Advertisement

Inequalities and Access

  • A. S. BhallaEmail author
  • Dan Luo
Chapter

Abstract

This chapter shows that while poverty has been reduced, income inequality has become worse in both China and India, which may be explained partly by very rapid economic growth during the past two decades. Non-income inequalities in the form of lack of access to education, health services and other public services have more adversely affected minorities than the majority population, which poses serious social and political problems. China has been more successful than India in ensuring access of minorities to education and health services. For example, it has practically removed child malnutrition which remains a serious problem in India.

Chapter  3 discussed rural and urban poverty which has been declining in both China and India. But income and other forms of inequality, the subject of this chapter, have been increasing in both countries. In India, urban inequality increased much more than rural between 1993–1994 and 2009–2010 (see Table 4.5). 1 But the situation in particular states is rather different. For example, in Jammu and Kashmir with 68% Muslim population, the rural Gini coefficient rose from 0.22 in 2004–2005 to 0.31 in 2011–2012. Similarly, urban Gini rose from 0.25 to 0.28 during the same period. Rural and urban inequality rose significantly in Assam and Kerala which also have substantial Muslim population (estimates supplied by Vishal More, New Delhi). Our main concern in this chapter is to empirically examine minority-majority income and non-income inequalities.

We use the Chinese Academy of Social Sciences (CASS) household survey data for 1988, 1995 and 2002 for China and the National Sample Survey Organization (NSSO) data (1993–1994, 2004–2005 and 2011–2012) for India to estimate minority–majority income disparities (Gini coefficients).

Methodology for Estimating Income Inequality

The Gini coefficient is widely used to measure income inequality among individuals or households. It can be decomposed in two different ways. First, we can divide the entire sample into different sub-groups according to such criteria as sex, region or ethnicity. The Gini coefficient for the whole population can be represented by the aggregation of three components: (a) an intra-class component arising from income variations within each class; (b) an inter-class component arising from differentials of mean incomes between classes; and (c) an overlapped component arising from the fact that poor people in a high-income class may be worse off than rich people in a low-income class. Second, if per capita income of the population can be divided into different sources, such as wage income and non-wage income, then the Gini coefficient can also be decomposed in the same manner.

Several formulae have been applied to calculate the Gini coefficient, which is derived mainly by calculating the area between the Lorenz curve and the diagonal line of a unit square. Pyatt (1976) used the concept of game theory and defined it as the ratio of the expected gain of a randomly selected individual in the population to the average income of the entire sample. The novelty of his approach is that it can be exactly decomposed into three separable components when the sample is divided into different classes. However, these earlier methodologies are quite complicated and rather impractical for empirical studies, as they use cumbersome techniques such as matrix algebra or covariance.

Yao (1999) developed a simple estimator of the Gini coefficient and presented a systematic procedure of decomposition by population class and income sources. With this method, all calculations can be done in a logically programmed spreadsheet without using any matrix algebra, integration, regression or covariance. This formula and decomposition method can be used for both individual data and evenly or unevenly grouped data. Below, we adopt the Yao approach, as it is suitable for analysing the household data at our disposal.

The basic equation for estimating the Gini coefficient of the whole population can be expressed as:
$$ \begin{array}{*{20}c} {G = 1 - 2\sum\limits_{i = 1}^{n} {B_{i} = 1 - \sum\limits_{i = 1}^{n} {p_{i} \,(2Q_{i} - w_{i} )} } } \\ {Q_{i} = \sum\limits_{k = 1}^{i} {w_{k} ,\;{\text{is}}\;{\text{cumulative}}\;{\text{income}}\;{\text{share}}\;{\text{up}}\;{\text{to}}\;i} } \\ \end{array} $$
(4.1)
in which the population is divided into n income groups, w i , m i and p i are the income share, per capita mean income and relative population frequency of the ith group, respectively. The sums of p i and w i are all equal to unity and both have to be strictly arranged following the ascending order of per capita incomes m i .
After deriving the Gini coefficient of the whole sample, one can further divide it into three components: intra-class (G A ), inter-class (G B ) and overlapped (G o ) (Eq. 4.2; Pyatt 1976). Here, the entire population is divided according to ethnicity; that is, Han and particular minorities:
$$ G = G_{A} + G_{B} + G_{o} $$
(4.2)
Equation 4.3 can be used to derive G B :
$$ \begin{array}{*{20}l} {G_{B} = 1 - 2\sum\limits_{I = 1}^{S} {B_{I} = 1 - \sum\limits_{I = 1}^{S} {P_{I} \,(2Q_{I} - w_{I} )} } } \\ {Q_{I} = \sum\limits_{K = 1}^{I} {w_{k} ,\;{\text{is}}\;{\text{the}}\;{\text{cumulative}}\;{\text{income}}\;{\text{share}}\;{\text{up}}\;{\text{to}}\;I.} } \\ \end{array} $$
(4.3)

S denotes the number of population classes, P I and w I are the population and income shares of the I th class (I = 1, 2, …, S) in the population. The format and explanation of Eq. 4.3 is similar to Eq. 4.1, with the exception that p I and w I in this case have to be sorted in the ascending order of class mean incomes m I .

G A can be calculated in Eq. 4.4 as follows:
$$ G_{A} = \sum\limits_{I = 1}^{S} {w_{I} p_{I} G_{I} } $$
(4.4)
where G I denotes the Gini coefficient of the I th sub-population.

The last component G o can be deduced by subtracting G A and G B from G. Recent experiments suggest that G o can be directly obtained from Eq. 4.1 if all the elements in the equation are sorted out by class mean incomes (first key) in the ascending order by household or group per capita incomes (second key) to obtain a concentration coefficient, denoted as G′ (Yao 1999). The difference between G and G′ equals G o .

We take decomposition one step further by decomposing the Gini coefficient of different population classes by income sources. The contribution of different income sources to inequality is quantified. We explore whether certain income sources influence inequality between minority and non-minority to the same extent. The basic equation for decomposing the Gini coefficient according to income sources is presented as:
$$ \begin{array}{*{20}l} {C_{f} = 1 - 2\sum\limits_{i = 1}^{n} {B_{fi} = 1 - \sum\limits_{i = 1}^{n} {p_{i} \,(2Q_{fi} - w_{fi} )} } } \\ {Q_{fi} = \sum\limits_{k = 1}^{i} {w_{fk} ,\;{\text{is}}\;{\text{cumulative}}\;{\text{income}}\;{\text{share}}\;{\text{of}}\;{\text{source}}\;f\;{\text{up}}\;{\text{to}}\;i.} } \\ \end{array} $$
(4.5)
G f is the Gini coefficient of a source of income f (f = 1, 2, …, F) and C f is the concentration ratio of source f. u f and u represent the means of source income f and the total income and w f  = u f /u measures the share of source income f in total income. The meaning of p i remains the same. Therefore, m i s w fi  = p i m fi /u f is the income share of i-th household in total source of income f. Again, in Eq. 4.5, p i s and w fi s must be sorted out according to the ascending order of per capita incomes m i s to derive C f and when calculating the Gini coefficients of per capita sources of incomes, G f , all variables in Eq. 4.5 have to be re-sorted according to the ascending order of per capita factor incomes, m fi . With S calculated C f s, G can be decomposed into its source components using Eq. 4.6:
$$ G = \sum\limits_{f = 1}^{F} {w_{f} C_{f} } $$
(4.6)
where F is the number of factor incomes and w f is the share of source f in total income. In other words, the Gini coefficient that measures total income inequality is the weighted average of the concentration coefficients of all income sources.

Empirical Evidence of Income Inequality

China

We use the CASS household survey data for 1988, 1995 and 2002 to illustrate the application of the above methodology. These surveys provide detailed information on the distribution of personal income in rural and urban areas for a number of provinces. 2 Table 4.1 summarizes basic features of the data.
Table 4.1

China: basic features of CASS household surveys, 1988, 1995 and 2002

 

1988

1995

2002

List of provinces

Beijing, Tianjin, Hebei, Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Shanghai, Jiangsu, Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Sichuan, Shaanxi, Gansu, Qinghai, Ningxia, Guizhou, Yunnan

Beijing, Heibei, Shanxi, Liaoning, Jilin, Jiangsu, Zhejiang, Anhui, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu

Beijing, Hebei, Shanxi, Liaoning, Jilin, Jiangsu, Zhejiang, Anhui, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Xinjiang

Rural sample

Number of households

10,258

7,998

9,200

Number of persons

51,352

34,739

37,969

Average household size

5.006

4.343

4.127

Number of provinces

29

19

21

Urban residents sample

Number of households

9,009

6,931

6,835

Number of persons

31,827

21,698

20,632

Average household size

3.533

3.131

3.018

Number of provinces

10

11

11

Urban migrants sample

Number of households

n/a

n/a

2,000

Number of persons

n/a

n/a

5,318

Average household size

n/a

n/a

2,659

Number of provinces

n/a

n/a

11

Source CASS (1988, 1995, 2002)

Table 4.2 summarizes the average rural income and Gini coefficients for urban and rural areas in China. Provinces are presented in descending order of average rural income. Incomes in provinces with a large minority population are generally at the bottom quintile and all are below the national average of 2,695 yuan per head. Meanwhile, the Gini coefficients of poorer provinces are higher than those for other more advanced provinces.
Table 4.2

Ranking of Chinese provinces by per capita rural income and Gini coefficient, 2002

Province

Average rural income per capita in descending order (yuan)

Urban Gini

Rural Gini

Integrated Gini

Zhejiang

5,074.64

0.3718

Beijing

4,932.56

0.2189

0.3642

0.2985

Jiangsu

4,564.99

0.3147

0.3139

0.3602

Guangdong

4,416.22

0.3364

0.3223

0.4336

Shandong

3,261.33

0.3583

Hebei

2,742.35

0.3068

Liaoning

2,659.59

0.2735

0.3518

0.3041

Hubei

2,560.72

0.2466

0.2906

0.3964

Jilin

2,483.95

0.3106

Jiangxi

2,416.24

0.2857

Chongqing

2,334

0.2882

0.2579

0.4277

Sichuan

2,319.76

0.3339

0.2667

0.4071

Hunan

2,300.17

0.3225

Henan

2,299.14

0.3041

0.2785

0.4025

Xinjiang

2,252.55

0.3403

Shanxi

2,249.35

0.3093

0.2888

0.4251

Anhui

2,157.29

0.2907

0.2724

0.4167

Guangxi

1,770.99

0.2298

Gansu

1,745.74

0.2567

0.3465

0.47

Yunnan

1,667.71

0.3034

0.4919

Shaanxi

1,641.52

0.316

Guizhou

1,441.44

0.3107

Source Based on CASS (2002)

Table 4.3 presents the Gini coefficients separately for seven Chinese provinces with a sizeable minority population. In Guizhou, Liaoning, Hunan and Jilin, these coefficients increased for the Han majority between 1995 and 2002, suggesting that income inequality increased during this period. In 1995, the average Gini coefficients for the Han and minority populations were 0.288 and 0.291, respectively, indicating that the Han incomes were more evenly distributed. However, in 2002, these coefficients rose to 0.342 and 0.292, respectively. An increase in the Gini ratio for the Han population was much faster than that for the minority. It grew by almost 20% in 7 years.
Table 4.3

 Gini coefficients for Chinese provinces with more than 9% minority population, 1995, 2002

Province

Year

Total Gini

Majority Han

Minority

   

Gini

Income (%)

Population (%)

 

Gini

Income (%)

Population (%)

Xinjiang

1995

Minority

(59.4%)

2002

0.333

0.309

28.0

17.1

Uygur

0.297

66.3

77.4

      

Hui

0.294

5.7

5.5

Guangxi

1995

Minority

(38.4%)

2002

0.35

0.344

73.1

63.2

Chuang

0.3

25.9

35.5

      

Others

0.268

1.0

1.4

Guizhou

1995

0.283

0.251

45.9

49.3

Minority

0.308

54.1

50.7

(37.8%)

2002

0.310

0.333

56.6

56.2

Miao

0.310

22.1

22.9

      

Others

0.294

21.4

21.0

Yunnan

1995

0.321

0.333

50.0

48.8

Minority

0.309

50.0

51.2

(33.4%)

2002

0.280

0.317

31.6

34.2

Yi

0.297

15.1

16.7

      

Hui

0.181

3.4

3.7

      

Others

0.24

49.9

45.3

Liaoning

1995

0.305

0.296

58.1

63.6

Minority

0.303

41.9

36.4

(16.1%)

2002

0.402

0.399

47.6

55.0

Manchu

0.379

44.5

35.4

      

Others

0.362

7.8

9.6

Hunan

1995

0.282

0.268

88.8

82.1

Minority

0.208

11.2

17.9

(10.1%)

2002

0.379

0.371

86.2

81.4

Miao

0.236

2.0

4.3

      

Others

0.369

11.7

14.2

Jilin

1995

0.292

0.289

94.4

94.0

Minority

0.327

5.6

6.0

(9.2%)

2002

0.316

0.318

96.3

95.3

Manchu

0.261

3.7

4.7

Source Based on CASS (1995, 2002)

Notes Figures in brackets under the name of the province represent the percentage of minority population based on the 2000 population census. The 1995 data used one single figure to represent minority population. It did not separate it into different nationalities. Income% = income share of total income; pop% = population share of total population

In Xinjiang, there is no difference in the Gini coefficients for the Uygur and Hui Muslims, suggesting similar income inequality. However, in Yunnan (home to several ethnic minorities) the ratio for the Yi is much higher than that for the Hui. Also in Guizhou, inter-minority inequality is wide: the Miao have a higher ratio than ‘other’ minorities (see Table 4.3).

