Jump to ContentJump to Main Navigation
School Choice InternationalExploring Public-Private Partnerships$

Rajashri Chakrabarti and Paul E. Peterson

Print publication date: 2008

Print ISBN-13: 9780262033763

Published to MIT Press Scholarship Online: August 2013

DOI: 10.7551/mitpress/9780262033763.001.0001

Show Summary Details
Page of

PRINTED FROM MIT PRESS SCHOLARSHIP ONLINE (www.mitpress.universitypressscholarship.com). (c) Copyright The MIT Press, 2022. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in MITSO for personal use.date: 27 June 2022

Public-Private Schools in Rural India

Public-Private Schools in Rural India

(p.90) (p.91) 5 Public-Private Schools in Rural India
School Choice International

Karthik Muralidharan

Michael Kremer

The MIT Press

Abstract and Keywords

This chapter presents results from a nationally representative survey of rural private primary schools in India conducted in 2003. Twenty-eight percent of the population of rural India has access to fee-charging private schools in the same village, while richer states have fewer rural private schools. States, districts, and villages with poor public school performance are each more likely to have private schools. Nearly 50 percent of the rural private schools in the author’s sample were established five or fewer years before the survey, and nearly 40 percent of private school enrollment is in these schools. This suggests a rapid expansion of private schooling, although it could also, in part, reflect turnover among schools in the sector. The chapter aims to show that private fee-charging schools play a significant role in the primary education sector even if primary education policy in developing countries is focused on increasing the resource base and the number of government-run schools.

Keywords:   private primary schools, India, fee-charging schools, private schooling, primary education sector, primary education policy, government-run schools

While the focus of primary education policy in developing countries such as India has largely centered on increasing the resource base and the number of government-run schools, the role of private fee-charging schools in the primary education sector has not been appreciated as much. However, as several recent papers point out (Kingdon 1996; PROBE Team 1999; De, Noronha, and Samson 2001; Tooley and Dixon 2003; and Mehta 2005), there is reason to believe that private fee-charging schools increasingly cater to a substantial fraction of the primary-school-going population in India. Most research on this subject to date comes from small-sample studies at the state or district levels.1

This chapter presents results from a nationally representative survey of rural private primary schools in India that the authors conducted in 2003. Twenty-eight percent of the population of rural India has access to fee-charging private schools in the same village. Richer states have fewer rural private schools. States, districts, and villages with poor public school performance are each more likely to have private schools. Nearly 50% of the rural private schools in our sample were established five or fewer years before the survey, and nearly 40% of private school enrollment is in these schools. This suggests rapid expansion of private schooling, although it could also in part reflect turnover among schools in the sector.

Private-school teacher salaries are typically one-fifth of the salary of regular public school teachers (and are often as low as one-tenth of these salaries). This enables the private schools to hire more teachers, have lower pupil-teacher ratios, and reduce multigrade teaching. Private school teachers are significantly younger and more likely to be from the same area as their counterparts in the public schools. They are 2–8 percentage points less absent than teachers in public schools (p.92) and 6–9 percentage points more likely to be engaged in teaching activity at any given point in time. They are more likely to hold a college degree than public school teachers, but are much less likely to have a formal teacher-training certificate. Children in private school have higher attendance rates and superior test-score performance, the latter true even after controlling for observed family and school characteristics.

The first section outlines the sampling methodology and how the data was collected. The second section presents results on the extent of private school prevalence and correlates of private school existence. The chapter then discusses the economics of private unaided schools and their sources of competitive advantage by comparing them with public schools on various measures including infrastructure, teacher characteristics, student characteristics, and student performance.

