positive for affected
cohorts (1976-1981). The coefficients are shown in Table 2, column (5). They are statistically
significant at the 5% level for the effected cohorts. Figure 2A plots the coefficients. It shows that
the difference in the effect of being born in a relaxed area on family size is zero for unaffected
cohorts and positive for the affected cohorts. The relaxation increased family size of first born
girls by approximately 0.25 children on average.
Effect on Sex Ratios by Birth Parity
This section evaluates the effect of the relaxation on sex ratios by birth parity. To observe
the effect of the relaxation on sex ratios, the sample must be expanded to include cohorts
born after the relaxation. Past studies comparing hospital birth records and population census
data, or by comparing sex ratios for the same cohort at different ages have found that sex
60
selection mostly occurs at very young ages, which is consistent with the lack of prenatal gender
revealing technology and tough government enforcement against infanticide (Qian, 2004; Zeng
et. al., 1993). Hence, any sex selection caused by the One-Child Policy should be observed for
cohorts born very close to 1980. I estimate the following equation using a sample of cohorts
born between 1962 and 1989 by birth order. Because of widespread under reporting of children
under one year of age, I exclude the 1990 cohort (Zeng, 1992). The reference cohort is composed
of individuals born during 1962-1968.
1989
maleitc= a
(relaxc x dil)3l -+-t + a + -, + Vitc
(2.3)
1=1969
This equation is similar to (2.1). The dependent variable indicates whether an individual is
male. Table 3 column (1) shows the estimates of l for first born children. They are statistically
significant. Column (2) shows that the estimates are robust to the addition of a control for
whether individuals are ethnically Han. Columns (3) and (4) show the estimates for second
born children. Columns (5) and (6) show the estimates for children of higher birth parity. The
coefficients for first, second and later born children from columns (1), (3) and (5) are plotted in
figures 2A, 2B and 2C with their 95% confidence intervals. The solid vertical line in the figures
indicates the beginning of the [initial] One Child Policy in 1978. The dashed line indicates the
beginning of the relaxation in 1982. Figure 2A shows that in areas that received the relaxation,
the fraction of males increased after the One Child Policy relative to other areas. It also shows
that the relaxation decreased the fraction of males.
Figures 2B and 2C show that the One Child Policy and subsequent relaxations did not
affect sex ratios of higher order births in relaxed counties differently from counties without
relaxations. The relaxation did not change the sex composition of siblings for first born children
born between 1972 and 1982. This is important because the exclusion restriction for using the
triple difference as an instrument for family size requires that the instrument does not affect
any right hand side variable other than family size. Dahl and Moretti (2004) and Ananat and
Michaels (2004) show that the sex composition of children has a direct affect on the divorce
rates of parents. Hence, if the relaxation also changed the sex composition of children in families
of the affected cohort, the 2SLS estimate will be biased.
61
To estimate the effect of the relaxation on sex ratios, I estimate the following equation using
the sample of first born children. The children are divided into three groups according to birth
cohort. The reference group is comprised of individuals not affected by the One Child policy
and the relaxation (born before 1978). The second group comprises of children born after the
One Child Policy but before the relaxation (1978-1981). The third group comprises of children
born after the relaxation (1982-1989).
3
maleitc = Z(relaxc X postil)jl + a + at + ~ + &itc
(2.4)
1=2
The probability of being male for individual i, born in county c, birth year t is a function
of: the interaction term between relaxc, and postil, a variable indicating the individual's cohort
group; ,',
county fixed effects and yt, cohort group fixed effects.
The estimate for 61 is shown in column
of Table 3. It shows that first born children
born in relaxed regions after the initial One Child Policy are 8% more likely to be male than
children born in un-relaxed regions. After the relaxation, first born children born in relaxed
areas are only 4% more likely to be male than children born in areas without the relaxation.
Both estimates are statistically significant at the 1% level.
It is interesting to note that although the One Child Policy constrained the family size
of individuals born as early as 1976, sex selection from the One Child Policy appears only in
cohorts born after 1978. This is consistent with past findings that sex selection in China mostly
occurs for very young children. In other words, once the policy is announced in 1978-1980,
parents were unwilling (or unable) to kill girls that were more than 1 or 2 years of age in order
to have a boy.
These results suggest that parents in counties which received relaxations reacted differently
to family planning policies than parents in counties without relaxations.
These differing re-
actions may reflect different preferences towards investment in children that is time, sex and
region specific, which will confound the two stage least squares estimate for the effect of family
size.. Excluding cohorts born after the relaxation (1982-1990) partially addresses this problem.
It has the additional advantage of excluding households which kept girls in order to have a
second child. The sample selection issue from the [initial] One Child Policy (1978-1981) will
62
be addressed by estimating the absolute value of the lower bound effect of family size with the
alternative sample.
Effect on Female Labor Supply
If the relaxation caused parents to have a second child and mothers to stay home to take care of
the child, the 2SLS estimate will confound female labor supply effects with family size effects.
To address this, I estimate the effect of the relaxation on mother's work status controlling for
mother's age. The results are not reported in this paper. They show that mothers of affected
girls were less likely to stay at home.
2.4.3
The Effect of Family Size on School Enrollment
OLS
The correlation between school enrollment and family size can be obtained by estimating the
following equation for the sample of first born children.
1981
enrollitc = sibsitcb + Xct
+
E
(urbanc x dil)6l + a + At + ,c + Eitc
1=1973
School enrollment for individual i, born in county c, birth year t, is a function of: sibsitc,
the number siblings he or she has; Xit, individual characteristics; the interaction term between
urban, distance to urban area, and dl, a variable indicating whether an individual was born in
year ; yt, birth year fixed effects; and V,, county fixed effects. The estimate in Table 5 column
(1) shows that on average, one additional sibling is correlated with 1.7 percentage point less
of enrollment. The estimate is statistically significant at the 1
level. Columns (2)-(5) show
that the OLS estimate is robust to controls for the full set of double interaction terms from
equation (2.2), a variable indicating whether an individual is ethnically Han, distance to urban
area and mother's education. 7 Panel B shows the OLS estimates using the constructed sample.
The point estimates are similar to those of the original sample and statistically significant.
7
The double interactions include the interaction term of relaxc and dij; the interaction term between girlitc
and di,;the interaction term between relaxc and girli; and girlitc. The reference group is comprised of cohorts
born during 1962.-1972.The dummy variable for the reference cohort and all its interactions are dropped.
63
Reduced Form Estimates
To illustrate the identification strategy, I will first estimate the effect of the relaxation on
enrollment separately for boys and girls. This can be characterized by the following equation.
1981
enrollitc = a
(relaxc x dil)/3l+ ca+ t + ?c + Vitc
(2.5)
1=1973
The reference group is comprised of individuals born during 1962-1972. The dummy variable
for the reference group and all its interactions are dropped. The coefficients for girls and boys
are shown in Table 2, columns (3) and (4). The estimates are statistically significant for girls.
Figure 3A plots the estimates for boys and girls. Cohort to the right of the solid line are those
affected by the relaxation.
The plot of the reduced form shows that for the affected cohort,
girls have higher education enrollment than boys, whereas for the unaffected cohort, girls had
lower school enrollment rates than boys.
The estimates in Figure 3A show that relative to areas without the relaxation, enrollment
for both boys and girls decreases after primary school. This is consistent with the hypothesis
that school provision and quality in relaxed regions relative to regions without the relaxation
declined during this period.
I control for this by comparing the effect of the relaxation on
enrollment for boys with the effect of the relaxation on enrollment for girls, which can be
characterized by an equation similar to equation (2.2) with school enrollment as the dependent
variable.
1981
1981
enrollitc = E (relax x girlitc x dil) l +
1=1973
1981
+ E
E
(relaxc x dil)3l
(2.6)
1=1973
(girlitc x dil)(lj+ (relax x girlitc)A + girliteK
1=1973
+ a +
t- + 4c + Vitc
The reference group is comprised of individuals born during 1962-1972. The dummy variable
for the reference group and all its interaction terms are dropped. The coefficients are shown in
Table 2, column (6). The estimates show that for older cohorts not affected by the relaxation,
individuals born in relaxed areas have on average 1% to 17% less school enrollment than areas
64
without the relaxation.
However, for cohorts affected by the relaxation, individuals born in
relaxed areas are on average enrolled in school 5% more than individuals born in areas without
the relaxation. The estimates are statistically significant at the 1% level. Figure 3B plots the
triple difference reduced form estimates. It shows that school enrollment in relaxed areas is
higher for girls of the affected cohort than for boys.
Two Stage Least Squares
Using the predicted residuals from the first-stage equation (2.2), I estimate the following second
stage:
1981
enrollitc =
sibsitcb + E
(relaxc x girl x dil)/3
1=1973
1981
1981
+ E (relaxc x dil)6l + E
1=1973
1981
+ E
(girli x dil)(
1=1973
(urbanc x dil)S1 + (relaxc x girli)A
/=1973
+X.tr
+
9 + o+
±ft V + Vitc
X:ct is a vector of individual controls (e.g. mother's education, ethnicity).
urbani is the
average distance to the nearest urban area. Column (7) in Panel A of Table 5 shows that
contrary to the negative OLS estimate, an additional sibling increases school enrollment by
20.5% in the actual sample. The estimate is statistically significant at the 5% level. I repeat
the estimation for the alternative constructed sample to estimate the lower bound effect of
family size on school enrollment. The result is shown in Panel B of column (7). It shows that
one additional sibling increases school enrollment of the first born child by 18.4%. The estimate
is statistically significant at the 1% level. Columns (8)-(10) show that the 2SLS estimates of
both the actual sample and the constructed sample are robust to individual and county level
controls.
65
2.5
Conclusion
This paper has two purposes. It evaluates the effects of the One Child Policy and the subsequent 1-son-2-child relaxation. Then, it uses exogenous variation in family size caused by this
relaxation to evaluate the causal effect of family size on school enrollment.
The One Child Policy is one of the most internationally controversial policies undertaken
by the post-Mao Chinese government. It reportedly increased female infanticide and led to a
generation of "spoiled children". However, the common misunderstanding that the One Child
Policy is uniformly enforced across China and the lack of local enforcement data has, until
recently, prevented researchers from measuring the causal effects of China's family planning
policies. The lack of transparency in the policy enforcement decision process added to the
difficulty of such studies.
This paper uses local enforcement data of the 1-son-2-child relaxation to evaluate the effects
of the relaxation and the One Child Policy on sex ratio and family size. It shows that although
the One Child Policy was enacted in 1978-1980, previous family planning laws which encouraged
birth spacing meant that the former was actually binding for cohorts born as early as 1976.
The results show that the 1-son-2-child relaxation was indeed implemented in regions where
sex selection was more severe after the initial One Child Policy. The relaxation decreased sex
selection from the levels immediately following the implementation of the One Child Policy, but
sex ratios in these regions did not return to their initial pre-One Child Policy levels.
I use the exogenous increase in family size of girls born in relaxed regions to evaluate
the causal effect of family size on school enrollment. The advantage of this method is that
it addresses the endogenous relationships between family size and parental preferences over
education and between family size and the quality of the first child. The results show that
school enrollment for girls from one-child households increased by 18-20% when parents had
an additional child. The findings reject models which predict that quality is monotonically
decreasing in quantity. However, the results are consistent with evidence from previous studies
which show that although family size is negatively correlated with education outcomes for
children from households of two or more children, only-children are disadvantaged compared to
children from two or three child families. Further research is needed to examine the family size
effect beyond the two-child context.
66
There are several hypotheses that can explain the only child disadvantage.
In a simple
framework where parents and children have the same preferences (or where parents internalize
child preferences), family size can increase school enrollment if children complement each other
in their respective production functions. Iacavou's (2004) finding that the only child disadvantage decreases as the child interacts more with children outside of school suggests that there are
comrplementarities in learning or development for children. In this study, I find that the family
size effect varies according to the age gap of the two children. Similarly, psychologists Zajonc
and Gregory (1982) hypothesized that children benefited from teaching younger children. This
hypothesis predicts that the younger child will be worst off relative to the older child. Exploration of this hypothesis is prevented by the fact that the second child in this study is 6 years
of age or younger.
There may also be economies of scale in schooling costs and learning (or other psychological
responses). In the context of a developing country, text books and clothes can be considered
as fixed costs for sending children to school. Then having a second child will lower the average
and marginal cost of school attendance for the first child as long as the secondary market for
these goods functions such that transferring the goods to children from other families is more
costly than transferring them to children within the household. 8
In summary, this paper presents strong empirical evidence that only-children are worst off
compared to children from two or three child families. More research is needed to understand the
8 f parents and children have different preferences, the results can also be explained by child behavioral
responses in regard to the decrease in her share of tangible and intangible resources within the household. While
there has been many studies about parents' decisions to reallocate resources in response to the decrease in average
resources, the effect of the first born child's behavioral response and parental response to child behavior has been
left unexplored. The results of this paper, however, suggest that this is worthwhile considering for there are
several ways that child behavior can cause parents to increase the first child's school attendance. For example,
if the first child dislikes sharing tangible goods and parental attention with the younger sibling, she may behave
badly and therefore increase her parents' desire to send her to school, away from the sibling during the day.
Simultaneously, being at school may have an added attraction relative to being at home for the first child as
a place where her position is not affected by the birth of the latter. In addition, the first child may be more
motivated to attend school because she feels that academic distinction will increase her stature in the household
relative to the younger child.
It is important to note that in China, there is no schooling beyond high school in rural areas. Universities
are highly concentrated in the largest cities. In fact, rural students with academic potential generally leave their
homes during high school, or even middle school to attend better quality schools in urban areas. The lack of
economic opportunities in rural areas means that such children do not return home after graduating from college.
Therefore, if parents desire to keep at least one child near them, they are more likely to encourage the child to
pursue higher education if they have a second child to keep near them. This will translate into lower drop out
rates for the first child. Like the hypothesis proposed by Zajonc and Gregory (1982), this explanation implies
that family size effects may be different for the children of different birth order.
67
family size effect beyond the one-child context and to understand the mechanisms underlying
this effect. In the mean time, the results show that the relation ship between quantity and
quality is not monotonically decreasing and a richer theoretic model is needed to understand
the effect of family size on child quality. Policy makers should note that "one" is not the optimal
number of children per household and consider administering aggressive family planning policies
in conjunction with programs that increase interaction between children outside of school.
68
Bibliography
[1] Ananat, Elizabeth O. and Michaels, Guy "The effect of marital breakup on the income and
poverty of children." MIT Working Paper, 2004.
[2] Angrist, Joshua D. and Evans, Williams N. "Children and their parents' labor supply:
Evidence from exogenous variation in family size." The American Economic Review, 88(3),
Jun. 1998, pp. 450-477.
[311Banister, J. China's changingpopulation. Stanford, California., Stanford University Press,
1987.
[4] Becker, Gary S. and Lewis, H. Gregg "On the interaction between the quantity and quality
of children." The Journal of Political Economy, 81(2) Part 2, Mar.-Apr. 1973, pp. S279-S288.
[5] Becker, Gary S. and Tomes, Nigel "Child endowments and the quantity and quality of
children." Te Journal of Political Economy, 84(4) Part 2, Aug. 1976, pp. S143-S162.
[6] Berhman, Jeremy; Pollak, Robert A. and Paul Taubman "Family resources, family size and
access to financing for education." The Journal of Political Economy, 97(2), Apr. 1989, pp.
389-419.
[7] Behrman, Jeremy and Taubman, Paul "Birth Order, schooling, and earnings." Journal of
Labor Economics, 4(3) Part 2, Jul. 1986, pp. S121-S145.
18]Black,
Sandra E.; Devereux, Paul J. and Kjell G. Salvanes "The more the merrier? The
effect of family composition on children's education." NBER, Working Paper 10720.
[9] Blake, Judith "Family size and the quality of children." Demography, 18, pp. 421-442.
69
[10] Brown, Phil and Park, Albert "Education and poverty in rural China." The Economics of
Education Review, Forthcoming.
[11] Conley, Dalton "What is the 'true' effect of sibship size and birth order on education?
Instrumental variable estimates from exogenous variation in fertility." New York University,
Working Paper.
[12] Croll, E., D. Davin, et. al., Eds. China's One Child Family Policy. London, Macmillan,
1985.
[13] Dahl, Gordon and Moretti, Enrico "The demand for sons: evidence from divorce, fertility,
and shotgun marriage." NBER, Working Paper 10281, 2004.
[14] Duflo, Esther "Evaluating the effect of birth-spacing on child mortality." MIT, Mimeo.
[15] Gomes, Melba "Family size and education attainment in China." Population and Development Review, 10(4), pp. 647-660.
[16] Goux, Dominique, and Maurin, Eric "The effects of overcrowded housing on children's
performance at school." CEPR Discussion Paper No. 3818.
[17] Greenlaugh, S. "Shifts in China's population policy, 1984-1986: Views from the central,
provincial, and local Levels." Population and Development Review, 12(3), 1986, pp. 493-515.
[18] Guo, Guang and VanWey, Leah K. "Sibship size and intellectual development:
Is the
relationship causal?" American Sociological Review, 64(2), Apr. 1999, pp. 169-187.
[19] Hannum, Emily "Education stratification by ethnicity in China: Enrollment and attainment in the early reform years." Demography, 39(1), 2002, pp. 95-117.
[20] Hannum, Emily and Park, Albert "Educating china's rural children in the 21st Century."
Mimeo.
[21] Hanushek, Eric "The tradeoff between child quantity and quality." The Journal of Political
Economy, 100(1), Feb. 1992, pp. 84-117.
[22] Iacavou, Maria "Family composition and children's educational outcomes." Institute for
Social and Economic Research, Working Paper, 2001.
70
[23] Kaufman, J., Z. Shang, et. al. "Family planning policy and practice in China: A Study of
four rural counties." Population and Development Review. 15(4), 1989, pp. 707-729.
[24] Kessler, Daniel "Birth order, family size and achievement: Family structure and wage
determination." Journal of Labor Economics, 9(4), Oct. 1991, pp. 413-426.
[25] Lee, Jungmrin "Sibling size and investment in children's education: An Asian instrument."
University of Arkansas at Fayetteville, Working Paper, 2004.
[26] Qian, Nancy "Missing women and the price of tea in China: The effect of income on sex
imbalance." MIT, Working Paper, 2004.
[27] Rosenzweig, Mark R. and Wolpin, Kenneth I. "Testing the quantity-quality fertility model:
The use of twins as a natural experiment." Econometrica, January 1980, 8(1), pp.227-240.
[28] Rosenzweig, Mark R. and Schultz, T. Paul "Fertility and investments in human capital:
Estimates of the consequence of imperfect fertility control in Malaysia." Journal of Econometrics,
1987, 36, pp. 163-184.
[29] Stafford, Frank P. "Women's work, sibling competition and children's school performance."
The American Economic Review, 77(5), Dec. 1987, pp. 972-980.
[30] White, T. "Birth planning between plan and market: The impact of reform on China's One-
Child Policy", China's Economic Dilemmas in the 1990's: The Problems of Reforms, Modernization, and Interdependence. Studies in Contemporary China. Armonk, N.Y.U.S.C.J.E.
Committee and London, Sharpe, 1992, pp. 252-69.
[31] Zajonc, Robert B. and Markus, Gregory B. "Birth order and intellectual development."
Psychological Review, 1982, pp. 74-43.