The estimates for Xinjiang are based on rural data as there was no urban survey undertaken by CASS. Similar inequality between different ethnic minorities may be due to their access to similar sizes of land holdings. As discussed in Chap.  6, Xinjiang is not a poor province. It is resource rich and is engaged in commercial agriculture, not subsistence farming. In this province, rural–urban income disparities are known to be smaller than in many other provinces. 3

Table 4.4 presents the Gini coefficients and their components. In 1995, the average Gini coefficient was 0.297, while the average of its components, intra-class Gini (Ga), inter-class Gini (Gb) and overlapped Gini were 0.187, 0.061, 0.090, respectively, accounting for about 55.1, 17.6 and 27.3% of the overall Gini coefficient. In 2002, the contribution of the three components remained roughly the same, with Ga increasing slightly to 62.0% and Gb decreasing a little to 11.7%.
Table 4.4

Decomposition of the Gini coefficients for Chinese provinces with minority population, 1995 and 2002

Province

Years

Gini

Decomposition of Gini

Contribution to Gini (%)

   

Intra-class Ga

Inter-class Gb

Overlap Go

Intra-class Ga

Inter-class Gb

Overlap Go

Xinjiang

1995

2002

0.333

0.168

0.116

0.049

50.40

34.90

14.70

Guangxi

1995

2002

0.350

0.187

0.100

0.064

53.30

28.44

18.27

Guizhou

1995

0.283

0.141

0.034

0.108

49.92

12.00

38.09

2002

0.310

0.135

0.009

0.166

43.40

2.97

53.63

Yunnan

1995

0.321

0.160

0.012

0.150

49.86

3.62

46.53

2002

0.280

0.096

0.048

0.136

34.42

16.97

48.61

Liaoning

1995

0.305

0.156

0.055

0.094

51.13

18.03

30.84

2002

0.402

0.167

0.094

0.140

41.56

23.49

34.95

Hunan

1995

0.282

0.200

0.067

0.015

70.95

23.62

5.43

2002

0.379

0.267

0.050

0.062

70.45

13.19

16.37

Jilin

1995

0.292

0.258

0.004

0.031

88.28

1.21

10.51

2002

0.316

0.292

0.010

0.015

92.22

3.08

4.69

Source Based on CASS (1995, 2002)

The decomposition of the Gini coefficients suggests that the major source of inequality comes from the groups themselves. Ga, which measures the income variations within each class, accounts for more than 50% of the overall Gini coefficient with only a few exceptions (Guizhou, Yunnan and Liaoning for 2002). In other words, income disparity between the Han and minority populations is not that large (Gb); instead, inequalities within the Han population and the minority populations contribute significantly to the overall Gini coefficient.

We compared our Gini estimates using the Yao (1999) method with those of Khan and Riskin (2005). 4 Both methods gave roughly the same ranking of provinces (for example, for the rural areas, both methods estimated the lowest Gini ratio for Guangxi and the highest Gini ratio for Zhejiang and Beijing).

India

Income inequality among Muslims needs to be compared with the incidence of poverty discussed in Chap.  3. We noted in Chap.  3 that rural and urban poverty has declined for Muslims (as for others) since 1993–1994. However, the picture regarding income (consumption) inequality is not necessarily similar. Urban income inequality among Muslims was lower (but their poverty incidence was higher) than that among Hindus (see Table 4.6). Increases in the rural and urban Gini coefficients from 1993–1994 to 2009–2010 were marginal. Between 1993–1994 and 2004–2005, the increase was relatively higher than that between 2004–2005 and 2009–2010. Among the religious groups, these ratios were lower for Muslims than for Hindus for all the 3 years considered. The category of ‘other’ religious groups which included smaller minorities (Christians, Sikhs and Jains) showed the highest Gini ratios. The higher increase in Gini coefficients is likely to be explained by an increase in the monthly per capita consumption of the higher-income groups. Among the social groups, the scheduled tribes (STs) had higher rural and urban Gini ratios than the scheduled castes (SCs). Gini coefficients for rural and urban areas, and for Muslims and Hindus are compared for two categories of state, those with a Muslim population of 25% or higher, and those with a Muslim population of lower than 18% (see Table 4.6).

A number of conclusions can be drawn from the Gini estimates:

  1. 1.

    For both periods, 1999–2000 and 2004–2005, rural and urban income inequality among Muslims is generally lower than that among the Hindu majority.

     
  2. 2.

    Results for states show wide variations. In 1999–2000, both rural and urban inequality was lower in Bihar, Haryana, Jammu and Kashmir, Madhya Pradesh, Maharashtra and Punjab. This was also the case in 2004–2005, but only for Bihar, and Jammu and Kashmir. In other states, urban inequality was higher; for example, in Assam, Haryana and Maharashtra. Rural inequality was higher for Muslims in Gujarat in 1999–2000 but it was lower in 2004–2005. In 2004–2005, rural inequality was higher for Muslims in Kerala, although it was lower in 1999–2000. In Assam, the rural Gini coefficients remained unchanged for both Hindus and Muslims. A note of caution is in order in making the above comparisons over time. The NSS estimates for 1999–2000 are known to have lost comparability with those for other years ‘due to the problem of using two reference periods for a few items for the same household. There was no unanimity in data adjustments, which may vitiate comparisons over time (private communication with Professor Radhakrishna, Hyderabad). Also for some states results may not be reliable on account of small sample size.

     
  3. 3.

    The estimate of the urban Gini ratio for Tamil Nadu for 1999–2000 is very odd. It shows exceptionally high urban income inequality for Muslims, 0.61 compared with 0.34 for Hindus. John and Mutatkar (2005, p. 1341) observe that the ‘HCR for Muslims is also more than 10% points higher than that for Hindus in spite of their average MPCE (monthly per capita expenditure) being higher than that of Hindus by 40 per cent’. They explain the high HCR for the Muslims by the unusually high income (consumption) inequality among them. But how does one explain a decline in the urban Gini coefficient for Muslims from 0.61 in 1999 to 0.32 in 2004? We believe that there may simply be a statistical error on account of a small sample for Muslims.

     
  4. 4.

    There is no significant difference in the size of the Gini coefficients for the Muslims in both categories of states. For example, Jammu and Kashmir with a Muslim population of over 68% had a Muslim rural Gini coefficient of 0.14 and Punjab with a Muslim population of 2% 0.15, suggesting low rural income inequality in both states.

     

In Table 4.6, the Gini coefficients are presented only for the Muslim religious minority in India. However, there are much smaller religious minorities (Sikhs and Christians, discussed in Chap.  2) for which the sample size is very small. They were left out of our estimates for 2004. John and Mutatkar (2005) have estimated rural and urban Gini coefficients for 1999 for Sikhs and Christians for a selected number of Indian states. For Assam, both rural and urban Gini coefficients for the Christians were lower than those for Muslims and Hindus, suggesting lower inequality. However, in Bihar their rural and urban inequality is higher. For the Sikhs, higher rural Gini ratios in Haryana and Punjab show rural inequality as similar to that among the Hindu majority. However, the situation regarding urban inequality is different. In Haryana, Himachal Pradesh, and Jammu and Kashmir, the urban Gini coefficient for the Sikhs is lower than that for the Hindus, suggesting less income inequality.

A comparison with Table 4.5 on the overall Gini coefficients shows that the Muslim rates are generally lower than the overall rates. Relatively low urban income inequality among Muslims may be because the Muslim community is not caste-ridden. Muslims form a very small proportion of the SCs and STs. Muslim castes account for nearly 41% of the OBCs and show slightly higher overall inequality (than that in other socio-religious groups) in both rural and urban areas (GOI 2006, p. 212).
Table 4.5

India: rural and urban Gini coefficients by socio-religious groups, 1993–1994, 2004–2005 and 2009–2010

Social and religious groups

1993–1994

2004–2005

2009–2010

Rural

All

0.284

0.299

0.306

Scheduled tribes (STs)

0.265

0.269

0.28

Scheduled castes (SCs)

0.253

0.260

0.258

Others (including other backward classes)

0.287

0.304

0.316

Hindus

0.281

0.294

0.298

Muslims

0.273

0.289

0.277

Others (including Christians, Sikhs, Jains)

0.319

0.345

0.396

Urban

All

0.345

0.376

0.401

Scheduled tribes (STs)

0.311

0.341

0.387

Scheduled castes (SCs)

0.303

0.316

0.334

Others (including other backward classes)

0.345

0.378

0.405

Hindus

0.341

0.373

0.396

Muslims

0.301

0.336

0.377

Others (including Christians, Sikhs, Jains)

0.386

0.373

0.413

Source Thorat and Dubey (2012)

Sources of income (consumption) inequality in India include a complex mixture of factors such as economic, caste, historical, geographical, gender and educational differences. Inequalities may be further explained by differences in land and other asset ownership, and discrimination in the labour and credit markets. For example, rural income inequality may be traced to the ownership of land. The Muslim minority in India is known to be less attached to land with limited ownership and small landholdings. Apart from land, other factors at work in rural areas may include the distribution of formal and informal employment, and availability of educational facilities and opportunities, occupational structures, and caste and social hierarchy (see Drèze and Gazdar 1997; Drèze and Sen 2013; Sengupta and Gazdar 1997).

In a quantitative analysis based on the National Sample Survey (NSS) urban data for employment and unemployment for 1987 (43rd Round) and 1999 (55th Round), Bhaumik and Chakrabarty (2010, p. 243) show that ‘between these two years, for each educational cohort, earnings differentials between SCs/STs and non-SC/ST castes had declined, while those between Hindus had increased’. This suggests that caste was less important than religion in determining earnings differentials. One important limitation of this analysis is the exclusion of self-employed people from the sample. Inter-religious results might be greater than shown above if Muslims figured prominently in the lower levels of income distribution.

A Comparative Perspective

The Gini coefficients for China and India show similar magnitudes for both rural and urban areas. The Chinese estimates are based on income data, whereas those for India are based on consumption data. In principle, this variation may reflect some differences. India does not have sufficiently long series for income data. That is why most estimates of rural and urban inequality are based on the (NSS) data on consumption.

A priori, consumption inequalities tend to be lower than income inequalities. The consumption Gini coefficients are generally lower than the income Gini coefficients. The fact that household consumption may exceed income (a case of dis-saving) at lower deciles of population may account for this. On the other hand, at higher deciles a higher proportion of income is saved.

Early overall estimates of the Gini coefficients for China and India for the 1950s to 1980s (see Bhalla 1995, pp. 162–165) show that, in the Chinese rural areas during the pre-reform period, rural and urban inequality remained unchanged. However, rural income inequality in India increased significantly between 1952 and 1978.

Using the Chinese Academy of Social Sciences (CASS) household survey data for 1995 and the Indian National Council for Applied Economic Research (NCAER) household survey data for 1994, Borooah et al. (2006) find that overall income inequality in China was slightly lower than in India. For both countries, the minority Gini coefficients were lower than those for the majority (ibid., Table 2). The decomposition of income inequality by different population groups shows (1) much greater between-group contribution to regional inequality in China than in India; and (2) greater influence of majority–minority differences in India on rural income inequality than in China. This latter conclusion may not be surprising, considering that Indian minorities (religious categories such as Muslims, Sikhs and Christians, and disadvantaged social groups constitute about 38% of the total population compared with only 9% in China.

More recent estimates for minority–majority income inequality (see Tables 4.5 and 4.6) show that between 1999 and 2004 in India, rural income inequality among the Muslims increased in Jammu and Kashmir, Kerala and West Bengal, the three states with a sizeable Muslim population. It also rose in poor states such as Madhya Pradesh and Uttar Pradesh. For the majority of the Hindu population, rural income inequality also rose during the same period.
Table 4.6

 Minority–majority consumption inequality in India , selected states, 1999–2000 and 2004–2005

State/religion

1999–2000 (NSS 55th Round)

2004–2005 (NSS 61st Round)

Rural

Urban

Rural

Urban

States with Muslim population 25% or higher

Jammu and Kashmir

Hindu

0.23

0.25

0.25

0.26

Muslim

0.14

0.19

0.24

0.22

Assam

Hindu

0.19

0.30

0.19

0.31

Muslim

0.20

0.37

0.20

0.34

Kerala

Hindu

0.29

0.32

0.37

0.44

Muslim

0.27

0.31

0.40

0.36

West Bengal

Hindu

0.23

0.33

0.28

0.38

Muslim

0.20

0.42

0.26

0.33

States with Muslim population below 18%

Uttar Pradesh

Hindu

0.25

0.33

0.29

0.37

Muslim

0.23

0.28

0.30

0.32

Bihar

Hindu

0.21

0.32

0.21

0.34

Muslim

0.20

0.27

0.19

0.24

Maharashtra

Hindu

0.26

0.34

0.32

0.36

Muslim

0.22

0.31

0.22

0.38

Gujarat

Hindu

0.24

0.29

0.28

0.30

Muslim

0.26

0.25

0.23

0.28

Tamil Nadu

Hindu

0.29

0.34

0.32

0.36

Muslim

0.28

0.61

0.27

0.32

Karnataka

    

Hindu

0.24

0.32

0.25

0.37

Muslim

0.26

0.28

0.25

0.33

Andhra Pradesh

Hindu

0.24

0.32

0.29

0.38

Muslim

0.21

0.27

0.29

0.29

Rajasthan

Hindu

0.21

0.29

0.25

0.38

Muslim

0.19

0.25

0.22

0.23

Madhya Pradesh

Hindu

0.24

0.32

0.26

0.38

Muslim

0.18

0.27

0.26

0.34

Haryana

Hindu

0.25

0.29

0.34

0.36

Muslim

0.17

0.15

0.20

0.42

Punjab

Hindu

0.25

0.27

0.30

0.42

Muslim

0.15

0.25

0.32

0.22

All India

Hindu

0.26

0.34

0.30

0.37

Muslim

0.24

0.33

0.29

0.34

Sources John and Mutatkar (2005) for 1999–2000; our estimates for 2004–2005 based on NSS 61st Round data

Some statistical differences of income inequality in China and India may simply be due to non-comparability of data. As we noted above, most recent estimates of the Gini coefficients in India are based on consumption per capita, whereas those in China are based on income per capita. Second, the Gini coefficient may not capture improvements in income distribution among the poor. As Khan (2008, p. 155) remarks, ‘a change in the Gini ratio caused by a change in income distribution at the upper end of the income scale leaves the welfare of the poor unchanged, just as an unchanged Gini ratio may hide an unfavorable change in distribution affecting the poor that is offset by a favourable change in distribution at the upper end of the income scale’. Differences in rural income inequality may also be due to differences in the ownership of land, as well as access to it. While there are no landless in rural China (although access to land may vary across regions), many Indian rural households are landless, especially among poor Muslims and SCs and STs.