5.1 Sampling Methodology and Data

The data used in this chapter was collected as part of a multicountry study conducted by us and coauthors on provider absence in schools and health clinics where India was one of the countries studied (the detailed results from the cross-country study are presented in Chaudhury et al. 2006).2 Within India, 20 states were selected, representing 98 percent of the population, or roughly one billion people. Using geographically stratified random sampling, 10 districts were selected within each state and 10 primary sampling units (PSUs) were selected in each district. The PSUs were allocated to rural and urban sectors in accordance with the population distribution within each sampled district.3 Rural PSUs (villages) within a sampled district were selected randomly without replacement with probability proportional to size (PPS).4

The survey focused on government-run primary5 schools but also covered rural private schools in villages where they existed. The definitions of school categories that we use are similar to those detailed in chapter 6 by Geeta G. Kingdon. The term government school refers to government-funded schools that are run by the government but does not include the government-aided schools that are privately managed. The terms public schools and government schools are used interchangeably in this chapter. The private schools referred to in the rest of this chapter are those that charge user fees and do not receive any financial support from the government. This includes both recognized (p.93) and unrecognized private schools, but does not include “private-aided schools” which are privately managed schools that receive funding from the government, and are typically forbidden from charging user fees.

Recognized private schools are required to conform to various government norms; the main benefit of recognition is that only recognized schools are eligible to issue “transfer certificates” (TCs) to their students (see chapter 6 for more details on the requirements for recognition). These TCs in turn are required for students to move across schools with credit granted for academic work done in the previous school. In practice, however, many of the recognized schools do not meet the stipulated norms (Kingdon, chapter 6 in this volume), and Tooley and Dixon (2003) argue that it is not uncommon for operators of private schools to have to pay bribes to obtain recognition status.

One response to the obstacles to obtaining recognition has been an increasing prevalence of unrecognized private schools that charge fees but have not obtained recognition and are not authorized to issue TCs. Unrecognized private schools circumvent this practice in several ways, the most common of which is double enrollment, whereby children are enrolled in both the government-run school (which is recognized by default) and in the unrecognized private school. Note that private unrecognized schools are more than just supplemental tuition centers and should be thought of as schools, because they usually run during the same hours of the regular school, and children typically do not attend both kinds of schools although they may be enrolled in both. Double enrollment is a convenient arrangement for all parties because the government school gets to show high levels of enrollment, parents and children get textbooks and other free supplies from the government school, and new private schools can operate without the burden of seeking recognition since TCs will be issued by the government school. However, this does lead to systematic underestimation of the relative size of the government and fee-charging private school systems in India, as discussed in Kingdon (chapter 6 in this volume).

In the rural sample, the survey covered all the primary schools in the village subject to a maximum of three (the maximum number of schools that could be covered during one day in the field). When the investigators reached the village, they listed all the schools present within a radius of two kilometers from the village center. In villages with fewer than three schools, all the schools were covered. In villages with more than three schools, three schools were surveyed; one (p.94) school was randomly selected in each of the three main categories of rural schools (government schools, private schools, and nonformal education centers). In cases where there was no nonformal school, but more than three schools in the village, enumerators selected two government schools and one private school or one government school and two private schools (the latter was the case only if there was only one government primary school but more than two private schools in the village).

Thus in addition to being representative of government-run primary schools, the dataset is also representative of the universe of private unaided primary schools in rural India because at least one private school was surveyed in any village that had at least one private school. Fifty-three percent of the private schools in our sample are unrecognized, suggesting that official sources of data on private schools significantly understate the extent of private school prevalence.6 While government surveys only include the recognized private schools, the random selection method is indifferent to the recognition status of the school and the sample here therefore includes both types of schools. Furthermore, the random selection of the schools within a village ensures that the distribution of school types in the sample is a reflection of the distribution of school types in the population. The remainder of this chapter does not distinguish between private recognized and unrecognized schools because they are both fee-charging schools that do not receive funds from the government, and this is the school category we focus on here.

Enumerators made three unannounced visits to each selected school over a three- to four-month time period from December 2002 to March 2003. Teacher absence was measured in all surveyed schools by physically verifying the presence of teachers on the school roster. In addition to recording teacher attendance, data was also collected on student attendance, school facilities, and teacher characteristics. Finally, the enumerators also administered a short test7 to 10 randomly selected fourth-grade children and collected basic demographic information on these children in all the schools that we surveyed.

