The extensive use of contract or para teachers in government primary schools is regarded as one of the most important policy innovation in the education sector in developing countries. But this policy has been highly controversial, with both supporters and opponents arguing their sides eloquently. Indian policy-makers seem to have taken the side of the opponents, as indicated by the provisions of the recently implemented ‘Right to Education’ Act. The RTE categorically states that there would be no contract teachers in the schools. All teachers would be regular teachers.

Before taking any sides, it would be nice to look at the empirical evidence. Unfortunately, there is hardly any rigorous work in the Indian context. But the two papers- ‘The Relative Effectiveness and Costs of Contract and Regular Teachers in India’ by Paul Atherton and Geeta Kingdon, and ‘Contract Teachers: Experimental Evidence from India’, by Karthik Muralidharan and Venkatesh Sundararaman, analyzing the effect of contract teachers on learning outcomes, which were presented at the 6^th Annual Conference on Economic Growth & Development (Dec. 16- 18, 2010), Indian Statistical Institute, Delhi, fill this gap.

These papers provide a strong and rigorous evidence of the huge positive impacts of contract teachers on student achievements. Both papers rely on different methods (non-experimental vs. experimental), and their underlying data comes from different states (UP- Bihar in case of the first, and Andhra in case of the second), which also suggests that the effects are likely to be very robust.

Empirical Challenges:

We are interested in figuring out the effect of contract teachers on student learning outcomes. So a simple regression would be

Y = β0 + β1.Contract Teacher + Ω.Other Controls + µ

But the coefficient β1 in the above setting is likely to be severely biased. The reason is that assignment of contract teachers to schools is likely to be correlated with the characteristics of the school. For example, the contract teachers may be assigned to the schools in remote areas. In that case, the dummy of contract teachers is picking up the effect of being a remote school as well. And there is no way we can figure out the size of that effect. We can merely guess its sign. Similarly, even within a school, the assignment of contract teachers to the kids may be non-random. For example,contract teachers may be assigned to weaker (or stronger) students. Then the dummy is picking up the effect of teaching to a particular type of students. In short, the ordinary least squares regression is not really useful in describing a causal relation between contract teachers and learning outcomes.

The above problem can be solved by either randomly assigning contract teachers to schools (as done by Muralidharan & Sundararaman) or controlling for unobserved school characteristics which are correlated with assignment of contract teachers and learning outcomes, by using school fixed effects (as done by Atherton & Kingdon).

Results:

I will merely discuss the result without going into the details of the dataset, the regression equations or the size and significance of the coefficients.

The results point in the same direction- Contract teachers are more effective than regular teachers. The evidence also indicates that the effect of contract teachers remains even after a reduction in the class size (and in the pupil-teacher ratio) due to addition of a teacher. Further, relatively disadvantaged students, and those who are in lower grades (such as grade 1, 2) are the ones who benefit more from contract teachers.

Why?

Why do contract teachers perform so much better than the regular teacher? A plausible reason would be the nature of their contracts- the contracts are annually renewable. Further, the contract teachers are hired locally. Thus, they belong to the same community, society, and thereby face strong accountability pressures. Further, being local implies the ‘social distance’ between the teacher and the student is also lesser.

Policy Implication:

The combination of low costs, superior performance measures than regular teachers on attendance and teaching activity, and higher positive impact on student learning indicates that the use of contract teachers could be a very cost-effective way of improving primary education outcomes in the developing countries. But it is not as straightforward as it seems. One can not rule out the possibility that once in large numbers, the contract teachers might demand permanent jobs and would be so granted just before the elections (as happened in Himachal Pradesh, the first state to recruit contract teachers). Then in the long term, there may not be any cost-effectiveness of contract teachers.

Nevertheless, the above evidence clearly indicates that one needs to seriously think about the recruitment, contractual terms and remuneration of teachers in the government schools. And this applies not just to teachers but to all government services.

Muralidharan and Sundararaman suggests one possible course of action- to hire all new teachers as contract teachers at the school level, and create a system to measure their performance over a period of time (six to eight years for example) that would include inputs from parents, senior teachers, and measures of value addition using independent data on student performance. These measures of performance could be used in the contract-renewal decision at the end of each fixed-term contract (or to pay bonuses), and consistently high-performing contract teachers could be promoted to regular civil service rank at the end of a fixed period of time. Continuous training and professional development could be a natural component of this career progression, and integrating contract and regular teachers into a career path should help to address most of the concerns above, including the political economy ones.

Ambrish Dongre is Senior Researcher with the Accountability Initiative.