(a) epw article: cross-tabs (wow -...

62
THE SPATIAL DIMENSION OF INTER-GENERATIONAL EDUCATION ACHIEVEMENT IN RURAL INDIA Anirudh Krishna Abstract A strong positive inter-generational trend has been found in relation to primary education in a group of 40 villages of two Indian states: Rajasthan in the north, and Karnataka in the south. This rising trend, however, falls sharply at the higher end of school education: fewer than 20 per cent of young adults have completed middle school and less than 5 per cent have completed high school, representing almost no increase upon the previous generation. An examination of who attends (and who does not attend) high school, by using both quantitative and qualitative methods, shows how—along with parents’ education, household wealth, religious and caste group—distance to middle and high schools, and transportation costs are consistently significant. An assessment of national data also shows how education in rural India more generally has a spatial dimension. The more rural one’s location, that is, the further one is located from towns. the less likely it is that one would be able to acquire a high school education. Enabling a larger share of the population to participate more effectively in the high-speed economy thus necessitates a reduction in spatial handicaps. Professor of Public Policy and Political Science, Duke University, Durham, NC, Email: [email protected] 1

Upload: others

Post on 09-Jul-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

THE SPATIAL DIMENSION OF INTER-GENERATIONAL

EDUCATION ACHIEVEMENT IN RURAL INDIA

Anirudh Krishna

Abstract

A strong positive inter-generational trend has been found in relation to primary

education in a group of 40 villages of two Indian states: Rajasthan in the north, and

Karnataka in the south. This rising trend, however, falls sharply at the higher end of

school education: fewer than 20 per cent of young adults have completed middle school

and less than 5 per cent have completed high school, representing almost no increase

upon the previous generation. An examination of who attends (and who does not attend)

high school, by using both quantitative and qualitative methods, shows how—along with

parents’ education, household wealth, religious and caste group—distance to middle and

high schools, and transportation costs are consistently significant. An assessment of

national data also shows how education in rural India more generally has a spatial

dimension. The more rural one’s location, that is, the further one is located from towns.

the less likely it is that one would be able to acquire a high school education. Enabling a

larger share of the population to participate more effectively in the high-speed economy

thus necessitates a reduction in spatial handicaps.

Keywords: Educational disparities, Inter-generational change, Distance to the nearest

town

India provides an awkward combination of widespread illiteracy alongside rapid

economic growth. As late as 2011, only 74 per cent of the population aged seven years

Professor of Public Policy and Political Science, Duke University, Durham, NC, Email: [email protected]

1

Page 2: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

and above was barely literate, with this figure being 17 per cent higher among men as

compared to women, which can be seen as evidence of a large and continuing gender

gap. The rural–urban gap also remains large; while 85 per cent of the urban population

was literate in 2011, the corresponding proportion in rural areas was only 68 per cent.

Large gaps in educational achievement also exist among other population groups. Only

55 per cent of the Scheduled Castes (SCs) and only 47 per cent of the Scheduled Tribes

(STs) were literate in 2001,1 figures which are considerably lower than the average figure

for all Indians, which in itself is low as compared to other countries at equivalent levels

of development.2

Despite an increase of nine percentage points in literacy between 2001 and 2011, more

than one-quarter of India’s entire population of over one billion people—and much

higher proportions of specific sub-groups—are unlettered, which is a shame, because it

seems unlikely that many among them derive much benefit from national economic

growth. “The importance of ‘credential capitalism’ has increased greatly in the era of

globalization” (Deshpande, 2006, p. 2441). Analyses have shown that acquiring merely a

primary school education no longer adds as much as once it might have to an individual’s

earning capacity (Mohanty, 2006; Psacharopoulos and Patrinos, 2004). At a minimum,

secondary education makes a difference to an individual’s earning prospects, with this

threshold expected to continue rising higher in the years to come (Sarkar and Mehta,

2010).

What should be done to extend educational opportunities, so that a larger proportion of

the population, studying through high school and beyond, can become active partners in

1 The term ‘Scheduled Caste’ (SC) refers to the former untouchables and Scheduled Tribe (ST) refers to what are, loosely speaking, India’s aborigines. These categories are recognized by India’s Constitution, which provides schedules listing specific castes and tribes as SCs and STs, respectively. The category of Other Backward Castes (OBCs) is a more recent administrative listing, and it refers to caste groupings that are neither upper caste nor listed in the schedules for SCs and STs. ‘General’ is a residual category that I use to include all the castes that are not formally denoted as SCs, STs or OBCs; these are the upper castes, including Brahmins and traders.

2 These figures, drawn from the official 2001 Census of India, are reported on the website of the Indian Government’s National Literacy Mission (www.nlm.nic.in). The literacy rates in comparable Asian countries are much higher: 93 per cent in Sri Lanka and Vietnam, and 95 per cent in the Philippines.

2

Page 3: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

building future growth while sharing equitably in these gains? This problem assumes

special significance in a society, such as that of rural India, wherein a very large number

of parents and grandparents are entirely unlettered or barely literate.

Rapid inter-generational change is required in a context wherein the factors associated

with such changes are barely known. Relatively few studies are available that have

examined inter-generational changes in education within India (or for that matter, which

have examined inter-generational changes in occupational status, incomes, or well-being,

in general).3 A disparate set of conclusions has emerged, providing evidence

simultaneously of stasis and rapid change. On the one hand, Jalan and Murgai (2008) find

encouragingly that “inter-generational mobility in education has improved significantly

and consistently across generations. Mobility has improved, on average, for all major

social groups and wealth classes.” Similarly, Azam and Bhatt (2012) find “significant

improvements in educational mobility across generations in India.” But on the other

hand, Emran and Shilpi (2012, p. 30) support a verdict of persistence, finding that

“educational mobility remained largely unchanged for a large proportion of Indian

children after a decade and a half of high economic growth. Between 1992-93 and 2006,

the only group that experienced significant improvements in educational mobility is

women in urban areas and more developed states.” Kumar, et al. (2002b, p. 4096)

conclude that “there has been no systematic weakening of the links between father’s and

son’s positions… The dominant picture is one of continuity rather than change.” In the

same vein, Majumder (2010, p. 463) uncovers “strong inter-generational stickiness in

both educational achievement and occupational distribution,” especially among SCs and

STs.

How can these contradictory positions be reconciled? Is there or is there not any

significant inter-generational improvement in educational achievement in India? More

importantly, what factors help and what others hinder progress?

In order to push forward the frontiers of available knowledge on these issues—by more

directly investigating change at the margin, among younger age –cohorts—a detailed

3

Page 4: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

examination of educational trends was undertaken in 2008 in 40 rural villages of two

states, Karnataka and Rajasthan. While Rajasthan has been traditionally backward in

terms of social development (in 2011, it ranked just above Arunachal Pradesh and Bihar),

Karnataka is close to the median among Indian states. These two states, therefore, present

a contrast, but not between extremes. A comparison of the rural areas in these states,

which are home to large numbers of SCs and STs, and other hitherto excluded groups,

and an assessment of a combination of quantitative and qualitative data, helped illuminate

some noteworthy trends.

The changes observed at the margin in these villages give reason for considerable

optimism, suggesting that illiteracy in rural India is largely a feature of the past, being

widespread among older people, but disappearing fast among younger ones. Younger

villagers, that is, those between 11 and 20 years of age, have completed primary school in

numbers that are more than twice as large as those for villagers aged 31-40 years—and

more than five times those for villagers aged 61 years and above. Women, as well as

those belonging to the SCs and STs, have shared in these gains. While the average for

these sub-groups remains behind the average for all villagers, their achievements have

also risen rapidly with every succeeding five-year age cohort. Among villagers aged 11-

15 years, the SCs in Rajasthan are now at par with the average for all groups, while those

in parts of Karnataka are slightly ahead of the overall average. If one is seeking to find

evidence of rapid inter-generational progress, the primary level of education is the best

place to look.