5.2 Private School Prevalence and Its Correlates

Twenty-eight percent of the villages in our sample have a private school. Since the villages were sampled on a probability proportional to size (PPS) basis, this implies that 28% of the population of rural (p.95)

Table 5.1 Private School Prevalence by State


% of villages with a private school


% of villages with a private school



Andhra Pradesh








Tamil Nadu














Himachal Pradesh


Uttar Pradesh


West Bengal








Madhya Pradesh


All India


India has access to a private school in the same village in which they live. But there is sharp variation in the prevalence of private schools across states, with Gujarat and Maharashtra having almost no rural private schools, while over 50% of the sampled villages in Rajasthan, Bihar, Uttar Pradesh, Punjab, and Haryana have a private school in the same village (table 5.1). Recent household-survey-based evidence presented in the Annual Status of Education Report (2005) confirms the increasing role of private schooling in rural India by showing that 15.5% of children aged 6–10 in rural India attend a private school and that over 20% of the children in this group attend a private school in several states.8

Table 5.2 presents results from ordinary least squares (OLS) regressions where the binary variable of private-school existence (at the village level) is regressed on potential predictors of private school existence. The first column includes the log of the village population, the log of the mean pupil-teacher ratio in the public schools in the village, and the mean level of teacher absence9 in the public schools in the village. The second column includes state fixed effects. The third column replaces the state dummies with the log of state per capita GDP. The fourth column includes district-level estimates of mean per capita consumption calculated from the fifty-fifth round of the National Sample Survey,10 and the fifth column includes district-level consumption as well as state fixed effects.11 (p.96)

Table 5.2 Correlates of Private School Existence at the Village Level

Dependent variable = 1 if village has a private school, 0 if it does not






Log village population











Log pupil teacher












Mean public school






Absence in village






Log state GDP/Capita



Log district

















State fixed effects


















Notes: Robust standard errors in brackets

*** significant at 1%;

** significant at 5%;

* significant at 10%

Villages with larger populations are significantly more likely to have a private school in all specifications. The most noteworthy result is that private schools are significantly more likely to exist in villages with a high rate of teacher absence in public schools. While the relation is very strong across Indian states, it is still significant at the 10% level after controlling for state fixed effects, and remains significant in all specifications. The surprising result is that states with a higher per capita income are less likely to have private schools in their villages. While a high pupil-teacher ratio (PTR) in the public schools in the same village is a predictor of private school existence across India, the correlation is not significant with either state income controls or state fixed effects, suggesting that the PTR in public schools is negatively correlated with the per capita GDP of the states. The final column shows that when we include state-fixed effects, richer districts are less likely to have a private school, though villages with high public-school teacher absence are more likely to have a private school.

Chaudhury et al. (2006) shows that higher-income countries and richer Indian states have significantly lower rates of teacher absence in (p.97) schools. Thus if private schools arise as a response to public school failure, we might expect richer states to have fewer private schools. On the other hand, since private schooling is likely to be a normal good we might expect the prevalence of private schools to be higher in the richer states.

The correlation between public school failure (as measured by teacher absence and nonteaching activity) and the likelihood of the existence of private schools can be seen clearly in figures 5.1a and 5.1b. While the two states with the highest incidence of private schools (Punjab and Haryana) happen to be among the richer states of India, it is quite striking that the two states with the lowest level of teacher absence in public schools (Gujarat and Maharashtra) have almost no rural private schools, even though these are two of the richest states in India.