[32] Zeng, Yi. "Is the Chinese family planning program 'tightening up?' Population and Development Review, 15(2), 1989, pp. 333-37.
[33] Zeng, Y., Tu, P. et. al. "Causes and implications of the recent increase in the reported sex
ratio at birth in China." Population and Development Review, 19(2), 1993, pp. 283-302.
71
I
Chapter 3
Income Inequality and Progressive
Income Taxation in China and India,
1986-2010
(Joint with Thomas Piketty, EHESS Paris-Jourdan)
3.1
Introduction
Current debates about policy reform in LDCs generally focus on improving the delivery of
social services, the design of market-friendly economic institutions, the effectiveness of poverty
reduction programmes, or the role of trade and market liberalization, and very rarely deal
explicitly with t ax reform and the need to develop modern income tax systems in those countries.
This is unfortunate for at least three reasons. First, poor countries tend to rely excessively
on highly distortionary tax instruments such as taxes on trade or indirect taxes on specific
consumption goods. The gradual shift towards modern and transparent income and payroll tax
systems is generally regarded as an important, efficiency-enhancing aspect of the modernization
process.
Next, many LDCs need to raise more tax revenues in order to properly finance education
and health investment, and income taxation can be part of the solution, especially in an international context characterized by sharp downward pressures on tariffs and various indirect taxes.
73
In countries like China and India, in spite of very rapid growth, tax revenues are currently
stagnating around 10%-15% of GDP, which is probably far too little. There is no example of a
country in the West that has been able to develop a proper education and health system with
total tax revenues around 10-15% of GDP. Improving the efficiency of social services is probably
a good idea, but might well be illusory in case those services are not properly funded.
Finally, many LDCs have witnessed a sharp rise in income inequality during the recent
period. Progressive taxation is probably one of the least distortionary policy tools available to
keep the rise in inequality under control and to redistribute a bit more equally the gains from
growth (it is less distortionary than more radical policy tools such as nationalization, minimum
wages or autarky). In India, the fact that many people did not benefit from the 5%-6% annual
growth rates advertised by the government and felt left out of "shining India" probably played
an important role in the recent electoral defeat of the BJP.
In this paper, we choose to focus on the case of progressive income taxation in China.
Although a progressive individual income tax system has been in place in China since 1980, it
has received very little attention so far, probably because the fraction of the population with
income above the exemption threshold was negligible until the 1990s (less than 1%). Using
annual, 1986-2001 tabulations from urban household income surveys collected by China's State
Statistical Bureau (SSB), we compute series on levels and shares of top incomes in China over
this period, as well as series on theoretical numbers of taxpayers and total income tax receipts
(based on actual tax law). We also make projections about the evolution of the number of
taxpayers and total receipts over the 2002-2010, assuming that constant income trends and
income tax schedules.
One additional motivation for computing theoretical numbers of taxpayers and tax receipts
is the fact that there is widespread presumption that official Chinese income tax law is not
being applied very rigorously by tax authorities. In particular, many observers seem to believe
that tax authorities make deals with large firms and autonomous regions or cities whereby the
latter offer a lump-sum payment to tax authorities and their employees and residents are not
subject to the official income tax schedule. Although at this stage there does not seem to exist
detailed tabulations of income tax returns by income brackets or tax liability in China (such
tabulations exist in most countries with an income tax system), we were able to use aggregate
74
1996-2001 incomrnetax receipts series (broken down by wage income, business income and capital
income for 2000-2001) and compare them with our theoretical series. It turns out that although
there is some evidence that the law is not fully applied, actual receipts and theoretical receipts
are reasonably close.
We were also able to compare our Chinese findings with similar series for India. Contrarily to
its Chinese counterpart, the Indian tax administration has been compilingdetailed tabulations
of income tax returns every year since the creation of a progressive income tax in India (1922).
Indian tax returns tabulations were recently exploited by Banerjee and Piketty (2003) to study
the long run evolution of top income shares in India, and we use their results for the 1986-2001
sub-period as a comparison point for our Chinese series.
Our main conclusions are the following. First, our general conclusion is that progressive
income taxation is about to become an important economic and political object in China and
India, and that income tax reform should rank high on the policy agenda in these two countries.
Due to high average income growth and sharply rising top income shares during the 1990s,
progressive income taxation is starting to hit non-negligible fraction of the population in both
countries (as more and more workers pass the exemption threshold, following what happened
in Western countries half-a-century ago) and to raise non-trivial tax revenues. According to
our projections, the income tax should raise at least 3% of Chinese GDP in 2010 (versus less
than 1% in 2000 and 0,1% in 1990), in spite of the 20% nominal rise in the exemption threshold
that took effect in 2004. The fact that progressive income taxation is becoming an important
policy tool has important consequences for China's ability to finance social spendings and to
keep under control the rise in income inequality associated to globalization and growth. Due to
faster income growth, to lower bracket indexation and to a higher fraction of wage earners in the
labor force, the prospects for income tax development look better in China than in India. This
potential is however limited by the fact that Chinese top wage-earners are currently severely
under-taxed relatively to top non-wage income earners.
The rest of the paper is organized as follows. Section 2 briefly describes the SSB data used
in this paper. In Section 3, we present our findings for the evolution of top income shares in
China, and compare them to the Indian series of Banerjee and Piketty (2003). The results of
our income tax simulations are presented and analyzed in section 4.
75
3.2
Data and Methodology
The Chinese data used in this paper comes from the urban household income surveys collected
by China's State Statistical Bureau (SSB). These surveys are designed so as to representative of
urban China. Between 13 000 and 17 000 households are being surveyed each year (see appendix
Table A1). The micro-files for these surveys are unfortunately not available for all years, and we
asked SSB to provide us with annual, 1986-2001 tabulations based on the micro-files. We asked
for two series of tabulations:
household tabulations and individual tabulations.
Household
tabulations report for a large number of income brackets (and in particular a large number of
top income brackets) the number of households whose total household income falls into that
bracket, their average total income and household size, as well as their average income broken
down by income sources (wage income, business income, capital income and transfer income).
Individual tabulations report for a large number of income brackets (and in particular a large
number of top income brackets) the number of individuals whose individual income falls into
that bracket, their average income and household size, as well as their average income broken
down by income sources. In practice, some forms of income cannot be properly attributed to a
specific individual within the household (this is particularly true for transfer income and capital
income), so that the total income aggregates reported in household tabulations are larger than
in individual tabulations, and various adjustments are necessary when one uses the latter (see
appendix Tables A1 and A2). However the important advantage of individual tabulations is
that China's income tax applies to individual income (rather than household income).
We used standard Pareto interpolation techniques to approximate the form of the Chinese
household and individual distribution of income, and we then used these structural parameters
to compute top fractile incomesand to make income tax simulations. The Chinesedata appears
to be very well approximated by a Pareto distribution (for any given year, Pareto coefficients are
extremely stable within the top decile), although there is some presumption that top incomes
are severely underestimated in the survey data (more on this below).
We did not attempt to use similar tabulations from rural household surveys, but given that
our focus is on top incomes and progressive income taxation this should not be too much of
a problem: average rural income was in 2001 more than 3 times smaller than average urban
income, so that there are probably very few rural households and individuals in the national
76
top decile, and even less so within the top incomes subject to progressive income taxation
(agricultural income is exempt from the income tax and is being taxed separately).
We did not use any new Indian data in this research. All our series regarding India are
borrowed from Barnerjee and Piketty (2003), who used Indian income tax returns tabulations
published in "All-India Income Tax Statistics" brochures (annually available since 1922) to estimate top income levels and national accounts to compute the average income denominator. Top
income shares estimates based upon income tax returns are likely to be higher than estimates
based on survey data (as the latter generally underestimates top incomes), but there is no obvious reason why the trends should not be comparable. Note also that the standard household
surveys used by economists working on India (NSS surveys) can hardly be used to compute
top income shares, as these are mostly expenditure surveys: except for particular years, and
contrarily to SSB surveys, NSS surveys contain no systematic information on incomes.
3.3
Top Income Shares in China and India, 1986-2001
Did income inequality in China increase as much as in India during the 1990s? Before we look at
our top income shares series, it is useful to recall one important difference between Chinese and
Indian incomes during the past 15 to 20 years. While real GDP per capita increased by almost
160% in China between 1986 and 2001 (6,4% per year), it increased by slightly more than 60%
in India (3,4% per year) (see Figure 1). According to the best available PPP conversion factors,
real per capita GDP was virtually identical in China and India in 1986 (less than 20% larger in
China), and it is almost twice as large in China as in India by 2001. Note that the growth gap
is even larger if we look at survey data rather than national accounts. While total 1986-2001
income growth is virtually the same in Chinese national accounts and household surveys, there
exists a well-known "growth paradox" in Indian statistics: real GDP per capita (as measured
by Indian national accounts) has increased by 64% between 1986 and 2001 (3,4% per year),
but real consumption per capita (as measured by NSS surveys) has increased by only 24%
(1,4% per year). According to official Chinese statistics, there exists no such growth paradox in
China: real GDP per capita (as measured by Chinese national accounts) has increased by 154%
between 1986-2001 (6,4% per year), and real per capita income (as measured by SSB surveys)
77
has increased by 140% (6,0% per year).
If we now look at the evolution of the top decile income shares in China over the same
period,,we find that income inequality has increased at a very high rate during the 1986-2001
period. According to our urban survey estimates, the top decile income share rose from about
17% in 1986 to almost 26% in 2001, i.e. by more than 50% (see Figure 2). The levels are
probably underestimated (they are even lower than in the most egalitarian developed countries,
e.g. Scandinavia), but the trend seems large and robust.
As we move up the income hierarchy, the trend gets even bigger. For instance, the top 1%
income share has almost doubled between 1986 and 2001, from slightly more than 2,5% in 1986
to over 5% in 2001 (see Figure 3). If we compare these results with those obtained for India,
we find that the levels are much lower in China than in India (the Chinese 2001 top 1% share
is lower than the Indian 1986 top 1% share), which again suggests that survey-based measures
underestimate top incomes, but that the trend is substantially larger in China. The top 1%
income share has increased by more than 90% in China between 1986 and 2001, and by less
than 50% in India (see Figure 4).
These results can be used not only to evaluate the prospects for progressive income taxation
in China and India (see Section 4 below), but also to shed some new light on the on-going debate
about globalization and the rise in inequality. Although our data does not allow us to identify
precisely the causal channels at work, and in particular to isolate the impact of globalization,
we note that the fact that the rise in income inequality was so much concentrated within top
incomes in both countries seems more consistent with a theory based on rents and market
frictions (see e.g. Banerjee and Newman (2003)) than with a theory based solely on skills and
technological complementarity (i.e. inequality rises in the South because low-skill southern
workers are too low-skill to benefit from globalization; see e.g. Kremer and Maskin (2003)),
which would seem to imply more gradual shifts in the distribution. To the extent that the skill
distribution is more unequal in India than in China (e.g. literacy rates are substantially higher
in China), the skill-based theory would also seem to imply that income inequality should have
risen more rapidly in India than in China, whereas we find the opposite (as far as the top 1%
income share is concerned).
78
3.4
Progressive Income Taxation in China and India, 1986-2010
VWenow come to the issue of progressive income taxation. Table 1 describes the evolution of
Chinese income tax schedules during the 1980-2004 period. The striking fact is that China's
income tax law has remained basically unchanged since its creation in 1980. The only major
change is that the nominal exemption threshold for wage earners has been raised from 9600
yuans per year in fiscal years 1980-1998 to 12000 yuans in 1999-2003 and 14400 yuans since
2004. Also note that the Chinese income tax systems treats wage income in a much more
favorable manner than business income and capital income: while wage-earners are subject to
the income tax only if their annual wage is high enough, all business and capital income is
subject to the tax.
In contrast to the Chinese income tax, the Indian income tax (which is much older, since it
was created in 1922) has always treated all income sources equally: the progressive tax schedules apply to total individual income, irrespective of where the income comes from. Another
important difference is that the tax schedule has been changed almost constantly in India during the 1986-2004 period, resulting into a general decline in tax rates and a continuous increase
in the exemption threshold (see Table 2).
From our perspective, the first important implication of these differing evolutions is that the
exemption threshold (for wage earners) has increased less than inflation (and much less than
nominal incomes) in China since 1986, while it increased approximately at the same rate as
inflation in India, resulting into a massive increase in the proportion of the population subject
to the income tax in China and a more modest increase in India (see Figures 5, 6 and 7). In
China, the exemption threshold in 1986 (9600 yuans) was about 7 times larger than average
individual urban income (1394 yuans), so that less than 0,1% of all wage earners were subject
to the income tax. By 2001, the exemption threshold (12000 yuans) was less than 15% larger
than average individual urban income (10787 yuans), so that 32,2% of all wage earners were
subject to tax according to our estimates. In India, the exemption threshold has always been
set around 2-3 times average income during the 1986-2001 period, and it is only because of
the rise in top income shares that the proportion of the population subject to the income tax
has increased somewhat during this period (from 0,7% in 1986 to 3,8% in 2001). This is an
important rise from an historical perspective (the proportion of the population subject to the
79
Indian income tax had been relatively stable around 0,5%-1% between the 1920s and the early
1990s), but this is clearly much less than in China: due to lower bracket indexation and higher
real income growth, the Chinese income tax has become a mass tax during the 1990s, while it
remains an elite tax in India. Assuming that China's 2004 income tax law applies until 2010
(i.e. there is no further rise in the exemption threshold after 2004) and the income trends
(both in average income and top income shares) continue after 2001 at the same rate as during
the 1996-2001 period, our projections indicate that almost two thirds of Chinese urban wage
earners (over 200 millions individuals) will be subject to the income tax by 2010 (see Figure 8).
One important question, however, is whether the Chinese income tax law is really being
applied in practice.
I.e. do all individuals who are supposed to be subject to the income
tax according to the law really pay the income tax? Many observers in and outside China
seem to believe that tax authorities make deals with large firms and autonomous regions or
cities whereby the latter offer a lump-sum payment to tax authorities and their employees
and residents are not subject to the official income tax schedule. Unfortunately, there does
not seem to exist any reliable statistics on the number of income tax taxpayers in China (let
alone tabulations of taxpayers by income brackets, similar to what is being published in India
and other countries), so we cannot compare our theoretical numbers of taxpayers with the
actual numbers. However we can use data on aggregate income tax revenues and compare it to
theoretical tax revenues in order to evaluate how strictly the law is being applied. We compiled
from China Tax Yearbooks aggregate income tax revenues series for 1996-2001, broken down by
income source (wage income, business income, capital income and other income) for 2000-2001.
This very useful decomposition of tax revenues does not seem to be available prior to 2000.
The comparison between actual tax revenues and theoretical tax revenues is given on Table 3.
The first conclusion emerging from Table 3 is that actual income tax revenues are reasonably
in line with theoretical tax revenues (as a first-order approximation), thereby suggesting that
income tax collection in China is somewhat less chaotic and arbitrary than what many observers
tend to assume. In 1996, actual income tax receipts made 0,28% of GDP, and theoretical receipts
0,33% of GDP; in 2001, actual income tax receipts made 1,02% of GDP, and theoretical receipts
0,66% of GDP (Table 3). If we look separately at receipts by income source for 2001, we find
theoretical receipts on capital income were equal to 40% of actual receipts (this reflects the
80
fact capital income in under-reported in surveys), and that the corresponding figure was 64%
for business income (business income is also under-reported in surveys, but less severely than
capital income) and 96% for wage income. The latter figure could be interpreted as saying that
wage income is fully reported in surveys, and that tax law if fully applied (all wage earners who
are supposed to pay the income tax do pay it).
Such an interpretation might well be misleading, however. There are good reasons to believe that top wages are under-reported in SSB household surveys, in which case the fact that
theoretical receipts (based upon under-reported top wages) and actual receipts coincide merely
reflects the fact that collection rate is less than 100%. If we adjust top survey wages so as to
obtain reasonable Pareto coefficients for the distribution, we find that theoretical receipts for
wage income are equal to 216% of actual income, i.e. the tax collection rate for wage income is
less than 50%. Although the problem is probably less severe than what many observers tend
to assume, these illustrative (and highly uncertain) computations suggest that there does exist
a tax collection problem in China.
It is also interesting to note that actual receipts have increased at a significantly higher rate
than theoretical receipts during the 1996-2001period. One interpretation could be that tax
collection has improved. Another interpretation
is that household surveys underestimate not
only the levels of top incomes, but also the upward trend in top income shares. In order to get
a sense of the likely magnitude of this effect, we computed by how much the upward trend in
top income shares needs to be scaled up in order to ensure that the trend theoretical receipts
does match the trend in actual receipts. We find that the 2001 top 1% share should be scaled
up by about 35% relatively to the top 1% share in 1996, which is substantial (see Figure 9).
Although there is some uncertainty about the quality of tax collection and survey data,
actual and theoretical tax receipts both show that income tax receipts (as a fraction of GDP)
have increased substantially during the 1990s. The contrast with India is particularly striking:
while Indian income tax revenues have stagnated around 0,5%-0,6% of GDP during the 1990s,
Chinese income tax revenues have been multiplied by more than 10, from less than 0,1% of GDP
in the early 1990s to over 1% of GDP in 2001 (see Figure 10). The stagnation of Indian tax
revenues reflects the fact that tax rates have been continuously reduced (see Table 2) and that
the proportion of individuals subject to tax has increased only modestly (see Figure 7). The
81
substantial rise in Chinese tax revenues reflects the facts that tax rates have remained the same
(see Table 1) and that the proportion of individuals subject to tax has increased enormously
(see Figure 7).
Note that Chinese tax revenues would be substantially larger in the absence of a preferential
tax treatment given to top wage earners over top business and capital income earners.
We
computed that if the business income tax schedule was applied to wage income as well, then
Chinese income tax revenues in 2001 would be more than 3% of GDP (instead of 1%). Although
this preferential tax treatment of wage income might raise serious political problems in the
medium run (as independent workers feel more and more disadvantaged as compared to top
wage earners in large firms), as it did in other countries where similar preferential tax treatment
was applied (such as France), removing this legal provision is however unnecessary to ensure
the growth of Chinese income tax revenues. Because of the phenomenal growth in average
incomes (and even more so of top incomes), income tax revenues should make much more than
1% of GDP in 2010. According to our projections, which are based on the assumption that
tax law will not be changed after 2004 and that income trends will remain the same as in
the 1996-2001 period, income tax revenues in China should make about 4,3% in GDP by 2010
(see Figure 11). The assumption that the exemption threshold will not be raised after 2004
does not seem unreasonable, given that the 2004 increase in the exemption threshold was fairly
high (from 12000 to 14400 yuans, i.e. 20%) and that inflation is currently very close to 0%.
Moreover our projected tax revenues estimates should be viewed as a lower bound, first because
we assumed that the survey-based trends and levels in top shares were not under-estimated
(in particular we did not make the adjustment reported on Figure 9), and next because we
assumed that there would be no improvement in tax collection (1996-2001 show that there has
been some improvement in tax collection and/or an under-estimated rise in survey-based top
income shares). In other words, there are good reasons to believe that the income tax will raise
at least 4% of GDP in China by 2010.
If this happens, then China will have gone through its fiscal revolution. As Table 4 illustrates, moving from an elite income tax raising less than 1% of GDP to a mass income tax raising
around 4-5% of GDP is exactly the kind of process through which Western countries during the
1914-1950 period (when their income levels were similar to current Chinese levels). Although
82
Indian income tax revenues will probably increase during the coming years, the prospects for
India look less good, both because of lower income growth and higher bracket indexation. One
reason why India faces more difficulties than China in making its income tax a mass tax might
also be that the proportion of formal wage earners in the labor force is ridiculously low in India.