This positive picture, however, loses much of its lustre when one considers higher levels

of school education. Nearly everyone attends school at the primary level, but very few

stay in school for more than seven or eight years, with less than five per cent going on to

high school. Even as the figure for primary education has increased by leaps and bounds,

that for high school education seems frozen in time, moving glacially across generations.

Simultaneous verdicts of rapid progress and no change are, therefore, concurrently

maintainable.

4

Page 5: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

What factors help explain this peculiar combination of encouraging and discouraging

results? How can rapid growth at one end of the school education spectrum be reconciled

with almost no growth at the other end? The data from the study villages in Rajasthan and

Karnataka helped identify a range of factors that make a difference to these results.

Prominent among these factors are sheer physical distances (from villages to high schools

and middle schools), which are consistently significant. The fact that a spatial dimension

attaches to educational achievement more generally across rural India is demonstrated

later with the help of data derived from two nationally representative sample surveys.

EXAMINING INTER-GENERATIONAL CHANGES IN RURAL KARNATAKA

AND RAJASTHAN

Alternative explanations corresponding to diverse demand- and supply-related factors are

examined in the grassroots study presented below. Field research was conducted during

the second half of 2008 by employing a combination of qualitative and survey methods.

Because of budgetary constraints, only two districts were selected in each state: Dharwar

and Mysore in Karnataka, and Ajmer and Udaipur in Rajasthan. Within each district, ten

villages were selected purposively in order to capture a range of diversity, including

population size, distances from markets and major roads, and differences in population

composition, including different caste and religious groups. While these villages are not

statistically representative of the state or even of any particular region, they nevertheless

capture a wide range of variations in terms of socio-economic and population groups, and

location considerations.

Two survey teams, each composed equally of men and women belonging to the local

areas, were trained intensively for two weeks. They made three types of inquiries within

each village surveyed. First, a census of village residents was taken, focusing upon the

educational achievement of each member of every household. A total of 26,134

individuals were resident in these Rajasthan villages, and 23,039 individuals in the

selected Karnataka villages, at the time of survey. The collection of information on the

5

Page 6: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

educational achievements of all these individuals facilitated a closer examination of inter-

generational change and transmission effects.4

Next, the survey teams selected a random sample of all villagers in the age cohort of 14-

22 years. Extensive semi-structured interviews were undertaken with these individuals,

and separately, with one or both parents. A total of 722 such parent–child pairs were

interviewed in Rajasthan, and 760 such pairs were interviewed in Karnataka. Semi-

structured interviews helped elicit more fine-grained information regarding the reasons as

to why individuals either persevere with or drop out of middle and high schools.

We also interviewed a sample of school-teachers in each of these 40 villages. A total of

40 teachers in Rajasthan and 38 teachers in Karnataka were interviewed. The teachers

and parents also gave their opinions about why some students had dropped out while

others continued with their education, and they also suggested improvements that could

help reduce the drop-out rates and improve the quality of education.

The various factors that according to analysts influence the demands of individuals and

families for education include the effects of wealth, gender, family size, parents’

education, and school and teacher quality. Greater income and wealth have been found to

be directly related to the demand for higher education, and poorer families are under

greater pressure to withdraw their children from school. A gender difference has also

been found to be important: the demand for educating girls is lower than the demand for

boys’ education in many contexts.5 Larger family size also tends to be associated with a

lower demand for education, as bigger families have less to spend on each individual

child.6 Inter-generational effects are reflected in the higher value that more educated

parents place on educating their children, and consequently, parents’ education is

expected to have a positive influence upon demand.7 In the Indian context, social group,

as denoted by caste and religion, is also important. Lower castes have traditionally been

less able or less willing to attend schools, as evidenced by the overall low achievements

of SCs and STs.8

6

Page 7: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

The relative effects of these factors are examined below. Unfortunately, one more factor,

related to school and teacher quality, could not be examined here. These effects have

been explored in different countries by, for example, Heyneman and Loxley (1983), and

Lloyd, et al. (2000). There is, however, little consensus either about how to measure

these variables or about the likely size of their effects.9 Teacher absenteeism and an

overall low quality of education in public schools are crucially important, as pointed out

by different analysts, and it stands to reason that working upon these factors would serve

to enhance the supply of education, thereby simultaneously helping raise demand.10

Unfortunately, we are unable to test for the relative effect of these factors; the data simply

do not show any considerable variation across schools and villages within each of these

states. We must leave for later consideration an examination of how school quality

matches up with the other effects examined below.

School location has been identified as another relevant factor. Cross-country analysis

shows that proximity to school can be a major influence upon enrolment rates (Kochar,

2004; World Bank, 2006). Primary schools are located within every village studied, but

secondary and especially high schools are usually located some distance away. While

convening focus groups of parents in each village, we questioned them about the distance

to secondary and high schools along with the availability of public transportation (which

in these villages consists mainly of public buses). We also inquired about some other

aspects of socio-economic development, such as the distance to the nearest market, the

type of road linking the village to the nearest highway, and general economic well-being.

These focus groups also itemized the expenses usually associated with each level of

education, that is, primary, middle and high school, respectively. Collectively, these

different items of information helped in the construction of a fairly detailed picture.

THE RISING TIDE OF PRIMARY EDUCATION

As noted briefly above, these data provide a clear indication of an enormous inter-

generational increase in functional literacy. The percentage of villagers going to school

for five or more years has more than doubled in both states over the preceding 20 years,

7

Page 8: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

with a faster rate of inter-generational increase in Karnataka villages.11 Table 1 provides

the aggregate figures for the villages of both the states.

PRESS!—Insert Table 1 here.

A reading of Table 1 from right to left makes clear the extent of inter-generational gains.

Among villagers aged 61 years and older, no more than 15 per cent in Karnataka and 11

per cent in Rajasthan are functionally literate, as defined above. Among those presently

in the age cohort of 31-40 years, who attended primary schools 20 to 30 years ago, no

more than 32 per cent in the Karnataka villages and 29 per cent in the Rajasthan villages

have acquired five or more years of school education. Contrast these numbers with the

corresponding ones for the age cohort of 11-20 years: 85 per cent in the villages of

Karnataka and 64 per cent in those of Rajasthan have acquired five or more years of

school education.

Inter-generational change in educational attainment is, therefore, quite remarkable. Two

generations ago, only a small minority of villagers—11 per cent in Rajasthan and 15 per

cent in Karnataka—had acquired functional literacy. One generation ago, the

corresponding numbers were 29 per cent and 32 per cent, which were still a minority. In

the generation that is now coming of age—those aged 11-15 years—a vast majority of

villagers, including both men and women, are functionally literate.

Increases in primary school attendance continue unabated, as the data for different age

cohorts make clear. In the age cohort of 11-15 years, 89 per cent of males and 67 per cent

of females have attended school for at least one year in these Rajasthan villages—up

from 36 per cent of males and just 4 per cent of females in the age cohort of 41-60 years.

In the villages of Karnataka, as many as 97 per cent of males (and an equal percentage of

females) in the age cohort of 11-15 years have attended school for at least one year—up

from 52 per cent of males and 21 per cent of females in the age cohort of 41-60 years.

8

Page 9: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Not all –village residents have, however, shared equally in these achievements. Table 2

separates the data for Rajasthan for functional literacy by age cohort, gender, district, and

major caste group. If one picks any row of this table and reads the numbers from right to

left, it can be seen that younger age cohorts have steadily higher percentages of

functional literacy. Across districts and caste groups, functional literacy has increased

steadily. Overall, 75 per cent of males and 54 per cent of females in the age cohort of 11-

15 years are functionally literate, signifying a huge increase over the corresponding

figures posted by the age cohort of 61-plus years, which are 17 per cent for males and

only 1 per cent for females.