Table 5.3 shows more related evidence by comparing teacher absence rates across different kinds of schools in India. The first column of table 5.3 shows the weighted average teacher absence by school type across the full sample of schools. Columns 3–5 show the difference in teacher absence relative to the government-run schools. While the weighted average all-India teacher absence in private schools of 22.8% is slightly lower than that of the 25.2% in government schools, this difference is not significant. However, with the addition of village/town fixed effects, the teacher absence rate is 3.8% lower in private schools relative to government schools and this is significant at the 1% level. The addition of school, teacher demographics, and visit-level controls increases this difference to 7.8%, which is over 30% of the observed absence rate in government schools (25.2%). This suggests that private schools are disproportionately located in areas with poorly performing public schools and that the efficiency of the private school (at least as measured by teacher absence) is even higher after controlling for school facilities (which are negatively correlated with teacher absence) and teacher demographics.

The higher prevalence of private schools in villages with high absence among public school teachers could be interpreted as suggesting that private schools enter where public schools are failing or as evidence that the establishment of private schools reduces political pressure for teacher attendance in public schools. However, to the extent that one might expect higher-income states to have more private schools, the finding that richer areas have fewer private schools suggests that poorly performing public schools rather than increasing incomes are the more important source of demand for private schools. (p.98)

Public-Private Schools in Rural India

Figure 5.1


Table 5.3 Absence Rate by School Type

Difference relative to government-run






Teacher absence

Number of observations

No fixed effects

Village/town fixed effects

Village/town fixed effects + controls*

Government-Run Schools



Non-formal Schools






Private Aided Schools






Private Schools






Notes: (*) Controls include a full set of visit-level, teacher-level, and school-level controls Bold numbers indicate significant differences at the 1% level

Finally, it is noteworthy that there is some evidence that large-scale prevalence of rural private schools is a recent phenomenon. This is suggested in previous studies of specific states such as De, Noronha, and Samson (2001), and Mehta (2005), but we are able to confirm this on a nationwide basis. Figure 5.2 plots the cumulative distribution function (CDF) of private school formation and enrollment over time, and we see that nearly 50% of the private schools in the sample have been established in the five years before the survey. Nearly 40% of the total private school enrollment is in schools that were less than 5 years old and over 60% of total enrollment is in schools that were less than 10 years old in 2003. Of course, these numbers will exceed the net increase in private school enrollment to the extent that other private schools exited over the period.

5.3 Economics of Rural Private Schools

5.3.1 School Infrastructure

Table 5.4 presents summary statistics on school infrastructure in public and private schools. While private schools are more likely to have an electricity connection and toilets for teachers, they are less likely to have libraries (book banks) and classrooms without mud floors. On aggregate there doesn’t appear to be a significant difference in the (p.100)

Public-Private Schools in Rural India

Figure 5.2 Private School Formation/Enrollment over Time (Cumulative distribution function)

Table 5.4 Private versus Public School Facilities




Difference with state fixed effects

Difference with village fixed effects

Fraction of schools with electric connection available






Fraction of schools with library available






Fraction of schools with covered classrooms available






Fraction of schools with non-mud floors available






Fraction of schools with teacher toilet available






Average school infrastructure index (0–5 scale)






Notes: Significance level: *** 1%; ** 5%; *1%

(p.101) infrastructure index between private and public schools, but the results with state and with village fixed effects suggest that conditional on being in the same village, private schools have poorer facilities and infrastructure than the public schools.

5.3.2 Sources of Competitive Advantage of Private Schools

Probably the single most distinguishing feature of the private schools in rural India is the fact that they pay much lower salaries to teachers than the government schools. While we don’t directly collect data on teacher salaries, we have data on the various fees charged by each school in our sample along with the total enrollment, which allows us to estimate the monthly revenue for the private schools (since they typically don’t receive any funding beyond what they raise in school fees). Median monthly revenue of a private school in our sample is around Rs4,000 per month,12 with the median fee being Rs63 per month and the median private school having an enrollment of 72 students.