83
Bibliography
[1] A. Banerjee and A. Newman (2003), "Inequality, Growth and Trade Policy", mimeo, 2003
[2] A. Banerjee and T. Piketty, "Top Indian Incomes, 1922-2000", mimeo, MIT and EHESS
(Paris-Jourdan),
2003
[3] S. Chen and Y. Wang, "China's Growth and Poverty Reduction: Recent Trends between
1990 and 1999", mimeo, World Bank, 2001
[4] A. Deaton, "Adjusted Indian Poverty Estimates for 1999-2000", Economic and Political
Weekly, January
25, 2003
[5] R. Eckaus, A. Lester and N. Qian, "Income Inequality in a Transitional Economy: China
as a Case Study", mimeo, MIT, 2003
[6] M. Kremer and E. Maskin, "Globalization and Inequality", mimeo, Harvard, 2003
[7] Piketty, Thomas (2003), "Income Inequality in France, 1901-1998", Journal of Political
Economy 111, 1004-1042
[8] Piketty, Thomas and Emmanuel Saez (2003), "Income Inequality in the United States,
1913-1998", Quarterly Journal of Economics 118, 1-39
[9] M. Ravallion and S. Chen, "When Economic Reform is Faster than Statistical Reform:
Measuring and Explaining Income Inequality in Rural China", mimeo, World Bank, 2003
[10] S. Tendulkar, "Organized Labour Market in IndiaSchool of Economics, 2003
84
Pre and Post Reform", mimeo, Delhi
[11] S.J. Wei and Y. Wu, "Globalization and Inequality: Evidence from Within China", NBER
Working Paper 8611 (2001)
85
o00
-o
-C
a1)
U)
0
CS a
Ca)
')c
.
-
a)-
a
3
a)
0m
E
C-l -0
0
ma
r
t
0
-c
U)
a)
I
'r
I
.
I
r
4
(B
I
iC
f
1
I
00
o00
S
n
.0
-co
C
a)
Co
U)
o
o
=
n
-d
._ ,CL
0
C-
0''
,L 0 ~
,
oa)
C)
o
=
0
Icu>5
D
E
o
CL C)
.- ':.-
Cu
Wa)
0aVa)
LC)=
W0 c
mC
*C',1
L
N
I rs 0n
0
c
C.
m.
co
m
-n
CD
-C
U)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~M.^
.:
w
C
0
0,
0
Um
C
.5
.,c
*0o0
C- Y
O
i
0)V
u
0.L2
0
_0.0
10
(co0
(0
C
CCU)
L)
Cu6
C O
a5 T
20U
0
CuCu
1 )
I-
-,
u,
<I-
m
0.
0.Q
Cu
Sr
c
h
gC
QL
c
n
00
cO
*
.s
;:
.
" ,
:
U)
0
m 1)
U))0
Qa,
Ii
0- 1
of'if.
0 .D
I
J!
C:
0
0
.
U)
=C
m0
_0
C')
a,
-c
c c tD
00
C
0) 0)
- C) $
Cu C3u 0
ii '
N
f=
0
o
0
-
OC4
co
CO
0m
0
0
i
m
m
Q)
E
ox
00
a)
D
s00o
a
0
00
-
o
-
>C)<
CD
U)
o0I
Cu
bC-
Ct.)
x
U)
(-
CO
U.
a)
-o
-
I-
a)
02)
EL
00
-r
U)
'C
U-
)
C')
C'
U
U
0
20
o20
20
20
20
6o
20
6
2
6
20
selel~l o uoloeJ_-
6
0
6
20
a)
6
z0
·-
a)
O D
a).o
(.)-
a,
-;
.
0
o
0
K
0
c
a0
0
c
-
m
co
-2
o
a)
occi U) .0)
)
-
cw
80
0
c- ,
c
o
C
"
c
Cu
I
N
I,
-C
=
cc
c0)c
' -
2
a)
L
clc
.2) 2
a
)
O-
a)
D
CD
N
-
~!a)
-
.?U1
C
N
a)D
Oa)
rD
-
a)
a2 )
cn
E)i
a)
0 ,
ur
)
0)
c
cc~~~
So ~
c
cO
ot.D o
o
cN
o0
o
00
o
co
occ
0-a-
0c-i
0
cc
0)~~~
o
cc
o
cc
0o0
.i
0
0
C'
_
c)
o0
Icn
~~~c
OO
-
Zo c
0
0
U)
.-c
o
r-
0)
-c
u)
0
0W_0
--
o.
a,)
Lo
o F
co
I-0
o
c~
0
.. =C
Co
o
O0
0
0)
o
Cu
1-
-
0
a:
-;
am-
=o
tn~'4
"00'a)
r-
o)
0)
cc
~~~~i7 ~A>
<Da)
en
N) a)
L
~~~~~D
O
>o~
U-.
I
)
CR
.to_
0)
.
ow
0)
~Cu
Ov
a)
cc
cc
tg
0O
00
0
c
0
O
L,
0
0
cc
0
cc
0n'
0
'i
0
c-i
0
0
0
0
CD
0
'I
0
c-I
0
t
ou
C"'
0
0
-
oo
Em
00.
00,0
0O
C
aC O
3Cl
.-o' a)
u ) C)
0
O
'o
O sa
mid
ia-
C3C
0
cc
a.)
0
Co
0)
C
=)d 0o
~
z (
2o
,
0,)
E
0CIO
O scm
-
00- C
r.0)
C)
.o
ira.
*-
ooE
Nz o
0c
Lo
'a.O*
z
oo o
Co
o
0
Z
0 0 0 0 0 0 0 0
o
o
0
o
o 0
o
-
N
0
-o
Li
*-Q
oE°o
|OmCl
o --
a)
(
o48
co
CCo
EC E
Co
o
o
Co
v
O
0
0
T
0
cm
0
I
N
0
O0
0
0
N
0
o
0
0
0
Co
0
6> 000ooo'o01.
/8Ai'
0)
o
-
(0
T
o
M
N
-
0
-
C C
-c
Co
2O
tL
-~'o
a)
aa _' o
0)
a '
0
Co
C O0D o O
o. U) ~.-
0)
a)
r-
*0
m
Lm
Co
0
OD
- 0
CD
C ~0Q
<im._ 0O
aI0
(L
V
--O i ')r-CYC-
LO
,U)
· 0)
°a
Co
ci)
-5
I- a)CD C
C0v
)
-0
.
"- Co
C
,
6
CO
elH/ H 000L
N
-
0
75
)
U0C
o
~o o
o
z
C
o oo . cO0)~l
CE
C~l ,-,~~~~~~~u)
>- --
-)
gV
C
E
U)(V~~~~U
a)
CO
C)(
'E co ._.
CD
co
D
(U)
~~
0)
.
0
cn
· ~ b E',,
.C
o
8
'a
r_
Ca)
ci
C
0o
(Lo
0
0 0 00
Co
0~~~~~~~~~~~~~~1)0
V
0
Cli
LO
N
N
C
-
o
0
-z
L(
0
Figure 5A- Fraction of Males in Counties which Plant
Some Tea and Counties which Plant No Tea
0.54
0.53
0.52
0.51
0.5
0.49
1962
1966
1970
1974
1978
Birth Year
-No
1982
1986
1990
Tea -Tea
Figure 5B - The Effect of Category 1 and 3 Crops on
Sex Ratios
Coefficients of the Interactions Birth Year * Amount of Category
1 Crops Planted and Birth Year * Amount of Category 2 Crops
Planted in Unrestricted Sex Ratios Equation
0.03
0.02
0.01
-0.01
-0.02
-0.03
-0.03
I
1964
I
I
1968
I
I
,
1972
,
i
I
1976
r
1980
i
Il
I
1984
I
I
I
1988
Year
-- Cat 1 .- Cat 3
93
Cu
c
0=
0
0
0
Cu
o0.
0
m
=0
a)
E .2
.C
S° O
Cuc
l\s
/
0IC
a)
_U
O
3CW
0a)
0~m
am
ca
,_
cn
Cd)0
Cn
C
0
W
4a)X
-oC C6
\
N
U
=00-0
.52c
C.)
.
_
00
~ o-x
.)
I
a)
EL0a)
o O
co
a)
I
o
N
-
o
0)
.-
Ecn
C.)
co
0)
m
e
0
Ci
0
6
6
O
C)
C'J
0
O
0
O
-O
N
Cm
0
0
0
o_ 2
m
CU)
C
i
7C--
0
r
C.-
xO
0)
XOo
U) <
Z,
c-w
oo
CO
a)6 C0
CD
0:
0
E
C-
U
0)n
*)
03)
fC-
a)
I...=
o
Cu
a)
oO .
C-
0-
01
m
0.O
- a)
Io; *5.
Cu
U
a)
;z
a)
=
.,-
,-
Ca0
T
= Cu
a)
CO
CO
0
c o
a)
C.)
CO
Cu
0;
6
0
0
o
6
o~~~~~~~~
0
6~
o0
C
90
0-
90
U)
0O
90
t3 ao)
o
Uw
._L
a
4)
.I-r
cm x
0o
o aO
<
0
r
CS
co
O
.2n O
W
4
>-
co
i
,a) C'.
n
D
U),
0)
,
0 C
0
C - I
(2=
C2)
U.
0
co
Ca)
-
C)w 0L
_0.
CD
o
~._W Cl
SO
/
~Cu
_)
.
i
t)
I-r
0
U)
,- t
t
a)
a)
.-
;a) .. 0V
m
C.
6
0o
6
0
0
6
c,
0
0
Figure 7 - The Effect of Planting Tea and Orchards on Girls'
Education Attainment
Coefficients of the Interactions Birth Year * Amount of Tea Planted and
Birth Year * Amount of Orchards Planted in Pooled Education Equation
0.15
-0.1
0.1
-0.3
0.05
-0.5
0
U,
° -0.05
-0.7
O -0.15
-0.9
-0.2
-1.1
..0
-0.1
(U
!--
-0.25
-1.3
-0.3
-0. 35
-1.5
1964
1968
1972
1976
1980
Birth Year
-
Orchards -
Tea
95
C
-1
~
oo CO
O ON
0 C0
0 0 ON
N-
C
CD0O
0
Co
0
0
0
oo 0 00 00_ 00 00 ON00
Ci)
N -D
N
0
CD
Co
0
CO
Co
0
)
666666
0
N)
·
N
CO
0
OO
r0
o;o
oo 000o 0O0
0 o'00
C) 0 0 0
CD C .t
CO LO
~''It
o
~d
N
Co
0
0 0C
U
0
N
C)
CD
0
-D
C mCD
)
* o
N 00 O) c
o
N C N 0C
o o N
o ooCoCC
'O
6666
N
CD
O
CD
'r-
0
c. O
~: tO
0
0)
o)
0
!O!! °°°Olrl
CO Duo
O
stt m
FD FC o-
tDC
0
C0
q)
0')
0 o 0 0
0 00
o0) 009
O9
0.
o o 0 0 0
o 0 0 0 0
000 o o
o66
0) 000
oOo0
)
E
6~
E)
O; ;
O0
o
U)
C
00
Co
0 C00
0
C)
~r~' 0
0
0
N
Co
66ooo o
Oa a a
0
U) u)
0
in0
.
q
0
0
COC
LO
0
N
0
0
0
o
CD
)
0
0
N
CD
0
0
oq 6
O
o
O) o~ q
sa.ss.as.
N
N
N
0
0
0
o
'-
N
CD
CD~
0
0
0
NCD
0
0
q0
C
N-
o
q
.
o
<0
0
0
~
O
-
gg
._'r
._
N-
C
0
0
0
Qo
~-o
O
CO 0 CO
>
CO CO D
0
SR
(D
CD
NOCO
O-
I
CO CO CO
ur
uR COUR CD
CD CD CD
CD
-t
It
-It
w
'I)
t
CD
uC1-CD
rCD
N-
N
Co
C
-
CD
CD
CD
CD
C 0
0
0
0
CN
, 0C
C N NO
CD N- C
O
U- 'C) co:) cl
CDt CD CD CD
N
co
(D
CD CD N
O0
O0t O O,
0 0 0 0
I-
C
U)
0
Co
CD
co
D
0o
CN
N
't
~~36
o
E
Rlo
C
0*0I0 D
0)Cu
~
CD CD CD
CO
C
0:
Co
COCo
CO
cN
N
"t N
6 X
)-
o
N >
0
-
.U)
V
'U
=g_
c) -ZX
:51
co
D C CD
.x II
~iLC
0
U)
0)
a)
s
F-
3U)
00
-
0
CU
U)
0.
0
>
0
CU
co
N
CD
0
E ·
U)
0
O
o
0
o g=
'=EU
_C-
w
L)
VS ) .c,°o s _~a)
C
CV.8---
I_
U)
Cu
0
Eoo-
* o)
-~
0o C)0
gU)
o
C
C UIO
0
0. U)
o-
a 4Li Co
. U)
00O
OL*,
o
.-
0._= m a)
4-
0
_) _
e uEEs°
C
0 0X
/)
U
<c-
-3
.
Cu
U)
c
6
C) >~ N
2
'sO
.0
CD
C
ii
0 cn
3
0n c CUC
C:
Cf) C- c
C:
0
0
§
Cu
CDi-
0
0
C/) Ca) W
Cu
a)
C0
a)
.)
E
in=
Cu 0 U)
wOO0
Table 2 - The Effects of Tea, Orchards and Cash Crops on Fraction of Males (Unrestricted):
Coefficients of the Interactions between Dummies Indicating Birth Year and the Amount of Tea, Orchards or
Category 2 Cash Crops Planted in the County of Birth
Dependent Variable: Fraction of Males
Tea
Orchards
(1)
Cat 2 Cash Crops
(2)
(3)
Birth Year
Coeff.
Std. Error
Coeff.
Std. Error
Coeff.
Std. Error
1963
-0.005
(0.013)
0.001
(0.005)
0.000
(0.002)
1964
0.005
(0.023)
0.003
(0.006)
-0.001
(0.002)
1965
-0.026
(0.013)
0.000
(0.005)
-0.003
(0.002)
1966
-0.009
(0.014)
0.003
(0.005)
-0.001
(0.002)
1967
-0.014
(0.015)
0.003
(0.005)
0.000
(0.002)
1968
-0.021
(0.014)
-0.003
(0.005)
-0.003
(0.002)
1969
0.001
(0.015)
0.000
(0.005)
-0.001
(0.002)
1970
-0.022
(0.016)
-0.007
(0.007)
-0.004
(0.002)
1971
-0.008
(0.011)
0.002
(0.006)
-0.002
(0.002)
1972
-0.012
(0.010)
-0.006
(0.005)
-0.003
(0.002)
1973
-0.022
(0.011)
-0.007
(0.006)
-0.004
(0.002)
1974
-0.019
(0.014)
0.000
(0.005)
-0.003
(0.002)
1975
-0.014
(0.012)
-0.008
(0.007)
-0.002
(0.002)
1976
-0.002
(0.019)
-0.005
(0.006)
-0.002
(0.002)
1977
-0.010
(0.018)
-0.003
(0.005)
-0.002
(0.002)
1978
-0.023
(0.014)
-0.005
(0.006)
-0.004
(0.002)
1979
-0.006
(0.011)
0.003
(0.006)
-0.002
(0.002)
1980
-0.031
(0.015)
0.000
(0.005)
-0.004
(0.002)
1981
-0.021
(0.015)
0.001
(0.006)
-0.004
(0.002)
1982
-0.024
(0.011)
0.010
(0.005)
0.000
(0.002)
1983
-0.029
(0.015)
0.003
(0.005)
-0.002
(0.002)
1984
-0.035
(0.018)
-0.003
(0.005)
-0.005
(0.002)
1985
-0.026
(0.016)
0.002
(0.005)
-0.003
(0.002)
1986
-0.028
(0.014)
-0.003
(0.005)
-0.004
(0.002)
1987
-0.016
(0.016)
0.003
(0.005)
-0.001
(0.002)
1988
-0.042
(0.012)
-0.006
(0.006)
-0.006
(0.002)
1989
-0.037
(0.019)
0.000
(0.005)
-0.005
(0.002)
1990
-0.037
(0.018)
0.010
(0.006)
-0.003
(0.002)
Observations
R-Squared
49082
49082
49082
0.14
0.14
0.14
All regressions include county and birth year fixed effects.
Standard errors clustered at county level.
97
Table 3 - The Effects of Tea, Orchards and Cash Crops
on Fraction of Males (Pooled):
Coefficients of the Interactions Between Dummies Indicating Birth Year and the Amount of Tea,
Orchards and Category 2 Cash Crops Planted in the County of Birth
Dependent Variable: Fraction of Males
Orchards
Tea
(1)
Cat 2 Cash Crops
(2)
(3)
Coeff.
Std. Error
(0.009)
0.000
(0.002)
(0.010)
-0.001
(0.002)
0.012
(0.009)
-0.003
(0.002)
0.011
(0.009)
-0.001
(0.002)
(0.018)
0.002
(0.009)
0.000
(0.002)
-0.014
(0.017)
0.003
(0.009)
-0.003
(0.002)
1969
0.013
(0.018)
0.011
(0.009)
-0.001
(0.002)
1970
-0.013
(0.019)
0.001
(0.010)
-0.004
(0.002)
1971
0.008
(0.014)
0.016
(0.011)
-0.002
(0.002)
1972
-0.003
(0.014)
0.002
(0.010)
-0.003
(0.002)
1973
-0.001
(0.013)
0.003
(0.010)
-0.004
(0.002)
1974
-0.003
(0.017)
0.014
(0.010)
-0.003
(0.002)
1975
-0.021
(0.016)
-0.012
(0.011)
-0.002
(0.002)
1976
0.003
(0.023)
-0.002
(0.012)
-0.002
(0.002)
1977
0.001
(0.021)
0.006
(0.009)
-0.002
(0.002)
1978
-0.008
(0.016)
0.008
(0.009)
-0.004
(0.002)
1979
0.009
(0.014)
0.015
(0.010)
-0.001
(0.002)
1980
-0.014
(0.017)
0.014
(0.009)
-0.004
(0.002)
-0.004
(0.002)
Birth Year
Coeff.
Std. Error
Coeff.
Std. Error
1963
-0.005
(0.016)
0.001
1964
0.019
(0.026)
0.015
1965
-0.013
(0.016)
1966
0.000
(0.016)
1967
-0.015
1968
1981
0.003
(0.018)
0.022
(0.010)
1982
-0.014
(0.014)
0.017
(0.010)
0.000
(0.002)
1983
-0.021
(0.018)
0.009
(0.008)
-0.002
(0.002)
1984
-0.016
(0.021)
0.012
(0.009)
-0.005
(0.002)
1985
-0.006
(0.019)
0.017
(0.009)
-0.003
(0.002)
1986
-0.016
(0.017)
0.006
(0.009)
-0.004
(0.002)
(0.002)
1987
-0.005
(0.018)
0.014
(0.009)
-0.001
1988
-0.025
(0.015)
0.008
(0.009)
-0.005
(0.002)
1989
-0.015
(0.022)
0.019
(0.009)
-0.005
(0.002)
1990
-0.013
(0.023)
0.029
(0.011)
-0.002
(0.002)
Observations
49082
R-Squared
0.14
All regressions include county and birth year fixed effects.
