The myth of the portable teacher

What happens when effective teachers transfer into tougher schools?

If you’re a teacher, it may seem obvious that your ability to teach is an inherent part of who you are. Like any other aspect of your character, surely you are who you are wherever you are? If you can explain clearly, build routines, insist on high standards and make students work hard in one school, why wouldn’t you be able to do the same somewhere else?

It’s easy to see why the idea of moving effective teachers into struggling schools has such appeal. For policymakers, it feels fair, practical and admirably direct: if some children have less access to strong teaching, then paying strong teachers to work where they’re needed most looks like a straightforward act of justice.¹ For teachers, the assumption is just as seductive. If we’ve been successful in one school, it’s natural to believe that success belongs to us, that our explanations, routines, standards and judgement will travel with us wherever we go. But as many teachers discover, moving schools can shatter that belief. Success in one setting doesn’t guarantee success in another. Not only are the students different, so too are the routines, the curriculum, the expectations, and — perhaps most crucially — the institutional backing we’ve taken for granted may be weaker or all together absent. Personal effectiveness may be, at least in part, a relationship between our expertise and the conditions in which we work.

Teacher quality is real

Teachers matter enormously. The difference between strong and weak teaching is one of the most robust findings in the value-added literature. A one standard deviation increase in teacher quality is associated with annual gains of at least 0.11 standard deviations in maths and 0.095 in reading, while other studies find that stronger teachers are linked not only to higher test scores but to longer-term outcomes such as college attendance and earnings.²

One of the strongest arguments for focusing on teacher quality is that there is often more variation within schools than between them. Hanushek, Rivkin and Kain’s work on teachers, schools and academic achievement used matched panel data to disentangle the influence of teachers and schools, with particular attention to the problem that students and teachers are not randomly distributed. Their findings helped establish a central point in the value-added literature: school-level averages can hide substantial classroom-level variation.

A student in a weaker school may learn more in the classroom of a highly effective teacher than a student in a stronger school learns in the classroom of a much less effective one. School quality shapes the odds, but it doesn’t determine the quality of every lesson. Students don’t experience school in the abstract; their experience is of particular teachers and lessons. Even though a school can be chaotic, individual classrooms can be oases of calm. Equally, a school’s impressive reputation can allow some teachers to coast.

Because teacher quality can — and often does — exceed school quality, it’s tempting to assume it must be stable, self-contained and portable. We imagine a good teacher as someone who embodies effectiveness, ready to be lifted from one school and slotted into another. But although teacher may be more effective than the school around them, they’re never independent of it.

What happens when strong teachers move?

Matthew Kraft and John Papay’s recent paper on teacher portability tests this assumption directly. It uses the Talent Transfer Initiative, a randomised US programme in which low-achieving schools were allowed to offer high-performing teachers from higher-achieving schools a $20,000 stipend to transfer and stay for two years. The design created a rare natural experiment, allowing the researchers to see what happens when teachers with strong records of effectiveness move into markedly different school contexts.

The headline result suggests the programme was a success: transfer teachers performed better than the teachers these schools would otherwise have been likely to hire. But, compared with their own previous performance, they became less effective, with their value-added impact falling by around 0.12 student standard deviations after transfer, enough to move the typical high-performing transfer teacher from about the 85th percentile of effectiveness to the 66th.

This makes clear that teacher effectiveness depends, at least in part, on the conditions in which teachers work. If effectiveness were simply a stable property of the individual teacher, the decline would be hard to explain. A great teacher would remain just as effective wherever they were placed, rather than having their impact altered by different students, routines, curriculum, leadership and school culture.

In the study, the nature of the work itself changed. The transfer teachers moved into classrooms where prior maths attainment was 0.40 standard deviations lower and prior English language arts attainment was 0.29 standard deviations lower, meaning they were having to teach students with substantially different starting points, gaps in prior knowledge and instructional needs. The proportion of students eligible for free or subsidised school meals rose from 68% to 92%. Teachers in the study also reported much weaker conditions for teaching: positive ratings of student motivation fell from 86% to 39%, student discipline from 81% to 52%, and parental involvement from 60% to 32%.

This confirms what many teachers discover when they move schools: expertise is both general and local. Some things we take with us — subject knowledge, habits of explanation, standards, routines and judgement — but some we learn from the students in front of us: what they know, what they struggle with, how quickly they grasp new ideas, which examples and explanations are likely to work.

We also learn the school’s habits: whether leaders follow through, colleagues share standards, the curriculum is fit for purpose, and assessment tells anyone anything useful. We learn how refusal is handled, whether consequences happen, whether line management clarifies or clutters expectations, and whether meetings solve real problems or merely produce evidence of effort. Some of this knowledge may be transferable, but most of it has to be reaquired around new students, routines and constraints.

Kraft and Papay argue that the decline in teacher effectiveness “appears to be driven by lower match quality, negative indirect school effects, and the loss of student-specific human capital.” In plainer language, teachers are more effective when they’re well matched to their school and students, when the school supports good teaching, and when they have built up knowledge of the students they teach. Change those things, and effectiveness changes too. As they put it, “Schools are critical in supporting or constraining teacher effectiveness.”

The measurement trap

For all their faults, value-added models (VAM) capture something important: whether students taught by a particular teacher make more progress than we’d expect, given their prior attainment and other observable characteristics.³ However, these models can tempt us to treat effectiveness as if it’s located entirely within the teacher. The danger isn’t the measure itself, but that we miss what it abstracts away. VAM is a proxy for teacher effectiveness; a statistical estimate of a teacher’s contribution to student test-score growth, not a direct observation of teaching quality. It depends on the effects of all the other teachers to have taught a student, as well as the personalities, beliefs and home lives of students themselves. And this is before we even begin to think about the school in which estimates are produced.