PRESS!—Insert Table 2 here.

Despite experiencing inter-generational improvements, STs remain far behind the other

caste groups. Among the 11-15 year age cohort, ST men are 28 percentage points behind,

while ST females are 30 points behind the average for their age group. In terms of

religious groupings, the figure for Muslims, 74 per cent, is not statistically

distinguishable from the average for these villages.

Large differences continue to persist between men and women in Rajasthan. In the age

cohort of 31-40 years, 46 per cent of the men but only 9 per cent of the women are

functionally literate. The size of the gender gap is 37 per centage points for this age

cohort. It increases slightly (to 38 per centage points) for the next younger age cohort (of

26-30 years), and it increases further to 50 per centage points for the age cohort of 21-25

years, before beginning to fall. In the age cohort of 11-15 years, the corresponding gender

gap is lower, at 21 per centage points.

Gender differentials continue to remain especially large within particular caste groups

and sub-groups. The largest gender differentials are found within Ajmer’s OBCs,

followed by SCs also of Ajmer district (27 per centage points). Interestingly, it is not the

SCs—the former untouchables—who are worst off in terms of educational achievement.

Some cultivating and livestock-rearing castes, not as low as SCs in terms of ritual status,

9

Page 10: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

nevertheless have considerably lower educational achievement, particularly among

females.12

The story in the villages of Karnataka is broadly similar but different in a few important

respects (Table 3). Overall, functional literacy is higher in Karnataka as compared to the

villages in Rajasthan—and this difference is mostly accounted for by the higher

achievement of women.

PRESS!—Insert Table 3 here.

The gender gap in Karnataka is much smaller than in Rajasthan. In the 11-15 year age

cohort, there is virtually no difference between men and women for OBCs, and among

STs and Muslims, a higher percentage of women than men are functionally literate in

rural Karnataka. Among SCs, the gender difference is small in Mysore but somewhat

larger in Dharwar.

A comparison across the two states indicates that differences across men of the same age

cohort are quite small but differences between women are large across the board. Women

in both the states are becoming more literate with each passing age cohort, but this

increase has been faster in Karnataka as compared to Rajasthan. The largest gender gap

in Karnataka, within the 11-15 year age –cohort, is 13 percentage points, which is

observed among the SCs of Dharwar district. Although large and troubling, this gap is

still substantially smaller than the gap of 40 per centage points observed among the OBCs

of Ajmer in the same age cohort.

In respect of Given the lower thresholds ofaverage figures for basic literacy and primary

education, these data, looking at changes at the margin, that is, among younger

generations, provide reasons for optimism. Other recent studies also report percentages

for primary level enrolments that are far higher than the average national literacy figures

(ASER, 2011), thereby representing a rising tide of primary education.

10

Page 11: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

This rising tide, however, depicts only one part of the school education story. The other,

darker part consists of the lack of progress at the middle, and especially the high, school

levels. Of the entire population in these Karnataka villages, only 19 per cent have

acquired more than eight years of education, while only 5.5 per cent have acquired more

than 11 years of education. In the villages of Rajasthan, these numbers are smaller still:

only 9.5 per cent of the population has acquired more than eight years of education and

only 3.1 per cent have been educated for 11 years or more. Worryingly, unlike the

numbers for primary education, those for high school education are not rising perceptibly

among younger villagers. Of all those currently aged 20-24 years, only 3.5 per cent in the

villages of Rajasthan and 5.7 per cent in those of Karnataka have graduated from high

school, which shows hardly any significant increase over the older generation.

WHY ARE THE NUMBERS SO LOW IN HIGH SCHOOL?

Why do so many villagers—though still not all—attend school through the primary level,

only to drop out at the middle school and especially the high school levels? What factors

are associated with villagers going to school and remaining in school? In this section and

the next one, we will look at some answers to these questions. First, let us consider the

sub-set of villagers aged 18 years and younger; it is within this age cohort that the largest

potential exists for enhancing education. We look first at the number of years of

education as the dependent variable, while examining the factors that are significantly

associated with spending extended periods in school. These results are presented first for

Rajasthan, and next for Karnataka. To the extent that the same explanatory factors

achieve significance in both states, a broader claim can be advanced.

PRESS!— Insert Table 4 here.

Table 4 presents the results for Rajasthan.13 The dummy variable, Ajmer, coded 1 if the

respondent’s village is located within Ajmer district (and zero if it is located in Udaipur

district), has a significant coefficient of -0.41, showing that everything else being the

11

Page 12: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

same, a resident of Ajmer district is likely to attend school for about one-half of one year

less than a resident of Udaipur district.

These results also show how the factors, distance to middle school and bus service, are

both significant and have the expected negative sign. Children from more distant villages

are less likely to attend school for one additional year, particularly if the village is not

served by a bus route.14 Among household level variables, household size is significant.

However, the related coefficient is very small.

Inter-generational transmission effects are significant and positive. Since an extended

family structure is quite common, especially in Rajasthan, we consider the effects of the

highest educated adults in a family, looking separately at male and female adults aged 21

years or older. Both variables—the highest level of adult female education and the

highest level of adult male education, corresponding to the highest education level

attained, respectively, by adult females and adult males of the respondent’s family—are

highly significant. The size of these effects is also roughly equivalent to those that Dreze

and Kingdon (2001) found earlier: each additional year of adult female education is

associated with a 3 per cent increase in a child’s education, while each additional year of

adult male education is associated with a 9 per cent increase. Since the gender gap in

education has always remained high in Rajasthan, these inter-generational effects are

much stronger for adult males as compared to adult females. In the case of Karnataka,

which is considered below, the education levels of adult males and adult females have a

roughly equal effect upon a child’s educational prospects.

Distinct caste and religious groupings have significantly different patterns of educational

attainment. The variable, SC, takes the value 1 if the household concerned belongs to an

SC; ST and OBC are similarly constructed; with General caste serving as the comparison

group. It is also important to examine differences on account of religion, and the dummy

variable Muslim has a similar construction, with Hindus, the majority religion, serving as

the relevant comparison group. Everything else being the same, OBCs and SCs tend to

have, respectively, 0.75 and 0.62 fewer years of education as compared to the General

12

Page 13: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

castes. The highest deficit of this type is suffered by STs, who have, on an average, 1.47

fewer years of education. Muslims tend to have about one year of education less than the

average for these villages.

Gender matters significantly in the villages of Rajasthan, The dummy variable Gender is

coded 1 for male respondents and zero for female respondents. Men tend to have, on an

average, an extra 0.70 years of education as compared to women, even though these

caste- and gender-based deficits have been declining in recent years.

Finally, wealth and age are both significant. The possession of each additional asset

(among a total of 12 assets examined here) is associated, on an average, with an extra

0.22 years of education. On an average, the households in these villages of Rajasthan

possess 4.3 assets and the standard deviation is 2.1. Thus, a difference of one standard

deviation in terms of wealth is associated with an additional 0.45 years of schooling, a

significant effect but not an excessively large one.

A similar analysis was conducted for the villages in Karnataka (Table 5). The factors

found to be commonly significant in both the states include distance to middle school and

bus service, household size, and adult female education. The sizes of the related

coefficients are also about the same across these states.

Another group of factors is also commonly significant, but the size of these effects is

considerably smaller in Karnataka. Thus, STs and Muslims suffer deficits but of much

smaller magnitudes, at 0.27 years and 0.29 years, respectively. Similarly, the wealth

variable also has a smaller effect, being about one-half of the size experienced in the

villages of Rajasthan. The variable for highest adult male education in the family also has

a lower effect in Karnataka.

PRESS!—Insert Table 5 here.

13

Page 14: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Most interestingly, two factors that are significant in Rajasthan—SC and Gender—are

not significant in Karnataka. Women are statistically indistinguishable from men in terms

of educational achievement. Similarly, SCs in Karnataka do not have significantly lower

(or higher) education levels than other caste groups.