Standard errors clustered at county level.
98
t Ea
-
0o o
oo
0000
0
0
co
-
(D
°,o
°
0l>
00
C
o
a)
a)
o
E 0
o
c
C)
m-_
cuz
0)
CUa0
c
030
4)
0
0)
a)
X
E U)
o,~*oo
CU C
o
) 8
o~*
CD
cD
tC0
*:O o
2
v
w
U)
C
0
0
0
-.
U)
C-3
C
0
0
sc
in
0u
.2
c
4) O-
.C
0
0
c
I
a)
C
o
(DRno
0) ot EF
l
0u
CD
CO
0es)0q
qC)
066 C)
0
00
O0
000)6C
2 °)
co
q0 qC)
0W
)
Z0
O
CU
U)
.
03 E
0
06
4
E)
V CUO
su
0
C
V
o
a)
--
a
0 4- C
E
0 .E
0
0
OCUE
C:
0
' cE
_
0
CU
a)
2
CU
8 X° a)
U
00
O4
0Q.C @
-
a)
0
0
<:OCU
C-,
0)
V
0
CU
0
0)
>
(n
0)
0;
9o
0 rI' -,
U)
96
LO
(D
>-
...
.).
0
a
0
-o
._o
O-
o:
9 0
o,
o
SI
'O
CQ
I-
C)
0~~
Z
LO LO
oC
z
CDz,
~E
00
-=
0
0
-C
JL
u0
0
.c
0CeD
U)
=0o m
C0D'
U)
0o
>-
-O , o
O
UsC\J
co
c~~~~~~~~~c~
(O0
==
C
m
ci
2
Cu
0
U)
.0
CU
*aO
)°E)
0c
~0
0I~.Co >
~
~o I"
C o
C
RI
Cu
Cu U)
0
a)
0
CL)
O
0
C
j3c
o
0) U)
-
Cu
. U)-
0)
0
sa
0
U-
js CD
eo
~~~~~(0
C-.
LO CDO
n
-Cl)
Cu
0
0
cn-J
9906 'z
-o0
0
0
cL )
Cs
oO
U)
o~
.x,
0
a C
U)0
C
iLa)
0-
-i
C
U)~~~~~~~~~~~~~l
I-''D
0C
U)u
0 _C
~0
0
o
-D
a)
0
o
C
C
0.
C
.C _
0
0
n.?J
0o
a)
co
0
0
0
0
cu
C
0
0 (
-o
<r1
C
s
1~
O
cn
a)M
E
E-
a
Ct
0)
q Z
0:)
co
CD
No4O
C'
CO
)
OD
o co
V
O9' ci
C-'
>C
p
0
cu
cu
0
'a
a)
c
It)
C
-
a;
CO)
oo
_- 'a
o o
oo
otU) C
o
0)
0
66~
C.)
CO
C)
0n 0
R
t
0~ca)
I0a)
ES~ -C
.cU )c=o(/3
E
.u' Ca) <
0,
CD E
0,
ILL
co
CO
oo:
u:
C)~
O)
,
o,'
oC
0 0.
0
00
It)c
It
O ".
5ooO9
-
co
~o
C~C
C
C.
Cu
1u
0
C
a
o,_
OD-
C
0
ooL
15
- (O
o,' C
CO
CO)
Ci
C)
N4 f
LO
C
Cu
o
a)
0
I
C
13.C
c - -~
= o-0
u
n
0
q-0~
0
,-- c'(9
C 0
C CD
o 'o
i0 oC) <tC
.0
MI
X
ECOO
CD
0,
z5 0(
c
CD
0
-
(0
(D
V2
6
o
o
0
u3
to
x0O
V
B'O
XC
a)
CD
~O
co
O O
9o
c)a)
C:'
O
0
06
Io
,t
COX
C
3
9o
069C
c
:
0).
'tC
co
c
0?.
Cu
CD
O0
0 I:
0.
(0
c3 o
cn
0
,
a)
0
E
:E
N
E
0,
IL
I'-
LO
C
6
C)
C
.oo
(N0
L
C
,.t
CN
N
o. oN)
,6
c0
L)
O
0
.I
0,~
xO
X
(9
m
iR
. O3
CDO CC
CD
·
t j.
oN
6o
CI
d
C1i C)
96c
a
o0
.0
(9
0,
0,
C
C
C
00,
C
0
2
co c;
(
CO
-C (3Ea)
o 0)0
0
*O.y
C V
Q-C
a)
'Q
0_
a)
I-
0)
0
0o
a)
0
cC
0 0
Cu
(N
°Zot
-0
V'a)
)
2 Onm 0,
Cu
0,Q a)
.0
.S
Cu1V)U
C1
c0
O
- t
C
(NI
U,
o.-
o
0
l----..-..--.----.......
........----.......
I
t
a
oL
C-,
C
o
I
f
O- 111
c
X
o
.c0d
V o <.
N
I
I
.72
co
a)
O)
)
aw cm0 t4.
ui
.
:2 I44)
-x 0)o fa
0
S
CU
oC muC: ,
I0
c
-0)
-
m
0
a)
t
C
VD
a)
w
C
i
CD 0)
c.
C:
a) -
°
0
6
o
s
r
U)
.'
V
'
0
6
C)
6
Cm
N
0
6
0
Z0
0
0O
6
r
|
CN
0
M
4
0
0)
r
-
Cl-
C
CII
_,0,
QW
- w
CL
0>
66
6
ci
6
6
. 6cm.
C,
CD
c
0
U-
x
xL.X
C')
0
0
99
a) X
Dc C
CnO )C1
O)
0
s 0O
C-
m -j
n (D
C-
/
02~~~
aL
OCU--
(n
c
0,
C
C0s
o)
.
0
(U
C,)O3
0
0)
3
(
m
(U
>-C
O03
0)4
°:E
O
: -0 c3
UWUXi
o0
c-
CD
o- 0
CD
I)i.
--
co
(4)
co
a)
.-
<CU
U)
q
0
0
U)U)
o
C\
0
C
Lo
'I
o
c"
o
o o o o o
oo N
o
C
o
o
V:
U)
0~
oC)
U)
0
oS
C) C)
U
aD
0
O
R
ooo6o 6 6 oo o666 6 oo
C)
o
C)
F3
C)
)
)
C)
C)
C)
O
C)
O co
C)D
6 E) 6 ZC) C
6
C)
C)
C) Ca
o
x- o ~o
66 o
v
C
N
U)C
Oq
N,
a) o
0
U-
D
i
-
O
C)~ C
C
i
U)
0
-I
v
O
ON ~ ON
Nq N
0
C
C
C
C)
D
0
N:-
NZ- ;
-
i
-
LO
-
(C
0
-
q Nq N,
-
-
-
C
)
N
0
0
N - -
'It
U)-
M-
6 9 9 9 9 9 9 9 9 9 9 9 9 9 9 O9 9 9 9 9 9 9 9
q
Nq- D CO U)
C
C)
U1)
-C
-5 -o2
0
Q
.,
0n
(
i
0
C
N
O) O O CO C) O)
a a) a
O OOO6
O O6O
C
)
C)
C)
0
0
C
C
-* 0 0 - '
o C)
C)
C
N-
-
O0 C
C
)
O0
N-- 0
0
C)
F );;C
co 0C) c50C)
c
6)C)
C)c
)
O
0
C)
C)
C
C
o0
0
C
C
-:1- N-,:
0
0
0
0
0
C
)
0
j-
C
6 6)o
G)
CN
IC)2
-6 C
0
N CD C) t
0 0
-
0
0 0 CD CD
ci)
ra
co c; c c
a0
-
a N
0 C0
C c c
;o
9
w
M0
ON
i
. CO
C) 0 0 N C0
;c o o
oo
N 0 CD LO
0
)C
N 0
U)
"t
)
0 N0-
o o o o o o o
0
~-
CA)
C
0N 0
0
C)
C)
C)
06
-
N
C)
C)
C)
a
C
0
C
C
C
v- U' CO m~ w- N-z N:0
m
C
C)
C C C c)c
0
C)
0
0
C)
aCa a
a)
Nq C
0
N
Oa a a
C)
C
C
C)
C
C
C
oc)666
, 0) R
C
)
C)
-
0
C
0
C
6 6
a
C)
0
C
0
C
R
N
v-
C
C
0 ~'
a aC
a)
0ci
C0ooU w LONtm 't
U)C(D
LO
0N
m
t C)
o
m
U~0C
m
' (D
NC)
0o
0
m
0
0tm N N
0 C)
9~~~
o
od
o 00
o o- 0Do00
o U)C o -- C) ) o- C) N d
C0 9 C) C 9 C C C) C) C) C) c) C) c) C) c) C) C C C 9 9 C 9 9 C)
oU)t
N
0
(I,
)
C)
9
C
CD
00 -j0
'O
o
ci
U0
w~
En
V6
0
~uC1
IE2
0 0) 0
C) 0
C)
C)
C)
C)
C
C
C)
C)
C)
C)
0 0 N- N- C)
c3 C)
o C)
:. oc3
C)
C) C)
C)
C)
C)
oa
C)
0n
C)
C)
N::- N-
N-,
N-:
C)
C)
C)
0 0 0 0
C)
NC:
C
C
NZ- N- 0a
C
C
C
C)
R-
o) c3)o
,)
~;N
C) ) C
CD
tt)
0
9~ c)
0 9
o
a
0
C)
U)
N-
CD 0
C)
C
C)
C) C) C)
9
C)
6 9) 9
99
C)0
.C)C
C)
9
0
qC)q
0
CD CD N-_ NC
C)
C)
C)
~
C)
C)
C)
0
C)
o
C)
9-
co
Io
w-CDo0C,C) N)
0
C)
C)
C)
C)
v6C)N
C CC)
C)
C)
C)
2a)
666
0)
00o CD
0D
D C)
-D
-C
c
0
O
O
o N
C
C)
C)
0 C6666a~
CD
U)
N
0
0
0)
0
N-
U)
CD
C) C)
a a aO a Cco
6 a a a 0o
a0a6
C)
C) C
C
o- ot C) C
C)
0
-
r-0
C
C)
-
a3
.11. C)
C) C)
0
LO
C)
6666o
C)
0U
N
N-- CC)
Cq
C)
C)C)C)6O6
Q
-C
CD
a)
0ai)
-
C)
!PCD
N
tt,
0
C0
N
U
C
C
ooooM
0 N C M t C N N N- C N
C
)
C
9
C
9
C
9
C)
C)
C
9 9 9
-
w'- C
Oq v
~
9
C)
C)
N
N-
9
C)
9
-
U
C)
C)
C
9
C)
C)
'o
.C
cC:_
co,
03
CoU;:
U)
U)
00 -j
) U
C
C
C
6 6 66
NU)
C)
CD
(O U)
CD
O CD
(O
C) C C
C
C)
C)
a
C)
~' N
C
C)
'It
CO
~0
666
C
C)
C U) No 0N 0- 0O co
C CD CO '
C)
C
C
a
C
)
C
C
o
a
C)oC
C
C
6 66
N U)
CD - CD U)
C
C) C) C
6
CD LO I- C
) C C)
0 00
00
LU
a)--
'-
=;
0c
.-C~
tCU
iE 3
0
M
N-
"I
c
Di
to
U)
LC
0C
N-
LO
U)
U) U)
~0 CD N-
-
D N
C
N
U) N
0
N N
0 U)0
909U) 9N- N C) 90)
9N 0 0)
9 C 9 9 9 t
6 6
C ) 6 6 6 6 6 6 6 6 6 6606
9 96 90 06 C C C C C
0
C
i
U)
i
CD
CD
Di
NC
i
0
c
Di
0
CD
i
C)O~
N- NDi
i
O
N
N.Di
0
Ni
t
NDi
U)
Ni
CD
NDi
NNi
0
NDi
0
C
Nl- 0
i
Di
0
i
N
0
Di
0
0
i
.0
Di
U)
0
i
CO
0
Di
N0
Di
0
0
D
Di
0
Di
C
6
g
ci
-C=
CU
-c
.0
0
CU)
0
C-
C)
Di
Di
A
o
Table A2 - Differences-in-Differences Estimates of the Effect
of Planting Tea and Orchards on Sex Ratios:
Coefficients of the Interactions between Dummies Indicating
Whether a Cohort was Born Post Reform and Dummies Indicating
Whether Any Tea was Planted in the County of Birth
Dependent Variable: Log Sex Ratio
Tea * Post
(1)
(2)
-0.039
-0.029
(0.010)
(0.011)
Orchards * post
0.027
(0.013)
Observations
R-Square
30355
30355
0.13
0.13
All regressions includes controls for category 2 cash crop*post, post and county fixed effects.
Standard errors clustered at county level.
104
Table A3 - Descriptive Statistics of 0.1% Sample of the 2000 Population Census
Counties that Plant no Tea
Counties that Some Tea
Obs
Mean
Std. Err.
Obs
Mean
Std. Err.
Fraction of Male
81774
53.31%
0.0017
25290
53.56%
0.0031
Fraction of Han
81774
93.47%
0.0008
25290
86.05%
0.0019
Years of Education
81774
7.14
0.0110
25290
6.89
0.0198
Male-Female Education
58590
0.55
0.0071
18034
0.55
0.0141
Fraction with Tap Water
81441
31.39%
0.0012
25182
37.60%
0.0021
Cohorts born 1962-1986
Birth Year x County Cells
105
U)
R
_
N O
2.
m0
(D
O
0
C
N C C
O~ OIOD- C O C
O - O - O
O ~" ;
2 ~'
C
N"~
0
O O O 6O O O O
0 O O O O O O
O_ O_~ O 0O
_)
v
ON )
,)N DU)0
O dO)O)
U)
N
C 0O 0C
OO O0
OC)
N
0
t O
D
C
(D
C) 1)
NMNN
NNN
- NN-)MNMN
C
-t
-
D
t
t
0
O CD
O tO
O N
O O O Ct
O OD C)
O O) O O O0OD O) N
O CD
O O) L)
O O U)
O
N
CDNCoD C) 0D
'a)
0
05
66
C
D-W)0ON
O
O
O
Cl)
N
O
U
0
O
C
0
OO
(D
-
N
U)
I
O
d
v)
C)6
N
)
C
N d N 0 C ) N CC'
N Nt CD
)DM 0oot000
UtD
)t6o6o666
u ur
) M
rC CD
:C
00t U)
0 Cl
000
00
ovu)
OC
'-E0o
.'m
. C
OC==
'U)
o -i
oo
c
<
O
O:o
0
Cu
Z
a)
c
0
Cu
I
'2
M
CD
Cl)O
CD
N1
Cl) C)
Cl Cl)
a)
~ ~
C
w()
N C N-
N
C))
"
C)
-
N
~ ~ ~ ~
'
-Cl
't
'
C)
N C)N
~
C
O
-'-C) U)
d~
~ ~ ~ ~
C
)
O
O
O
N
CD Cl
U) C)
Cl) CD
~
t- -
1
~ ~ ~ O
0
CO
C)
C')
t-
D t CCD
U) 00
U)
U
C)
~ ~ ~
O
~ ~
U'
U)
cu
U)
0CNl
)NCDN
NNNC(DCsd--CDtF
N O 0N l)
N
O O
O
0
O U
m) IOdOODC CD
O Cl) NO 0O)CD
m CD
N N
O 0UDN
CO N
0O
O
O
O OO OO OO vO
t
o
_0
a)
0
U)
.T
m~
0U)C
0o
U)
OD d
0
N
c CD
N U) t s
N
N
a u)
U-)
CD C)!
O
ov
6666666666O
F
N
-C
CD C) C--
N U) :-
0 C') LO)N 0) LO
O O O O
.....
o. .o7
OS
O
aoao
_ ON
_6_6
CuQ)
a)
s-C
C a0
a
a
0
x
696o
t
o
oM
t
m o (o oooo6O6ooo9o9O6
w oD
0
rccm
Oru
LU)
Cu>
0m)
CS
r-
C)
Cu
0-ocn
O
a)
o
m
m
o
666666666666666666
N
~U)
0 C-
a
N CD0OO0
-00
~
v
cn
=Z
aO
n
(O
CD C
N 0) - 0
U) N
N N N N N N
.)
N
N C) 0
C
N N N
Cl)
U)
L
CD U)0 C)
U
N- 0'
N
C
OOOOO°°°°°°°°°°°°°°_
n
C
N N N N
0
)
C)
N mc)N N
N
0
CDCl) 0N
- r
N U N
,) )
o
D CD
U)
0
CD
N
O
9969699969699969999
~0
a)
rC
nC-f
0
(n
C-)
.0
C
cua)e0-
0
O,
U
-D
U)
00
CD
a)
0
0
N
OU
-
CD
C-) N
0 0
D
U
00
-
Cl
N'-
C
t°
CO
Oo
D
00
00,
C)
U) CDN
m
0
C0
o6
U)
C
0
N'-
Ut C
C
t
D
U)
N
U)
0
U
CD
CD CD
N
U)c
CD-C)
N
,0
U)
_
0)
Cm
ND C DU)
N CD
IN U)
CD CD CD
N- 0I
CD U) N
U) Cl CD0
U^)~
cq
Cl) (D U)
C
CD)
0 CD '0w C'O cDt
t0D
N
oo
d
U 0) N
0 (O
~ Ci°
O) 0NsNe
0 (O
~ N(D 0d0 0~ 00
) 0 tUO
o 'o I U)
D C
C)
C)
CD CD C
CD N
C)
C)
C)
C)
N
CJ) C)
C)
N
N
N
N
N
C)
C)
C)
C)
C)
N
N
CD CD CD
C)
C)
C)
C)
0
* CY
-0)I
-
In
0
(
0
N
02
)
)
m
-
-
.-
w
m
t
O
-
-
0
-o
0
O
0
0
0
N
N
N
N
0m) 0
CO
N
uO
0
o -
0
a)
a
0
r
(U
0
o(Da)
C>
0
N
.
ONO mOCOO
O- C
O O cC
O 0O NO ) .-O
O CDO
O- 0)
O
O
'')
U) 0)
C 0)
N
ND OO o OD-O O U)
C) NO
CO N-
C)X
N
C-
N
N
N
N
N
C)
CD O)
CiC) - N
N N
N
C
)
O
0
)
C)
C,
0
I..
a0cn
a)
In
a)
E2
0
0m)
C)
C
C-2
)
u)
NO
t
)
N
C'-
N
N
0
1
O) C)
U)
N
O
CMO
1) t)
(0
N CO
U) C d
O
NU)
C) .LO)
oo666666666666666666
0
C')
0) UJ
C0) N
C')
N
C)
N
N-
C
00)
a0) C)
N
D
N
N CO N
C)
CO
U&) C)
NC
N- CD N- N
N
)
c o
LOCD
-1
00C6
-
0O
CD
a
U
0
'
T
.)
Ut)
CO
.