We can calculate an upper bound for teacher salaries in private schools assuming that all the revenues of the private schools are used to pay teacher salaries. We calculate the upper bound on median teacher salary to be less than Rs1,000 per month and the upper bound on the mean teacher salary to be less than Rs1,750 per month. The mean salary for a regular government school teacher in a typical state like Andhra Pradesh (where we have actual salary data13) is around Rs7,500 per month. We can see that the typical total monthly revenue of a private school is often less than the monthly salary of one government school teacher. Even conservatively, rural private school teacher salaries are typically around one-fifth of that of regular government teacher salaries and they are often as low as one-tenth of the salaries of regular government teachers. The differences are even more pronounced when benefits are included because government teachers are guaranteed a pension after retirement, while private school teachers rarely have such provisions. This allows the private schools to hire more teachers, reduce multigrade teaching, and have significantly lower pupil-teacher ratios.

Table 5.5 clearly demonstrates these points. The average PTR in the private schools of 19.2 is less than half the ratio of 43.4 in public schools. This gap of 24.3 widens to 29.6 with state fixed effects, and to 34.4 with village fixed effects. Thus conditional on being in the same village, the private school has nearly 35 fewer pupils per teacher than the government school in the same village. Doing the calculation using (p.102)

Table 5.5 Sources of Private School Competitive Advantage




Difference with state FEs

Difference with village FEs

Mean total enrollment






Mean number of teachers






Pupil-teacher ratio






Log pupil-teacher ratio






Multigrade teaching






Average grade of starting teaching English






Fraction of teachers engaged in teaching activity






Average student attendance






Notes: Significance level: *** 1%; ** 5%; *1%

logs, we find that the PTR of a public school is 2.85 times higher than the PTR of a private school in the same village. The lower PTR in the private schools also translates into lower levels of multigrade teaching (the practice of one teacher simultaneously teaching multiple grades in the same room).

Field interviews with parents of children attending rural private schools suggest that two of the major attractions of private schools are the fact that they start teaching English early and that there is more teaching activity in these schools. The last two rows of table 5.5 confirm that these differences do exist. Private schools on average start to teach English a whole grade earlier, with the effect being even more pronounced with state and village fixed effects. Private schools also have significantly more teaching activity going on, and again the magnitude of the difference increases with state and village fixed effects.

One reason for this is likely to be that head teachers in private school are much more likely (and able) to take disciplinary action against shirking teachers than their counterparts in the public schools. We found that only one head teacher in the nearly 3,000 public schools we surveyed reported ever dismissing a teacher for repeated absence.14 On the other hand, 35 head teachers in a sample of around 600 private schools reported having at some point dismissed a teacher for repeated absence, and therefore shirking teachers in the private sector are around 175 times more likely to have disciplinary action taken against them!

(p.103) If we consider the cases with village fixed effects (which is the relevant case when considering the choice faced by a parent with regard to choosing between a private and public school in the same village), we see that combining the effects of a lower pupil-teacher ratio and a higher level of teaching activity leads to a child in the private school having three to four times more “teacher contact” time than in the public school.

The better performance of the private schools is also reflected in the fact that student attendance rates are also substantially higher in private schools (as seen in the last row of table 5.5). Pupil attendance is 11.3% higher in the all-India sample, and 13.4% higher with village fixed effects. If we think that the true measure of the relative role of the private and public sectors is attendance as opposed to enrollment, then the true share of rural children taught in the private sector will be even higher after adjusting for the differential attendance rates.

5.3.3 Teacher Characteristics

A key issue that follows the discussion on teacher pay in private schools is to understand who the private school teachers are and the reasons for their willingness to work at such low salaries. Field visits suggest that the availability of these inexpensive teachers in the villages is being driven by local educated young people who are typically unable to find jobs, unwilling (and usually not needed) to work in agriculture, and not looking at teaching as a long-term career. Teaching suits these young people well because the short working day of four to six hours allows them the time for further study via correspondence (distance-education) courses or in colleges that follow a different shift. The short working days also allow them to look for other longer-term jobs on the side. And finally, teaching provides them with both income and respectability while they look at other long-term options.