Together, these results show that reducing distance to middle schools and extending

public transportation would help raise educational attainment significantly. The education

levels of parents and those of other adult family members also make a significant

difference. However, these variables, together with some others found to be significant

(such as caste), are immutable, especially in the short- to medium-term, and are not as

easily amenable to policy intervention as are distance and transportation.

WHO STUDIES BEYOND THE PRIMARY LEVEL?

The importance of reducing distance to schools and of public transportation was also

highlighted when those acquiring higher levels of school education—those most at risk—

were separately analysed. This section present the results of binary logistic regressions

undertaken to identify the factors associated with not attending school beyond eight

years.

What individual and village characteristics are associated with attending school for more

than eight years? Are the same factors that we found significant for education overall also

significant in terms of these higher levels of education?

Table 6 reports these results for Rajasthan, while examining the sub-group of village

residents aged between 12 and 18 years. Modelling the probability that an individual does

not attend school for more than eight years, it is found that the dummy variable for Ajmer

is no longer significant. At higher levels of education, there is no significant difference

across these two districts of Rajasthan.

14

Page 15: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

PRESS!— Insert Table 6 here.

The variables for bus service and distance to middle school continue, as before, to be

significant, and the variable for distance to high school additionally gains significance in

this analysis. The size of the odds ratio for bus service indicates that, everything else

being the same, the availability of bus service in a village reduces by 24 per cent the odds

that an individual would not attend school beyond eight years. An extra distance of one

kilometer to middle school raises by 6 per cent (and an extra one distance of one

kilometer to high school raises by 16 per cent) the odds that an individual would will not

study beyond eight years.

Among household level variables, the education levels of adults, gender, wealth and caste

are all found to be significant in Rajasthan. Each additional year of education achieved by

the highest educated female in the family lowers by 8 per cent, and each additional year

of education by the highest educated male lowers by 16 per cent, the chances that an

individual would not study beyond eight years.

As compared to the General castes, Muslims are eight times more likely to not study

beyond eight years, while STs and SCs are respectively, 6.3 times and 88 per cent more

likely to not study beyond eight years.15 The odds that a male does not study beyond eight

years are 62 per cent lower as compared to females. Wealth also matters. Each extra

household asset is associated with lowering by 46 per cent the odds that a member of the

household does not study beyond eight years.

Table 7 provides the corresponding results for the villages of Karnataka. Bus service and

distances to middle and high schools have the same significance as in the villages of

Rajasthan, showing that these variables can and should be operated upon more generally

for raising education levels.

PRESS!— Insert Table 7 here.

15

Page 16: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Inter-generational transmission effects are also of roughly the same magnitude as in

Rajasthan. As before, the size of caste and gender disparities is much smaller in

Karnataka, and the wealth effect is also relatively smaller in this region. The dummy

variables for SC and Gender are not significant in this analysis, showing that even in

terms of higher education levels, females and SCs in Karnataka do not suffer from any

significant deficit as compared to the other groups. On the other hand, STs and Muslims

do have significantly higher odds of not progressing beyond eight years of education, but

these odds ratios are also much lower than the comparable ones for Rajasthan.

In general, regardless of how the data are analysed, two sets of factors are consistently

revealed as being important. First, caste, wealth, parents’ education, and gender-related

factors are significant; they matter more in Rajasthan and less in Karnataka.

What is more relevant for short- to medium-term policy intervention is reduction of

distances to middle and high schools, or at least, making regular transportation available

in all villages, which would considerably help enhance enrolments beyond the primary

level. While primary schools are located within nearly all villages, middle and especially

high schools continue to be located some distance away. Among the 40 villages

considered here, the longest distance to middle school is eight kilometers (from Satyagala

village of Mysore district) and the longest distance to high school is 10 kilometers (from

Veerapur village of Dharwar district). It is not easy for a young student to travel such

distances every day on foot. The deterrent effect of longer distance is particularly severe

for girls; once they reach puberty, parents are increasingly unwilling to let them travel

very far from their homes. The data show that the gender gap is consequently the greatest

in villages located furthest from towns and high schools.

Improving physical access would go a long way toward raising the currently low rates of

middle- and high-school attendance. Apart from its deterrent effects in terms of time and

trouble, distance also has monetary implications, which is especially burdensome for

lower income groups.

16

Page 17: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Even where school education is notionally free or when a nominal tuition fee is levied by

the state, families still have to incur considerable costs on sending children to school

(Nambissan, 2001; Tilak, 2002). While the costs associated with primary school are

relatively small, the costs rise higher for middle and especially for high schools, with

transportation constituting a sizeable component, especially in the villages located at

greater distances from towns and from high schools. The focus groups convened in each

of the 40 villages helped put together estimates of the average expenses incurred by

families. An average annual amount of Rs. 845 is spent by families in the villages of

Rajasthan on sending a child to primary school. In Karnataka, the corresponding figure is

Rs. 875. Books and stationery (costing Rs. 275), together with school uniforms (costing

Rs. 335 in Rajasthan and Rs. 430 in Karnataka), make up the bulk of these expenses. For

sending a child to middle school, families spend an average of Rs. 1,225 in villages

where transportation costs are not involved and up to another Rs. 700 when children have

to be sent by bus to middle schools. The average annual costs for high school come in at

about Rs. 2,600, which was equivalent at the time of investigations to almost an entire

month’s earnings for a day labourer in these villages. A very large part of these costs is

accounted for by transportation expenses, which on an average, account for 33 per cent of

the total, rising to 67 per cent in the more distant villages.

While it is important, cost does not, by itself suffice to explain why so many children,

relatively rich and relatively poor, drop out after completing primary or middle school.

Parents in Rajasthan and Karnataka have a strong desire to send their children to school.

In their list of priorities, expenditure on education far outranks many other investments

and expenditures. Our surveys revealed that a little less than 90 per cent of the parents in

Karnataka (and more than 90 per cent in Rajasthan) would rather spend a given amount

of money on children’s education than on either home improvement, or purchasing

livestock, or digging irrigation wells. Less than 30 per cent of the students and parents

mentioned high costs (or the need to get paid work at an early age) among the three

principal reasons for dropping out of school. In comparison, many more mentioned

distance and lack of suitable transportation. Other parts of the qualitative data upheld a

similar conclusion. Among a randomly selected sample of students and parents who were

17

Page 18: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

asked for suggestions about ways to raise enrolments in middle and high schools, an

overwhelming majority picked locating middle and high schools closer to their villages.

Their second highest priority was for ensuring reliable and safe bus routes.

NOT JUST IN THESE VILLAGES: OVERLAPPING DISABILITIES OF

DISTANCE

Distance matters, not only in these states and villages, but more broadly across rural

India. Three pieces of evidence derived from two nationally representative data sets

(NCAER and DLHS-3) show commonly how education achievement falls off at greater

distances from towns.16

Towns in India represent locations wherein higher education facilities and other social

and physical infrastructure tend to be co-located.17 Table 8 shows how the distance to

town correlates with the distance to all higher educational facilities: as the distance from

town increases, the distances to the nearest middle school, secondary school, and higher

secondary school also tend to increase.

The average distance to the nearest higher secondary school from the villages located ten

or more kilometers from town is a daunting 13.4 kilometers. Young people from such

villages who wish to continue with school at this level must be willing and able to take on

this additional hurdle twice a day.

PRESS!— Insert Table 8 here.

It is not only schools, however, to which rural people living further from towns have

greater difficulties in obtaining access. A whole slew of physical and social infrastructure

is more accessible in villages located closer to towns—and less available in the villages

located further away. Distances to public health centres and bank branches, to take just

two examples, increase monotonically with greater distance from town. Electricity and

canal irrigation are more often available in the villages closer to towns, and villages at

18

Page 19: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

greater distances are progressively less well served. Telephone connections also tend to

dissipate at greater radial distances from towns. Partly because these services are harder

to access, and partly because of lower educational achievements, as shall be seen below,

poverty is higher within the more distant villages.