Educational Access and Attainment

So far, we have considered income/consumption variables to explain income inequality. However, such non-income variables as literacy, education and health can also influence income inequality. How do the minorities fare in China and India with respect to these non-income variables? We discuss educational access below before turning to health access.

Access to higher and better education is linked directly to economic and social well-being. Several factors determine educational attainment: low education of parents or household head, per capita income, location, and ethnicity or minority status. Educational attainment is an investment in human capital which is a route to a higher living standard. A higher level of education commands higher earnings leading to higher incomes. The reverse relationship may also hold; that is, a rise in per capita income may enable an individual or household to enjoy greater access to education. Furthermore, as we shall discuss below, differentials in educational access and attainment can explain, inter alia, rural–urban, caste-based and minority–majority income inequalities. Educational gaps between different religious and social groups and communities can partly explain their marginalization and social exclusion.

China

We undertake regression analysis to determine the influence of the above factors on educational attainment. We also estimate the marginal effects which capture an increased (or decreased) probability that a child will complete four or more years of schooling, given a one-unit increase in the independent variable. The results, statistically significant for 1995, show that ethnicity has an important negative effect on educational attainment, as does household income. But the results for ethnicity are not significant for 2002 (see Table 4.7), which suggests that minority status is no longer important in determining the years of schooling received.
Table 4.7

Determinants of children’s education in rural China, 1995 and 2002

 

1995

2002

 

Logit regression

Marginal effects

Logit regression

Marginal effects

A. Minority status

−1.442***

−0.092***

−0.765***

−0.029***

(0.169)

(0.017)

(0.173)

(0.008)

B. Educational and other characteristics of household head

4 or more years of education

1.223***

0.069***

0.818**

0.034*

(0.211)

(0.017)

(0.321)

(0.018)

1–3 years of education

0.096

0.003

0.351

0.009

(0.274)

(0.009)

(0.409)

(0.009)

Working in agriculture

−0.154

−0.052

0.079

0.002

(0.189)

(0.006)

(0.159)

(0.005)

Male

−1.043**

−0.024***

−0.293

−0.008

(0.418)

(0.006)

(0.412)

(0.009)

Communist Party member

0.413**

0.012**

0.233*

0.010*

(0.219)

(0.006)

(0.124)

(0.006)

C. Location and geography

Plains

0.424***

0.013***

(0.157)

(0.005)

Hilly region

  

  

Sub-urban

1.056***

0.049***

(0.198)

(0.013)

D. Per capita income

Log (income)

0.535***

0.0188***

0.090

0.003

(0.097)

(0.003)

(0.098)

(0.003)

Constant

−0.595

 

2.616***

 

(0.768)

 

(0.813)

 

Number of observations

4479

 

5468

 

Pseudo R-squared

0.15

 

0.11

 

Source Based on CASS (1995, 2002)

*, ** and *** means significant at 10, 5 and 1% level respectively

Note In the 2002 analysis, we did not include observations with less than RMB100 average income

The locational factors (whether a household is located in the plains, a hilly region, the suburbs, or a designated poverty region) and membership of the Communist Party have a weak influence on educational attainment. Location or geography is not a sufficient factor to identify or target those who need better education. This result is consistent with what Riskin (1994) describes as the ‘impoverished regions’ approach to poverty programmes and how it is increasingly ineffective in reaching the poor.

Some other studies (for example, Hannum 2002) have confirmed the role of poverty (low household incomes) in explaining educational disparities among ethnic minorities. A rural component of the 1992 National Sample Survey of the Situation of Chinese Children examined disparities in school enrolment which were associated with ethnicity, poverty and gender. The Hannum study concluded that, in general, the gender gap among minorities was less marked than among the Han. However, minority girls were more likely to drop out of school for reasons of poverty and the need to provide help within the household.

In China, minorities in rural areas generally have lower levels of education and schooling than the Han majority. Their rural and ethnic features put them at a double disadvantage. But in the more advanced urbanized regions (Beijing, Hebei and Tianjin), contrary to expectations, minority literacy rates are actually higher than those for the Han Chinese. This may be because it is the better-educated minority people who move to the urban areas, 5 which suggests that urban areas are inhabited by minority groups which are more successful and place greater value on education.

Economic factors may better explain higher education of minority people in some urban areas. These people might have originated in less-poor families with educated parents. Some minority groups (the Miao, for example) may attach greater importance to education than others. Knight and Song (1999, p. 132) argue that one reason for urban minority people to acquire more education ‘could be a process of self-selection whereby successful minority people have spread to cities throughout China’. Another plausible reason they give is affirmative action programmes in favour of minorities. If this were indeed the case, why would the rural minorities (who are also covered by preferential policies) attain less education than the Han? We believe that the explanation lies more in the overall urban bias of the Chinese authorities than in affirmative action (see below).

Of course, we should not ignore the fact that, although the minority literacy rate in urban or advanced areas was higher than that for the Han Chinese, the share of minorities in the total population of these areas was quite small.

There is gender disparity in adult literacy rates, especially in the five autonomous regions and poor Western provinces. But the female/male ratio in Xinjiang is very high and above the national average. The lowest female/male ratios are for Tibet, Guizhou and Yunnan.
Table 4.8

China: educational levels (6 years and above) by ethnicity, 2010 (%)

 

No schooling

Primary education

Secondary education

Tertiary education

Han Chinese

5.0

28.7

56.7

9.5

Tibetan

4.7

27.8

57.7

9.7

Yi

30.6

45.9

18.1

5.5

Qiang

14.3

53.8

28.1

3.8

Hui

7.0

42.2

42.4

8.3

Uygur

8.6

35.6

46.4

9.4

Mongolian

3.5

41.6

48.6

6.3

Source GOC China Census 2010

Secondary education = junior plus senior middle; Tertiary education = college plus university plus post-graduate

The educational attainment of different ethnic minorities varies a great deal (see Table 4.8). Mongolian and Tibetan minorities have the lowest shares of those with ‘no schooling’, and Yi have the highest share. The Chinese government has made massive investments to provide schooling of Tibetans in Tibet and other regions which explains its small share. Yi, who are much poorer especially in Sichuan, are less educated than the Tibetans. Tibetan share of those with tertiary education (university and higher) is as high as that of the Han Chinese.

The situation of the Yi was even worse: only a little over 1% had tertiary education. The Hui and Mongolian minorities outperform the Han Chinese: over 5% of Mongolians had tertiary education compared with about 4% Han Chinese. Nearly 55% of Mongolians had secondary education compared with 53% Han Chinese.

In Chap.  1, we noted that the ethnic minorities in China are concentrated mainly in the Western region and the five autonomous regions, which are also poorer than the Eastern and Central regions. Minority literacy rates in many poor Chinese provinces in the south-west and west of China are very low and below the national average. This concentration of illiteracy mainly among minorities partly explains their poverty and backwardness.

Data on ethnic minorities from China’s Ethnic Statistical Yearbook enables us to estimate (1) ethnic minority enrolments from 1995 to 2006 at primary, secondary and higher levels of education; and (2) the number of ethnic teachers for the same period. The first indicator shows that minority shares of enrolments at the primary level rose very modestly, especially from 2002 onwards (see Fig. 4.1). The pattern of enrolment at secondary level was somewhat similar except for a sudden spurt between 2002 and 2003. Enrolment shares at the tertiary level actually declined before starting to rise in 2004.
Fig. 4.1

China: minority shares in educational enrolments, 1995–2006.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

The number of ethnic minority teachers at the primary level rose until 2001. Between 2001 and 2006, it remained constant (see Fig. 4.2), which may suggest one or both of two things: that the number of primary schools during this period hardly changed, or that the number of ethnic primary schools (which would hire mainly ethnic teachers) remained unchanged. However, the number of ethnic teachers in regular secondary schools between 1995 and 2006 rose steadily. The number of ethnic teachers at institutions of higher education is quite small and rose rather modestly. The number of primary students from ethnic minorities rose from 1979 until 1997 but has been declining steadily since then (see Fig. 4.3).
Fig. 4.2

China: number of ethnic teachers at different educational levels, 1995–2006.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

Fig. 4.3

China: primary school students from ethnic minorities, 1979–2006.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

Minority students are generally at a disadvantage on account, inter alia, of language difficulties. Linguistic minority students who are not well-versed in Mandarin (Tibetans, for example) may particularly suffer while passing through the highly competitive school system. Students from primary schools have to pass an entrance examination in order to enter junior high and middle schools. Then, at the end of secondary schooling, they have to pass another entrance examination to enrol for university-level education. In theory, the Chinese Law of Regional Minority allows minority students to sit for examinations in their native languages. However, in practice this may rarely happen. The policy of awarding bonus points to minority students taking examinations in the Chinese language may help them advance to university education but rarely does it make up for their educational handicaps.

The CASS household survey data show that only 65% of the minority population has completed four or more years of schooling compared with 80% of the Han population. There are similar gaps for both lower-middle and upper-middle schools. Males among minorities have a much higher rate of educational attainment than females.

There is also the problem of higher drop-out rates among minority students. For example, less than 10% of the Tibetan children in primary school advance to secondary education. A number of factors may explain this situation. First, minority students and their parents may perceive that education would not help them much because of their ethnicity. Second, the school curriculum may be biased in favour of the Han and against minorities. In the interest of promoting national unity and identity, the Chinese authorities may not attach much importance to the inclusion of such subjects as religion and minority languages in the national curriculum. Third, low rates of minority enrolment may be explained by a lack of interest in bilingual education. The diversity of minority languages makes the spread of education in remote regions very difficult.

Below, we present a detailed case study of Tibet, one of the five autonomous regions of China.

Tibet: A Special Case

Tibet is an important autonomous region for political as well as economic reasons. During recent decades, its economic growth has been impressive, thanks partly to massive central subsidies, and large-scale construction and other developmental programmes. It is, therefore, interesting to study whether income and non-income inequalities in this predominantly minority region have narrowed or widened. Equally, it would be important to explore whether the educational attainment (including literacy) of local Tibetans has improved.

A Chinese scholar based in Hong Kong (Zang 2015, pp. 162–163) maintains that ‘Beijing has been unable to develop an effective policy towards its minority nationalities in the post-1978 era. In comparative perspective, the PRC did a better job in Mao’s China’. There were few ethnic riots in the pre-1978 period because the country was more egalitarian. In the post-Mao period, the benefits of economic growth have accrued disproportionately to the Han majority and often at the expense of ethnic minorities in Tibet and Xinjiang. Rural–urban income disparities in Tibet widened significantly between 1990 and 2014. However, they had been narrowing from 2000 until 2010, when the ratio of rural to urban incomes declined again (see Table 4.9). This was a period of relative peace and quiet in Tibet when few riots took place (see Chap.  8, Table  8.2).
Table 4.9

China: rural–urban income disparities in Tibet, 1990–2014

Year

Per capita net rural income (yuan) (1)

Per capita urban disposable income (yuan) (2)

Ratio of (1) as % of (2)

1990

582

1,613

36.1

1995

878

4,000

30.9

2000

1,331

6,448

20.6

2005

2,078

8,411

24.7

2010

4,138

14,980

27.6

2011

4,904

16,196

30.3

2012

5,719

18,028

31.7

2013

6,553

20,394

32.1

2014

7,359

22,016

33.4

Source GOC, Tibet Statistical Yearbook; China Statistical Yearbook

Literacy rates in Tibet are extremely low. Between 2001 and 2006, they were much lower than those in Gansu, Qinghai and Sichuan, the three provinces that are major sources of inward migration to Tibet. Thus, the migrant population enjoys an educational advantage over the local Tibetan population (Fischer 2014; see also Chap. 5).

Raising literacy rates in Tibet is particularly challenging because of the difficulties of recruiting teachers locally and of attracting Han teachers who may have no knowledge of the Tibetan language, or who may be unwilling to stay in Tibet given its severe climate and high altitude.

