Online AppendixThe Right to Education Act:

Trends in Enrollment, Test Scores, and School Quality

By Manisha Shah and Bryce Steinberg

1

2 PAPERS AND PROCEEDINGS MONTH YEAR

-.05

0.0

5.1

Atte

nds

Scho

ol, A

ges

6 to

16

(Rur

al)

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

135

140

145

150

Num

ber E

nrol

led

(in m

illion

s), G

rade

s 1

to 8

(Rur

al)

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (DISE data)

Figure 1. RTE and Rural Enrollment

Note: This Figure replicates Figure 1 using NSS and DISE data using only the rural sample.

Source: NSS Rounds 62, 64, 66 and 68 (2005-2012), DISE 2005-2014

VOL. VOLUME NO. ISSUE RIGHT TO EDUCATION 3

-.12

-.08

-.04

0.0

4.0

8.1

2At

tend

s Sc

hool

by

Age

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

Ages 6 to 12 Ages 13 to 16-.1

2-.0

8-.0

40

.04

.08

.12

Cur

rent

ly E

nrol

led

by A

ge

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Ages 6 to 12 Ages 13 to 16

4060

8010

012

014

0N

umbe

r Enr

olle

d (in

milli

ons)

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (DISE data)

Primary Upper Primary

Figure 2. RTE and Enrollment by Age: A Snapshot from 3 datasets

Note: This Figure shows βt from an OLS regression of Equation 1 with their respective 95% confidence intervals ontwo measures of enrollment by age group (NSS and ASER) and raw enrollment numbers by primary/upper primaryenrollment (DISE). “Attends school” is equal to one if the child lists attending school as his or her primary activity,and zero if he or she lists another primary activity. “Currently Enrolled” is equal to one if the child reports beingenrolled in school, and zero if he reports having dropped out or never enrolled. “Number enrolled” is the averageof the total number of students enrolled. NSS and ASER regressions contain district, age, and sex fixed effects andare clustered at the district level. 2008 is omitted base year. 95% confidence intervals are shown for ASER and NSSregressions.Source: ASER 2005-2014, NSS Rounds 62, 64, 66 and 68 (2005-2012), DISE 2005-2014

4 PAPERS AND PROCEEDINGS MONTH YEAR

-.1-.0

8-.0

6-.0

4-.0

20

.02

.04

.06

.08

.1C

urre

ntly

Enr

olle

d by

Typ

e

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Government Private

4060

8010

012

014

0N

umbe

r Enr

olle

d by

Typ

e (in

milli

ons)

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (DISE data)

Government Private

Figure 3. Private vs. Government School Enrollment

Note: This Figure shows βt from an OLS regression of Equation 1 on government vs. private school enrollment(ASER) and raw government vs. private school enrollment numbers (DISE). The ASER regression contains district,age, and sex fixed effects and are clustered at the district level. 2008 is omitted base year. 95% confidence intervalsare shown for the ASER regression coefficients.Source: ASER 2005-2014, DISE 2005-2014

VOL. VOLUME NO. ISSUE RIGHT TO EDUCATION 5

-.15

-.1-.0

50

.05

.1.1

51s

t qua

rtile

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

−.1

5−

.1−

.05

0.0

5.1

.15

1st quart

ile

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

-.15

-.1-.0

50

.05

.1.1

52n

d qu

artil

e

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

-.15

-.1-.0

50

.05

.1.1

52n

d qu

artil

e

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

-.15

-.1-.0

50

.05

.1.1

53r

d qu

artil

e

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

-.15

-.1-.0

50

.05

.1.1

53r

d qu

artil

e

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

−.1

5−

.1−

.05

0.0

5.1

.15

4th

quart

ile

2005 2006 2007 2008 2009 2010 2011 2012Year (NSS data)

−.1

5−

.1−

.05

0.0

5.1

.15

4th

quart

ile

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Figure 4. Enrollment by 2008 Enrollment Quartiles

Note: This Figure shows βt from an OLS regression of Equation 1 on two measures of enrollment (NSS and ASER)by enrollment quartile in 2008. The regressions contain district, age, and sex fixed effects and are clustered at thedistrict level. 2008 is omitted base year. 95% confidence intervals are shown for the regression coefficients.Source: ASER 2005-2014, NSS Rounds 62, 64, 66 and 68 (2005-2012)