One result of greater distance is lower educational achievement. Table 9 shows how the

percentage of individuals across rural India who have acquired education up to middle

school or higher levels tends to fall off at greater distances from towns.

PRESS!— Insert Table 9 here.

Across income quintiles, the share of villagers who have acquired middle school or

higher levels of education is lower in villages located more than 10 km from towns.

Poorer villagers are less likely to have acquired education up to middle school as

compared to richer villagers. However, regardless of the levels of poverty and wealth,

educational achievement is lower at greater distances from towns.

Not just the quantity but also the quality of learning is lower at greater distances from

towns. Table 10 shows how test scores related to different learning abilities—reading,

writing, mathematics, and English proficiency—are all progressively worse at greater

distances from towns.

PRESS!— Insert Table 10 here.

Access to diverse sources of information is also lower in the villages located further

away. NCAER data show that the share of households who listen to radios regularly falls

from 49 per cent in villages located within 5 km of towns to below 43 per cent in villages

located more than 10 km from towns. The corresponding proportions for TV viewing and

newspaper readership fall, respectively, from 62 per cent to 56 per cent, and from 41 per

cent to 34 per cent. Across the board, residents of more remote villages face disabilities,

19

Page 20: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

and the hardships that they face in gaining access to higher education levels make these

overlapping disabilities a particular burden for future achievement.

Education in rural India thus has a distinct spatial dimension. Not only are rural areas less

well-represented among the educated population but the more rural one’s location—that

is, the further one is from towns—the less likely it is that one would be literate, well-

informed, or otherwise prepared to participate effectively in the high-speed economy.

CONCLUSION: THE SPATIAL DIMENSION OF EDUCATIONAL

INEQUALITY

The demand for education is at an all-time high in Indian villages. Considering a sample

of 1,375 households selected from 236 villages of five northern states (Uttar Pradesh

[UP], Madhya Pradesh [MP], Bihar, Himachal Pradesh [HP], and Rajasthan), the PROBE

report (1999) found that as many as 98 per cent of the parents regard it as important for

boys to be educated, while 89 per cent regard education important for girls. Our study in

Rajasthan and Karnataka also shows how as compared to a variety of other uses of

money, large majorities prefer to incur expenditures on their children’s education.

As a result of these increasing demands from below, coupled with the supply of primary

education within, or at least, very close to every village, primary school achievement in

rural India is at an all-time high. In rural Karnataka, more than 90 per cent of all

individuals aged 11-15 years have completed at least five years of education, a state of

affairs that would have been inconceivable 20 (or even 10) years ago. Among villagers

who are 20 years older than this sub-group, being 31-40 years at the time of

investigations, no more than 43 per cent of the males and 22 per cent of the females had

acquired five or more years of school education. In Rajasthan, a state that many regard to

be educationally and socially backward, a quantum leap in primary education has been

experienced over the past 20 years. While only 9 per cent of the females and 46 per cent

20

Page 21: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

of the males aged 31-40 years have acquired five or more years of school education, fully

54 per cent of all females and 75 per cent of all males aged 11-15 years have completed

primary school. The remaining liabilities arising on account of gender, caste status,

religious group, and poverty are being dissipated steadily, as each succeeding age cohort

of children registers a higher enrolment figure for primary school.

The achievement of primary education is much higher for the younger generation as

compared to their mothers and fathers, and especially, to their grandmothers and

grandfathers, thereby signifying spectacular, widespread and unprecedented gains.

Similar gains have been made across the nation: 46 per cent of the mothers of children

currently in school across rural India did not attend school themselves (ASER, 2011). As

further evidence of large inter-generational gains at the primary level, the share of 6-14

year-olds not in school fell to a low of 3.3 per cent in 2011.

The rising tide of primary education, however, falls sharply when one assesses

enrolments in high school, which have remained static at excruciatingly low levels. Only

about 3 per cent of the population in the villages of Rajasthan and less than 6 per cent in

the villages of Karnataka have acquired 11 or more years of school education. These

trends are not rising fast enough; in fact, they appear to have flattened out. Even among

the sub-set of younger villagers, that is, those who are currently 20-24 years of age, no

more than 3.5 per cent in Rajasthan and 5.7 per cent in Karnataka have 11 or more years

of school education.

A number of factors are implicated in understanding as to why the level of high school

attendance remains puny in rural areas. Among them, distance is of considerable

importance. Making it physically easier to get to high schools—over the longer term by

locating schools closer to villages, and more immediately by ensuring the availability of

transportation—would considerably help raise high-school enrolments. One reason as to

why children are attending primary schools in such large numbers has to do with the

proximity of such schools to nearly every village. The currently vast distances that the

residents of many villages face from high schools constitute a considerable deterrent. To

21

Page 22: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

be sure, distance and access are not the only factors that matter. We saw above how caste,

gender, the education levels of parents and household wealth can also make a difference

under different circumstances. Separately, the effects of school and teacher quality,

factors that have not been examined here on account of lack of reliable data, need to be

seen in juxtaposition with other effects. And what students learn by attending schools,

that is, the quality of learning outcomes, is a separate and equally important concern

(ASER, 2011). The consistently significant effect of greater distance cannot, however, be

wished away or attributed to other influences. Policies to improve physical access must

thus form part of the thrust to raise educational achievements in rural India, especially

beyond the primary level.

Table 1

Percentage of Villagers with Five or More Years of School Education

(All Residents of 20 Villages Each of Karnataka and Rajasthan)

Age Cohort (Years)

State 11-20 21-30 31-40 41-60 61+

Karnataka

(n=23,067)85 55 32 25 15

Rajasthan

(n=26,124)64 45 29 18 11

Source: Original data collected by the author.

22

Page 23: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 2

Percentage of Rajasthan Villagers with Five or More Years of School Education

(All Residents of 20 Villages of Ajmer and Udaipur Districts)

Caste Group

District

Age Cohort (Years)

11-15 16-20 21-25 26-30 31-40 41-60 61+

F M F M F M F M F M F M F M

General Castes

Ajmer 84 80 76 96 65 93 47 82 31 73 15 62 - 58

Udaipur 92 88 84 98 64 93 47 94 35 82 16 63 1 40

Other Backward Castes

Ajmer 38 78 21 76 10 76 8 56 3 39 1 27 1 11

Udaipur 65 80 50 87 26 77 10 71 7 52 1 25 1 11

Scheduled Castes

Ajmer 50 77 20 78 12 59 1 56 4 36 1 13 0 4

Udaipur 72 75 38 81 26 81 5 77 4 48 0 36 - 9

Scheduled Tribes

Udaipur 24 47 7 47 2 33 2 26 0 16 0 4 0 3

Muslims

Ajmer 53 74 10 74 67 11 69 0 55 2 33 - 22

OVERALL 54 75 38 79 23 73 14 62 9 46 4 30 1 17

Notes:

1. Cells with less than 30 people were left blank.

2. M=Males; F=Females.

3. OBCs = Other Backward Castes; SCs = Scheduled Castes; STs = Scheduled Tribes.

4. Hardly any STs reside in these villages of Ajmer district, and very few Muslims are found in the

villages of Udaipur district.

Source: Original data collected by the author.