Table 5.6 provides summary statistics consistent with this view. The private school teachers are on average over ten years younger than their counterparts in the public sector and are twice as likely to be from the same village where the school is located. They are more likely to have a college degree but also much less likely to have a professional teaching certificate, which suggests that even though they are more educated, they are not looking at teaching as a long-term career option.

This probably helps to explain why teacher absence is not even lower than it is in the private schools given the high likelihood of action being taken for repeated absence. Since the private school (p.104)

Table 5.6 Teacher Characteristics




Difference with state FEs

Difference with village FEs

Average age of teachers






Fraction of college graduates among teachers






Fraction of teaching certificate holders among teachers






Fraction of female teachers






Fraction of local teachers






Notes: Significance level: *** 1%; ** 5%; *1%

teachers are being paid a much lower wage and are often looking at other long-term options, there is little “efficiency wage” cost of being fired. Thus, if pursuing other opportunities requires a certain level of absence (and an accompanying probability of action being taken), this is a tradeoff that the private school teachers probably are willing to make. However, despite the low wages, we see that private schools have lower teacher absence and higher teaching activity than the public schools—especially in the same village.

5.3.4 Parent Characteristics

Given that public schools are free and private schools charge fees, we would expect that the students attending the private schools come from more socioeconomically privileged backgrounds. Based on the random sample of children in the fourth grade whom we test and collect demographic information on, we can compare the family backgrounds of children in both types of schools. Table 5.7 provides these comparisons, and as we would expect, the children attending private schools come from more advantaged family backgrounds. They have more educated parents and indicate possessing a higher level of assets. However, it is worth noting that the absolute level of education of the parents of the children attending private schools is actually quite low. For instance, 20% of the private school students are first-generation learners, which while lower than the 30% finding in public schools, is still quite significant. Thus while private schools cater to the more affluent in the rural areas, many of their students come from disadvan-taged backgrounds. This is consistent with the results of (p.105)

Table 5.7 Household Characteristics




Difference with state FEs

Difference with village FEs

Average number of rooms in house






Average fraction of children taking tuition






% of literate fathers






% of literate mothers






% of fathers with education 10 grades or higher






% of mothers with education 10 grades or higher






Notes: Significance level: *** 1%; ** 5%; * 1%

Tooley and Dixon (2003), who mention that the majority of private schools in India cater to the poor (though their observation is based on an urban study) and the findings reported by Andrabi, Das, and Khwaja (2002) that private schools in rural Pakistan are affordable to middle- and even low-income groups.

5.3.5 Performance of Private Schools

As discussed earlier, private schools have lower teacher absence and higher levels of teaching activity. They also exhibit significantly superior performance on the test that was administered. Table 5.8 shows the test score performance advantage of private schools (in standard deviations). While controlling for family and other characteristics reduces the size of the private school effect, it is still strongly significant and of considerable magnitude (0.4 standard deviations on the test). Of course, we cannot rule out that some of these results are being driven by unobserved heterogeneity among the students. Similarly, as discussed earlier, student attendance is around 11 percentage points higher in the private schools (75%) relative to the public schools (64%). This could partly be due to artificially inflated enrollment figures in the government schools.

5.4 Conclusions

We find that private unaided fee-charging schools are widespread in rural India, particularly in areas where the public system is (p.106)

Table 5.8 Performance Differentials of Private Schools

Regression of mean student test score (in std. deviations) on school type and controls






Private School







Family demographics and private tuition






School facilities






State fixed effects






Village fixed effects


















Notes: Significance level: *** 1%; ** 5%; *1%

dysfunctional. The number of such schools appears to be growing rapidly with both demand-side variables (desire for English-medium education, less multigrade teaching, smaller classes, more accountable teachers) and supply-side variables (availability of educated unemployed young people) playing an important role in this rapid growth. Salaries paid by these schools are only about one-fifth of those paid by public schools, but these schools have many more teachers relative to the number of pupils, and the private school teachers are more likely to be teaching than public school teachers.

Our results have a number of implications. First, efforts to improve the quality of education in India should consider the private as well as the public sector—especially since private schools are disproportionately located where the public system is failing. For example, policy makers might consider the possibility of offering short training courses to raise skills among private school teachers.