The 2010 Population Census of China enables a breakdown of educational attainment in Tibet and Xinjiang by city, town and rural areas (see Table 4.10). Female illiteracy was significantly higher than male illiteracy and females have lower educational attainment at primary and secondary levels. However, at senior secondary levels and above, it is quite close to that of males. It is interesting to note that the female illiteracy rate was higher in the city than the towns and rural areas. High female illiteracy rates in the city deny women access to the job market, and this exclusion from employment deprives them not only economic benefits, but also dignity and self-respect. But the average illiteracy rate is somewhat lower than the town rate and substantially lower than the rural rate as expected. The average educational attainment at primary level is below the average for the city and town but above the average for the rural areas. At the senior secondary school level and above, the city and town rates are substantially higher than the average rate, but the rural rate is much below the average.
Table 4.10

China: rates of educational attainment (6 years and above) by city, town and rural area in Tibet and Xinjiang, 2010

Education level

Average

City

Town

Rural area

Rate

F/M

Rate

F/M

Rate

F/M

Rate

F/M

Tibet

Illiterate

37.3

1.41

20.1

2.01

24.98

1.55

40.4

1.37

Primary school

42.8

0.80

31.99

1.03

35.02

0.89

44.7

0.78

Junior middle school

12.1

0.73

18.68

0.91

15.5

0.77

11.7

0.70

Senior secondary school and above

7.8

0.92

29.2

1.01

24.5

0.86

3.77

0.92

Xinjiang

Illiterate

3.4

1.19

2.37

1.27

2.58

1.18

3.71

1.19

Primary school

40.79

1.00

25.8

0.99

31.88

1.02

44.75

0.99

Junior middle school

41.1

0.91

29.0

0.95

33.75

0.94

44.34

0.90

Senior secondary school and above

14.7

1.04

42.76

1.16

31.79

1.03

7.19

0.94

Sources Compiled jointly by the Department of Population and Employment Compiled jointly by the Department of Population and Employment, National Bureau of Statistics (NBS) and the Department of Economy and Development, State Ethnic Affairs Commission of the People's Republic of China, 'Tabulation on Nationalities of the 2010 Population Census of China’'

F/M Female-to-male ratio

Tibet and Xinjiang are two large autonomous regions with large ethnic minority populations. A comparison of educational attainment in the two shows that literacy rates were much higher in Xinjiang than in Tibet. The literacy gaps are accompanied by gender gaps. Gender disparities (reflected in the F/M ratio) were much higher in Tibet, while they were much less significant in Xinjiang.

High illiteracy rates (or low literacy rates) in Tibet have a historical origin. Before China took control of Tibet in 1951, its educational situation was much worse than that of the rest of China. 6 A low starting point, poor natural conditions and very low population density were partly to blame for the government’s failure to raise educational profiles quickly. 7 The central government not only made efforts to establish a modern education system in Tibet by sending teachers from other provinces, but it also mobilized educational resources and facilities in other parts of China for the benefit of Tibetans. In 1974, it issued written instructions to other provinces to select teachers for secondary schools and colleges to work in Tibet. These teachers were to stay in Tibet for 2 years, after which they were to be replaced by others.

Recognizing the difficult climatic and socioeconomic conditions in Tibet noted above, in the mid-1980s the central government decided to request 19 other provinces to set up Tibetan schools and classes in Tibetan in their territories for Tibetan students. Three Tibetan schools were set up in Beijing, Chongqing and Lanzhou; Tibetan classes were also introduced in other provinces. In 1997, 7,000 Tibetan pupils were sent to other provinces to study (Iredale et al. 2001, p. 161). While some returned to Tibet, others went on to study further, or to take up jobs in other provinces. The number of Tibetan-language schools also increased from the late 1980s onwards (Upton 1999) . On the basis of fieldwork conducted in 1996, Upton (1999, p. 307) observes that ‘contrary to Western and Tibetan exile rhetoric’, the textbooks in schools ‘do contain a fair amount of material drawn from Tibetan sources and relevant to Tibetan cultural life in the broad sense’.

In more recent years, the educational indicators for Tibet have improved (see Table 4.11). Enrolment of school-going children and the number of primary-school graduates substantially increased between 2000 and 2014. Other educational indicators such as female teachers have also improved. However, high enrolments rates may conceal high drop-out rates, reflecting low quality of education, difficulty in passing Chinese language examinations and the poverty of Tibetan parents forcing children out of school. This situation is reflected in a significant decline in the proportion of graduates entering senior secondary schools between 2009 and 2014.
Table 4.11

China: educational indicators for Tibet, 2000–2014

Indicator

2000

2006

2009

2014

No. of primary school teachers (000)

13.2

16.0

18.7

37.8

Primary student enrolment (000)

313.8

329.5

305.2

526.7

No. of primary school graduates (000)

37.0

48.6

50.8

120.2

Enrolment of school-age children (%)

85.8

96.5

98.8

128.6

Graduates of primary schools entering senior secondary schools (%)

55.0

92.0

98.4

60.0

Graduates of junior secondary schools entering senior secondary schools (%)

82.5

42.5

55.2

92.2

Female students as % of all students

46.2

47.3

44.4

45.8

Female teachers as % of all teachers

42.6

48.4

47.7

51.8

Source GOC, Tibet Statistical Yearbook

Adult literacy rates increased rather modestly, from 54.5% in 2000 to 60.4 in 2009. Even earlier data show that, between 1990 and 1997, Tibet’s literacy rate rose from 31% (combined rate for the Han and minority population) to 46%, an increase of nearly 48%, which is quite significant.

Drèze and Sen (1995, p. 66) state that ‘literacy rates in Tibet are not only abysmally low (even lower than in the educationally backward states of North India), they also show little sign of significant improvement over time’. This claim is not fully justified, as is shown by the above discussion.

India

Educational disparities in India at both basic and higher education levels have their roots in income, caste and gender differences between religious groups and social communities. A comparison between different religious minorities in India shows that the Muslim community has consistently lower levels of mean years of schooling. On average, a Muslim child goes to school for only 4 years. Nearly 25% of Muslim children aged 6–14 have either never attended school, or have dropped out of it. Their drop-out rate is the highest at the primary and secondary levels (GOI 2006, pp. 56–62). At higher levels of education (university, for example), fewer Muslims are enrolled, and Muslim university graduates have difficulties in finding jobs. Unemployment rates are the highest among Muslim graduates.

In India, the overall rural literacy rates for the Muslim minority in 2011–2012 were only slightly lower than those for the majority Hindu population. But there were wide variations across states in Muslim and non-Muslim rural literacy rates (see Table 4.12). For example, in Assam, Bihar, Haryana, Uttar Pradesh and West Bengal the Muslim rates were lower than the Hindu rates. But in many other states such as Andhra Pradesh, Arunachal Pradesh, Chhattisgarh and Orissa, they were higher. The Muslim rural literacy rates improved significantly since the National Council of Applied Economic Research (NCAER) 2004–2005 household survey. In fact, 16 states and union territories registered higher rural rates for Muslims than for Hindus.

The rural literacy rates for Muslims are the highest in Kerala and the lowest in Bihar, Haryana and Rajasthan. The lowest rates are not necessarily in the poorest states, since Haryana is one of the richest Indian states. This may be explained more by caste discrimination than any economic factor. It is remarkable that Kerala, which also suffered from caste discrimination, has achieved the highest rural literacy rate for the SCs and STs throughout India. Ramachandran (1997, p. 274) observes that ‘the worst forms of untouchability in the country were practised in Kerala and the persecution of people of the oppressed castes took savage forms’. So, how did the situation change? Several factors may have been at work: public provisioning of health services by the state, the role of civic society, missionaries and the matrilineal system (women made exemplary progress in education and health) explaining a change in social attitudes in overcoming caste discrimination. The F/M ratios reflect inter-regional gender disparity among minorities (see Table 4.12), which is lowest in Kerala and Tamil Nadu, and the highest in Haryana and Rajasthan.
Table 4.12

India: rural and urban literacy rates by state, ethnicity, social groups and gender (7 years and above), 2011–2012

State

Rural

Urban

Rural

Urban

Hindu

Muslims

Hindu

Muslims

Scheduled castes/tribes

 

Total (%)

F/M

Total (%)

F/M

Total (%)

F/M

Total (%)

F/M

Total (%)

F/M

Total (%)

F/M

All India

70

0.76

68

0.80

87

0.88

78

0.86

64

0.73

80

0.81

Rajasthan

62

0.60

61

0.61

83

0.78

65

0.83

56

0.55

69

0.64

Bihar

66

0.70

60

0.70

84

0.82

74

0.96

53

0.62

70

0.76

Madhya Pradesh

68

0.75

66

0.77

86

0.86

80

0.89

61

0.73

77

0.80

Uttar Pradesh

66

0.69

59

0.74

83

0.86

63

0.79

60

0.64

71

0.77

Haryana

75

0.76

65

0.53

88

0.87

65

0.69

73

0.74

73

0.72

Himachal Pradesh

81

0.82

83

0.85

91

0.90

79

1.23

77

0.80

90

0.84

Punjab

82

0.87

60

0.86

85

0.88

79

0.85

68

0.87

72

0.75

West Bengal

75

0.82

65

0.85

90

0.91

76

0.92

68

0.80

84

0.87

Gujarat

69

0.77

81

0.78

88

0.91

88

0.88

64

0.72

80

0.77

Maharashtra

77

0.78

78

0.86

91

0.89

88

0.90

68

0.72

87

0.84

Andhra Pradesh

59

0.74

62

0.89

84

0.86

81

0.85

52

0.74

75

0.78

Karnataka

70

0.78

70

0.81

89

0.90

84

0.90

64

0.77

81

0.84

Kerala

93

0.95

94

0.94

97

0.97

94

0.94

85

0.91

93

0.95

Tamil Nadu

75

0.81

78

0.91

88

0.89

89

0.89

71

0.82

83

0.85

Orissa

71

0.81

76

0.83

83

0.83

83

0.77

62

0.75

70

0.74

Jammu & Kashmir

73

0.70

70

0.71

86

0.90

77

0.79

69

0.68

81

0.86

Jharkhand

66

0.71

64

0.75

85

0.87

79

0.79

62

0.75

73

0.79

Chhatisgarh

74

0.77

92

0.83

84

0.83

86

0.83

73

0.73

77

0.78

Uttaranchal

81

0.78

73

0.89

92

0.93

68

0.75

77

0.76

89

0.88

Assam

86

0.89

82

0.85

94

0.95

87

0.88

86

0.87

94

0.94

Arunachal Pradesh

76

0.89

90

0.86

93

0.89

67

0.77

74

0.91

92

0.94

Sikkim

87

0.91

87

0.58

95

0.95

97

0.88

85

0.86

91

1.01

Nagaland

97

0.93

0

 

98

0.96

100

1.00

92

0.99

98

0.98

Goa

87

0.90

81

0.68

92

0.90

78

0.89

71

0.80

88

0.79

Delhi

96

0.92

100

1.00

89

0.87

80

0.77

88

0.75

83

0.79

Source NSS 68th Round (2011–2012) (GOI 2015)

Note There are some odd exceptions as in the case of Delhi where Muslim urban rates are lower than the rural which is 100% (which even Kerala has not achieved!) with no gender disparity. Disaggregation of data into religious and social groups often creates problems due to small sample size, which distorts results

So far we have considered only rural literacy rates. Urban literacy rates are generally higher than the rural, as expected. The Muslim urban rates are generally higher than the rural except in Arunachal Pradesh, Delhi, Goa, Haryana, Himachal Pradesh, Punjab, Uttaranchal and Uttar Pradesh. F/M ratios for Muslims in urban areas are generally higher, suggesting lower gender disparity, except in Andhra Pradesh, Arunachal Pradesh, Delhi, Orissa, Tamil Nadu and Uttaranchal.

Bhalotra and Zamora (2010) undertook a logit regression analysis and estimated marginal effects (or probabilities) similar to the one reported for China. Their results show the importance of such variables as religion, education of the household head and access to rural infrastructure. Their detailed empirical estimation leads them to the following main conclusions:

  1. 1.

    Ownership of wealth and assets such as land has a positive effect on schooling of boys and girls, particularly among upper-caste Hindus.

     
  2. 2.

    Similarly, years of schooling of the most educated household member has a significant and positive effect on school attendance as would be expected. However, this effect is weaker for Muslim girls.

     
  3. 3.

    Children are more likely to be at school when the head of household is a female, suggesting that women have a greater commitment than men towards educating children.

     
  4. 4.

    In large households, the probability of children staying at school is lower than among smaller households. Since Muslims generally live in larger households, a priori, the effect of this variable should be stronger for the Muslim children. However, the results show that ‘the household size effect is smaller for Muslim boys and low-caste girls’ (ibid., p. 184). The reasons for this result are not clear.

     

A study on the educational attainment of adults up until 1979 (Deolalikar 2010) has constructed average years of completed schooling for different age cohorts of Muslims and non-Muslims, both male and female, on the basis of the NSS 55th Round. It shows that Muslim males have lagged behind upper-caste Hindu males in schooling, whereas Muslim women have narrowed the gap in schooling with upper-caste Hindu women.