6 PAPERS AND PROCEEDINGS MONTH YEAR

-.4-.3

-.2-.1

0.1

.2.3

.4M

ath

Scor

es, A

ll, A

ges

6 to

16

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

-.4-.3

-.2-.1

0.1

.2.3

.4R

eadi

ng S

core

s, A

ll, A

ges

6 to

16

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Figure 5. RTE and Test Scores for All Children

Note: This Figure replicates Figure 2 using all children including those who are currently enrolled, dropped out andnever enrolled.Source: ASER 2005-2014

-.4-.3

-.2-.1

0.1

.2.3

.4M

ath

Scor

es b

y Sc

hool

Typ

e

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Government Private

-.4-.3

-.2-.1

0.1

.2.3

.4R

eadi

ng S

core

s by

Sch

ool T

ype

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014Year (ASER data)

Government Private

Figure 6. RTE and Test Scores, by School Type

Note: This Figure shows βt from an OLS regression of Equation 1 on test scores for students in public (government)and private schools. Students in other types of schools (such as madrasas) are omitted. The regressions containdistrict, age, and sex fixed effects and are clustered at the district level. 2008 is omitted base year. 95% confidenceintervals are shown for the regression coefficients.Source: ASER 2005-2014

VOL. VOLUME NO. ISSUE RIGHT TO EDUCATION 7

-.1-.0

50

.05

.1Te

ache

r Abs

ence

Rat

e

2007 2009 2010 2011 2012 2013 2014Year (ASER School data)

Figure 7. RTE and Teacher Absenteeism

Note: This Figure shows βt from an OLS regression of Equation 1 on teacher absenteeism from the ASER schoolsurveys. 95% confidence intervals, clustered at the district level, are shown as bars. The survey was not conductedin 2006 or 2008. 2007 is omitted base year as there is no ASER 2008 school data. Regression contains district fixedeffects.Source: ASER School Data 2007, 2009-2014

8 PAPERS AND PROCEEDINGS MONTH YEAR

Table 1—Enrollment and Test Scores Pre- and Post-RTE

Panel A: Enrollment All Children 6-16

Dep. Var : Attends School Enrolled Ln(Enrollment)Source: NSS ASER DISE

Post-2008 .0645∗∗∗ .0145∗∗∗ .0914∗∗∗

(.0021) (.0013) (.0193)

Observations 437,854 5,117,613 340

Mean DV .877 .932 14.1

Panel B: Enrollment by Gender

Dep. Var : Attends School Enrolled Ln(Fem. Enrollment)

Source: NSS ASER DISE

Post-2008 .0547∗∗∗ .0105∗∗∗ .0965∗∗∗

(.0022) (.0012) (.0190)

Post-2008 X Girl .0209∗∗∗ .0087∗∗∗

(.0023) (.0008)

Observations 437,854 5,117,613 340Mean DV .877 .932 13.4

Panel C: Test Scores

Dep. Var : Math Score Read ScoreSource: ASER ASER

Post-2008 -.217∗∗∗ -.070∗∗∗

(.010) (.009)

Observations 4,669,247 4,690,067

Mean DV 2.60 2.77Note: This table shows coefficients from an OLS regression of schooling outcomes on a dummy for post-2008. PanelA shows measures of enrollment or school attendance from three seperate data sources. “Attends school” is equalto one if the child lists attending school as his or her primary activity, and zero if he or she lists another primaryactivity. “Enrolled” is equal to one if the child reports being enrolled in school, and zero if he reports having droppedout or never enrolled. “Log Enrollment” is the natural logarithm of the total state enrollment of children in primaryand upper primary school. Columns 1 and 2 contain district, age, and sex fixed effects, and are limited to childrenaged 6-16. Column 3 contains state fixed effects. Panel B shows enrollment by gender. Columns 1 and 3 are identicalto Panel A, but add an interaction of post-2008 with female. Column 3 reports the log of total female enrollmentby state. The dependent variables in Panel C are test scores, where math and read score range from 0-4. The Rightto Education Act was passed in 2009, though not fully implemented until 2010. Standard errors, clustered at thedistrict level for all regressions except Panel A column C and Panel B column C, clustered at the state level, areshown in parentheses. ∗ 10% significance ∗∗ 5% significance ∗∗∗ 1% significance.Source: ASER 2005-2014, NSS Rounds 62, 64, 66 and 68 (2005-2012), DISE 2005-2014