Second, the disparities between private and public schools highlight some potential areas for reform in the public sector. The huge salary differential suggests that many public school teachers may be receiving enormous rents.

Finally, there may be scope for public-private partnerships in education, whether in the form of voucher programs or otherwise. One issue with voucher programs is whether there will be an adequate supply response, but the evidence suggests that private schools are already widespread in rural areas and that new schools can be created rapidly.

(p.107) There is substantial scope for carefully designed policy experiments aimed at leveraging the private sector for universal quality education, and it is important to follow these experiments with rigorous evaluation to provide systematic evidence for future policy decisions in this regard. The recent draft of the “Right to Education Bill” that is expected to be introduced in Parliament mandates that 25% of seats in private educational institutions be reserved for “weaker sections” of society. It also goes on to say that for each such admitted child, the “government shall reimburse to the school at a rate equal to the per-child expenditure in state schools/fully aided schools, or the actual amount charged per student by such school, whichever is less.” The discussion around this legislation would be an opportune moment to think about the most efficient institutional forms for delivery of primary education in India.


We thank Nazmul Chaudhury, Jeffrey Hammer, and Halsey Rogers for their collaboration and insights on the global study that generated the data that this chapter is based on, and Konstantin Styrin for valuable research assistance. We offer thanks to the staff of the Social and Rural Research Institute, New Delhi—and especially to Chhavi Bhargava, Navendu Shekhar, A. V. Surya, and Aditi Varma—for conducting and overseeing the fieldwork for the primary surveys. We also thank participants at the PEPG conference, two anonymous referees, Rajashri Chakrabarti, and Paul Peterson for comments and suggestions. All errors are our own.


Bibliography references:

Andrabi, Tahir, Jishnu Das, and Asim Khwaja. 2002. The rise of private schooling in Pakistan: Catering to the urban elite or educating the rural poor? Mimeo. Kennedy School of Government, Harvard University.

Annual Status of Education Report (ASER). 2005. Mumbai: Pratham Resource Center.

(p.109) Bashir, Sajitha. 1994. Public versus private in primary education: Comparison of school effectiveness and costs in tamilnadu. PhD dissertation, London School of Economics.

Chaudhury, Nazmul, Jeffrey Hammer, Michael Kremer, Karthik Muralidharan, and F. Halsey Rogers. 2006. Missing in action: Teacher and health worker absence in developing countries. Journal of Economic Perspectives (winter): 91–116.

De, Anuradha, Claire Noronha, and Meera Samson. 2001. India: Private schools and universal elementary education. South Asia Education Sector, Technical Working Paper No. 3. World Bank, Washington, D.C.

De, Anuradha, Manabi Majumdar, Claire Noronha, and Meera Samson. 2002. Private schools and universal elementary education. In India Education Report—A Profile of Basic Education, ed. R. Govinda. New Delhi: Oxford University Press.

Govinda, R., and Varghese, N. V. 1993. Quality of primary schooling in India: A case study of Madhya Pradesh, International Institute for Educational Planning, National Institute of Educational Planning and Administration. New Delhi: NIEPA.

Kingdon, Geeta. 1996a. The quality and efficiency of private and public education: A case study of urban India. Oxford Bulletin of Economics and Statistics 58, no. 1: 57–81.

Kingdon, Geeta. 1996b. Private schooling in India: Size, nature and equity effects. Economic and Political Weekly 31, no. 51 (December).

Kingdon, Geeta Gandhi, and Mohd Muzammil. 2001. A political economy of education in India- I: The case of UP. Economic and Political Weekly 36, no. 32 (August: 11–18).

Kremer, Michael, Karthik Muralidharan, Nazmul Chaudhury, Jeffrey Hammer, and F. Halsey Rogers. 2004. Teacher absence in India. World Bank: Washington, D.C.