Rural and urban disparities in India persist even at higher levels of education and among religious and social groups. For example, the shares of graduates in the population aged 15 and above vary widely. A study by Deshpande and Yadav (2006) showed that in 1999–2000 the most disadvantaged in the rural areas were the SCs, STs and Muslims and Hindu Backward Classes (OBCs). In urban areas the SCs and Muslims had the lowest shares, which is rather surprising considering that the SCs enjoy positive affirmative action in education which the Muslims do not (see below). The government policy focuses too much on affirmative action through reservation at higher educational institutions, considering that social disadvantages for deprived social groups begin as early as primary school (Desai et al. 2010, p. 86). According to the NSS 68th Round, Muslims had a higher proportion of non-literates than Hindus, Christians and Sikhs in 2011–2012. The share of Muslim graduates was also much lower than those for Hindus, Christians and Sikhs (see Table 4.13). Student enrolments in colleges and universities confirm this picture. The enrolment of students in government and private (unaided) institutions for Muslims increased somewhat between 2007 and 2014 but it declined in private unaided institutions. In the latter, the share of Muslims was lower than those of other minorities and the Hindu majority. The shares of the SCs and STs in these institutions were lower than that of the OBCs. These social groups depend mainly on government institutions of higher education (Thorat and Khan, forthcoming).
Table 4.13

India: distribution of population by education level and religious groups (15 years and above), 2011–2012 (%)

General educational level

All

Hindu

Muslim

Christian

Sikh

Not literate

30.2

30.2

34.3

14.9

26.4

Literate and up to primary

9.0

8.8

10.8

8.0

5.3

Primary

11.7

11.1

14.8

12.4

14.5

Middle

16.6

16.5

16.6

19.6

13.1

Secondary

13.9

14.0

11.7

17.0

19.4

Higher secondary

8.8

9.1

6.0

11.1

12.2

Diploma/certificate

1.3

1.3

0.7

4.2

0.8

Graduate

6.2

6.4

3.3

9.6

6.1

Postgraduate and above

2.0

2.1

0.7

2.9

1.9

Source NSS 68th Round (GOI 2015)

Gross enrolments vary according to monthly per capita expenditure. For example, the enrolment ratio for the lowest quintile of income (0–20%) was about 10% compared to 74% for the highest quintile in 2014 (ibid.). The results for 1995, that is almost 10 years earlier were similar. Ideally, net enrolments, which exclude over-age and under-age pupils, would be a better indicator but such data are not available.

The Indian government emphasizes the quantitative aspects of education (for example, enrolment rates and years of schooling) far more than the qualitative ones (for example, school attendance and retention rates, wastage in education, teacher absenteeism and so on). Panagariya et al. (2014, p. 271) note that pupil-teacher ratio in elementary education started rising especially since 2002–2003. Although there was a significant increase in enrolment (thanks to an increase in the number of schools), there has not been a proportionate increase in the number of teachers. There was a decline in the quality of education in government schools, in part due to teacher absenteeism. Increasing preference for private elementary schools may reflect parents’ dissatisfaction with the quality of teaching at government schools.

The level of attendance and non-enrolment indicates the participation of young people in educational institutions. One striking feature of the NSS 71st Round data for 2014 (GOI 2016b) is the high ratio of those in rural areas who never enrolled for any education, not only among Muslims (over 13%) but also among Scheduled Tribes (STs) (14%), SCs (nearly 12%) and even Hindus (9%). Non-enrolment ratios are the lowest for Christians followed by Sikhs (see Table 4.14). The net attendance ratios for Muslims are the lowest at primary, secondary and above higher secondary levels of education, which suggest high drop-out rates. It is the highest for Christians and Sikhs. In the case of rural Muslims, the share of those not enrolled (over 50%) was greater than that of ‘currently attending (48%) but that of urban Muslims was lower (ibid., pp. 29–30).
Table 4.14

India: non-enrolment and net attendance rates by religious and social groups (5–29 age-group), 2014 (%)

Religious/social group

Ratio of those who never

enrolled

Net

attendance

ratio

Spatial

dimension

Level of

attendance

Rural

Urban

Rural +urban

Primary

Secondary

Higher

Religious group

Hindus

10.4

4.7

8.9

84

54

13

Muslims

15.4

10.0

13.4

79

39

7

Christians

4.9

2.0

3.9

87

63

18

Sikhs

5.3

3.4

4.8

86

55

15

All

10.9

5.6

9.4

   

Social group

SC

12.8

7.6

11.7

82

49

9

ST

14.8

8.1

14.0

83

46

7

OBC

10.8

6.3

9.5

83

51

12

All

10.9

5.6

9.4

83

52

12

Source NSS 71st Round (GOI 2016b)

The reasons for dropping out of school vary. In the case of male students, 31% of the sample gave ‘engagement in economic activity’ as the major reason whereas 30% of female students gave ‘engagement in domestic activities’ as the major reason. As expected, other reasons include marriage for females and financial considerations for both male and females in urban and rural areas’. The survey also showed that 33% male and 27% females in the 5–29 age-group in rural areas ‘were not interested in education’. It is unclear why. Do rural people need labour for farm and non-farm activities to earn a minimum of livelihood? Or does the lack of interest in education spring from a feeling that education would not improve their economic situation? In urban areas, lack of enrolment was clearly due to financial constraints.

In recent years, elementary education in India (that is, of children aged 6–14 covering primary and middle school) has received serious attention after a long period of neglect. The government’s policy on elementary education is enshrined in the Right to Education Act (2009) which adopted free and compulsory elementary education as a fundamental right.

A Comparative Perspective

China’s advantage over India in the social sectors, often emphasized in the literature, is confirmed by Table 4.15. China scores over India in every respect. Adult literacy rate in China is much higher than that in India. China’s population with at least secondary education is also substantially higher. Except in tertiary education, its gross enrolment ratios at primary and secondary levels are much higher. Indicators of the quality of education in China, namely, drop-out rates and pupil-teacher ratios, are also much better than those in India. However, a disaggregation of the population by minority and majority in the two countries presents a somewhat different picture.
Table 4.15

China and India: a comparison of educational indicators

Indicator

China

India

Adult literacy rate (15 years+) (%) (2005–2013)

95.1

62.8

Pop. with at least secondary education (% ages 25 and older) (2005–2013)

Mean years of schooling (no.) (2010)

Gross enrolment ratios (%) (2008–2014)

- Primary

- Secondary

- Tertiary

65.3

7.5

128a

89

27

42.1

4.4

113a

69

25

Drop-out rate, all grades (% of primary school cohort) (2005–2008)

0.4

34

Primary completion rate (%) (2008)

96

94

Pupil/teacher ratio (primary) (2008–2014)

18

35

Public expenditure on education (%) (2005–2010)

-

3.1

Sources UNDP Human Development Report; World Bank World Development Report and World Development Indicators

aThe higher than 100 ratios at primary level for both China and India suggest that pupils enrolled are older or younger than the population in the relevant age-group. The gross enrolment ratio measures the number of pupils enrolled at the primary level as a percentage of the population in that age group

Drèze and Sen (1995, p. 65) discuss literacy and basic education in the two countries and conclude that China has done much better. Does this conclusion hold when one considers majority–minority literacy rates separately? Adult literacy rates show greater gender disparity in India than in China where the disparity is most glaring in Tibet but the F/M ratios are generally above 0.90, suggesting very low gender disparity. Only in Tibet, Inner Mongolia, Guizhou, Qinghai, and Yunnan, the F/M ratios are lower. Gender disparity is the highest in Tibet followed by Qinghai.
Table 4.16

Rural literacy rates for China and India

Socio-religious groups

1994

2004–2005

Rate (%)

F/M

Rate (%)

F/M

India

Muslims

49

0.64

59

0.74

SCs/STs

52

0.52

56

0.67

Hindu majority

53

0.60

68

0.71

China

1995

2002

Minority

22

0.93

12

0.94

Han majority

15

0.96

8

0.92

Sources For India, Shariff (1999) (estimates based on NCAER 1994 rural household survey) for 1994; NCAER rural household survey for 2004–2005. For China, CASS (1995, 2002)

Generally, adult literacy rates in Indian states with a sizeable Muslim minority population are much lower than those in most Chinese provinces (see Table 4.17). The only exception is Tibet where it is closer to those in Jammu and Kashmir, Uttar Pradesh and Bihar. In India, Kerala is the only state in which the adult literacy rate is close to the rates in Guangxi, Inner Mongolia, Xinjiang and Sichuan in China.
Table 4.17

China: and India: adult literacy rates (15 years and above), 2014

Province

Overal Literacy

Rate (%)

Male literacy rate (M) (%)

Female literacy rate (F)

(%)

F/M ratio

China

National

95

97

93

0.91

Five autonomous regions

Guangxi

96

99

94

0.91

Inner Mongolia

95

97

93

0.89

Ningxia

92

95

89

0.90

Tibet

68

76

59

0.74

Xinjiang

97

98

96

0.92

Western provinces

Gansu

89

93

85

0.89

Guizhou

88

94

83

0.85

Qinghia

87

92

82

0.83

Sichuan

93

96

91

0.93

Yunnan

92

95

89

0.87

India

Selected states with a sizeable Muslim population

National

71

80

62

0.77

Jammu & Kashmir

69

80

57

0.71

Assam

84

88

79

0.89

West Bengal

75

80

69

0.86

Kerala

95

97

92

0.95

Uttar Pradesh

64

75

52

0.69

Bihar

59

71

46

0.65

Sources GOC, China Statistical Yearbook 2015; For India, NSS 71st Round (GOI 2016b)

The Chinese minority literacy rates are low, especially in rural Tibet (46% for 6 years and above, according to the 2000 China Population Census). The rural rate for Tibet is even lower than that for the SCs and STs in Orissa (57% for 2004–2005—see Table 4.14), one of the poorest Indian states. It is also lower than the rural literacy rates for the Muslims in Uttar Pradesh and Bihar (51 and 56%, respectively), other poor Indian states. Thus, China’s record looks much less impressive when one considers the case of minorities. To be fair, Tibet’s exceptionally low rural literacy rate is unique in China. The rural literacy rates are much higher for other provinces. For example, in Xinjiang, which has a sizeable Muslim population of the Uygur and Hui, the rural literacy rate is 97%.

Gender disparity among the minorities is high in both China and India, but it is more serious in India. The female-to-male ratios (F/M) for minorities are much lower than those in China. Within India, the average F/M ratio is slightly higher for Muslims than for Hindus. But the F/M gap is quite wide in Rajasthan in favour of Hindus, and in Uttar Pradesh in favour of Muslims (see Table 4.12).

In India, access to education is often determined by caste considerations. For example, caste-based inequalities in rural literacy widened between 1994 and 2004 between upper-caste Hindus and SCs and STs (see Table 4.16). The hold of caste is stronger in some states than in others. For example, in the poor caste-ridden states such as Bihar and Uttar Pradesh, the rural literacy rates were the lowest in 2001, 27 and 44%, respectively. The Muslim rate for Bihar was better than that of the STs and SCs, but it was the same in Uttar Pradesh (GOI 2001).

The Hindu majority in India had the highest rural literacy rate in 1994 and 2004–2005, but for China, for 1995 and 2002, these rates were higher for the minority (see Table 4.16).

In Uttar Pradesh, there is evidence of caste-based differences in educational attainment even after controlling for differences in income levels (Drèze and Gazdar 1997, pp. 82–87). Cases have been cited of discrimination against the scheduled caste settlements in the location of schools.

In China, there is no parallel to the caste factor determining educational access. However, although caste has no relevance, educational inequalities may be explained by class and social structures such as the bureaucrats and the party cadres, especially at the local level (Bhalla 1995). The Chinese cadres and their children are known to enjoy privileged access to education, particularly to institutions of higher education.

To redress imbalances in educational access, both China and India have introduced affirmative action in favour of disadvantaged groups and minorities to alleviate their social exclusion and marginalization (see Chap.  2). These preferential policies are intended to provide equality of opportunity generally in education and the labour market.

In China, policies to promote greater access of ethnic minorities to education have included lower entrance requirements at different levels of education, exemption from payment of school fees and ‘bonus points’ for taking examinations in Chinese instead of a minority language (Bhalla and Qiu 2006, pp. 95–97; Iredale et al. 2001, pp. 70–85). It is not clear whether these policies have actually helped narrow the gaps; indeed, some evidence suggests that these gaps have not narrowed (Sangay 1998). At best, the policies may have prevented the majority–minority gaps from widening.

In India, affirmative action extends to disadvantaged social groups but not to religious minorities. However, affirmative action has been extended to Muslim castes among OBCs in Karnataka and Kerala, for example (GOI 2006, p. 198). A historical caste-based disadvantage (which deprived ‘untouchables’ of access to education), rather than religion, formed the underlying principle for preferential treatment. Thus, the Indian position regarding affirmative action towards religious minorities is different from that in China, where all ethnic minorities are protected.

Health Status and Access to Health Care

Poverty alleviation and social inclusion require a multi-pronged attack going beyond income-based approaches. Apart from the important role of education discussed above, the health status of an individual and his or her access to health services are important ingredients of an anti-poverty strategy. How healthy an individual is determines his productivity in employment and his ability to earn a living. Low standards of living in a minority group or population may, indeed, reflect low indicators of health status (for example, high infant mortality and morbidity rates, and low life expectancy) (Gupta and Mitra 2004).

China

We do not know of any data on child mortality or morbidity by ethnicity for China. However, there is some empirical evidence to suggest that ethnic minorities enjoy less access to health services than the Han majority. The CASS Household Surveys for 1988, 1995 and 2002 provide some information on health status and access indicators for both rural and urban households.