23

Page 24: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 3

Percentage of Karnataka Villagers with Five or More Years of School Education

(All Residents of 20 Villages of Dharwar and Mysore Districts)

Caste Group

District

Age Cohort (Years)

11-15 16-20 21-25 26-30 31-40 41-60 61+

F M F M F M F M F M F M F M

Other Backward Castes

Dharwar 93 93 74 80 52 73 37 59 26 50 16 41 5 37

Mysore 94 95 80 92 60 78 42 66 21 46 18 39 6 27

Scheduled Castes

Dharwar 76 89 46 69 20 75 6 41 5 25 3 16 0 0

Mysore 92 95 76 88 49 72 27 55 16 38 6 25 2 11

Scheduled Tribes

Dharwar 83 75 61 66 34 70 9 44 9 29 9 27 0 25

Mysore 88 86 60 57 22 49 17 36 6 20 2 15 0 7

Muslims

Mysore 91 90 78 74 76 77 45 56 40 31 8 31 8 30

OVERALL 92 92 74 82 51 74 33 58 22 43 14 35 4 27

Notes:

1. Cells with less than 30 people were left blank.

2. M=Males; F=Females.

3. The caste group “General” is not represented in this table because a very small number of people

fall within this category in these districts.

4. Hardly any Muslims were found in the villages of Dharwar district.

5. OBCs = Other Backward Castes; SCs = Scheduled Castes; STs = Scheduled Tribes.

Source: Original data collected by the author.

24

Page 25: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 4

OLS Regression on Years of Education

(All Residents of 20 Rajasthan Villages Not Older Than 18 Years)

Intercept -0.63**** (0.11)

Independent Variables

Village Level Variables Ajmer (dummy) -0.41**** (0.05)

Distance to middle school (kilometers) -0.05** (0.02)

Distance to high school (kilometers) 0.00 (0.12)

Bus service in village 0.13* (0.05)

Household and

Individual Level

Variables

Household size (number) -0.05**** (0.006)

Highest adult female education (years) 0.03** (0.009)

Highest adult male education (years) 0.09**** (0.005)

Muslim (dummy) -1.03**** (0.13)

Other Backward Caste (OBC dummy) -0.75**** (0.07)

Scheduled Caste (SC dummy) -0.62**** (0.09)

Scheduled Tribe (ST dummy) -1.47**** (0.08)

Gender (dummy) 0.70**** (0.04)

Age (years) 0.44**** (0.00)

Wealth (number of assets) 0.22** (0.09)N 10,394R2 0.596Adj- R2 0.595F-value 956.80F-probability <0.0001

Note: *p<=.05; **p<=.01; ***p<=.001; ****p<.0001.Source: Original data collected by the author.

25

Page 26: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 5

OLS Regression on Years of Education

(All Residents of 20 Karnataka Villages Not Older Than 18 Years)

Intercept -2.18**** (0.10)

Independent Variables

Village Level Variables Mysore (dummy) 0.49**** (0.05)

Distance to middle school (kilometers) -0.02* (0.01)

Distance to high school (kilometers) 0.004 (0.007)

Bus service in village 0.17** (0.06)

Household and

Individual Level

Variables

Household size (number) -0.02*** (0.006)

Highest adult female education (years) 0.03**** (0.006)

Highest adult male education (years) 0.03**** (0.005)

Muslim (dummy) -0.29** (0.11)

Scheduled Caste (SC dummy) 0.001 (0.05)

Scheduled Tribe (ST dummy) -0.27**** (0.07)

Gender (dummy) -0.04 (0.04)

Age (years) 0.50**** (0.004)

Wealth (number of assets) 0.11* (0.08)N 8,615R2 0.754Adj- R2 0.754F-value 1,883.54F-probability <0.0001

Note: *p<=.05; **p<=.01; ***p<=.001; ****p<.0001.Source: Original data collected by the author.

26

Page 27: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 6

Results of Binary Logistic Regression for

Education > 8 Years in Rajasthan Villages

(Not Attending School for More Than 8 Years Is the Dependent Variable)

Coefficients Odds Ratios

(95% Wald Confidence

Limits)

Intercept 11.91****

Ajmer -0.22 n.s.

Bus service in village -0.24*** 0.76 (0.62-0.89)

Distance to middle school 0.05** 1.06 (1.01-1.10)

Distance to high school 0.15**** 1.16 (1.13-1.19)

Household size (number) 0.09**** 1.09 (1.05-1.13)

Highest adult female education

(years)

-0.09**** 0.92 (0.88-0.96)

Highest adult male education (years) -0.15**** 0.86 (0.84-0.89)

Muslim (dummy) 2.10**** 8.19 (3.34-19.7)

Other Backward Caste (OBC

dummy)

0.48*** 1.61 (1.22-2.13)

Scheduled Caste (SC dummy) 0.63** 1.88 (1.25-2.85)

Scheduled Tribe (ST dummy) 1.84**** 6.31 (3.48-11.5)

Gender (dummy) -0.73**** 0.48 (0.38-0.61)

Age (years) -0.65**** 0.52 (0.50-0.55)

Wealth (number of assets) -0.61*** 0.54 (0.42-0.78)

-2 Log Likelihood 4,824

Likelihood Ratio Chi-square 2,478

Pr>Chi-Square <0.0001

N= 10,394

Note: *p<=.05; **p<=.01; ***p<=.001; ****p<.0001; n.s.: not significant.

27

Page 28: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Source: Original data collected by the author.

Table 7

Results of Binary Logistic Regression for

Education > 8 Years in Karnataka villages

(Not Attending School for More Than 8 Years Is the Dependent Variable)

Coefficients Odds Ratios

(95% Wald

Confidence

Limits)

Intercept 11.61****

Mysore -0.78**** 0.46 (0.37-0.57)

Bus service in village -0.42** 0.66 (0.50-0.87)

Distance to middle school 0.04**** 1.04 (1.02-1.06)

Distance to high school 0.02** 1.03 (1.01-1.07)

Household size (number) 0.05*** 1.05 (1.02-1.08)

Highest adult female education

(years)

-0.08**** 0.92 (0.90-0.95)

Highest adult male education (years) -0.06**** 0.94 (0.92-0.96)

Muslim (dummy) 0.91*** 2.50 (1.57-3.99)

Scheduled Caste (SC dummy) 0.05 n.s.

Scheduled Tribe (ST dummy) 0.34* 1.40 (1.02-1.92)

Gender (dummy) 0.01 n.s.

Age (years) -0.67**** 0.51 (0.49-0.53)

Wealth (number of assets) -0.21** 0.94 (0.90-0.98)

-2 Log Likelihood 7,135

Likelihood Ratio Chi-square 3,549

Pr>Chi-Square <0.0001

N= 8,615

28

Page 29: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Note: *p<=.05; **p<=.01; ***p<=.001; ****p<.0001; n.s.: not significant.

Source: Original data collected by the author.

Table 8

Greater Distance and Lower Provision across the Board

Distance from the Nearest Town (km)

<=2 km >2 to <=5 km >5 to <=10 km >10 km

Percentage of All Villages 8.1 16.3 25.7 49.9

Percentage of All Households 13 18 27 41

EDUCATION FACILITIES

Average distance from middle school (km) 1.6 3.1 3.8 5.6

Average distance from secondary school (km) 2.7 4.3 5.9 8.6

Average distance from higher secondary school (km) 3.3 6.2 8.2 13.4

OTHER SERVICES

Average distance from basic health facility (sub-centre/PHC/private

clinic)

3.7 5.0 6.3 8.6

Percentage of villages not electrified (domestic purpose) 11 13 16 19

Percentage of households with electricity connection 66.5 61.4 59.5 57.8

Percentage of villages with canal irrigation 15.8 15.7 15.0 12.6

Percentage of villages with bank branches 24.8 10.9 14.3 12.8

Average distance from bank branch (km) 4.6 5.0 7.1 11.4

Percentage of households with TV (colour or black & white) 41.1 35.7 33.6 29.0

Percentage of households with telephone (mobile or fixed line) 37.0 32.8 30.5 25.5

Percentage of households with BPL cards 31.7 33.7 34.5 37.6

Source: DLHS-3 Data (2007-08) (see Endnote 16).