O .4
L
CO U)
C,)
0
~
'C ._
a)0
1-
I_---
cm
mog
C
6666 0 0 0
0
0
O
N -O)
D 0
L
O
O
CO
O
N
CO U)
~
N
0) CO
)U)
0
)
)
O
O
O
0
~
.4
U
O
0It
N
N
00
*
U
)
CO
0
N
C)
0)
0)
0)
O
0)
M
CO
.-
0)
0
.4 N
C)0 w
-
~
O
~
O
C)
U)
.m
N
to
00
)
N
CdO
t
0t 0 )
Of
O
M
C) - cO
L
0
CO
0)
u)
0)
Ct
0
CO NN
U)
0)
N
0
0
O
~
NC
CO
C
) U
C
N
0 s0 O
00
~
~
~
N-
U')
N
U
0
.4
0
~
~
t
Uc)
-
.4
)
N
D
)
Cl)
'O
m
-c
m
In
0o
a)
a
U)
00
0
a
(U
W,
00
0
0
O
0
Nl
0
co
0)
O
O
O
CO C
N
)
NO
t0
0 O)
N
)
0)
CO 0)
Nc
N
0).N N
o
I) . t
0
co
C
NO
a)
L
Ca
0
'- a,
o_r *X
0
'O
0
o=
°a0<
rn
a,
C
0
(U
C)
0
t
N
.
)
20
0)
U)
0
4
.-
0)
N-
0)
0)
m
U)
N.) C)
N
()
N
cO
CO(0(0
.
C
N
0)
0
CO
-
0
0
0
C')
CO
C')
0
LO
)
CD C
C
0
CD
CO 0)
CO
0
.-
N-
.4
C'
0
C')
.4
CO
)
.4
NO
co
:3a)
*0>CD
a,t
a')
O
N
N-
O
C'
00
wi
.C
0
0
in
0
N-
U)
C'
O
CO N
CO 00
C'
0
N.4
o
.4
N
CO LI)
c 0) -C
'O X
C')
co)
0
CO U)
C') C )
666666
Oo Ooooo6666ooo oo
(U
ua
0
CO C')
U)
Ue
Vu
LU
L
cO
NCO .
a,
¢
-
a)
F-
N
N.4
N
a) ,N U)CO
0N
N-U)
N
0 N N (0
0
CO 0)
"'
0
CO) CO) CO) C'
0
0 0
0
.
c
N
CO~
)
') Id'
N CO) C
6
0 N NN u
I'
6s
)°
-
C'
)
N
N 0
0XCa
Ca
.
0
'a)
.t
N
1
t
-
-
(0
0)
I:
U
CO N
-
-
N
'-N
0
:._
-D
U)
C')
o I4- C')
C"
0
0
00a
U)0- En
oc2
oC'CO
N
N
Co
0)
96690006666006066066
C~ .
>
s)
)C
o.
o o o C: N~ "; -c
(D CO , *t
a) -2a)
En
a)
c U)
C
U) I-
a).
c')
C)
S
-
UI)
CO
CO0 )
(0
CO C ( 0) (00)
CO 0) 0) 0)
C0
N0)
-1
S
tC'
U) 0
N-
C
m
)
0
c
N
-0) -0 -0) 0)- 0)N- NN- NN- O CO
0) 0)
0) 0) 0) 0)
0)
o
U)
0
c
a:
-
<C)
Ca
Figure 1A: The Effect of Relaxation on Family Size
Coefficients of the Interactions between
Born in a Relaxed Area * Birth Regions
0.5
0.45
0.4
0.35
0.3
- Girls
-- Boys
0.25
0.2
0.15
0.1
0.05
0
1973 1974 1975 1976 1977 1978 1979 1980 1981
Birth Year
Figure 1 B: The Effect of Relaxation on Family Size
Coefficients of the Interactions between
Dummy for Girl * Born in a Relaxed Region * Birth Year
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
-0.05
1973
1974
1975
1976
1977 1978
Birth Year
1979
1980
1981
108
Figure 2A: The Effect of Relaxation on Sex Ratios
of First Born Children and 95% Confidence Intervals
Coefficients of the Interactions between
Born in Relaxed Region * Birth Year
0.3
0.25
0.2
0.15
0.1
0.05
0
-0.()5
-0.1
1970
1974
1978
1982
1986
1990
Birth Year
Figure 2B: The Effect of Relaxation on Sex Ratios
of Second Born Children and 95% Confidence Intervals
Coefficients of the Interactions between
Bomrn
in Relaxed Region * Birth Year
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0. 5
1970
1974
1978
1982
1986
1990
Birth Year
109
Figure 2C: The Effect of Relaxation on Sex Ratios
of Later Born Children and 95% Confidence Intervals
Coefficients of the Interactions between
Born in Relaxed Region * Birth Year
t
A
U.,
0.3
0.2
0.1
0.0
-0.1
-0.2
-0.3
1970
1974
1978
1982
1986
1990
Birth Year
110
Figure 3A: The Effect of Relaxation on School Enrollment
Coefficients of Interactions between
Born in Relaxed Region * Birth Year
A n
.,.,J
0-0.05
-*-Girls
-0.1 -
-Boys
-0.15
-0.2
-0.25
1973 1974 1975 1976 1977 1978 1979 1980
Birth Year
Figure 3B: The Effect of Relaxation on School Enrollment
Coefficients of Interactions between
Dummy for Girl * Born in Relaxed Region * Birth Year
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
1973
1974
1975
1976
1977
1978
1979
1980
Birth Year
111
Table 1: Descriptive Statistics
CHNS 1989 and 0.1% Sample of China Population Census
Obs
Mean
Han
13271
Siblings
Std.Err.
Obs
Mean
0.944
(0.002)
15500
0.949
(0.002)
13271
1.153
(0.009)
15500
0.922
(0.008)
Sisters
13271
0.504
(0.006)
15500
0.511
(0.006)
Brothers
13271
0.649
(0.006)
15500
0.411
(0.005)
Enrollment
13271
0.473
(0.004)
15500
0.456
(0.004)
Mother's Education
12862
6.063
(0.037)
14890
5.668
(0.035)
Father's Education
12134
8.058
(0.034)
14239
7.628
(0.033)
Mother is Housewife
13271
0.119
(0.003)
15500
0.139
(0.003)
Relaxed Area
13271
0.254
(0.003)
15500
0.242
(0.003)
A. By Sex
Female
B. By Family Size
Std. Err.
Male
Siblings
Only Child
Sex
19038
0.502
(0.004)
9733
0.611
(0.005)
Han
19038
0.941
(0.002)
9733
0.958
(0.002)
Enrollment
19038
0.399
(0.004)
9733
0.591
(0.005)
Mother's Education
18488
5.300
(0.029)
9264
6.952
(0.048)
Father's Education
17623
7.476
(0.027)
8750
8.530
(0.044)
Mother is Housewife
19038
0.134
(0.002)
9733
0.121
(0.003)
Relaxed Area
19038
0.279
(0.003)
9733
0.186
(0.003)
No Relaxation
C. By Relaxation
Some Relaxation
Sex
10828
0.544
(0.005)
17943
0.535
(0.004)
Han
10828
0.968
(0.002)
17943
0.934
(0.002)
Siblings
10828
1.048
((0.010)
17943
1.016
(0.007)
Sisters
10828
0.512
((0.007)
17943
0.505
(0.005)
Brothers
10828
0.536
(0.007)
17943
0.511
(0.005)
Enrollment
10828
0.437
(0.005)
17943
0.480
(0.004)
Mother's Education
10454
5.034
(0.040)
17298
6.345
(0.033)
Father's Education
9914
7.439
(0.036)
16459
8.058
(0.031)
Mother is Housewife
10828
0.111
((0.003)
17943
0.141
(0.003)
Relaxed Area
10828
180.299
(1.256)
17943
147.335
(1.195)
Distance to Urban
9460
2.041
(0.017)
17943
11.849
(0.087)
Agriculture
10818
0.720
(0.004)
17903
0.569
(0.004)
Distance to Primary School
Distance to Middle School
10828
10827
0.230
1.008
(0.006)
1(0.009)
16672
16672
0.399
1.584
(0.004)
(0.011)
Distance to High School
10827
4.920
(0.084)
16672
4.506
(0.067)
Sample of cohorts born 1962-1981
112
Co
N
C) 0) (D
to
u.
=E EF
COil
~~ 0)C ~~'
Co ~~ c'.0
J O
CD
(D
0
In
°,
a
O
cN R
It N
O ,
O°'
O
R
M
0)
Nt
O
Oa
z 0)
N
OO,OOO
OCD
w
O
-t
e-
(0
C
0
-
It)
-)o
U)
ao
#
9 C. o ) o
°9O6
C)
o
N CD. O1 O6666OO6OO6OO6O
O
o
O
o
c'.
oc C)j
6O
- COc )
0a)
toz
0O
r
Cs
0DO
6O
-0
C
.=
X
._
O:._=
DC
.0
C-O
O
=
c-
cu
C
'
,a
CD
OX
m
't
0)
O
LO
0)
o)
O
o
N~
0)
0)
0)
0)
o
0)
OU)
0)
0)
1~~~~~~~
~
x
0
.-
CD
a)
U)
o
0
N a) E>
CU
_0
a)
ca
-O
U
a)
&~
t-
U)
a)
.0
aU X
a)
*a)
a C
C
a)
o
0
a
O
a)
)
aa)
C
a)
a
-o
U)
C
No
co
:.
O °
C)O
C
(D
C
OC
co
°
o
O
°aO
o
o
cN
a
96
O
°
O
o g
E
.
-
op 00 a
9696 9 9 9 69 6996
°
w
U)
- .0
Co CO
C)
oo
C)
CO
v-~~~~~
O
~~
~
(NJ ~~N
N
N aoC C)
) R~ It) It
0)
(N
o o N- Coo,oCoo C) oc Co)
o Co)oCo o o co
(NJ
Lo)
-
m~
()
m~
m~
a)
C)
LO
C
C)
LCO a)
CO
C)
W U) X
a) c
_
o o-0NOOo
CO
cN
't. O,OO o0IO
O O
O
O O O~ (O O, O
1O O
~ O
O CO
LO
m~
It)
CX C
Cu
0
(NJ U) U)
0)0
C,
U)
r
it
C
.5
0ca)
_
CC
O '-O ~-O - --- O ~-O ,N
--~-
O,
NC) C
C
666666-6666O
L
a
a C
aO- CD O ,r
LO O
' C:
0 0~1 ,r
CO
v~- LO
~~- LO
O CO~' 'O,
~- CO
~ O
~- LO
O v-
T°
EE
-s-s
C)
C)
._C
Cu.n
C
.2C---
~CO
:
C C
i 0i
U) U)
oo~
0
a)
a)
a)
t)
a
, 02
0cn
C-)
o,
a)
CD
U
-5
CO
0)
LO
)
-- 0) -- 0)
CO
0)
-
- - 0)~
O
0)
0)
0)
0)
0)
-
-
-
U)
0)
-
0
a)
-C
a)
It
U) 02
o0
0
7Cu3
_C:
C/)
)
*0
'-
wo
t
C2
OO
U)
W)C)
a1)
0
0
10
C/,
m
Nm
6a
66
o
O
C
C)
_
6
O
C)
C)
a
_0
N-
C)
C)
C)
o o o o
cN *
co
o
sao
_4
st
'4
C
N
U)
C
o
N;-
_
CO q
C)
(
q
CD
C)
C)
O
0
N
Cy)
'T
C
o
C
O
°,
°,)
S
)
0
°
N(
C
c
Cq
O
'4
a7
)
°,
O
N
(0
_
a0
tC)
C)
I
C
C)
N
C)
'
U)
m
C
q
-
0
CD
C
o
oN
O
°
O
r_
I_
U)
N
C)
t
C)
It
w
C
)
6
6
C
6
U)
'0t
NCD
U)
U)
C
NU)
C)
6 6 6 6
a)~
CD
0
(0
CC)
O
~C
LO
O)
U)'
C)i
U)
00
LO
>- Ot
`
O
U)
C
C)
6 C6 66
o
)
C
CO q
C)
0
(
C)
aa a a a
C
66_
C)
66660n
C)C
(0
00
CO
SS q
) C) C) N
C) NC) U)
O °°
- 't
)
)
)
) P P
)
O O
O O O ° ,
C) U)
N0
°U) a t ) (0)
NO (0
' . w °N t NP
P CO
O
O °, O O O ° O O O O
N
w
(~0
'4.
N-
C)
(0
N
w I''4
0
0
N
(D
0
0
0
C)
C)
6
.
-
C
I.
. I
6
0
O~ O~ ;
U
C
c0
C)
a
- (0
0 (D 0
U)
(
C)
q
q
a
a
LO
(O
(O
U) )
C
(0
C
0
C)
U
C)
N
N- '4
U)
(0
U)
C)
'4.
(O
C)
C) C)C
O) O
-
C)
C)
U)
U)
(0
N-
(0
'4
(0
C)
(0
U)
C
(0
'4
)
(0
C)
C
Z
CO (D
U)
(0
C)
0
0
O) C)
,-C)
'4
C
C)
C)
a a q_ q
6 6 6 6
a
0.
0I
0.
t
CO
C)
U)
q
C)
0.
0.
60
U)
N
(0
C)
)
O
C)
.
CO
CD
60
N
'
4 C) U) '
t
'-
't
0.
0.
0
C
(0
'T
C
(0
(0)
C)
60 60
0I
.
toi
666 ~6a
q_ O_
_t
_
CD
C)
c C)
z
0
.
N
R
0 '4 (
- N C
N
U)
(
N
6
0
cO
.
.
.
a
(
C
0
't
C)
(0
0D
(0
C
C
(0
C)
CN
U) C)
0.
(0
60 60
.
.~~~~~
0
0.
(0
0
C
cn
_ C
I0
'C
a
0
0
O
(I,
0=
CD
co
n
oo_
C
Cu
0
oo
~0
E
E
0
o4
o)
o)_CC_C)
Nr
(D U)
U CD
N N
N 'tU) (D ) C) N
N C N
N 6 N '- N
a ao (s
,It U)
(0
(0
(O
C) C) C)
(0
N-~:~'-
-
X
CU3
U)o
Cu
U)
U)
LO
)
6 6
CO C
C)
N-
( C)
C)
C)
C)
66666666O
C
U)
C
'
C)a6a666
U) (00 00 N-0
C)
C)
C)
C
(
C)
C)
)
NC
NNC)
>-
N
rlP
'4.
a)
EFE
a1)
0
.a)
C
4t
(
(0
0
(D
(D
(0
)
OC)
)C)O
C)C)66
(D
C)
o
N
)
CO Is
q_ ac.
Q)
0
U)
N
L)
C) (0D 'O - O-
N- N0
CO
CM
Cq
C)
0
O
a)
N (
(N03
0
U-)
oCC) o C o6 6 6 6 a6 o6 6 C C)C)
-
0:
0
0)
U) N(0
(0
C
)c
N (0
C) C
66O
'T
(0
)C)
N
NIw
C)
C)
CC)
m
em ( 0
U)
C
(0
w
C
'4
t
'-
U)
6 C)6 0C66)C
N- U )
N
't
oC)i-C)C
C)
U)
C)
(
(0
(0
C)
a)'
a
_
a
'4 ILOU)
C) C) U)
66O
(
(O
(
t
C)
(0
N-
(
It .
C)
C)
C
N
'
CC
C)C)C))
-
't
)
C)
(0
(0
U
(0
00 0(0
C)
C)
x -0
LE
C)
C)
cu
CD a)
C)
z
(
(
(0
C
C)C)
'I
I4t
N(0
:)
,I.
N-
C
C
X
la0
0
C)
N-
_0
C
a)
.C
C)
C -8
=
a)
o
a)
a)
in
0
0
In
C
0
Cu
a)
'C
(0
N
C)
~~~~~~~~~~~I
'4
U)
(D
N-
(0
)
C
C)
C)
C)
c)
C)
C) O
C)
(0
C)
N
(0
C
(
(0
C)
t
(0
C)
U) (D
(0 oo(0
C)
C)
N(0
c)
(
(0
C)
D
C)
C
c
a)
)
N
CD
)
C)
C5)
C
C
2
a)
.0
0
a,
C
.
U)._n
o
0
In
a)
0
a)
Cu
C
a) m
- Q1)
Table 4: The Effect of Relaxation on Sex Ratios for First Borns
The Effect of Relaxation on Sex Ratios for First Borns
Dependent Variable: Dummy for Male
_.
(1)
Born in relaxed region * Born 1976-1981
Born in relaxed region * Born 1982-1989
0.106
0.098
(0.024)
(0.025)
0.037
0.035
(0.021)
(0.021)
N
Y
44234
44234
0.00
0.00
Han, Han * Birth Cohort
Observations
R-squared
(2)
Regressions include county and birth cohort fixed effects.
Standard errors clustered at the county level.
115
c o
0) 0)
0)
o0
F(D C
i
q
C)o O
04
o
I.co
oo
LO
COF
C0 0
LO fl
014
oc-o
Co
-J
Co)
m
w-- C
00
.
No
C-
c o
C:) Ea
(-4
(i
c
o 0o
0
W
0
-
_
66 Ci
c)
U)
0
U)
a)
Un
a)
u)
CD
in
0
U)
o 0C)
in
c-
O 0
a
a
c
o
6o
~o d
o i
in
U)
E
O
o
U)
in
0
CO
I-LO
04
z z z
0
i
a
w
La
N0
0
0
co
iO
in
0E
oo
0
U)
iDn
O
o
LO
oD c
co
o 0l(c
c
0
0
0
1I-
C)
a)
9~o
0
oo
tC
Co
-i
C
--
96
0006
_
)
CO
C
U)
z
U)
c.'
CD
U)
s
0
0
U)Z
)
U)
in
0 0 z0
z0 zz
00
0
on
in
o 6R
O
>
n
-C
C
C.-
U')
.-
ts
0N
.a
D
._)
cn
a)
t-
r (-4
CU
U)
o
~C
o
in60.
C)
9 C
I-
0
>
CD
0
N
0
z
z
U)o U)
oRo
C'z0)
Co
C-
CU
zzz
0
0
9
-j
0
CO
04 °:
t-- (
R
r-
0
0
(q
C)
0
a
w
co
o
a
a)
E
CO
N
U)
U)
LO
Iin~
oo
00C)
0
vl; c)0
C-- i
c
.
.a
w
>a)
w-0
0
N
',_
~o
00
c
N
co
> >
0 0 00
z z
0 U?
t-
E
._
0
U
r
a
o
0' T)
U)
V o©
C
x
cn
CU
c _2
C)
-
x <~
io
C
n
i o)
CD
m
U)
p
._
U)
0
-_
a)
E
U)
w
.m
sn
U)
C)
tn
a)n
u
m
U0
~~~.o , 0=mw
D
U) C
U) zo 7
Un
C)
.0
.,
w
tn
>s
00
a)
- m
un
._
o2
.0-
C
00
c
0C
U1)
C
a)
C
DI=
..._
C:
C
I
C.i0
'
a)
cn
C
.0
F
-0 0
,C)
U;)
_<
C
: v
Figure2: Thetop 10%incomesharein China,1986-2001
27%
25%
23%
21%
19%
17%
15%
13%
CD
W
e0
CO
c>
D
0)
~
01
0)
w
c
~
x
o
o~
Source:Authors'computations using urbanhouseholdsurveys tabulations(TableA5, col. (1),ind.income)
Figure 1: Real per capita GDPin Chinaand India, 1986-2001 (1986= 100)
260
250240-
230
220 -
4-China --- India
--
210
200
-
190
180
170
160 -_
150
140 -130 L--________.