There were substantial disparities between minorities and non-minorities in 1995 for households with or without rural health clinics in their village (see Table 4.18). While 50% of the rural minority households had a health clinic in their village, nearly 86% of rural non-minority households had a health clinic. For urban households, there are differences in the access of urban minorities to sanitary facilities in both 1988 and 1995, and the gap widened during this period. Also there were sizeable disparities in access to running water in the 1988 survey, and there were still some disparities in 1995. In rural areas health indicators were worse for minorities than for the majority in 1995.
Table 4.18

Minority–majority health situation/status in China, 1988, 1995 and 2002 (%)

Health variable

Minority

Majority

1988

1995

2002

1988

1995

2002

Urban

Access to running water

72.8

92.3

 

78.1

94

 

Sanitary facilities

Lack of sanitary facilities

45.0

49.2

 

34.7

25.2

 

Shared sanitary facilities

18.1

5.5

 

20.5

8.7

 

Have toilet, lack bath

34.6

25.1

 

40.2

36.4

 

Have bath and toilet

2.3

20.2

 

4.8

29.8

 

Medical expenses per capita (yuan)

57.3

64.7

372.5

46.9

70.4

453.7

Child care expenses per capita (yuan)

35.0

53.0

29.8

76.7

  

Rural

Access to running water

16.6

 

28.1

 

Villages with health clinics

50.3

 

85.7

 

Medical expenses per capita (yuan)

2.1

11.4

 

4.4

16.2

 

Child care expenses per capita (yuan)

52.9

30.6

  

Source Based on CASS (1988, 1995, 2002)

There is not much quantitative difference in medical expenses per capita between 1988 and 1995 for both minorities and non-minorities. Minorities spend less in both years, but the difference is less than 5 yuan, which may be statistically significant given the large number of observations.

In the case of rural areas, public health insurance for minorities declined from 1.5% of households in 1988 to only 0.2% in 1995. Self-financed health insurance is the main mode of insurance in rural areas. It is generally acknowledged that rural health insurance in China has declined in the wake of economic and social reforms (Bhalla 1995). The cost of rural health services is rising, and the staffing is poorer (Bloom and Fang 2003).

The proportion of minorities in households which finance their own medical costs is higher than that of non-minorities in both years, although the gap closed somewhat in 1995 (see Table 4.19). This could be, in part, due to the fact that in 1988 the questions were asked in such a way that households could report some members being covered publicly and others being self-financed, whereas in 1995 the entire household was classified in only one way. Nevertheless, there is still an important differential between minority and non-minority households. There is not much difference between minorities and non-minorities in urban areas (especially in 1995) regarding public health insurance. However, the differences are more significant in the cases of semi-public health insurance and private health insurance.
Table 4.19

China: health insurance of rural–urban and minority-majority households, 1988, 1995 and 2002 (%)

Type of insurance

Minority

Majority

1988

1995

2002

1988

1995

2002

Urban

Public health insurance

48.4

54.1

41.7

50.0

54.8

46.8

Semi-public health insurance

14.9

 

21.5

 

Private health insurance

5.3

5.2

8.6

2.5

Self-financed health insurance

22.4

38.2

46.7

17.0

29.4

47.8

Rural

Public health insurance

1.5

0.2

 

0.8

0.7

 

Self-financed health insurance

86.0

87.5

 

92.8

87.6

 

Private health insurance

0.3

 

0.4

 

Source Based on CASS (1988, 1995, 2002)

A regression analysis based on the 1995 and 2002 CASS household data for rural China (see Table 4.20) shows that minority status, low per capita income, household heads engaging in agriculture, living in mountainous areas, and living in designated minority areas are important factors that reduce access to health services, defined as the distance to a village health clinic. A regression on the determinants of access to clean drinking water generates similar results.
Table 4.20

Determinants of household access to health clinics in rural China, 1995 and 2002

 

1995

2002

Logit regression

Marginal effects

Logit regression

Marginal effects

Minority status (ethnicity)

−0.859***

−0.106**

−0.032

−0.002

(0.15)

(0.02)

(0.165)

(0.0124)

Logarithm of per capita income

0.332***

0.031***

0.073***

0.005***

(0.06)

(0.01)

(0.023)

(0.002)

Head of household is illiterate

−0.251*

−0.026

−0.373**

−0.032*

(0.13)

(0.01)

(0.182)

(0.018)

Head of household is male

0.231

0.023

0.135

0.010

(0.17)

(0.02)

(0.165)

(0.014)

 Party member

−0.097

−0.009

0.060

0.004

(0.10)

(0.01)

(0.098)

(0.007)

Working in agriculture

−0.314***

−0.027***

−0.063

−0.005

(0.10)

(0.01)

(0.077)

(0.006)

 Sub-urban areas

1.219***

0.074***

−0.998***

−0.105***

(0.29)

(0.01)

(0.111)

(0.016)

 Plains areas

1.513***

0.141***

0.391***

0.029***

(0.10)

(0.01)

(0.087)

(0.006)

 Hilly region

0.219***

0.020***

0.836***

0.055***

(0.09)

(0.01)

(0.110)

(0.006)

Designated minority region

−0.901***

−0.113***

−1.017***

−0.104***

(0.15)

(0.02)

(0.162)

(0.022)

Constant

−0.958*

 

1.562***

 

(0.51)

 

(0.220)

 

Number of observations

7967

 

9200

 

Pseudo R-squared

0.15

 

0.06

 

Source Based on CASS (1995, 2002)

Notes ***, ** and * denote 1, 5 and 10% significance levels respectively

The Autonomous Regions

For the five Chinese autonomous regions, we compare urban and rural doctor-to-population ratios (
Fig. 4.4

Xinjiang: rural–urban doctor-to-population ratios, 1997–2004 (number of doctors per 10,000 population).

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

Fig. 4.5

Tibet: rural–urban doctor-to-population ratios, 1997–2004.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

see Figs. 4.4, 4.5, 4.6, 4.7 and 4.8). This ratio is often used to measure adequacy (or otherwise) of health services. However, it cannot indicate access without an income dimension. One would need information on the proportion of minority population actually treated by doctors to obtain any idea about the satisfaction of their basic need for good health. In the absence of such detailed information, we present the doctor-to-population ratios for the Chinese autonomous regions for the 1997–2004 period.
Fig. 4.6

Guangxi: rural–urban doctor-to-population ratios, 1997–2004.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

The number of doctors at the county level (we have no information on county population to estimate the ratios) is allocated between urban and rural in the proportions of 30 for urban and 70 for rural. Since 1999, the gap between urban and rural ratios has been widening in Xinjiang (see Fig. 4.4). The Tibetan case is interesting: from 1997 to 1999, the gap between urban and rural ratios disappeared but started widening after 1999 (see Fig. 4.5). Can this be explained by the lack or absence of urbanization in Tibet, which consists mostly of counties and rural areas? Inner Mongolia shows the narrowest gap between urban and rural ratios for doctors (see Fig. 4.8).
Fig. 4.7

Ningxia: rural–urban doctor-to-population ratios, 1997–2004.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

In 2014 in the prefectures of Xinjiang, the ratios of hospitals and health centres per 1000 population varied from 59 to 85 whereas that for hospital beds varied from 55 to nearly 70. The minority-dominated prefecture (Kizulu Kirgiz) showed the lowest ratios. At the county level also these ratios varied significantly but the county with 93% minority population did not show the lowest health ratios (see Chap.  6, Table 6.19).
Fig. 4.8

Inner Mongolia: rural–urban doctor-to-population ratios, 1997–2004.

Source Based on data from GOC, China’s Ethnic Statistical Yearbook

The ratios of medical technical personnel and registered nurses in 2014 were the highest for Xinjiang among the five autonomous regions. They were the lowest for Tibet (see Chap.  6, Table 6.18).

India

Most health indicators for India, especially infant and child mortality, child malnutrition and per capita expenditure on health, compare unfavourably with those of China (see Table 4.24). India has been widely criticized for its relatively poor performance regarding child malnutrition which prevails not only among the poor households but also among richer ones (Tarrozzi 2008) . 8 In China, private expenditure per capita on health is four times as large as India’s and the physicians per 10,000 population are twice as many. However, this is not to suggest that India has not made any progress in the post-reform period (since the early 1990s) when its economic growth accelerated and per capita incomes started rising. One reason for India lagging behind is that China grew much faster for a much longer period. There are wide variations in the performance of different Indian states regarding the improvement of different health indicators, but all made progress during the last few decades (Panagariya et al. 2014) .

The differences in income and education between different social classes and religious groups discussed earlier in the chapter account for, inter alia, disparities in health status, and access to and utilization of health resources. They are reflected in inter-state, rural–urban and class variations. According to the NSS 71st Round on health in India conducted in 2014 (GOI 2016a), greater amount of expenditure was incurred for non-hospitalized treatment by urban population than rural. The expenditure on hospitalization is positively correlated with the monthly per capita expenditure irrespective of the type of expenses. The data also show that the bulk of the rural and urban population remains uncovered by any government or private health insurance schemes, a situation similar to that in China where only about 10% of the rural population is covered by any health insurance. Yet in China the amount spent on private pre-paid health plans as a proportion of private health expenditure in 2012 was more than twice as high (see Table 4.24). China has introduced significant health insurance reforms including the introduction of a new Rural Cooperative Medical System, which is aimed at making rural health care affordable for the poor. Under the system, the enrolment rate expanded from 91% in 2008 to 99% in 2014. The number of beneficiaries of the scheme nearly tripled during this period ( China Statistical Yearbook, 2015).

The infant and child mortality rates are higher for Muslims than for Christians and Sikhs, but lower than for Hindus according to the National Family Health Survey (NFHS-3) for 2005–2006 (see Table 4.21). The infant (IMR) and child (UMR) mortality rates for 1981, 1991 and 2001 (the three census years) for Muslims, Christians and Sikhs are compared with those of the Hindu majority. In 2001, the Muslim IMR was equal to the national average for all religions; but the UMR was below the average. The Christian IMR was higher than the national average although the UMR was much lower.
Table 4.21

India: infant and child mortality rates by religion, (number per 1,000 live births)

Religion

Census data

NFHS data

 

1991

2001

1998–1999

2005–2006

Average for all religions

IMR

74

72

73

57

U5MR

96

98

101

74

Hindu majority

IMR

74

73

77

59

U5MR

97

99

107

76

Minorities

Muslims

IMR

68

72

59

52

U5MR

101

95

83

70

Christians

IMR

58

77

49

42

U5MR

70

77

68

53

Sikhs

IMR

55

a

53

46

U5MR

67

82

65

52

Sources Census of India for 1981, 1991 and 2001. IIPS (2000, NFHS-2; 2007, NFHS-3) for 2005–2006

aThe series was too erratic so the IMR is not shown. IMR = Infant mortality rate, defined as the proportion of children dying before their first birthday

U5MR = Under-five mortality rate

An unusual phenomenon concerns child mortality among Muslims. It is consistently lower for Muslims than for the upper-caste Hindus. A statewise breakdown of the child mortality rate for 1998–1999 confirms the Muslim child survival advantage in a number of states: Bihar, Kerala, Uttar Pradesh, Madhya Pradesh and Maharashtra. The lower rates of Muslim child mortality defy economic and intuitive logic, which would suggest higher rates because the Muslims in general are poorer and less educated (especially Muslim females) than upper-caste Hindus. How can one explain this situation? Are there cultural/religious differences (for example, earlier breastfeeding by the Muslim mothers) or nutritional differences (the Muslims eat more meat than the Hindus, who are often vegetarians) which explain child survival among the Muslims? Bhalotra and Zamora (2010, p. 3) suggest that ‘some of the Muslim advantage may stem from their lower degree of son preference, their closer kinship, their more non-vegetarian diet, the better health of Muslim mothers and their lower propensity to work outside the home’.

Greater urbanization among the Muslims (they tend to live in urban areas more than the Hindus) may be another factor explaining some of the survival advantage. Child mortality rates are generally lower in urban areas.

A probit analysis of the determinants of child mortality examines the influence of such factors as religion, ethnicity, education of the mother, household features and so on. It shows that religion and ethnicity (or social group) does not significantly change the probability of a child dying in the first year of its life. However, there are religious and social differences for under-5 mortality. Muslim children are less likely to die before the age of five than non-Muslims (Deolalikar 2010, pp. 75–76). Maternal education tends to lower the probability of a child’s death, as would be expected. However, it is surprising that rural residence does not have any significant adverse effect on infant mortality, although it does on under-5 mortality.

