No one disputes that good teachers are crucial if students are to perform well. And there is certainly plenty of evidence to back this up. High-performing teachers are important for both short-and long-term outcomes, such as test scores, grades, university attainment and earnings. Increasing the effectiveness of teachers could generate significant lifelong benefits.

And that is not all: the positive effects spill over into the rest of society. For example, higher educational attainment reduces the likelihood that young people will commit crimes later in life, which is good for everyone. Higher student test scores are also linked to economic growth and technological advancement. In economics, such side benefits are usually referred to as “positive externalities”, and they suggest that teacher effectiveness is important for improving students’ lives and society at large.

So how can we ensure that teaching is effective? The approach in England has generally been to put regulatory demands on state schools to hire teachers with a certain level of education. At the same time, unions have been fighting attempts to stop automatic pay rises based on number of years of service and strict employment criteria. The strategy in the majority of schools so far has been centralisation, not flexibility.

This could, theoretically, make sense - if we knew what made teachers effective, we could impose strict requirements on schools, resulting in higher performance system-wide. But is it really viable for us to regulate characteristics such as qualifications as a proxy for teacher effectiveness?

Well, no. The empirical research is consistent across different countries (including England): there is basically no evidence suggesting that teachers’ qualifications affect students’ performance. Similarly, experience has an impact only for the first couple of years at most, after which the effect disappears entirely. It turns out that better trained and more experienced teachers are not necessarily better teachers.

This has profound implications for the viability of teacher qualification requirements, as well as centralised employment terms and pay based primarily on years of service. If we are interested in increasing teacher effectiveness, we have no reason to latch on to these approaches. Rather than promoting better teaching, they are, in fact, likely to be directly harmful to students: requiring teachers to have specific qualifications risks preventing many perfectly good educators from entering the market. And, similarly, basing teacher pay on experience as an indicator of effectiveness diverts us from testing other principles that might matter more.

At the same time, some evidence exists that teachers’ subject knowledge does make a difference to their students’ performance. There may therefore be a case for minimum subject knowledge requirements, which could be established through a diagnostic test. But the impact should not be overstated and, clearly, other factors also determine teacher effectiveness, although they are difficult to quantify. Because of this, any diagnostic test should be used only to weed out those who are obviously not up to scratch in their subject.

Experimental approach

So what is a viable alternative once the worst apples are removed from the barrel? It is pretty simple: decentralise requirements, employment terms and pay to the school level. Headteachers are more likely to be able to separate high-performing from low-performing teachers than any centralised regulator.

In the context of uncertainty over what factors increase teacher effectiveness, it is better to allow a decentralised process in which professionals can try different approaches. Individual-level contracts are the way to go, since they allow headteachers to experiment with different remuneration schemes, including various performance-related pay structures, to further advance teaching quality.

Of course, academies and free schools have much greater autonomy over employment requirements and teacher pay than other publicly funded schools. Yet there is little reason why we should not extend this autonomy to all institutions. The government is correct to allow state school headteachers more flexibility in how pay increases are determined. We are still waiting for it to abolish qualification requirements for teachers.

The fact that good instructors matter does not mean that we know what makes them good. Indeed, the quest to find characteristics that can explain effectiveness has largely proved to be fruitless. Only by allowing stronger experimentation - which requires lax central regulation - will it be possible to raise the overall quality of educators.

Gabriel H Sahlgren is director of research at the Centre for Market Reform of Education, based at the Institute of Economic Affairs, and author of Incentivising Excellence: school choice and education quality.