Variable
|
Baseline land ≤ median
|
---|
Budget share of total education expenditure
|
Budget share of school fees, uniform, and private tuition
|
Budget share of books and transport
|
---|
(1)
|
(2)
|
(3)
|
---|
Mother is working
|
0.038*
|
0.031*
|
0.007
|
|
(0.023)
|
(0.018)
|
(0.012)
|
Other control variables
|
Yes
|
Yes
|
Yes
|
Household fixed effects
|
Yes
|
Yes
|
Yes
|
Observations
|
1998
|
2000
|
2010
|
Number of households
|
999
|
1000
|
1005
|
Weak identification stat (Kleibergen-Paap rk Wald F)
|
3.256
|
3.254
|
3.437
|
Stock-Yogo weak identification test critical values
|
5.45 (25 %)
|
5.45 (25 %)
|
5.45 (25 %)
|
Over-identification stat (Hansen J)
|
0.150
|
0.296
|
0.009
|
Over-identification p value
|
0.699
|
0.586
|
0.921
|
- Notes: Robust standard errors clustered at the household level in parentheses. Significant at *10, **5, and ***1 %. Additional control variables are as follows: average age of children in the school-going age (5–17 years), number of boys in school-going age, number of girls in school-going age, square of number of boys in school-going age, square of number of girls in school-going age, household income, household size, number of males in the age group 16–60, average age of household members, land owned by household in 2007 interacted with time, household’s asset (index) in 2007 interacted with time, time trends depending on whether the district is a NREGS phase 1 district, mandal level total night lights in 2006, and baseline enrolment rate in the district. Instruments are as mentioned in Table 5