The distribution of hospitalization cases for childbirth in 2014 indicates that public hospitals are utilized more than health centres in both rural and urban areas by all the religious and social categories (see Table 4.22). Public health centres are less frequented by religious minorities and the Hindu majority. Muslims use rural public hospitals more than the private ones as do the Christians and Sikhs. But in urban areas, private hospitals are more popular among the Christians than other religious groups. In rural areas, STs used the public centres more than the private hospitals. For both SCs and STs rural public hospitals were utilized more than either rural public health centres or rural private hospitals. In urban areas, the SCs utilized public hospitals for childbirth more than the private whereas the STs utilized the latter slightly more.
Table 4.22

India: distribution of hospitalization cases for childbirth by religious and social groups (2014) (per 1,000 no. of hospitalized cases by level of care)

Religious and social groups

Publichealth centres and others

Public hospitals

Private hospitals

Rural

Urban

Rural

Urban

Rural

Urban

Religious groups

Hindus

18.1

3.4

51.8

43.4

30.1

53.2

Muslims

17.9

5.7

53.4

45.2

28.7

49.1

Christians

20.5

2.2

45.8

37.9

33.7

59.9

Sikhs

5.5

3.7

56.7

40.6

37.8

55.7

All

18.0

3.8

52.0

43.6

30.0

52.5

Social groups

SCs

17.8

4.9

60.6

58.1

21.6

37.0

STs

27.0

4.5

57.9

46.6

15.1

48.9

OBCs

18.9

3.4

46.5

43.9

34.6

52.7

All

18.0

3.8

52.0

43.6

30.0

52.5

Source NSS 71st Round (GOI 2016a)

In general, the role of social class and religion may be an important factor in explaining malnutrition among children and women, 9 which is particularly bad among SCs, STs and OBCs of all religious backgrounds (Thorat and Sabharwal 2011). Discrimination and exclusion may be partly responsible for this situation (Thorat and Sadana 2009). The nutritional status of Hindus and Muslims is worse than that of Christians and Sikhs. Poverty is another causal factor that explains malnutrition. However, poverty reduction is necessary but not sufficient to eliminate malnutrition, which has remained acute despite poverty reduction (Radhakrishna 2015).

Despite some nutritional improvements, nearly half of Indian children continue to suffer from malnutrition, which is much more widespread than income poverty. Malnutrition not only retards physical and mental development but it also contributes to poverty. Radhakrishna et al. (2011, p. 17) argue that ‘the existence of mother-child-mother malnutrition has resulted in intergenerational transmission of poverty’.

Aggregate data often conceal useful information. Therefore, special tabulations were prepared for GOI (2006) on small, medium and large villages to determine differentials in the availability, accessibility and utilization of such public services as primary schools and health clinics by religious and social groups at the village level. This micro-analysis shows that invariably, villages with over 40% Muslim population in the six states with large Muslim population had fewer health facilities. This conclusion holds for small, medium and large villages (see Table 4.23). In Kerala, 33% of Muslim concentration medium-sized villages had health facilities compared with 60% with a Muslim population of less than 9%. In Assam, Bihar and J&K, all Muslim concentration villages (small, medium and large) had the smallest number of health facilities. As expected, large villages have better health facilities than medium-sized and small villages. However, the largest proportion of facilities is reported in villages in which the Muslim population ranges between 10 and 39%.
Table 4.23

India: health facilities for Muslims at the village level, 2001 (%)

State/share of Muslim population

Small villages (<1,000 population)

Medium villages (1,000–2,000 population)

Large villages (>2,000 population

% of villages having facility

% of villages having facility

% of villages having facility

Assam

<9%

16.0

31.4

50.2

10–39%

18.7

31.0

55.3

Over 40%

13.2

29.0

49.3

Kerala

<9%

50.0

60.0

98.6

10–39%

100

96.0

Over 40%

33.3

96.5

Jammu & Kashmir

<9%

35.5

60.9

86.1

10–39%

36.1

58.0

87.3

Over 40%

29.0

54.6

83.6

West Bengal

<9%

34.4

60.0

81.2

10–39%

42.0

61.6

80.7

Over 40%

37.5

56.6

73.1

Uttar Pradesh

<9%

14.4

31.8

54.8

10–39%

17.4

34.4

57.1

Over 40%

13.5

31.0

56.6

Bihar

<9%

8.3

17.1

37.0

10–39%

8.7

17.6

37.9

Over 40%

8.2

12.9

27.9

Source GOI (2006)

GOI (2006, p. 144) concludes that ‘in most of the states, the proportion of Muslim-concentration villages with medical facilities is somewhat lower than the proportion of all villages with such facilities, suggesting a bias in public service provisioning in Muslim concentration areas’.

A Comparative Perspective

China is widely credited with having made more rapid progress in providing health services than has India. For example, such health indicators as infant mortality, maternal mortality and child malnutrition are far better for China. We do not have information on mortality indicators by ethnicity. However, these indicators are likely to be less favourable for the Chinese ethnic minorities than for the Han majority considering that they are generally poorer, less-educated and enjoy limited access to health services.

In both countries, access to public health services, especially for the minority communities and particularly in rural areas, is relatively limited. The minority status of the population has a role to play in determining access (or lack of it) to such public health services as drinking water supply and sanitation, not to mention nutrition. The access of minorities to health clinics and other facilities is lower than that of the majority.

Minority–majority differences in China and India tend to be due to an urban bias in the provision of health services. Even in rural areas, there may be a bias against public provision of services for minorities, as noted above. Generally, in urban areas, health facilities are more readily available and are within easy distance. On the other hand, in rural areas people often travel long distances to seek medical assistance on account of lack of local health facilities.

The high and rising direct cost of health services for the poor in general, and rural minorities in particular, limits access in both countries. We noted above that, in China in 1995, rural medical expenses per capita were much lower than urban, and rural child expenses per capita were one-fifth of the urban. Similarly, minority expenses per capita were lower than those of the majority in both rural and urban areas (see Table 4.18). Caste and class-based inequalities also exist in India, and they have worsened with greater privatization and a decline in government health care expenditure (Sen et al. 2002 ; Balarajan et al. 2011). For health care in rural India, people have to depend on private doctors who often do not have any medical training. Fraud and exploitation become easier when rural patients are ignorant about what sort of medicines doctors prescribe. A study undertaken by the Pratichi Trust in India revealed exploitation of poor and ignorant patients who were made to pay for treatment they did not receive (Sen 2011).

In both countries, out-of-pocket health expenditure is quite large, suggesting an increase in private health care provision (see Table 4.24). Over the years, private health care has become more important, which may be explained by a relative decline in the availability of public health facilities and by a perception of better quality of private health care.
Table 4.24

Health indicators for China and India

Health indicator

China

India

Infant mortality (per 10,000 live births) (2013)

10.9

41.4

Under-5 mortality rate (per 10,000 live births) (2013)

12.7

52.7

Maternal mortality rate (per 100,000 live births) (2013)

27

174

Life expectancy at birth (2012)

73.7

65.8

Male (years)

70.4

57.7

Female (years)

75

77

Child malnutrition (% of children under 5) (stunting moderate or severe) (2008–2014)

9.4

47.9

Physicians per 10,000 (2007–2013)

Nursing and midwife personnel per 10,000 (2007–2013)

14.9

16.6

7.0

17.1

Hospital beds per 10,000 (2005–2012)

39

9

Population without access to improved services (2008) (%)

Water

11

12

Sanitation

45

69

Per capita expenditure on health in PPP$ (2012)

578

196

Total expenditure on health as % of GDP (2013)

5.6

4.0

Total health expenditure as % of GDP (2007)

  

Out-of-pocket expenditure as% of private health expenditure (2012)

78.0

87.2

Private pre-paid plans as % of private health expenditure (2012)

7.0

3.3

Sources UNDP Human Development Report; WHO World Health Statistics; World Bank

World Development Report and World Development Indicators

Poor people in rural areas, especially those from minorities who tend to be even poorer, are denied access to health services for lack of resources. For the rural poor, health care has to compete with many other urgent demands. They often do not go to a doctor unless they are seriously ill because medical expenses form a large proportion of a rural household’s budget. In China, 44% of the poor fall into poverty because of medical expenses (WHO and DRC 2005). Medical expenses are responsible for more than half of India’s households falling into poverty (Balarajan et al. 2011).

Reasons for the reduced access of minorities are complex but are part of a more general problem of lack of public financing and health provisioning in both countries, particularly in rural areas. In China, these resource constraints have become more serious with the devolution of fiscal responsibility from the centre to county and village levels.

Malnutrition of children under five is a glaring problem in India, while it is very limited in China. In India, it is particularly bad among minority groups. Although we do not have comparable information for the Chinese minorities, malnutrition is unlikely to be a serious problem for Chinese minority children. China’s record in providing adequate nutrition is far superior to that of India. The production and distribution of food are much better integrated in China than in India, which may have helped in reducing malnutrition among the poor. Failures of food security in India may also explain poor health care (Radhakrishna et al. 2011). We believe that the reasons for China’s better performance in nutrition and health care are to be sought in faster growth over a longer period, more rapid reduction of poverty, greater political commitment and better implementation of programmes. 10

Bhagwati and Panagariya (2013, p. 154) observe that ‘the Indian government has had an extremely poor track record in delivery’ which is glaringly obvious in education and health sectors. The government may be able to hire better-qualified doctors for its medical and health services but it often fails to make them deliver better service (Das and Hammer 2007). The system of public delivery of health services in India suffers from many deficiencies, for example, absenteeism of medical and health personnel, lack of accountability, low quality of clinical care and mistrust of the system leading to the growth of private services. More and more potential beneficiaries are increasingly turning towards private services even if they are more expensive, which imposes a major burden on the poor (Hammer et al. 2007).

Concluding Remarks

In this chapter, we have argued that inequalities in both India and China need to be examined from their economic, social and cultural perspectives. Income inequalities narrowly defined do not provide a meaningful analysis of minority–majority differences. We show that income inequalities are also influenced by non-income issues such as educational attainment, health status and access to health care.

An abundant literature exists to show that China has been far more successful than India in promoting social and human development (Dev 2008; Drèze and Sen 1995 ; Rao 2011). Empirical evidence provided in this chapter confirms China’s superiority in terms of social and human indicators. However, both countries suffer from growing rural–urban and inter-regional inequalities, which persist and keep widening due, partly, to rapid economic growth in both countries (see also Chap.  1).

Growing income and non-income inequalities are slowing down the process and speed of poverty reduction. They are also contributing to the social exclusion and marginalization of the poor. A more rapid rate of poverty reduction in both India and China would have been achieved had income growth been more equitably distributed. Income growth in both countries has been quite rapid but the benefits of this growth have not trickled down to the poor as quickly as expected. Rising income inequality is not an inevitable outcome of rapid economic growth, which can go hand-in-hand with equity if appropriate redistributive measures are adopted.

Our main concern in the book is to examine rural–urban, inter-regional and gender inequalities between minority and majority populations, as well as among different religious and ethnic minorities, and disadvantaged social groups excluded from the mainstream of society. Here, an aggregative picture hides similarities between China and India. In both countries, the economic and social situation of minorities is generally much worse than that of the upper-caste Hindu majority in India and the Han majority in China.

Above, we noted the existence of bias against minorities (for example, Muslims) in the public provision of health facilities. Is such a bias symptomatic of a more general economic and social discrimination discussed in Chap.  1? Many observers believe that it is. For example, Hasan (2009, p. 231) argues that ‘a vast majority of Muslims suffer double discrimination by virtue of being Muslim and poor’. She claims that ‘their under-representation in the political, administrative and security structures of the state’ (discussed in Chap.  7) is caused by these factors.

In China also, some observers have found evidence of economic and social discrimination (see Chaps.  1 and  6). Such minorities as the Uygur and Tibetans feel that they are discriminated against by the Chinese government and the Han majority.

The socioeconomic situation of the Chinese ethnic minorities, at least in rural areas, is somewhat better than that of the minorities in India. There are no landless workers in China (thanks to land reforms after the Revolution), whereas many Indian minorities and disadvantaged social groups consist of a large number of landless workers with no assets except their labour. However, growing social discontent and protests in both countries suggest that economic growth is not participatory or inclusive, and that it is bypassing the poor and marginalized minorities. In Chap.  6, we discuss whether this discontent is due to large and rising inequalities and persistent poverty, or any other non-economic social, cultural and political factors.

Both China and India are aware of the unfavourable social and political consequences of rising income and other inequalities in terms of discontent, civil strife and violence. In 2000, China introduced the Western Region Development Strategy to reduce regional inequalities by raising the living standards of the poor in the Western region. While the Chinese government has taken a step in the right direction, it is unclear whether the Strategy has succeeded in reducing rural−urban and minority−majority income and social gaps.

We are not aware of any similar regional strategy in India focusing particularly on minorities (except affirmative action for the social groups, which does not extend to the Muslim minority). The anti-poverty programmes that do exist (programmes for employees and the self-employed, and the public distribution system, for example) are targeted at the poor generally, rather than at minorities. The effectiveness of these programmes has also been questioned (Dev 2008; Radhakrishna and Ray 2005). A comprehensive, targeted and effective development strategy is needed to check inequalities between minorities and the majority populations.

Notes

  1. 1.

    For India, Radhakrishna (2015, p. 63) estimated growth elasticity using NSS consumer expenditure data for 1993–1994, 2004–2005 and 2009–2010. The elasticity was positive, which indicates that more rapid growth aggravates inequality. However, growth elasticity of poverty was negative, implying a positive impact of growth on poverty reduction.

     
  2. 2.

    However, we need to bear in mind that despite improvements over the official SSB (CBS) surveys, the CASS household surveys are not without some weaknesses. Adjustments made to the official surveys do not remove some of the basic problems of the official surveys, which use a narrow definition of income. For example, they do not estimate the rental value of housing. Furthermore, the rural household surveys of the 1980s did not adequately cover non-agricultural or rural activities. Bramall (2001, p. 698) notes: ‘revised estimates remain seriously deficient, largely because they are still based on the data collected by the State Statistical Bureau (SSB). Some of the deficiencies produce a continuing under-estimation of true income inequality’.

     
  3. 3.