120
-
__
__
110-_
-
_
100 -
90-
--
_-N
Source:Authors'computationsusing nationalaccounts(see TableAO,col. (5) and(16))
117
Figure3: Thetop 1%incomesharein Chinaand India,1986-2001
10%
9%
8%
7%
OZ
6%
-|-- Top 1% share(China)
5%
-- Top %share (India)
4%
3%
2%
1%
distribution);
(TableAS,ccl. (4),md.
householdsurveystabulations
computationsusing urban
authoss
China:
Source:
0%
O
N
o
-
>
of
O
8
O
0
O
N
'lN~~~~~~~~~~~~~~~8
Source:China:authoss'computations
usingurbanhouseholdsurveystabulations(TableAS,col.(4),ind.distribution);
India: auhors'computationsusing ncometax returnsdata(seeBanerjeeandPlketty(2003,TableA3.col.(1)))
Figure 4: The top 1% income share in China and India, 1986-2001 (1986 = 100)
200
190
180
170 160
+Top
-4-
1% share (China)
-- D-Top 1% share (India)
150
140
J /M-
130
/
'
/
120
110
100
90
Source:China:authors'computationsusingurbanhouseholdsurveystabulations(Table AS,col. (4),ind. distribution);
India: authors'computationsusing ncometax returnsdata(seeBanerjeeand Piketty(2003,TableA3,col.(1)))
118
Figure5: Incometax exemptionthreshold,averageincomeand P99thresholdin China,1986-2001
(currentyuans)
40,000
36,000
32,000
28,000
24,000
20,000
16,000
12,000
8,0001
4,0001
0
-
I
I
I>
Ix
Ix
o
)
o
In)
Ix
o
Source:Exemptionthreshold:Chinesetax law (seeTable1); averageincomeand P99threshold:authors'computationsusing urbanhouseholdsurveystabulations
(TableA1, col.(10),andTableA4,col. (15))
Figure 6: Incometax exemptionthreshold,averageincomeand P99thresholdin India, 1986-2001
(current Rs)
100,000
90,000
80,000
70,000
60,000
50,000
40,000
30,000
20,000
10,000
0
Source: Exemptionthreshold: Indiantax law (see Table2); averageincomeand P99threshold: authors'computations using nationalaccountsandincometax returns
data (see Banerjeeand Piketty(2003,TableAO,col. (7),andTableA, col. (9))
119
Figure7: Thefractionof individualssubjectto the incometax in ChinaandIndia,1986-2001
36%
34%
32%
30%
28%
26%
24%
22%
20%
18%
16%
14%
12%
10%
8%
6%
4%
2%
0%
~ ~~ ~ ~ ~ ~~~~~
X~
eo
0.
0.
1X1
0.
0.
0
I
0
0
0
Source: China: authors' computations using urban household surveys tabulations (Table A6, col. (1));
India: authors' computations using tax returns data (see Banerjee and Piketty (2003, Table A, col.(4)))
Figure8: Projectedfractionof individualssubjectto the incometax in China,1986-2010
(assumptions:
tax lawunchangedafter2004;post-2001incometrendssimilarto 1996-2001)
70%
65%
60%
55%
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%
Source: China: authors' computations using urban household surveys tabulations (Table A6, col. (1));
India: authors' computations using tax returns data (see Banerjee and Piketty (2003, Table A0, cl.(4)))
120
Figure 9: Using 1996-2001
Tax Receiptsto Re-Evaluatethe Riseof Top IncomeSharesin China
--- r
z/u
260
250
240
230
220
210
200
190
180
170
160
150
140
130
120
110
100
90
Source:Authors' computationsusing urbanhousehold surveystabulationsand actualincometax receipts
Figure 10: Incometax revenuesas a fraction of GDPin China and India, 1986-2001
4 '02
1. /0
_.
1.1%
1.0%
0.9%
0.8%
0.7%
0.6%
0.5%
0.4%
0.3%
0.2%
0.1%
0.0%
CD.
Source: China: 1996-2001:
1-
°o
actual tax receipts
0
0
0
from China Tax Yearbook
0
(see Table 3); 1986-1995: adjusted
0
simulated
tax receipts
0
-
N
(see Table A6, col.(15));
India: actual
tax receipts fromAll-India IncomeTax Statistics (see Banerjeeand Piketty(2003))
121
Figure11:Projectedincometax revenues(asa fractionof GDP)in China,1986-2010
(assumptions:tax lawunchangedafter2004;post-2001incometrendssimilarto 1996-2001)
A{An!A
4.0%
3.6%
3.2%
2.8%
2.4%
2.0%
1.6%
1.2%
0.8%
0.4%
0.0%
-.
I
-2
9
. 9. .
t, o.
.
NA
9
.
.
Or rD
.
. s,. oM9 s2°; 2° .
°
.
°
. °I..
I
.
I
gI o
N
Source: 1996-2001: actual tax receipts from China Tax Yearbook (see Tab 3);
1986-1995 and 2002-2010: adjusted simulteded tax receipts (see Table AS, col(15))
122
E'~
0
x
C,
0
I
CU
O
0X)
ECD)
-)
'-
C 0..C
y- O >,
a)
E
-2
0
.C
00 0
0^
U)
0 0
0 00!0C
a)
C
Un
a)
C0
.CD
0)
C
C
00~
O
O
~_0
. EZ
o
,C a En
) r)C
0V
O0
.) C
le
CCU
Cu:3
7
N
0-Z
C
co
0) 0
0
0)
C
000O 0O
O
*_X
CuC
m
C°
"O
0
N
C=
00)
dD
C)
OC
CO
0) 0
(00a
0
aC CF)
N
0)0)
-'
a)X D -0CC- 0
Ea)0
* c aCu
.
-1 -0 0
CU
00
Cu_
o
0 0 0 0 C 00 00 o
CC
cu:3
0D00
CO
0
OO
C
O
N N
C
N N- N
C)O N
CN
00D
) CD
0 0 o00
0 0N 0CO0(0 N N 00C1
0 CN
NO
CO
O 0)
O CD
gq
N
N8
'a)
00
0
-CuE
0)
aCu
0
en
0f
c
CC
°
0LO U'- U--
Cu0) 0
>< C4
C
a>D
no
->0e
C.
.0) V
ca
C Cu
Co
M3
:3
Un a)
15
E
.~C 0
0 0.)
TC:C
co
0))0
(0
(0
0 ( 00C
O 00) 00
) (0
O 0
N 0
(0
0
D
OL
CO
O5
C)
C)
co
CO(
C>
( 0)O
(O0
(D
COO
ND (D
(0)
CD
-
0
cm40
a)
Cu
0.3
CC.-Dc~CD 0)a)C
(0
2
0 0 0
E
0C
'CuC5)V
Zi UC0
aE
L
CO
O) 0-0T
O
N UO0C
0o U)
C)
(
N
0)
)
:~- B<
X
a
Cu0)
-C_
L.
0
-a
V C
CO
0')
0
C
C
)
00
CuC
cO
C)
0)
0a)
co*
CM)
CD CD
w0'U)0))
.0a)
(U
C.)
)
C)
EL
0)
XC
c- _
z
EX
V5 ,C
C F
.C
o0- 0.
x ,cs n 4-In a)) >
a)
E ct
0 0
c, 0
c)
~0
U) Ca
) CDc
0C O
C
V
WE
E
0
I%
n
a)
X
C)
CCa
Cu
0)
CO
N
mu
I-
U)
E!
I'
cC
0) 'c 0.
.Zo
M0E
o
0
CD
t
C0 )
t
U
a)
0,
0)
0 O0)O 0OO 0OO
-C
°
C)
Yco
0)
U
xU,
0
U)
N
U)
Q
C UC 0om
E
U)0C
0
C.)
c~
c:n
Eo
04E -E-E0
oc
en ~.
U0)0
0
CC
En~
a) C
L
00
a
.' (!
0
0
O )L
1
0O
cn a)0)0
a)
O C)C
-2
CD 0)
Cu QO =
U) 0) oo U)
-E
C D
CO9
X0-f
-C
a .0
C) C
C
o0 (CaC
CN
o
Q0
N CD
0
(0
0)
u
UC
02c)
E .a
U)
C0)
CD
C
C,
EtC6
Vu 0 C
Table2: Progressive
IncomeTaxSchedules
in India,1986-2004
1986-1988
1989-1990
1991
1992-1993
Brackets
of annual Marginaltax Bracketsof annual Marginaltax Brackets
of annual Marginaltax Brackets
of annual Marginaltax
income(Rs)
rate
income(Rs)
rate
income(Rs)
rate
income(Rs)
rate
0-15000
0%
0-18000
0%
0-22000
0%
18000-25000
25%
18000-25000
20%
22000-30000
20%
25%
25000-50000
30%
25000-50000
30%
30000-60000
30%
30%
50000-100000
40%
50000-100000
40%
60000-100000
40%
30000-40000
35%
over100000
50%
over100000
50%
over 100000
50%
40000-50000
40%
50000-70000
45%
70000-100000
50%
0%
15000-20000
20%
20000-25000
25000-30000
over100000
0-18000
55%
1994
1996-1997
1995
1998
Brackets
of annual Marginaltax Bracketsof annual Marginaltax Brackets
of annual Marginaltax Brackets
of annual Marginal
tax
income(Rs)
rate
income(Rs)
rate
income(Rs)
rate
income(Rs)
rate
0-28000
0%
0-30000
0%
0-40000
0%
0-40000
0%
50000-100000
20%
50000-100000
20%
40000-60000
20%
40000-60000
15%
50000-100000
30%
50000-100000
30%
60000-120000
30%
60000-120000
30%
over100000
40%
over100000
40%
over120000
40%
over 120000
40%
2000-
1999
Brackets
of annual Marginaltax Bracketsof annual Marginaltax
rate
income(Rs)
rate
income(Rs)
0-40000
0%
0-50000
0%
40000-60000
10%
50000-60000
10%
60000-150000
20%
60000-150000
20%
over150000
30%
over 150000
30%
Note: India'sincometax appliesto individualincome,not to householdincome(exceptfor HinduUndividedFamilies).The general
principleis that all incomesourcesare subjectto the sametax rates(theprogressive
tax scheduleappliesto the sumof all individual
incomes,whateverthe source).Thereare howeverspecialexemptions
for particularformsof interestincome,transferincome,etc.The
tax schedulesreportedonthis tabledo notindude"temporary"
tax surcharges
(forinstance,a 10%tax surcharge
hasbeenappliedto all
incomesabove60000Rssince2000,sothat theeffectivetoprateis 33%ratherthan30%).
124
Table 3: Simulated vs Actual Income Tax Revenues in China, 1996-2001
Actual Income Tax Revenues
Total
Receipts
Wage income
Receipts
Busines income
receipts
Capital income
receipts
Other
receipts
(billions current yuans)
Total
Receipts
(% GDP)
1996
19.3
0.28%
1997
26.0
0.35%
1998
33.9
0.43%
1999
41.4
2000
66.0
28.3
13.3
19.0
5.5
0.74%
2001
99.6
41.1
16.0
34.8
7.7
1.02%
Other
receipts
Total
Receipts
0.51%
Simulated Income Tax Revenues
Total
Receipts
Wage income
Receipts
Busines income
receipts
Capital income
receipts
(billions current yuans)
(% GDP)
1996
22.2
12.0
2.2
8.0
0.33%
1997
32.0
18.6
3.3
10.0
0.43%
1998
37.6
22.1
4.0
11.4
0.48%
1999
36.5
19.7
4.9
11.9
0.45%
2000
48.5
28.0
8.3
12.2
0.54%
2001
63.7
39.6
10.3
13.8
0.66%
2001b
'147.3
88.8
16.0
34.8
7.7
1.52%
Ratio Simulated/Actual Income Tax Revenues
Total
Receipts
1996
115%
1997
123%
1998
'111%
Wage income
Receipts
Busines income
receipts
Capital income
receipts
1999
88%
2000
73%
99%
63%
64%
2001
64%
96%
64%
40%
2001b
148%
216%
100%
100%
100%
Source: Actual receipts: China Tax Yearbook, various issues (1997-2002); Simulated receipts: authors'
computations using urban household surveys tabulations (see Table A6)
Note: Simulated receipts for 1996-2001 have been computed by applying the relevant tax schedule to the
individual distribution of wage income, business income and capital income estimated from urban household
survey tabulations and reported on Tables A2 and A3. The 2001b estimates have been computed by inflating
business, capital and other income so as to matach actual tax receipts, and by inflating survey-based top decile
wages by 50%, so as to obtain a realistic Pareto coefficient for the wage distribution.
125
Cl
\'q
-
EL
(f001-
C) >)
O
L
- 00)
L
C)
0)
6
"I
N
CO
0)
N
O
'"t
(O
-a,
-.
s d)
)
Cr) c-
Oc
c
xxi
=3
COCa)
cD )
Om
CD
0a)()
oa c
, E
CL C
Co
O -s
Lo
*
C)
oO
C)
>_
e-
0
.
N
U'
C
m 7
M
10 10
-o
o
-a)
0 0
U)
O-
C
d)
. (0.o
'-
U
o
w-
00
0
U)
e
=o
xQ
D
- -
6i0
a..Cqa~(
N:
CN
C0(N
.L
C
(L)-
n)-
E
0
I
(n
o
a)
>
.00m
cE
0X E
CLo
c)
(U)
0
c
C
CDo
°0
E
0o
u
O:
o
N
-
U')
C:
0 0
o0
.Q)
Oo 0CCa)
x) 2-
> Ua)
0
0
-a
N
0
.I_
0
.0
(U
I-
(NL((
o(
o=
I--
cN
t
o ;
0 a
DCL
(va
C0
b(0
C
O
a) a
0 0)
(0 0 (N
-
(n
a)
0
V
a)
D
0
LO
C)
-
00
0
'
0 )0
0)
0C
vO
aD
a)
CDD
(55-
V
'0
a
-
0C~c
a)
Co>>
a)
O
N-
.
XE
0a)
vOO
00 LCO
(N
C
L.
L
LL
0)
0)
O
O
O O
O
O
O
O) O
O OO)
O 0
a)CNO o oCN
.
S
.
a)
_0
01O 0O
a)
U
nF
0
C
co
0 0
a)
oo o0 0
(0 o C)
oq 0
LO
-
.
S
.
S
0C 0O c
C
O) O -
-)
O
C
oq
v'C~
aO (N
Oq
C
(DCI
.
S
C:
C
r
a) g
00 0
cn 0- '
r--
A
o
0
o
N
(0 (0
Lo
a)
Lo
(0
Lo
g
o
I
(nC/
O 0
U
0
In
U)
>
Ln
o
( N
t
(0 CD 0
0) n
N
) <(0
(0 (0 (D (0 CD
C0 (0 C)
rNN
In 0 (0
c
0
t
N
co
N
0
In C) w
ni U)
) 0
O
0
0n (0
(D -t- Ni
0
) N
o-(0 I 0
(0 (0 (0
Int
tD(
U
m
m'
N
t
1-
) - a)
c)
(a 0 0)
C
N
~ a)
N
to
U
N-
0
,I
Lo
C
N
In)
t
I¢
N 0
N
N- (0 '
v
N N
02
In
In
N
(0
co
N-
IO
(0
N
U) 0
0
(0
N N
a
C - 0
a ..)04a) Q-.
C
~
I
L
0
(0 U)I
Ci
a)
o
-$
N0
0
NIn C) a)
N
T N 0 N
0 (
(0
D U) N
N U) N C) (0 U) N- C)
CS
In (0 C) N- (0
0
0
N N N U) t In In In)U(0 (0 N- 0
t
U)
C)
II
--CD
N
(0
In)
C)
o
o
I-
,I -
0
In
0
N-
0
In
In C)
ts
N
o
0U'
0n tN
a) .
O
U
On N
(0 0
(0
N
U)
U) 0
)
t C)
IN
N
(0 0)
N- (
N
In (0
In (
N
-
0
n
0
c
0N- (0
0 I
'-E6
o0. X°
j2.
a)8
(N
IN In
(0
U
Co
->CIR.
_
N
aD
N
N
In
FL
0
,0
N-0 In C)O)
O
co6
£0
0
0
>
-
ID
(0
C)j
CD~
-
o
ao o, x
8
U)
-
75
0
CL
In) NN- N-
0
N
N 0) In N C)
0n 0n 0n 0n 0
0
0
N
co
0
)
In N- C)
C) C) C) C)
v
)
a)>a)
a 0 I
CO
o cE
> a)
o
OEn
--
N
ao
C
£-
S
2
00
N
o
N-
1-i
-
e
°O '
-5
C
i-I
Ul
r-
0
-
0U)
-
t
t
(0 I
U)
-
C
(
(0
o
C)
N U) 0
N
N
I (
N N
M
C
N
0
I 0
a
t
-
t
0
0
N- N
.C,
In
(
U
T
0…NM
0
D
a)
0
a)
~0
0.
- t(0
r- 0
0
N- 0
0
0
- N
N
n rNi C
0
N N N N ….
.vN
0
zo
0
_
oo
N (0
C) C)
cn
U)
U)
-D
C
U (0
IN -
0
(0
m0 In N
O
O
N
co
(0
CO t
O
C) N 0) (0
U) N- U)D
C
)Un
aw~~~~
C
a0
C
0
0
0
N
N
N N
NN
(0
0-
N
C) (N
txN
0
-
0OC
o
n N Co'),
o
n
q>-o
0-
U) (0
°
I
.0 OS
(0
u-)r
N
In
o
N)
N
8
--B
C
'ci
0.
B
.,
CD
-eg ft
>
E
(0 N
N 0 N
NCO)
>
F
C)
U°
(0 C) CD
C) 0 C) U
( nN Nto s_ ssa)
1
c
d
C)
nr
N- U)
)
…VIU0~
n(0(0V
I
In
-
,
N
-2 C 0.
I
F
a)
CD
) o
o
o
o
N
N
N
N
N
N
In
(0
0
0n
N-
t
m
0
0
U)
In 0
N LO
N U)
U) U)
o
N
o
N
o
N
o
o
N
o
I
o
m
m
a)C
Coa)
(0
0
.0
C
In I0 (
In U) .n
0
C) U)
(0
- . N N N N N n c~~~~~0 m
a
C)¢
N (0
0 N- 0 m
) (0 N
U) U) U) t
0
N- N U)
N
(
C) m
N0 U (0 0
t In In) In (0
I
·
>r
.0
N
U) 0
C N
4
N
CC
CC
In
t
0
0
N
N
~0)In.
Cq NN
(0 UN
(0 N N
NC))nnN
U
N
*0
E Co
,Ct
C CL
n
0
N C) U) IN
2
U (0
I(
(0 (0 N- N-
a))c
I)a)
a)0
a)U1 a
x
-(0
Ia)2
CD
C) C)
o§
a
M
0)
m
~-
In
0OC)
(0 N-
2
tD
0
-
0
ci_ (-)
(0 N
0
U) N N O
0n N- 0n 0
U
C.
oa)
CD
Ca) >t
.
r-
a)
a)
.
X
N C)
tIn N
n0 (U0
D1--
c
0
.
a
.
-(0
BeE
CC0L
N IN10!P
In (0
0O! 0 O 1D
-C
NC) ,
~t
n 0~0
co
0
C.
U)
*0
U °C
In C
…oNUU)UU0U
a) C-^
C
N
0}.
