Do Private Tuitions Improve Learning Outcomes?
Despite increased attention to school based learning over the past decade by policy makers, the ASER reports have shown that the learning levels of children in the Indian education system have remained consistently low and have, in fact, declined over the past 8 years. The latest ASER report shows that only 41% of children in the age group of 6-14 can read a standard 2 text (ASER 2013). Consequently, critical and rigorous analysis of policies surrounding provision of school-based education has received much-deserved attention. In the process, the role of additional educational inputs provided by households, such as private tutoring, has remained neglected.
Private tutoring is defined as fee-based tutoring that provides supplementary instruction to children in academic subjects that they study in the mainstream education system. This phenomenon is widespread across many developing countries, including India. As per the latest ASER (ASER 2013), approximately one-fourth of children enrolled at elementary level (Std. 1 to 8) in rural India attend private tuitions. They pay on average, Rs. 170 per month, amounting to slightly above Rs. 2000 per annum to attend these tuitions.
An important question that arises in this context is: do learning outcomes of children improve if they attend tuition? Finding a difference in learning outcomes of those who attend tuition and those who don’t, and attributing it to private tuitions is misleading. Part or all of the difference in learning outcomes might be due to different characteristics of children who attend tuition. There are observable and unobservable differences between the two groups of children, which make it difficult to figure out the effect of tuition, if any. To give an example, ASER data indicates that children belonging to richer households are more likely to attend tuitions. Richer households are also likely to provide more support to a child in the form of other material inputs. Data also shows that children of more educated parents are more likely to attend private tuition, but more educated parents are also in a position to help the child with studies. This makes it difficult to disentangle the effect of tuition from the effect of other material inputs, or the effect of having educated parents.
There are many techniques available to overcome this problem. Interested readers can refer to the relevant literature. Choice of technique ultimately depends on availability (or not) of appropriate dataset, time and money at hand, and feasibility of data collection. We use household fixed effects (FE) technique to estimate effect of tuition on learning outcomes. Household FE utilizes variation in status of children on private tuition within a household. To give an example, suppose there are two children in a household. One attends private tuition, and other doesn’t. Then, the difference in the learning outcomes of these two children would be attributed to private tuition. Note that all other observable and unobservable factors at the household or village level affecting learning outcomes are controlled for in this technique. Hence, household FE approach reduces self-selection problem substantially. But one must remember that it doesn’t eliminate the problem completely since it can’t control unobservable child-specific differences such as motivation, intelligence, dedication etc.
We use this technique due to the availability of ASER dataset for 2011, whose underlying sampling strategy is such that pre-determined number of villages from each district and pre-determined number of households from each selected village are surveyed. A unique characteristic of ASER dataset is availability of learning outcomes for reading and math.
In order to estimate learning levels, we developed a standardized aggregate score. For this, we sum up reading and math scores for each child, and then standardize it by subtracting a child’s aggregate score from the mean aggregate score of all students, and then dividing by the standard deviation of aggregate score for that year. This standardized aggregate score has been used as the dependent variable in our empirical analysis. The key independent variable is whether the child attends tuition. Other independent variables are whether the child attends government or private school, age and gender of the child, class in which the child is studying, and finally both parents’ age and education. We have other controls at the household and village level, but they are not relevant in a household FE model.
What do the results show? Household FE estimation results indicate that attending private tuition has 0.14σ effect on learning levels. How large is this effect? Comparing the coefficient on private tuition with that of standard/class in which child is studying or that of type of school reveal that the effect of attending tuition is as large as an additional year of education or attending a private school instead of a government school. Interestingly, results also show that the effect of tuition is almost twice as high for children enrolled in government schools, compared to those who are enrolled in private schools. Further, children who attend tuition and whose parents are less educated, benefit more from these tuitions. Effect of tuition is also higher for children who stay in non-pucca households compared to those who stay in pucca households. Given that children attending government schools or having less educated parents or less well-off have lower learning levels, private tuitions clearly are benefitting disadvantaged students.
There is significant variation in prevalence of private tuition across states. States like West Bengal, Tripura have 67-69% children at elementary level attending private tuition, while Bihar and Orissa have 40-50% children at elementary level attending private tuition. And we find that the effect of tuition is higher in these states. In Bihar and West Bengal, attending private tuition has 0.22σ effect on learning levels, while in Odisha, attending private tuition has 0.18σ effect on learning levels.
Why do private tuitions have a positive effect on learning outcomes? One straightforward explanation is that those who attend tuition spend more time studying. Though ASER doesn’t capture time spent at tuitions, analysis of IHDS data indicates that those who attend tuition spend, on average, 9 hours in tuitions. That would mean 1.5 extra school days per week. Another explanation could be remedial teaching in the sense that tutors might be making some efforts to identify the child’s weakness, and teach accordingly. And finally, as Dr. Wadhwa points out in the ASER report, the link between incentives and accountability – if someone is paying for a service, the onus is on the service provider to deliver, because the consumer can always ‘vote with her feet’.
Bray, Mark. 2007. The Shadow Education System: Private Tutoring and Its Implication for Planners. UNESCO: International Institute for Educational Planning, Paris
Duflo, Esther, Glennerster, Rachel, & Kremer, Michael. 2007. Using Randomization in Development Economics Research: A Toolkit. Handbook of Development Economics, edited by T. Paul Schultz and John A. Strauss, Vol. 4
French, Rob & Gandhi-Kingdon, Geeta. 2010. The Relative Effectiveness of Private and Government Schools in Rural India: Evidence from ASER Data. DOQSS Working Paper No. 10-03, Institute of Education, University of London
Muralidharan, Karthik. 2013. Priorities for Primary Education Policy in India’s 12th Five Year Plan. India Policy Forum 2012-13. Vol. 9, pp1-46
Wadhwa, Wilma. 2014. Private Inputs into Schooling: Bang for the Buck?, ASER 2014. Available at http://www.asercentre.org/Keywords/p/205.html.
 See Muralidharan (2013).
 See Bray (2007).
 There are Statewise variations. For details, see Wadhwa (2014).
 This is referred to as ‘self-selection’ problem in empirical economics.
 Duflo et al. (2007).
 Approach is similar to French & Gandhi-Kingdon (2010)
 We perform same analysis using ASER 2012 data as well. Since results are fairly similar, we report findings obtained from using ASER 2011 only.
 Details can be found in any ASER report available on ASER website.
 Baseline is a child in government school not attending private tuition.