    Wiemer (2004, p. 177) undertook a regression analysis on the basis of the 1998 county data on (i) GDP per capita; and (ii) the percentage of non-Han people in the population. He included the share of population engaged in agriculture as a control variable because one would expect GDP per capita to be lower in agricultural counties regardless of any influence of ethnicity. The regression results based on 82 counties show that: (i) ‘For a given agricultural share in a county’s population, every percentage point increase in the non-Han share in population is associated with an expected decrease in GDP per capita of 44 yuan’ (p. 177); (ii) As the non-Han Uyghur population tends to be more concentrated in agricultural counties, ‘the actual disparity between Han and non-Han localities is greater than reflected in the ethnicity coefficient alone’ (pp. 177–178).

     
  4. 4.

    Khan and Riskin (2005) use the CASS household survey data for 1995 and 2002 to compare China’s household income and its distribution. They estimated some decline in rural and urban income inequality from 1995 to 2002. They found that a decline in rural income inequality was caused by a reduction in intra-provincial and inter-provincial inequality and an improvement in wage income and farm income. A decline in urban income inequality is explained by a fall in inter-provincial inequality and better distribution of imputed rental income and net taxes.

     
  5. 5.

    On the basis of the 1988 CASS household survey, Knight and Song (1999) also found that, in urban areas, the minorities enjoy a higher level of education than the Han majority. This finding is confirmed by Sautman (1999), who notes that in Jiangsu province Huizu (Muslim Chinese) minorities have a higher level of education than the Han.

     
  6. 6.

    There were only 20 schools run by the Tibetan local authorities and 96 small private educational institutions. The total number of students was about 3,200 (Zhang 1989) . The first new school (Changduo Primary School) was established in 1951 and, by the end of 1959, the number of primary schools had increased to 462. There were also three secondary schools. By 1984, three universities had been established; the number of secondary schools had increased to 89, and primary schools to 2,526 (ibid.).

     
  7. 7.

    In 1951, Tibet’s total area was more than 1.2 million km2, but its population was 957,000; that is, 0.8 persons per km2. Tibet’s population doubled between 1950 and 1980 but, in 1982, the population density was still only 1.6 persons per km2; that is, 1/60th of the national average. It was very difficult and costly to provide adequate schooling facilities and educational infrastructure with such low population density. Second, Tibet is 4,000 m above sea level, with an annual average temperature below 0 °C. Low pressure and thin oxygen make it very difficult for people from outside Tibet to stay there for long periods, which reduces the possibility of attracting externally the large number of teachers required.

     
  8. 8.

    Tarrozzi (2008) shows that 22% of Indian children in richer households (richness measured by wealth, educated parents and urban residence) are stunted. Some scholars (for example, Panagariya et al. 2014) have criticized the WHO methodology for estimating malnutrition. However, the fact remains that even after allowing for the methodological flaws, child malnutrition remains a serious problem in India.

     
  9. 9.

    UNDP (2003) shows that India’s performance in terms of nutritional outcomes is worse than that of less-developed countries in Africa.

     
  10. 10.

    See Radhakrishna et al. (2011) and Panagariya et al. (2014) on the links between poverty and malnutrition in India.

     

References

  1. Balarajan, Y., Selvaraj, S., & Subramanian, S. V. (2011, February 5). Health care and equity in India. Lancet, 377, 505–515.Google Scholar
  2. Bhagwati, J., & Panagariya, A. (2013). Why growth matters. New York: Public Affairs.Google Scholar
  3. Bhalla, A. S. (1995). Uneven development in the third world: A study of China and India (2nd revised and enlarged edition). London: Macmillan.Google Scholar
  4. Bhalla, A. S., & Qiu, S. (2006). Poverty and inequality among Chinese minorities. London: Routledge.Google Scholar
  5. Bhalotra, S., & Zamora, B. (2010). Social divisions in education in India. In R. Basant & A. Shariff (Eds.), Handbook of Muslims in India: Empirical and policy perspectives. New Delhi: Oxford University Press.Google Scholar
  6. Bhaumik, S. K., & Chakrabarty, M. (2010). Earnings inequality: The impact of the rise of caste-and religion-based politics. In R. Basant & A. Shariff (Eds.), Handbook of Muslims in India: Empirical and policy perspectives. New Delhi: Oxford University Press.Google Scholar
  7. Bloom, G., & Fang, J. (2003). China’s rural health system in a changing institutional context (Institute of Development Studies (IDS) Working Paper 194). Brighton: IDS.Google Scholar
  8. Borooah, V. K., Gustafsson, B., & Li, S. (2006). China and India: Income inequality and poverty north and south of the Himalayas. Journal of Asian Economics, 17(5).Google Scholar
  9. Bramall, C. (2001) The quality of China’s household income surveys. China Quarterly, 167.Google Scholar
  10. CASS (Chinese Academy of Social Sciences). (1995). Household survey 1995. Beijing: CASS.Google Scholar
  11. CASS (Chinese Academy of Social Sciences). (1988). Household survey 1988. Beijing: CASS.Google Scholar
  12. CASS (Chinese Academy of Social Sciences). (2002). Household survey 2002. Beijing: CASS.Google Scholar
  13. Das, J., & Hammer, J. (2007). Money for nothing: The dire straits of medical practice in Delhi, India. Journal of Development Economics, 83(1).Google Scholar
  14. Deolalikar, A. B. (2010). The performance of Muslims on social indicators: A comparative perspective. In R. Basant & A. Shariff (Eds.), Handbook of Muslims in India. New Delhi: Oxford University Press.Google Scholar
  15. Desai, S. B., Dubey, A., Joshi, B. L., Sen, M., Shariff, A., & Vanneman, R. (2010). Human development in India: Challenges for a society in transition. New Delhi: Oxford University Press.Google Scholar
  16. Deshpande, S., & Yadav, Y. (2006). Redesigning affirmative action: Castes and benefits in higher education. Economic and Political Weekly, 41(24).Google Scholar
  17. Dev, S. M. (2008). Inclusive growth in India: Agriculture, poverty and human development. New Delhi: Oxford University Press.Google Scholar
  18. Drèze, J., & Gazdar, H. (1997). Uttar Pradesh: The burden of inertia. In J. Drèze & A. Sen (Eds.), Indian development: Selected regional perspectives. Delhi: Oxford University Press.Google Scholar
  19. Drèze, J., & Sen, A. K. (1995). India: Economic development and social opportunity. Delhi: Oxford University Press.Google Scholar
  20. Drèze, J., & Sen, A. (2013). An uncertain glory: India and its contradictions. London, Allen Lane, an imprint of Penguin Books.Google Scholar
  21. Fischer, A. M. (2014). The disempowered development of Tibet in China: A study in the economics of marginalization. Lanham, MD: Lexington Books.Google Scholar
  22. GOC (Government of China). (2015). Report of the work of Xinjiang government. Beijing.Google Scholar
  23. GOI (Government of India). (2001). Census of India. New Delhi: Registrar General of India.Google Scholar
  24. GOI (Government of India), Cabinet Secretariat. (2006, November). Social, economic and educational status of the Muslim community in India: A report of the prime minister’s High Level Committee (chaired by Rajinder Sachar). New Delhi.Google Scholar
  25. GOI (Government of India). (2016a). Ministry of Statistics and Programme Implementation. National Sample Survey (NSSO) 71st round on health in India 2014. New Delhi: NSSO.Google Scholar
  26. GOI (Government of India). (2016b). Ministry of Statistics and Programme Implementation. National Sample Survey (NSS) 71st round on education in India 2014. Kolkata: NSSO.Google Scholar
  27. Gupta, I., & Mitra, A. (2004). Economic growth, health and poverty: An exploratory study for India. Development Policy Review, 22.Google Scholar
  28. Hammer, J., Aiyar, Y., & Samji, S. (2007). Understanding government failure in public health services. Economic and Political Weekly, 42.Google Scholar
  29. Hannum, E. (2002). Educational stratification by ethnicity in China: Enrollment and attainment in the early reform period. Demography, 39(1).Google Scholar
  30. Hasan, Z. (2009). Politics of inclusion: Castes, minorities, affirmative action. New Delhi: Oxford University Press.Google Scholar
  31. Iredale, R., Bilik, N., & Wang, S. (2001). Contemporary minority migration, education and ethnicity in China. Cheltenham: Edward Elgar.Google Scholar
  32. John, R. M., & Mutatkar, R. (2005). Statewise estimates of poverty among religious groups in India. Economic and Political Weekly, 40(13), 1337–1345.Google Scholar
  33. Khan, A. R. (2008). Growth, inequality and poverty: A comparative study of China’s experience in the periods before and after the Asian crisis. In B. Gustafsson, S. Li, & T. Sicular (Eds.), Inequality and public policy in China. Cambridge: Cambridge University Press.Google Scholar
  34. Khan, A. R., & Riskin, C. (2005). China’s household income and its distribution, 1995 and 2002. China Quarterly, 182.Google Scholar
  35. Knight, J., & Song, L. (1999). The rural-urban divide: Economic disparities and interactions in China. Oxford: Oxford University Press.CrossRefGoogle Scholar
  36. Panagariya, A., Chakraborty, P., & Rao, G. (2014). State level reforms, growth, and development in Indian states. New York: Oxford University Press.CrossRefGoogle Scholar
  37. Pyatt, G. (1976). On the interpretation and disaggregation of Gini coefficients. Economic Journal, 86(342).Google Scholar
  38. Radhakrishna, R. (2015). Well-being, inequality and poverty and pathways out of poverty in India. Economic and Political Weekly, 50(41), 59–71.Google Scholar
  39. Radhakrishna, R., & Ray, S. (Eds.). (2005). Handbook of poverty in India. New Delhi: Oxford University Press.Google Scholar
  40. Radhakrishna, R., Ravi, C., & Reddy, B. S. (2011). State of poverty and malnutrition in India. In M. Mohanty (Ed.), Council of Social Development, India: Social Development Report 2010. New Delhi: Oxford University Press.Google Scholar
  41. Ramachandran, V. K. (1997). On Kerala’s development achievements. In J. Drèze & A. Sen (Eds.), Indian development: Selected regional perspectives. Delhi: Oxford University Press.Google Scholar
  42. Rao, C. H. H. (2011). India and China: A comparison of the role of sociopolitical factors in inclusive growth. Economic and Political Weekly, 66(16).Google Scholar
  43. Riskin, C. (1994). Chinese rural poverty: Marginalized or dispersed? American Economic Review—Papers and Proceedings, 84(2), 281–284.Google Scholar
  44. Sangay, L. (1998). Education rights for Tibetans in Tibet and India. In J. D. Montgomery (Ed.), Human rights: Positive policies in Asia and the Pacific rim. Hollis, NH: Hollis.Google Scholar
  45. Sautman, B. (1999). Expanding access to higher education for China’s national minorities: Policies for preferential admissions. In G. Postiglione (Ed.), China’s national minority education: Culture, schooling and development. New York: Falmer Press.Google Scholar
  46. Sen, A. (2011). Quality of life: India vs. China. New York Times Review of Books, 58(8).Google Scholar
  47. Sen, G., Iyer, A., & George, A. (2002). Structural reforms and health equity: A comparison of NSS surveys, 1986–87 and 1995–96. Economic and Political Weekly, 37.Google Scholar
  48. Sengupta, S., & Gazdar, H. (1997). Agrarian politics and rural development in West Bengal. In J. Drèze & A. Sen (Eds.), Indian development: Selected regional perspectives. Delhi: Oxford University Press.Google Scholar
  49. Shariff, A. (1999). India Human Development Report. New Delhi: Oxford University Press.Google Scholar
  50. Tarrozzi, A. (2008). Growth reference charts and the nutrition status of Indian children. Economics and Human Biology, 6(3).Google Scholar
  51. Thorat, S. K., & Dubey, A. (2012). Has growth been socially inclusive during 1993-94–2009-10? Economic and Political Weekly, 47(10).Google Scholar
  52. Thorat, S. K., & Sabharwal, N. (2011, February 10–12). Inter-group inequalities in malnutrition in rural India: Need for socially inclusive policies. Paper presented at the IFPRI International Conference on ‘2020 Vision: Leveraging Agriculture for Improving Nutrition and Health’, New Delhi.Google Scholar
  53. Thorat, S. K., & Sadana, N. (2009). Discrimination and children’s nutritional status in India. Institute of Development Studies (IDS) Bulletin, 40(4).Google Scholar
  54. UNDP. (2003). Human development report. New York.Google Scholar
  55. Upton, J. L. (1999). The development of modern school-based Tibetan language education in the PRC. In G. Postiglione (Ed.), China’s national minority education. New York: Falmer Press.Google Scholar
  56. Wiemer, C. (2004). The economy of Xinjiang. In S. F. Starr (Ed.), Xinjiang: China’s Muslim borderland. Armonk: M.E. Sharpe.Google Scholar
  57. WHO (World Health Organization) and China State Council Development Research Centre (DRC). (2005). China: Health, poverty and economic development. Beijing: WHO; DRC.Google Scholar
  58. Yao, S. (1999). On the decomposition of Gini coefficients by population class and income source: A spreadsheet approach and application. Applied Economics, 31.Google Scholar
  59. Zang, X. (2015). Ethnicity in China. Cambridge: Polity Press.Google Scholar
  60. Zhang, T. (1989). Change in Tibetan population. Beijing: China’s Tibetan Studies Publishing House.Google Scholar

Copyright information

© The Author(s) 2017

Authors and Affiliations

  1. 1.CommugnySwitzerland
  2. 2.University of ReadingReadingUK

Personalised recommendations