CL
_-
31
Io
O
0
~
O
N
0
-
(D
I N
t
n tD
(0 In -D
C) C) C) In (0 (0 (0 N- t
) C)
R-C)
In
0 In ( 0 IC
C
,rN N M It LO0 (D
InU)
N
C)
0+ N- U 0 I
-
I
N In C
-
I
-
I
N
(0 I
U) st
In
It
U0
N
M 0a
I-
c
c
0a)
,, Y 0
0
C)
C)
0
C)
0
C)
C)
-N
C)
C)
U)
C)
C)
C)
C)
tI
C)
C)
C)
C)
(D N 0
C)
C)C
C)
C)
C c
)
0
C)
R)
0
g
c
N (II
…~~~~~~N'
N
..
)
C)
C
1
(a)
mo
cn0
0
N 0
0
0
C)
C)
a)-
N.
C
(0
iV a
>
C
(c
(0
aCC
)
0CO
N
C -
~CDa
)))
.g
00
C-,l
*~~0 8
.E
a,
E
.o
2
8I
E;5
0
0
0
U)U)U~~~~U
0a
C') CO H
(n
q'
0
0
N
,
CO
0…N~~~~~~~~~~~~~~~~~~~CIC4
1
-~~~c
=
~~ ~~~
...
6E 6
U
C')
Oo oo 0)
_
C
..,-, CU
U
U
-
61-444
64
2
0.
CL
00
<0
CO CO CO
'
)
0)
0
0)
0
0
0)
X0
)n
,C
.C a,
c7
N@
2>E . C
-
CD a)
0)
:5ux
.1
C
e
a
:Coo-
8.a~ 2
0>,C°
C
8
E
0
O °
tC) Ua) O
CO0
N
)
H
N 0
N
o L-o"E C u.)
.E
,o= -o',
It
C
.0
Cu
V
V
Co
m
.z.2
,-o
o
U)8
E
>
c,
~D
'2
E->
l:
Z
8
Iq
R
0) CO
0'
<
5
0
CO
CO
N
o ,c
Ct
0)
(C')
N
C )
D CO
CO 0
N N~
:)NtO
NlcCCO IL
IR
Cm
00
0l
I)st
O
t CO 0
st
CO H
N0)
N0)
C 0
C')
CO CO CO CO CO
N Nx N N N
CO
Nt 0
H CO
N N
(R
Cli
N
H0
N
CO
0D
CO
N
Cw0)- CO
N CO
tD
CO tO
H O0
N N
CO CO C)
0)) N gC CD H
N N N
C
O L)
-a
8
0
o
C,
a,
O
8 F"
F-' 'I,:''~,~S
a
C E
0)C
.20
C
~
l
V
O
o -
CL
g
cu
_
CZCD0
· ) .C
-CoU~E,C o r-
E
u_8
-
°--
°'
N
0
66
.:
o
-N
-
0
O
-)t '
o
o
NN
O
,…
-
o
o
O
0)0
NN N
wN
a
-o
o
aE
0,
u
a,
8
2
o
oS
o
o
-°o
-
a,~
a 0 ~a'n
D
C ~o
oOa
Cuu5
o-
>) .
~CC.) 5.o
~
o -oc
C0
--
02a)
.0
E
CD.
,.
CO
OO
a
Cu
_
.g
C
0
a,
3
.0C
0
.0)
a,
0
Z
0 ) -o
,
,-) O-
oc')
oC') oS
oCO
oSo
o~
~
N C) ~ 0 0
0
C oN
0
0 )C- H
") 0
HD 0
H r-HH-O0
s0 0 sH
sNO s
sCO<)N
s
s
F sN 1_~0
I'
s2
I'
N CO CO
4~:-
>
E08
m-
4
N
N
-
4
L
L)
N
t
CO H
N
1:
S
c
-
CO N C'O
-0N
N
v
C'
0
-s CD C
I
-P
O
CDCCm a,2
_')
aC.U)-
N
CO CO CO CO CO
H
) ) c0 c0
-
l' oM
C
0 )o
L-
Z;
CO
0) C
a) o;
)
0
t
a)
t
C
)
c0)
C'
c)
N N
0l) c)
O
o
a)
0) N
O 1-
0-D
CO
N
c)
o
N
c0)
co
-
-
c)
0
o o
0
_
.2CU
E
EO
0) N CO H O H
H N co
O N
CO CO H N c
D
)
H
H
0N
- N
oc)0
e
.0-- U,
''U,
-P
p 0 )0 ,
a,
.0
C0 CO C
O 0
N
o
)
0 0)
(D
O CO
C0')
C
C) 0
CO0
O
0)
0
O
0
0
C 0
U,
CCm
C.)C.D
N
-
C:
R
0CD r
Q5)
-o
~
0
I._ ==1
C
CO 0D
UM, a,
O >E ,
a,
2
a0
CD
co
0)
H O
O O
0) 0)
0)
0
0)
o
a)
0)
0)
N
a) 0
0)
a)
0)
It
a)
0)
u)
a)
0)
U
E>'~
a
CDCuO
O
0)
)
-
N
CO H -O 0) O
a) a) a) 0)
c
0) 0)~ o 0O
0) 0)cs zO
- CuC
.0CC2CO
.0
Cl
ci
..
a- Q)
a)
c
0
a
C
U( N o
F
51, CD
2_
(0N
o
N~
c(N
eII
N
N
6d t
cN
n
:
s
EE
)
(-D CDo
(
°R
°
)
a)
N
0)
N((((
N
r0.
0U
C U) >. o
a
3
a
-2
'M
0
Co
C0)
~_~~~
C)
(N
~
CD
C
(0
~
N-
CC)
~ ~
C
N
a)o "O
o0.C
.,.
0)
.
U)
C
m-
0
.0
~o
O
cn
E
C
0a)
E)
en (DL
E
0
c~)
1
COIn6)
't
0
O02 OO
cdO C)
)
L)
0
)
CO
N-
N
(N
-t
)
(
D
(N
)
0 oC=(D
0
N
N-
o
c)
O
't
(0
m
)
N
m<
L) t
)
m
I
t
r-
c)
I-
o
o
r
co
m)
o
c
a)
CD o
O
D
NNl-
Fr. CD
E
;R,coa)E
0_
C
V
I-
CDC
ca.0
E.
0
0O
0
0
E
o
N
R
_~z.
E
0eD
cn
.'
-,
CD
mO a)
CD
)
N(N
t
-
cO
N N
O O
m CD
CO N
o) C
N-C (D
(N N
N
0
c
c
-
oo°
N
O
-
O
Co
N
C
(N
L
UC) )L
(U
C
°
CD-
--:
E.~r0 ~.o
ou
E
u
ao
(
C O
)- cJa.L) On
L N- 0)
N
>
CO
D
~ O
)
CD
Cu
C)
y )
C1
C
N-
00
CDI
CD , a -
o ,,
U,
,
(0
6&6O COC) oDiu °t
C)
C
U)
a
a(N
C
-0
(0
-
N-
(c
t
)
N-
C)
(N
)
'I :
C
)
s
(0
N-
0)
0
(
N
N
N j
Cl
.~=u'
OCCU)
O -))
8 a)
nS
_) - =
.0
Ia)
Cu0ns
C >a0
D u
a).c
) SCo)
a)
=
II'
0E
E
CY
°,
~,
a'E
co
0
C.
CL
L)
uEO )
.a
V
o Oo
0
o
o '
. >,
U)
tO
C) co )
U0 a8,, 8
(N
In.CD
D
CD
.0
C
W
C)coD
)
~~
U)
= CC)
m-
0
na'
C> ro
~
s
~
z(N t.
c)
I)
(N
)O
C)).
(NoNCNNo
CC
_
o
Cu
C
-,
.Q
a,
ID
CD
0
E
I-- .
U)
a
0
NCO
C
(N
(N
0
C)
0)
c
C)
N
O
CC)
Lm
.
0
N
(N
0
0)
L5)
NC
D
0
)
O
0
(N
CO
O
C) N0
It)O
v)
-~aco...a) a)
0 a)c.
CuD0
2.
CU a,
.o
Cu
co
Cuu>
0co
.
a,
)_
0
-a
..
n
-a
Coc
0) CD
0
a)
ou,
.)E
>
'-
E
E
Ecn'
C)
NN
C
Ir
E
a)
In )0N)
m o
oo
CO 0
N
uI
CN
(3D
cO Iq
o
0
O.~ O.
O
D-Z 0-
C.0
--
CZ
a:I
E
-
0
F
C
~~
u(DcO)
tIt
=
C0N
-.
~-(NCO'FJ-COCOCOC)C
0
0~~
0.
0,
C
-
2
cn
CIO
>,
(C
)
0
Nco
C)
0
)
0)
0)
0
0)
)
0)
(N
0
0)
>'.0
Cto
cn
m
0)
0)
0)
0)
CO
0)
0)
C)
0)
0)
N-o
0)
0)o
C)
0)
0)
O
)
0)
0
0
0
0
N
a,
C.
c
Co
Z
2
a~. ) Co
…
)C
_
IliF
a)
a)
0
0
N
"6
2!
)
_
=D
0D
9!E.
CO
-
n
_ _ _i
'!
(O
0
_2
A
-
t
N 0t
U)
0
N
"
-
-2
a)
CD
N-
a)
c.)
a)
V~ 0
I
N
0N
tNIliN
0
w
Iw0
C'
a) 0
UM N
0
a
0
(0
rD
N
1~
U) a
c
m00
N
_ .I O In m
Ci l> lc
114 No
t
C')
N
Cn
U
n
c N
O
Ir
o
1
0 o
I
I
Cn
CC
wg
0
0
N
U)
)
-
I
U U)
Om
c
N
M
N
N
:'
1
O
ur
U)
) tN
O U)
0
oW
_
N0
0
N
.)
aT
ol~
0
0.;
o
0) t
i
N
U) N -1a)
03
O a)0
N
N
0
_: N
2!
0 m)
N1 C1
0C
a)
a.
U)
03
N
N
'-C
N
I
C)
st
NN-
a)
)
A
I
U)
N
U)
N
N
tD
tD
a)
N-
In
0
tO _
N-
U)
m_0 uw
m
N
.
0
N
r
O
C¢
OL) 0
t
_- A
) N1 a
l tD
Cm
O-
'InD
0
uO
a)~~~~~~~~~~~~~~~~~~~,
I-
0
0
0) (
N
a)
0
O
a. _
N
a)
.
Nr
N
tD
10
fl:
N
0
0
U)
- rU) U)
-
0'
U) Cn
N
a
N
a)
st U)
-
N-
0_ta-))
N
ID
N
N-)
N
0
1t O
- 0
U)
nNN
O
N
0
a)O
.
-,; C
N
-. N atZ)
N
.
IT_
I
N
(D0
0
N
t
w
(D 0
I-
a)
oU
(D a)
O
U) NO U) Na__)
Q) -
'!
a)
a)
l
Z;
a)
-
a> °
N) 0
U-
o i,,
N
N
M
N
0
a
) NU
U)ULNtD
a)
S
cli
o
a) a
.
U) N
-00
U--U)03'NO03)
_O
CD N
.
('
a)
N-
ur s
O6C
N-
-
U)
---
N
-
0
N
N
C'
)
U)
a)
On)
CO O a
N
0
a)
-
U
U)
url
i N- cli
O 0_C
C'
m
N-
C)
e
! 0N
CO _ o66
N C
0
'
U-
IN-
a
U) -;
Q- =
0
.
PNN
N
NNCN
3! NN*N =X cq (V (
U N
N
)
C
oN~
)C03)~
-
N
N
D
C)-
2
N
0I
0N--N0-0
CD
a
CD 0
N
C
0
FZ
di
Ili
U
V
aa
-
h-
-
rf
L)
a)
C'4
lo
0
.0
Lo
a).
m
(
"_ d
O
cl
N:
-
Lo m0
,
a o) U' - CU
o
o
a
0
0m
a
O
QN
t
At C' C00
-
m
2
CC
O~a) N0
-~)
_
vt (CN
.
_N
) N
Uq
0 -
'T
16 1
_;
0
.
0
.v .
mo
mm
o0) ( )'oU - U1)oN-Noa
0
N
t
N,
LO
N
N_0
)'_
C )
~
N
ll0S
R
>_m
eV "t
mV
Na ) CC CC
D
-
~
)
a)
0
0 a) U
o
0
'-
)N-a)0N
-
a)
N
U)
a)
UN
m) m
)-
m
-,
4
Ld
N-~ N
1)
U-
0
N
t
N N
o
L a
':
°-
N'
U- C'
C'
D
&
')U)
a
a
U-
°
o
I-
0
m) I- 'I,) 12 I-
0
N- N
a)
0'
0'
0'
0
(3
_
O
U)
U)
t
o
In
a0>~~~
a)
O0
C!
tv
0
'Ioq
T)
_
I
N
"'
03 N9
U)
a)
C
03
0 C
oU) o
a)
)U)
CD
U)
1-mt U
U)
0
0
0
_
N
N
N
0C0O0OCON
a) E
_
N
a) vx
Uv0
a
0
Q iz
N
U'
N
')O(0
-
U)
0
N
It
It
U0
'
a)
C
0
N
0
N
'ITN
0
N
C
06
0C
C0
0
tv
U)
0
0
a)
N
I
a)0
N-- 0a
N
C'
-N-1 NN
v
N
N
No
N ')')o
N
,D
-
U)
-
dIN-
c.
N
6
6
Uz
m
U- 03
No a)D U)
CD Nsta0)
r.
t
CD
1U) c)N
0
U-
U) 0 U) N
U N- C N 0 0
S
!u
N
0
U-
.
-gh
U.)
C0
COS
-m O)1rU--a).
U)
-
C' a
I
I.-vs
N
N
O
"
o
N-
N-U)
,:v:
CO?!~
°6 06
a)
S
cC
_
!C
,N-
V
_
CI
('C
O
lo)
a)
0N
)
a)
I_
N
N
(D
0t
N
m
N
")
N
_'
N
U) 0
O.N
o
UM
St
'
a)
N
('C
tD
.
N
NaU)
0
a)
a,
N
N
CN0
_
0
0
a) N
N-
-
N
N
a
U)
N
C
U) a)
_
It
-
)o('3'0
a)
0
I
0
U) 0
0
U
)
0 N- 0 a
03 NN
IQ
vO o^
dT
)
o
iu)
U- N-
N
0
0
I
C') IDC
a) 03 )
N
a)
N
a
;
'I
CO a)
U)
Oat
a
N
w
a
1.
a)
N
a0
C)
N
U)
D
t a)
oUZ
0_a)
a)
XNN a) a)
ED
N
a) N-
C!
C
0
t
N
N CU)
0na) C
a)
(C
,~
o
c3i
U0 U
I'
('
U
U)
a)
s
0
_
_N
-C
os
_ a)
U) N
CO 1- 0
CO
g
0
0t
0
I
-N 100
- )v00 U)U0
N -tt 0 N
-tr
1 - O L
M
)
N
N
O.
ci9
, _ NOt UNi -1Oi O
N o)6C
N *
C'6a
---N
-UU~
a)
In
) )
~2
vi
N
C;
, N-.,
M
.-
O 1N-
.
-
3
N
u)
O
NU)
C
U
0a)
C'N
CM
)(3U
Nt
…NN
N_- ° N
-C'C
C')0
U)C U )0
,,m 0l "IlO U)o
>N - N- lN- t U-t 0 0 30l
l
l tv
03o 0
Na )- c N
' i -6
66- i U)
u N
i
a)
0
-
0
N-w a)
I)C°
C
o
N0
N 3a)
O
O
O- a
a)
0
0
0
Lo
('C
W
0
ciU- -
')
N
N-~ a)-
0
0
U) a)
U- N -
N.
)
t~t
N
Ott
tt U- 0 U)
0
C0) U- a)
a)
a)
0
a)
No
0
C'
oU) ItU- m
ID.) o0, ID
D
w
N
U- rD
N
CD N
(0
a)
0
U)
N
Cd
0'
0
CC
L('
9!
,L
9!
g_
o
CL
N-
_
C
a)
N-
L^
a)
N0
LN N l_
Ev
s^CS
'nN N-
C O
')
N
CO C
IQU^
l)
C)
6 um
g
N
0
UoU)
cct
.
N
N- CC C'
CO N 1s
-
nN-0
. -
C^ N-
Cu^
N
O
N
L^)
U)
0v 0
_
a
a
U) N- U)
_ CD
qN
I lo
O- 0d N 0 0
0
wDs
0
0
C'I
t
~~~ ~.
,
N
O
Cv^.' ) O
V: C'!) IQ
L6s
6
I
v:
a)
0
0 N _ n "-1- O
ON
0)N NC 'C
_-_-N
N 0
m c9
.
U)
M
-
a)
a
C
-n
N
N
M
M
0
CO
Cv W) C(v "
0
'T
-
M
't
Cv
L^
a
U
lc^
A
',
C) 1
L^
tv
I
c')
ao
t
-t
LU N
C C.
N) CO
N
N
t0 N CO o
a) C
I 1_ 0
.
iU - U ) N - N . C
C
N
~m~~~~~~m~
~ ~~ c
c i
c6
'L
R
'L
L)
O
CN
')
N
N
CO
l0
_
- N 0
O
O
t t
'
N
UC'
N
CvM
g
Lo
N'
L,
O :
o o
N
O
:I
-
0.
N
U
C
N
N
C')
0
CN
N
U-
N-
N-
0
a)
a)
0
0
N
(
0
a
U
CD C-)
N.)
U-
)t'
CO 0
a
N
N- 0
N
O
"C
N1)
t
N
C') tO) -
N
N
!! 6
N
'-
t^l
N
N
N.
L
N
CO a)
u ;iu~
C')
C
_'
C') L^'
N
O C
0 nNU- 03 LU) N
Ut
t
vN
Lo
U- L^. L^. C^
Ihi
.
.
C
^
-
t^
03
(,
a
t
tv
-
-
C
2E
00o
cl Cv~L
N -°
0
UNCO
R-O
tv
N
CJ
N-
-
N
-
-
-
N
C')
a)
C
I
LU) CO N
a )
CC CC
0) a
C a)
a) C
0
CC a ) O0
CC0 C 0
to
IFc> )
)
a)
CC
a)
0
C
C CO0 0
tO000000
In
a
)
!~
C
c
C
-
)
_,
0C
a
0C
a
_~
0
a
_,
_)
_
o
0
c0
C_
C.^
co
N
CO NC3 a )
CC CC
a)
a)
a
a
a)
0)
0
a)
C
_ N
a)
a
0C a)
0
P
r-.
C')
-
L
a)
3
8
o0
o
0
!
Cv
a(.N
N
N
- InU
LCC LU
C
:i
L
a)
IN - NC
I'l
m tN s
C) N
a)
v cO
v N
1-N
^
c')
o
L^
CO
C') UT U-1 U-t L^ U)
N-U
-N .
0,U C
C
-
-r-
L, wb
a)
U)
0
N
N
i
0N
_
C-
'
C'
-
'a
('
C'
U-
.
N
0)
a
0
aU- a)
N-
U)C' t )03UO
vb (v OvT~1
N- 0
O) C
^0
0
03
a)
U-
C)
NU') N
O
-
L
U
L
L^
N
U)
U- N
N
Lo
_ (D
0D
U-
0)
a)
w
N -
N-
N-I'I^
a)
- ;:~,
No
a)
N
N
O14 Ol: ON
c')
0
0
N
CO
L
t
N tO
N
N
Ut
0
CN U-
a)O N
N
( NC a)
) N
CO
CC
C3
O
N
O,
N,
.6
F7
a_
L.
m:,
N
4
a,
- - '! I^ A 0
N-
M
0
-
A CD
7
M
N
N
N
C
I_ 0
N
D
_
U1
N0t
. .
t
C°
_
D C
0
I^
N
C
CDN-
M
C
-
C
,-,~~2
CD 0
C
0
Mi
M-, si
ON0CDOC
0
CD 0
CD0
N
-'
0
n
t
CD
O
IQ
Iq
-
N-
n
EL
.