29

Page 30: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 9

Percentage of Adults with Middle School or Higher Level of Education

Income

Quintile

Distance from the Village to the

Nearest Town

<5 km 5-10 km >10 km

Lowest 29 30 23

Q2 26 30 25

Q3 39 33 33

Q4 52 47 38

Highest 67 66 60

AVERAGE 43 41 36

Source: NCAER data for 2004-05 (see Endnote 16).

30

Page 31: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Table 10

Distance and Education Outcomes

  Learning Ability of Children between Ages 8 and 11 (%)

Distance from

the Village to

the Nearest

Town

Reading

Ability (if

the child is

able to read

a word or

more, not

merely just

a letter, in

any one of

the given

languages)

Writing

Ability (if

the child is

able to

write the

sentences

given with

two or less

mistakes)

Computational

Ability (if the

child knows

basic addition

and

subtraction)

English

Language

Proficiency (if

the child is

able to read a

word or more,

not merely just

a letter)

<=5 km 74.6 65.9 46.7 3.5

5-10 km 72.8 63.4 44.8 2.7

>10 km 70.0 59.7 40.4 1.8

Source: NCAER data for 2004-05 (see Endnote 16).

31

Page 32: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

References

Anh, T.S., John Knodel, David Lam, and Jed Friedman (1998). “Family Size and

Children’s Education in Vietnam”, Demography, Vol. 35, No. 1, pp. 57-70.

Asadullah, M. Niaz and Gaston Yalonetzky (2012). “Inequality of Educational

Opportunity in India: Changes over Time and across States”, World Development, Vol.

40, No. 6, pp. 1151-63.

ASER (2011). Annual Status of Education Report (Rural) 2011, Available at

http://images2.asercentre.org/aserreports/ASER_2011/aser_2011_report_8.2.12.pdf,

Accessed on 23 April 2013.

Azam, Mehtabul and Vipul Bhatt (2012). “Like Father, Like Son? Inter-generational

Education Mobility in India”, IZA Discussion Paper No. 6549, Bonn, Germany.

Available at http://ftp.iza.org/dp6549.pdf, Accessed on 23 April 2013.

Buchmann, Claudia and Emily Hannum (2001). “Education and Stratification in

Developing Countries: A Review of Theories and Research”, Annual Review of

Sociology, Vol. 27, pp. 77-102.

Chaudhury, Nazmul, Jeffrey Hammer, Michael Kremer, Karthik Muralidharan and

Halsey F. Rogers (2006). “Missing in Action: Teacher and Health Worker Absence in

Developing Countries”, Journal of Economic Perspectives, Vol. 20, No. 1, pp. 91-116.

Deshpande, Satish (2006). “Exclusive Inequalities: Merit, Caste and Discrimination in

Indian Higher Education Today”, Economic and Political Weekly, 17 June, Vol. 41, No.

24, pp. 2438-44.

32

Page 33: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Dostie, Benoit and Rajshree Jayaraman (2006). “Determinants of School Enrolments in

Indian Villages”, Economic Development and Cultural Change, Vol. 54, No. 2, pp. 405-

21.

Dreze, Jean and Geeta G. Kingdon (2001). “School Participation in Rural India”, Review

of Development Economics, Vol. 5, No. 1, pp. 1-24.

Dubey, Surendra N. (2001). Education Scenario in India—2001, Authors Press, Delhi.

Emran, M. Shahe and Forhad Shilpi (2012). “Gender, Geography and Generations: Inter-

generational Educational Mobility in Post-reform India”, Policy Research Working Paper

6055, World Bank, Washington, DC.

Filmer, D. and Lant Pritchett (1999). “The Effect of Household Wealth on Educational

Attainment: Evidence from 35 Countries”, Population and Development Review, Vol. 25,

No. 1, pp. 85-120.

Fuller, Bruce, Judith Singer and M. Keiley (1995). “Why Do Daughters Leave School in

Southern Africa? Family, Economy and Mother’s Commitments”, Social Forces, Vol.

74, No. 2, pp. 657-80.

Gomes, M. (1984). “Family Size and Educational Attainment in Kenya”, Population and

Development Review, Vol. 10, No. 4, pp. 647-60.

Heyneman, S.P. and W.A. Loxley (1983). “The Effect of Primary School Quality on

Academic Achievement across Twenty-Nine High- and Low-Income Countries”,

American Journal of Sociology, Vol. 88, No. 6, pp. 1162-94.

33

Page 34: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Jalan, Jyotsna and Rinku Murgai (2008). “Inter-generational Mobility in Education in

India”, Paper presented at the Fourth Annual Conference on ‘Economic Growth and

Development’, 17-18 December, Indian Statistical Institute, Delhi, India, Available at:

www.isid.ac.in/~pu/conference/dec_08_conf/.../RinkuMurgai.doc, Accessed on 23 April

2013.

King, E.M. and A. Hill (1993). Women’s Education in Developing Countries: Barriers,

Benefits and Policies, Johns Hopkins University Press, Baltimore.

Knodel, John and G.W. Jones (1996). “Post-Cairo Population Policy: Does Promoting

Girls’ Schooling Miss the Mark?”, Population Development Review, Vol. 22, No. 4, pp.

683-702.

Knodel, John, N. Havanon and W. Sittitrai (1990). “Family Size and the Education of

Children in the Context of Rapid Fertility Decline”, Population and Development

Review, Vol. 16, No. 1, pp. 31-62

Kochar, Anjini (2004). “Urban Influences on Rural Schooling in India”, Journal of

Development Economics, Vol. 74, No. 1, pp. 113-36.

Kremer, M., N. Chaudhury, F. Rogers, K. Muralidharan and J. Hammer (2005). “Teacher

Absence in India: A Snapshot”, Journal of the European Economic Association, Vol. 3,

Nos. 2-3, pp. 658-67.

Krishna, Anirudh (2010). One Illness Away: Why People Become Poor and How they

Escape Poverty, Oxford University Press, Oxford, UK.

34

Page 35: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Kumar, Sanjay, Anthony Heath and Oliver Heath (2002a). “Determinants of Social

Mobility in India”, Economic and Political Weekly, 20 July, Vol. 37, No. 29, pp. 2983-

87.

——— (2002b). “Changing Patterns of Social Mobility: Some Trends over Time”,

Economic and Political Weekly, 5 October, Vol. 37, No. 40, pp. 4091-96.

Lloyd, Cynthia B. and A.K. Blanc (1996). “Children’s Schooling in Sub-Saharan Africa:

The Role of Fathers, Mothers, and Others”, Population and Development Review, Vol.

22, No. 2, pp. 265-98.

Lloyd, Cynthia B., Barbara S. Mensch and Wesley H. Clark (2000). “The Effects of

Primary School Quality on School Drop-out among Kenyan Girls and Boys”,

Comparative Education Review, Vol. 44, No. 2, pp. 113-47.

Majumder, Rajarshee (2010). “Inter-generational Mobility in Educational and

Occupational Attainment: A Comparative Study of Social Classes in India”, Margin—

The Journal of Applied Economic Research, Vol. 4, No. 4, pp. 463-94.

Mare, Robert D. and Vida Maralani (2006). “The Inter-generational Effects of Changes

in Women’s Educational Attainments”, American Sociological Review, Vol. 71, No. 4,

pp. 542-64.

Mohanty, Mritiunjoy (2006). “Social Inequality, Labour Market Dynamics, and

Reservation”, Economic and Political Weekly, 2 September, Vol. 41, No. 35, pp. 3777-

89.

35

Page 36: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Montgomery, M.R. and Cynthia B. Lloyd (1997). “Excess Fertility, Unintended Births

and Children’s Schooling”, Population Council Policy Research Division Working Paper

No. 100, Population Council, New York.

Motiram, Sripad and Ashish Singh (2012). “How Close Does the Apple Fall to the Tree?

