Edward Glaeser points to an interesting assessment of such programs for teachers under implementation in 26 US states by Stuart Buck and Jay P Greene of University of Arkansas, who find that standard implementation problems that bedevil reforms in all sectors come in the way of the success of merit-pay projects. They claim that entrenched vested interests work towards "blocking, diluting or co-opting" the program by either preventing its enactment and diluting the focus or evaluation methods or outcome standards.
They write that in the absence of competition, "even if wise and benevolent state actors manage to get the incentives right at a particular moment in time in a particular place, their actions can always be undone by immediate successors" under the influence of powerful vested interests,
"With greater educational competition, however, state bureaucracies, local school boards, and building principals face a sharpened incentive to figure out how to structure their own pay systems so as to extract excellent performance from their employees. If they fail to do so, students may be more likely to leave the state school system altogether, thus diminishing the public schools’ budget... As a result, competition enhances the possibility that merit pay could reach its full potential, rather than being co-opted by interest groups whose goal is to reward whatever it is that they already happen to be doing. Put another way, competition between schools diminishes the political power of interest groups that are dedicated to undermining merit pay."
Their concluding comments about the problems faced with successful implementation of merit-pay is strikingly appropriate about many public institutions
"Public schools are not primarily educational institutions where policies are organized around maximizing student achievement. Instead, public school systems have to be understood as political organizations organized around the interests of their employees, their union representatives, and affiliated politicians and other interest groups... difficulty with attempting merit pay in education is that it consists of imposing a market-based practice in a non-market environment. None of the forces that cause organizations to seek effective merit pay systems, as well as maintain and alter them effectively over time, exist in public education. Established interests will forcefully resist and undermine merit pay, making its success doubtful."
However, it may not be correct to read this as a general endorsement against designing public policies with built-in performance-based incentives in non-market based environments. It only cautions about the presence of numerous adversarial factors which need to be mitigated when implementing such policies. I am skeptical of the success of teacher-based incentive pay, especially in developing country environments, for the following reasons
1. In many schools, the baseline is so low that even small interventions can make a dramatic difference in performance. For example, ensuring higher teacher and student attendance, teachers attendance for trainings, systematic syllabus coverage, periodic homework assignments, conduct of examinations and prompt evaluation of examination papers will all by themselves contribute substantially towards improving school performance. Compliance with these basic requirements, a function (and basic pre-requisite) of routine supervisory over-sight, is poor due to the virtually non-existent field presence of the administrative machinery.
The logic behind performance-based financial incentives is to kindle the inner competitive/motivational urge within teachers to improve their productivity and thereby increase performance outcomes. In an environment with such low baseline, which can be raised with small and routine administrative interventions, the incentive would then get inefficiently frittered away. It would be the equivalent of using powerful antibiotics on common-cold.
2. This can have many perverse consequences. For example, it could end up diverting attention from critical administrative deficiencies and ensuring that adherence to even routine duties and responsibilities are now made dependent on financial incentives. Taken to its logical conclusion, these incentives become an extension of salaries and lose their productivity improvement value.
These incentives could also end up generating perverse "entitlement effects", where teachers could start demanding incentives for even complying with their routine duties and responsibilities (for which they draw their salaries in the first place). While this would generate perceptible improvements in the immediate period, the medium and long-term moral hazard effects could be detrimental to the system as a whole.
3. In view of the aforementioned low-hanging fruits, did the mere fact that the schools were being treated for merit-pay based performance improvements itself contribute to actual performance improvements? In contexts where the baseline is itself very low, even small and cosmetic steps towards improving monitoring and supervision can yield substantial results. Merely being part of an experiment, where performance would be monitored more closely than in the business as usual case, is itself likely to shake-up the establishment and ensure better performance outcomes. Further, the nature of the experiment means that it is not possible to have double-blind treatments to control for the Hawthorne effect. Therefore, it may not be correct to conclude that the same outcomes would result when the policy is implemented on scale.
4. Even assuming the effectiveness of merit-pay, how do we know what is the optimal incentive amount. The wide variations in the socio-economic environments of different schools means that the optimal incentive structures are like to vary vastly. Given the aforementioned politics of such decisions, there exists the strong possibility of over-paying for achievement of sub-optimal results (which would have been achieved even without any incentives).
5. Crowding-out effect - Only the good (or already motivated) teachers are likely to be influenced by incentives. They enroll (or "crowd-in") for the incentive programs. This has the perverse effect of further widening the standards between the good and poor teachers, thereby causing even more distortions within the school system.
6. Incentives without disincentives - All existing merit-pay systems involve one-way incentives with no disincentives. Even if implemented, assured of their monthly salaries, a large share of teachers are not likely to respond to any of these incentives. Given the large salaries drawn by teachers, the quantum of incentives is not likely to be large enough to make substantial cognitive impact on the entrenched inertial psyche of majority of the teachers.
I am inclined to the opinion that financial incentive structures that reward improvement in school (as against teacher) performance may be more effective than teacher-based incentives. However, this too comes up against formidable challenges like the wide variations in socio-economic environments that determine performance outcomes. Further, the challenge with designing such incentive structures is to identify quantifiable and objective parameters that are reliable indicators of school performance outcomes.
But with appropriate normalization among the few commonly observed determining factors, it may be possible to normalize and arrive at an appropriate measure of the true base-line for all schools. In any case, financial incentives for school performance is easier to implement for the following reasons
1. It is likely to face less political opposition since it does not directly target teachers at an individual level.
2. By not directly focusing on individual teacher-specific outcomes, it does not create much of the aforementioned incentive distortions.
3. School-based incentives creates a great opportunity for schools to access funds for meeting their infrastructure and other requirements. This is likely to increase its popular acceptability.
4. School-level outcomes can be measured with much less distortions and are a more accurate reflection of student learning outcomes. Such macro parameters are also less vulnerable to being manipulated and subverted.
5. Involving the local community, an important factor in ensuring political acceptability and sustainability, is easier with such macro-level interventions.
6. Given the vast differences in school backgrounds, a school-level incentive program gives each school the flexibility to tailor school-specific incentives targeted at teachers and others.
Another alternative is to start with incentives for principals than teachers, as suggested by Jim Stergios, the executive director of the Pioneer Institute, a public-policy research organization in Boston (via Ed Glaeser). This arguement works on the assumption that school performance could improve dramatically if highly motivated principals had the resources to attract the best teachers and the strength to move the worst ones into other sectors. Even intutively, a large enough incentive for principals, could have the potential to get principals to manage their schools optimally and generate greater bang for the buck.