0
i
a
U)
It
0
)
CD
CD
CD
w
0
C
N
st
*
N
N
_
O
0
OR
N
"t
DC -
d,
N
Cli
-
a
ci.-Z
m
N
-^, N
N CDN
CDC
0
ID
D
_
_
---
CD
t,
ID
NN-
CD C
t_ DC
CD -
m
o
w
)
m
C
N
N
N
)a,:t
°mU:
0 0
~o"I
o
!O
o
0C
tC 0 0i
~o
'
~
V
CD'
t°
to
N
CD
_D
-
0
CD
0
C CD
-DC t-
-
F
t
0_
Dt^C
0t
'0
N
-
C
N0
N
C0
0
q
0C
O NC
I
N
(
z2
N
N
1oi
CL
-
N
CD
O
N
N
D
t
N
O
L
0
CD
N0
CR
~i
-
LO
N
CD
N
CD
mm v
N
0 CD
N
t0
D
Z
O
0
N
00
N
_
CD
T
C
O
C
N
D
Cw
CI
s
D
_
1-
-
aOs '"_,N
v0CCD
't,t I~N
N DDC 0
N
R
U),F
R
CD
0.
0OCDN
C
N
0
N C
N
C
CD
D
Wm
mm
0
0s
L
CD
N
CD
0
CD
N
N
DC
CD
-
_
N
N
0
0
N
N
CD
-
M
0
N
AO
o_
N
~
N
CD
0
N
', D.
N
_ NCD
N
0
CD
st
t
Z C 0
0
CD
C
CD
C
CD
CD
DC
CD
CD
_
CD
1-
m^
m
N
At
1PR0
N
t'7
r
Cy
-
N
N
CD
N
w
-
CN-
0 C
C
CD
CD, CD
0.
N
CD
DC
-
L
C^ 'T ONC
N-
't -:
N
ID
NCD
N
DC
0r^
N
CD
DC
0
N
Lo
m
t
0
-It
I
t'
I-CD
0
CD
CD
0
CD
CD
_ N
t7 t t
C
Dc CD
, ,
D
0
N
0
N
CD D
N N
0
DCN-
N
…NNCD
0
N
t O
- …- --C0
O0
0
!…
0
a,
;
N
CD NLN)
2
CD
_
Lo
N
…
-,N
I
NU
-
~2M
0
C
0
N
N
Nl
CD
o
CD
o
CD
N
CD
D
CD
C
DC FC
U7
0' Cz
0
CD
NCD
N!
t0 1-
0
0
0~
-
ND
N1- t
NN
I
I'd
N
D
N
N-
o
0
Ct^D
CD
0
-t
CD
C
CD
0
NCD, CD
CD
t0
"It
r-
t
"
t0:
D
0
N!
O
CD
ICD
N
0
P<
0
CD
DC
N
Id:
0
N-MCDUCDNt-D
CD)
CD
0 N
C
!
…N
o
0
CD
C
t
O
t
aC
~0
0C
N N. t
2
CD ww. .2 N2 _
A
,:6 UT
-
T
N0
_2
00 - 0
0
CD Cttw C
D
CD _
N CD
U
CD
4- 0
N-DCC
D
0 0
CD
C
D
ti
j
0 cD' N CD
t N CD
°
S
1
0
0
a)N-D
u)U
D-CD 'C
t
m m
N04
D N-C
_ICD
P
Nc'
I",0
U°
o
v
o
Ct
N
L6
a.
-tO
N' 10
.
4
C
t CC co
CD N-
CD
C
DI'-c
0; -~
C)
D
-
pO oa)o CD
w
C4
N
C
tco
N
ot
N
N-
CDN
t°s N--. ') NU 0°
N4
N
D CD N
Fjoiq o
(o(,
d
~o
CD
O
CDY C N
co
C
~
N
CD
~----N
0lo
CD
r-L) 0v
C,
~
ND
0N
N'
C
f~o
gt
Ot0
CD
OO
N
N
Cl
N-
-
CD
~
1
D
.;
-
-
N
t
n D0
-
0 N
CD N
.
N NNCD D
tD N
CD O
NC :
C LN
r.t
O
O N
tD
O
D
10
C D0~C
C
CD
t
o
I
D
clN
0D
ND C
0;
o
0
tN
D0
CD
-CD^CD.^ ON
N¢
0
-j
-
0DCCD ,
O
t
'N
N
i
tVj
cLi
CO0D
N
C
N
N-
mDN -
C
CD
N-
DC
2~ N
-DC
CD
t CD
DC
NCD
t0
NN
1M
CD
q
q~
R~
0
(
0
O
CD
t
Ct
U
N I-t-I -
N
-
CD
N
C
D
0
DC
DC
DC
DC
DC -
N
C
D-
DC
DC
CD
N
DC
DC
C
DC
DC
M 0r
I I
C~~~~NC1NC
N
oN N
LoI'
^
N c
0 NCD0
N
UD
-
0
…N
0D
-! W
N
-
CD 6
-~
Dc
-D DC CD
.
oCt
N
IQ
t0D
oN
N
0tD O -
-N
N
NO
N-
CD CD
D
~
0
0
Lo
Uo
tT
<ti
_
N
N)
tO
-l
toi
DC
0
I
o
U'
Lj
_~~~~c
-
.
D
CDDCO
O
tO
oi
^
q^
r0
0
w
-
-
O t°N
0
CD
CD D
DC 0
CC
L6
C
C -
C
t>O0
D
N
N
04
tON
C
to 0
N0CD
00-
0
.Z
Ca
0
oL
CD
-
o
C
C
CD
10 (D
IU) Um
m
n
N-
N0
't
s
0
~Ko
-
D
.
CD CD
N- C D
0z (et
4~~~~~
4s
'd_
)
0
0
CD
tO
C
C
O .2
2
^
0N
NCD
0
n
t
D~C0sttD1'
CD N
CD CD
CD CD
Cl
CD - t
0
U^
)
C CD
N CD
tO
w
___N
CD C
OtO-O
N::
Uii
tOt^ Lo
t
I
"
ClN
N
0
0
i
r
-tN
0
N
N
.
CD '
0
0
0
CD
U
oC
CD DC
0
N
O n
-4
4-
0
C
a
DC
N- C
C
C
DC2 DC
0DC 0
C
D
D
DC
D
DC
N
D
DC
C
D
C
DC! DC
CD CD
D 0
DC DC
N
DC
DC
M 6S
-
CD
cc
t
t
…
CD C
0N
0
CD
N
N
a
st
-
N
0
0
0
cN
…NoN~C
"i ,
-
NC
C
D 0
D
D
DC2 DC
0
R0
DC
~(
tD~cj
°-
CD
0
0c1
r 0
0
,
Cj N0
N -C
NNDOtCDO -1
0
t<)
CD
_ N_
~o0 0
CD C
oj
CD
N- N
00
NN _ N
CD CD N
CD DC CD
N-t N-
0i
O
cC
2 Dco
U
t°
tODCDDCCDN-DC
C
cniO
DOCDCDDC
OD0
Il0ND I
NC0
6 El
g
VVV~CD
~~o~ ~ ~~~~~~~~~~~u
~
C4
,
N~ N~
r
IN
8_
N
Lo a
cDDCo
'IT
-
-
C
Ea
-
tO Nt
0
0C NtO N-. CD
M
CD
-q
T
w
No
j
0
0
d :s
qO
4CD
6a
t
-
Ot
N-
iz
tL
N ctO- D~
0
CD
NO
0
1-
tq
R
51
a-
a)
N
0)
a.
-CD
c6 -a
,:K
(I
c
CD
Ct
CD
.0-
NDC
CD tC
N N
co
1-r
N
t
CD Ct
CN
…
t
a
:
I
0
NN
P
.0
CD CD
CD CD
-
t0
CD
CD
m
0
t
CD
-
-
g
I
D
- t I
U
N
N
D N- -
_
CDC
0
NW
CD
N0
N
N
*M-- rC
~N-CDDCN-'tqt
N-
wN I
C
N
^
O
N
C
I
CD
0_0
t
O
(L
N
0
-N-_-.DDNN
Nv
c
mm
t
IT
^ts
N-
C
-
't
r.
N
CD--~
C
C
0o
-
D
U
0
N
t)0
M
DC 0) CM0D
C CDD0
q
N
N
m
wD
CDO0
N- N
0
0
DN0O
CD
-
S0
0
If)
N
0
V
-
- C)
CD
_1
CD
NNl-
D
Q
LCD
0-
N
-NNNN
*N
D(
DCDC
C
N
…
0
'P N-
q
;,c"Ao0
,
t^
0£
CD N- '0t
0 C^
CD
C
L
"
NCD
CD
D CD D
l
N' CD Nl- "I
CD
0a0 -N!coto
~CD
MCD,1,tN
M
_NNC
:-o
ND
D0
t
CD
~
N-
tU o:,
0
N
_C
0
0t
,4
0
s
^
N
_
M
co<
DCC
C_ ^
-CDO00
CD DC
N
O
w^
_
CD- C C
0
DCN
NC
N
CD
0
0C -t_- tCD CD
C
CDi cD
CD
r:
tO
t_
N
.
!~0 CD:
O
0,
CD
0
N
NCDInNDl
N
0
tLF
0
DC
CD
N
-
CD
C
N
~
_
d
CD DC NN~00NN--D
o
DC
C
°
0
0
CD
DC CDi 0
CD
N--RCC0 CD
_D CD
0
N
N-~
6
w
CD CD
-o
-
N
-
.
i
sg
C
ze
A11
N
-
DC N-
c
N
-
N
CD
-CN
"
° _
N
N
-
N-4 DC - - -
L
g
0
M
N
0
D
N
CDw N
0M C
_
D C
10 CD
0
N
NN
O
tL
t
N
0
CD D
C
0C CD
0 CDD
CDCDDCDCCDCD1
2
0
--
C
M
C°
C
t
I
N-
N-
-
D
Z-O CD) CD
°^
_7
UD
0
CD
t
C
r-
A.
0
D0
N
N
0
N
t
St
IQ
0
-
EL-=
_li
D
0
C
- -,.-,o,,
N
r-
IN
I
.
C
c
0
_D
On
-
N
CD
TableA5: Topfractilesincomessharesin totalincomein urbanChina,1986-2001
household
distribution
P90-100
(1)
P95-100
(2)
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
18.7%
18.5%
18.9%
19.7%
19.7%
19.8%
21.1%
22.6%
23.4%
23.0%
23.3%
23.8%
23.8%
24.2%
24.5%
24.9%
10.6%
10.5%
10.8%
11.6%
11.6%
11.7%
12.7%
13.7%
14.3%
14.0%
14.2%
14.5%
14.6%
14.9%
14.9%
15.3%
individual
distribution
P90-100
(2)
P95-100
(3)
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
17.4%
17.8%
19.2%
19.7%
19.3%
19.5%
20.6%
22.6%
23.6%
23.3%
24.0%
24.8%
24.7%
24.9%
25.5%
25.9%
9.8%
10.0%
11.1%
11.7%
11.4%
11.6%
12.5%
13.8%
14.4%
14.3%
14.8%
15.3%
15.2%
15.4%
15.8%
16.1%
P99-100 P99,5-100 P99,9-100
(3)
(4)
(5)
2.8%
2.7%
3.0%
3.3%
3.3%
3.5%
4.0%
4.2%
4.4%
4.2%
4.3%
4.5%
4.6%
4.7%
4.6%
4.8%
1.6%
1.6%
1.7%
1.9%
1.9%
2.1%
2.4%
2.5%
2.5%
2.4%
2.5%
2.7%
2.8%
2.8%
2.7%
2.9%
0.5%
0.5%
0.4%
0.5%
0.5%
0.6%
0.7%
0.8%
0.7%
0.7%
0.7%
0.7%
0.8%
0.8%
0.8%
0.9%
P99-100 P99,5-100 P99,9-100
(4)
(5)
(6)
2.6%
2.7%
3.3%
3.4%
3.3%
3.4%
4.0%
4.3%
4.4%
4.4%
4.7%
4.9%
4.8%
4.8%
5.0%
5.1%
1.5%
1.5%
2.0%
2.0%
1.9%
2.0%
2.4%
2.6%
2.6%
2.6%
2.8%
2.9%
2.9%
2.8%
3.1%
3.1%
0.5%
0.5%
0.6%
0.6%
0.6%
0.6%
0.8%
0.7%
0.8%
0.7%
0.9%
0.9%
0.9%
0.9%
1.0%
1.0%
P90-95
(6)
P95-99
(7)
8.1%
7.9%
8.1%
8.2%
8.2%
8.1%
8.3%
8.8%
9.1%
9.0%
9.1%
9.3%
9.2%
9.3%
9.5%
9.6%
7.8%
7.8%
7.9%
8.3%
8.2%
8.3%
8.8%
9.5%
9.9%
9.8%
9.9%
10.0%
10.0%
10.2%
10.3%
10.5%
P90-95
(8)
P95-99
(9)
7.6%
7.8%
8.1%
8.0%
8.0%
7.9%
8.1%
8.8%
9.2%
9.0%
9.2%
9.5%
9.5%
9.6%
9.7%
9.8%
7.2%
7.3%
7.7%
8.3%
8.0%
8.2%
8.6%
9.5%
10.0%
9.9%
10.1%
10.4%
10.4%
10.6%
10.8%
11.0%
P99-99,5 P99,5-99,9 P99,9-100
(8)
(9)
(10)
1.2%
1.2%
1.3%
1.4%
1.4%
1.4%
1.6%
1.7%
1.8%
1.7%
1.8%
1.8%
1.8%
1.9%
1.9%
1.9%
1.1%
1.1%
1.2%
1.4%
1.4%
1.4%
1.6%
1.8%
1.8%
1.8%
1.8%
2.0%
2.0%
2.0%
1.9%
2.0%
0.5%
0.5%
0.4%
0.5%
0.5%
0.6%
0.7%
0.8%
0.7%
0.7%
0.7%
0.7%
0.8%
0.8%
0.8%
0.9%
P99-99,5 P99,5-99,9 P99,9-100
(10)
(11)
(12)
1.1%
1.2%
1.3%
1.4%
1.4%
1.4%
1.6%
1.8%
1.8%
1.8%
1.9%
2.0%
2.0%
1.9%
2.0%
2.0%
1.0%
1.0%
1.4%
1.5%
1.4%
1.4%
1.6%
1.8%
1.8%
1.9%
2.0%
2.0%
2.0%
2.0%
2.1%
2.1%
0.5%
0.5%
0.6%
0.6%
0.6%
0.6%
0.8%
0.7%
0.8%
0.7%
0.9%
0.9%
0.9%
0.9%
1.0%
1.0%
Source:Authors'computations
basedontop fractilesincomeslevelsreportedon TablesA2 and A3
132
-
Qm
CD
.-I
0
-
0006
666666666666
O
.
N N
O
O
0
0
0
0
0
0
0
-
-
1-
-
U~0q (P N'
N
C
(2
oE
MD
,o
.0
rD
)-
(
0.
:3-
O
d,
lyCL
N
OC-
O
O
N
ODLo
-
-i
0)
t
N
t
-
0) N
1
,ob O--N~
I
-
-
C- -0
N1 1
O
0
O0
0
)
0
~
C
o6
C
w
(
C
C'O
t
C
0
r19
N
') !t
)
t-
(NCO
"I'
-
0
0~ 02 c-N
1
t
N
M
CO
N
0
N
(0 0) 0
- 11
Ot
t (-
( CD (
N"
Q
t
N
(0
ci)
E 5 ) -E
C)*0)
0
C
ON
wD
rN
-o
W ._ °
E
cN
0
0Co Cj
o
0 N
6
OOC'
a-
(0
COC-N)
(0(0(0(0( C-0)
CO)
J
0
C
-
-(§ 0
'~
-CN
NC')C
) (0
C)CDC0(00)
(CO0
)
0)000
C C'°
'-0).
Co
O
6a-
c\ cq r,:o
c~ c~ op~
~o
)
n000e
Iz)
In
sJ
eD
h-I
eD
0(000
sc
s
~
~
N
N0
.0
U
C
0
d,
aw
O
°
0
°
CL
N
N
C
° O
°
o
(
0
CO
0
N
~ci566O0NC
r)
W
C'
000
0
0
Ui
(0
N
(0
04
CO (0 V
Ol0N) t
o
L~
22 °0
.)
u
R
6K
C
0o
0q
0
oS
0cS
0o
R
CO (0
7
u? CO
c°
(0
(R
(0
0GF
0
Otot
CO
(0
O1=
(0oD
(Pqo=
9=
01
Cu °
C
(
0=
a'
N
C-
._
OCC)
0
E
x
o-
X
0
O-O7
00000
C5
CL
0
O O
O66666666
0O 0…O N
No
O O
0)
(D
N
-
NO
O
O
O
OO
-
0C?
N
N 0
X
O
it
(0
0
?0
OO
PC~
C
LO to
C -ui=
CD
<ccU) X oiJ
-o
E
0
A
C
0O
01
C
O
O
O
O
O
O
O
N0
JC~(~(
o
O
6Q
6t
O
O
C-
O
00
0
C
6
O
C
N
(0 CO
N C
cc
-,
0)
c.
o '
C.0
cci
o'O
m
]N
0O 0o 0C-
0
N
N
C1
0) ( 0
,
N 0
°N
0)
N
(0
N
O
-
N
ID
(
0,
c0U)1D0.
N
0 (0
(0
e
.0'
L)
u>
-5
8QCC
'
c
0
0
O ° (0
0
(0
oooO6oc
.oo
_=
(
0
(C
-
I_
o
C
6 ~ a)
co
o
(
(m
c.0.0.
EjEE~
c
D0N O
0 C) q
04
-
)D OOl
N
(O
N
o o~o o c) 6
~
-
0 0
C'
0
0)
-E0
0l )
ci)
)X
oEQ
.0
(0
u
coi~.
(0 N't
w
0.
2.
0
(
0(0
oocD
I-
0
N
(
0
(0
q
N'Itm-
---
N
O
N
0
0
O
CO
(
-
(0
-
(0
0)
C- 0
N
0
N
(0
s
(0
--
(0
N
(O
(0
N
C
(0
0.2 00 D
e
-
C
C ')
__,.
E
-
O
q
6~~~N
c
(q
ci
ci
g~~~~~"
7
a-R
2D
m EuO
0
N
(0
(0(0(00
C- (0
0)
0
)
0)
0 N
)0))
C
t
(
0)
0)
0)
0
i
c
2N
N
N
N
N
(0
-
(
0)
0
0)
0)
0)
oS
cI c
6oSI-
)00)00)0)0)0
ci
oo
0)
ci~
ci
N
0
CO
00
ci
N
C
N
N
"
N
-
a
O
C
t
0
N
0
00
q
Lo
t
t
( C
00
N
0
N
0
r-
I
a
o
o
-4
(0
(0
(0
-
(0
0)
C
0
0
N
0
C
N (\
(
U
000
N
?
-
__C
00
D
ccipo