Some Evidence on Inter-generational Occupational Mobility in India”, Economic and

Political Weekly Vol. 47, No. 40, pp. 56-65.

Muralidharan, K., J. Das, A. Holla, M. Kremer and A. Mohpal (2012). The Fiscal Costs

of Weak Governance: Evidence from Primary Education in India, University of

California, San Diego.

Nambissan, Geetha B. (2001). “Social Diversity and Regional Disparities in Schooling: A

Study of Rural Rajasthan”, in A. Vaidyanathan and P.R. Gopinathan Nair (eds), Primary

Education in India: A Grassroots View, Sage Publications, New Delhi, pp. 459-517.

Patrinos, H.A. and G. Psacharopoulos (1996). “Socio-economic and Ethnic Determinants

of Age Grade Distortion in Bolivian and Guatemalan Primary Schools”, International

Journal of Educational Development, Vol. 16, No. 1, pp. 3-14.

Pong, S.L. (1997). “Sibship Size and Educational Attainment in Peninsular Malaysia: Do

Policies Matter?”, Sociological Perspectives, Vol. 40, No. 2, pp. 227-42.

PROBE (1999). Public Report on Basic Education in India, Oxford University Press,

New Delhi.

36

Page 37: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Psacharopoulos, George and Harry A. Patrinos (2004). “Returns to Investment in

Education: A Further Update”, Education Economics, Vol. 12, No. 2, pp. 111-34.

Sarkar, Sandip and Balwant Singh Mehta (2010). “Income Inequality in India: Pre- and

Post-Reform Periods”, Economic and Political Weekly, 11 September, Vol. 45, No. 37,

pp. 45-55.

Sathar, Zeba and Cynthia B. Lloyd (1993). “Who Gets Primary Schooling in Pakistan:

Inequalities among and within Families”, Pakistan Development Review, Vol. 33, No. 2,

pp. 103-34.

Shreeniwas, Sudha (1997). “Family Size, Sex Composition and Children’s Education:

Ethnic Differentials over Development in Peninsular Malaysia”, Population Studies, Vol.

51, No. 2, pp. 139-51.

Tilak, Jandhyala B.G. (2002). “Determinants of Household Expenditure on Education in

Rural India”, NCAER Working Paper Series No. 88, National Council of Applied

Economics Research, New Delhi.

World Bank (2006). World Development Report: Equity and Development, World Bank,

Washington, DC.

37

Page 38: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

NOTES

3 See, for example, Asadullah and Yalonetzky (2012); Jalan and Murgai (2008); Krishna (2010); Kumar, et al. (2002a); (2002b); Majumder (2010); and Motiram and Singh (2012).

4 Some individuals were not available within the village at the time of the interview, as they had migrated for work or left for some other reason. Information about such individuals was ascertained from other household members. It was only in rare cases, fewer than 2 per cent in all, that entire households were not available in the village.

5 Analyses that examine wealth and gender effects in developing countries include Filmer and Pritchett (1999); Fuller, et al. (1995); King and Hill (1993); Knodel and Jones (1996); Patrinos and Psacharapoulos (1996); and Sathar and Lloyd (1993). A survey of this literature is provided by Buchmann and Hannum (2001).

6 See, for example, Anh, et al. (1998); Gomes (1984); Knodel, et al. (1990); Montgomery and Lloyd (1997); Pong (1997); and Shreeniwas (1997).

7 For example, Lloyd and Blanc (1996); and Mare and Maralani (2006).

8 See Dostie and Jayaraman (2006); Dreze and Kingdon (2001); and Dubey (2001). See also the Sachar Committee Report, which considers educational achievement by different caste and religious groups (Available at: http://ncm.nic.in/pdf/compilation.pdf, Accessed on 23 April 2013).

9 Reviewing competing findings, Buchmann and Hannum (2001, p. 87) report that, “family factors are more important predictors of educational achievement than are school [quality] factors in most countries”, especially developing ones.

10 A survey undertaken in 2004 found that 25 per cent of public school teachers in India are regularly absent from work. In the two states considered here, teacher absenteeism is a little below this average figure, at 21.7 per cent and 23.7 per cent, respectively, in Karnataka and Rajasthan (see Chaudhuri, et al. 2006; Kremer, et al., 2005). A more recent survey (ASER, 2011), however, reports much lower figures for teacher absenteeism, and there is also additional evidence that school quality has improved (see Muralidharan, et al., 2012).

11 We employ five or more years of school education as a crude measure of functional literacy. In general, at least five years of school education are required for an individual to acquire functional competence in reading, writing, and basic numerical calculation. With less than five years at school, most individuals are not yet functionally literate.

12 In order to probe this peculiar discrepancy, OBCs in Ajmer were further disaggregated. Four specific named castes that constitute the bulk of the OBC grouping in the villages of this district were examined separately. Within the age cohort 11-15 years, the largest gender gap exists among Jats (52 per cent), followed by Gurjars (44 per cent), and Rawats (39 per cent). For the fourth caste in this grouping, the

38

Page 39: (A) EPW ARTICLE: Cross-tabs (wow - Sites@Dukesites.duke.edu/krishna/files/2013/10/THE-SPATIAL-DIMEN…  · Web viewTable 1 provides the aggregate figures for the villages of both

Rajputs, the gender gap is actually reversed: 89 per cent of the women and 82 per cent of the men in the 11-15 age cohort in this caste have five or more years of school education. There seems to be some validation here for the result reported by Dostie and Jayaraman (2006, p. 413) that “livestock ownership significantly reduces the probability of older girls’ school enrolment”. Jats, and especially Gurjars, own large herds of livestock. Only 22 per cent of the Gurjar females and 30 per cent of the Jat females in the 11-15 year age group have five or more years of school education. In the 21-25 age-group, these numbers are infinitesimally small: only 4 per cent of Gurjar females and 3 per cent of Jat females are functionally literate.

13 In order to test for possible serial correlation among members of the same household, a separate set of regression models using robust standard errors was also run, but the coefficients for all independent variables were identical in all cases reported here and below. The Variation Inflation Factors (VIFs) for all independent variables are within the range of 1.01-2.81, showing that multi-collinearity is not a significant problem in this analysis. Similar checks for collinearity were performed, with similar results, for other regression results reported below.

14 As we will see later, a very small number of villagers attend high school. Thus, it is not surprising that Distance to High School is not significant in this analysis. In the analysis reported in the next section, which looks specifically at the sub-group aged 14-22 years, this variable becomes significant.

15 These results are comparable to those found in a nationwide sample survey conducted by the National Council of Applied Economic Research (NCAER) and reported by the Sachar Committee. See http://ncm.nic.in/pdf/compilation.pdf, Accessed on 23 April 2013. 16

? Two waves of sample surveys representative of the rural areas of 16 major states—the India Human Development Surveys—undertaken by the National Council for Applied Economic Research (NCAER), constitute the first data source. These multi-dimensional surveys, conducted in 1993-94 and 2004-05, respectively, encompass a wide range of human development issues. A stratified random sampling design resulted in the selection in 1993-94 of 33,230 households from 16 large states. About one half of these households, that is, 13,459 in all, were selected at random for a re-survey in 2004-05. A second set of data were obtained from the District Level Household and Facility Surveys (DLHS), conducted by the Mumbai-based Indian Institute for Population Sciences (IIPS) and published by the Ministry of Health and Family Welfare of the Government of India. DLHS-3, the third in this series of surveys, was conducted from 2007 to 2008. It provides information related to 7,20,320 households from 28 States and 6 Union Territories of India. A total of 78 per cent of the surveyed households (5,59, 663 households in all) lived in rural areas, and we focus upon this part of the DLHS-3 sample.

17 We follow the official definition of town. See http://censusindia.gov.in/2011-prov-results/paper2/data_files/India2/1.%20Data%20Highlight.pdf, Accessed on 23 April 2013.

39