Kremer, Michael, Karthik Muralidharan, Nazmul Chaudhury, Jeffrey Hammer, and F. Halsey Rogers. 2005. Teacher absence in India: A snapshot. Journal of the European Economic Association 3: 658–667.

Mehta, Arun. 2005. Elementary education in unrecognized schools in India: A study of Punjab based on DISE 2005 data. NIEPA, New Delhi.

PROBE Team. 1999. Public Report on Basic Education in India. New Delhi: Oxford University Press.

Shah, Parth. 2005. Equity in education. Business Standard, July 13.

Tooley, James, and Pauline Dixon. 2003. Private schools for the poor: A case study from India. Reading, U.K.: CfBT Research and Development. Available at 〈http://www.cfbt.com/PDF/91001.pdf〉.

Topalova, Petia. 2005. Trade liberalization, poverty, and inequality: Evidence from Indian districts. NBER Working Paper 11614, Cambridge, Mass.


(1.) Notable among these are Bashir (1994) in Tamilnadu; Kingdon (1996b) in Lucknow (Uttar Pradesh); Govinda and Varghese (1993) in Madhya Pradesh; Tooley and Dixon (2003) in Hyderabad (Andhra Pradesh); and Mehta (2005) in Punjab. As Kingdon (1996a) mentions, “given inter-state variations in the structure and organization of education in India, evidence from a single state will be illustrative but not necessarily representative.”

(2.) See Chaudhury et al. (2006) for detailed results from the cross-country study.

(3.) Thus a district with 90% of its population in rural areas would have 9 rural PSUs and 1 urban PSU, whereas a completely urban district (as is the case when the randomly picked district is the state capital, for example) would have 10 urban PSUs.

(p.108) (4.) See appendix A of Kremer et al. (2004) for a detailed description of the sampling procedure.

(5.) Covering grades one to five in most states, and grades one to four in some states, depending on the classification of primary schools in the concerned state. The focus of the study was completely on primary schools, and so the usage of the term school should be understood to mean primary school unless stated otherwise.

(6.) Unrecognized schools are also more recently established, with an average age of 7.6 years as opposed to recognized private schools with an average age of 9.9 years. The fraction of schools in this sample that report being run by a religiously oriented group is quite small (15 out of 592 or 2.5% of schools). Schools run by religiously oriented groups form a larger share of the private-aided schools that get government grants and are not allowed to charge tuition fees (33 out of 152 or over 20%).

(7.) Since the survey was done across several states with different languages, the test was weighted towards math as opposed to language. The test was short but the items used had been pretested for validity. The test consisted of 12 arithmetic questions and 2 verbal questions (that asked the students names in the local language and English respectively). See appendix B of Kremer et al. (2004) for a detailed description of the test as well as the procedure by which it was administered, graded, and coded.

(8.) These states include Andhra Pradesh, Haryana, Kerala (including private aided schools), Punjab, Rajasthan, and Uttar Pradesh.

(9.) A teacher was considered to be absent if, at the time of a random visit during school hours, he or she could not be found anywhere in the school premises. See Chaudhury et al. (2006) and Kremer et al. (2005) for details on how absence and teaching activity were measured and on the various steps we took to measure these accurately.

(10.) We thank Petia Topalova for making her calculations of district-level consumption estimates available to us. See Topalova (2005) for details on these calculations.

(11.) Robust standard errors clustered at the state level are reported for specifications with state-level right-hand-side variables and likewise for district-level variables, where the standard errors are clustered at the district level.

(12.) The approximate exchange rate at the time of publication is Rs45 = U.S.$1.

(13.) Direct data on teacher salaries in Andhra Pradesh has been collected in a different ongoing study by one of the authors. The salary figures would be even higher if we included benefits, the largest portion of which is the present value of a defined benefits retirement pension. Private school teachers typically receive no benefits.

(14.) See Kingdon and Muzammil (2001) for more details on the power of public-school teacher unions and how it has evolved over the years (based on a case study of the state of Uttar Pradesh).