Approached with caution, this can be useful. Modelling makes it clear that teachers differ in quality and that students learn more in some classrooms than they do in others. As such, it helps separate a teacher’s contribution from the advantages or disadvantages students bring with them. Without it, we risk mistaking high prior attainment for effective teaching, or difficult starting points for weak teaching. The problem begins when an estimate of a teacher’s contribution is treated as the teacher’s essence.

School improvement narratives have long been dominated by a strangely individualistic model of teaching. Because weak teaching is treated as the property — or fault — of weak teachers, improvement becomes the process of correcting, coaching, incentivising, moving or replacing individuals. Of course, sometimes that’s necessary. Like all other human attributes, teaching ability distributes on a bell curve and some teachers need more support than others.

The problem is that this account is hopelessly incomplete. Teaching quality is produced by teachers working within a particular environment. Schools can make it easier or harder for teachers to be effective. In some schools, even the weakest teachers can teach effectively due to how efficiently systems operate. In others, even the most skilled are limited by the chaos or indifference that surrounds them. If we fool ourselves into believing teaching is purely the output of teachers, not only do we miss the most effective levers to pull to improve teaching, we contribute to the misery so many teachers experience as the day-today reality of their professional lives. Any measure is misleading if it encourages us to ignore the extent to which the system produces the thing being measured.

When we behave often behave as though improvement can be installed in individual teachers while everything around them remains unchanged we are likely to pour resources into a ‘solution’ that makes everything harder for teachers and has school leaders operating in the dark. This is the coaching trap: identifying the teacher as the unit of improvement when it’s school culture that’s the source of the problem.

The Coaching Trap: Why Better Teaching Depends on Better Schools

My point isn’t to abandon instructional coaching — it’s still a good bet in schools with strong cultural foundations — but to stop expecting it to create those foundations without addressing the systemic pressures which make it hard for teachers to be their best. As Kraft and Papay’s portability study makes clear, if school conditions shape teacher effectiveness, then trying to improve teaching one teacher at a time is, at best, inefficient and, at worst, irresponsible.

Kraft and Papay’s earlier work on professional environments points in the same direction: teachers improve more, and more quickly, in schools with stronger professional conditions: better order and discipline, purposeful collaboration, effective leadership, useful professional development, coherent culture and fair evaluation. And even the best teachers can be diminished by a change of context.

Systems and forces

In case it seems that I’m just blaming leaders for poor teaching, I need to make it clear that principals are not responsible for the conditions in which their schools have to operate. Schools sit inside wider systems of accountability, inspection, funding, recruitment, safeguarding, SEND provision, curriculum policy and public expectation. Many of the pressures that distort school life arrive from outside the school gates. Heads may create local systems, but they do so inside larger systems that reward some behaviours, punish others and make certain kinds of leadership much harder than they should be.

It’s helpful to think of schools as systems inside larger systems, constantly buffeted by external forces. Systems include those things over which leaders have at least some control: routines, structures, habits, processes, incentives and norms. Forces are those things imposed on schools from outside.

Forces are the weather. Funding settlements, inspection frameworks, teacher supply, local demographics, statutory duties, parental expectations, political priorities, social media storms, safeguarding demands and the changing complexity of children’s needs all press on schools regardless of a headteacher’s decisions. We can’t avoid these forces; they are the background against which we must contend.

Systems, on the other hand, are the responses we engineer in order to operate within those forces we can’t control. They’re the routines, structures, habits, processes and norms we build to make it possible to survive — and, ideally, thrive —despite external pressure. Children’s natural tendency to prefer not to be made to do algebra on a Friday afternoon requires reliable behaviour routines. Overstuffed exam specifications and students’ uneven prior knowledge require an efficient, coherent curriculum. Uncertainty about what students know requires assessment that reveals patterns and suggests workable solutions, rather than merely replicates what we already know as is typical with most internal assessment.⁴ Organisational complexity requires line management that balances trust and accountability in ways that allow teachers to do more than merely comply with explicitly stated expectations, and our finite capacity for attention requires systems that protect teachers from the illusion that all demands can be treated as equally important.

If we blame heads for the weather, we’ve simply moved crude individualism up a level. Blaming teachers for the effects of poor school systems is wrong, but blaming heads for the effects of poor national systems is no better. Headteachers inherit shortages, incentives, legal duties, inspection anxieties, budget constraints and political noise. They are not sovereigns. They are operating inside a system that often rewards visible compliance over genuine improvement.

Once effectiveness is imagined as something possessed by exceptional individuals, school improvement can look like a matter of importing the right person. This was the logic behind the cult of the ‘superhead’: find a charismatic turnaround leader, drop them into a difficult school, give them licence to act, and wait for improvement to arrive.

There have been heads who took on chaotic schools, created order, raised expectations and improved results. It would be foolish to deny that. But the superhead story has always been more complicated than its admirers wanted to admit. Too often, it treated leadership as a portable personal force rather than a set of conditions, relationships, routines and judgements built over time.

The cleanest caution here is not scandal but sustainability. In 2016, Schools Week reported on research into ‘superheads’ which found that although results rose during their tenure — often due to a combination of exam gaming and off-rolling — scores fell by an average of 6 percentage points after they left, with larger falls when the head had been in place for longer. Worse, the subsequent clean-up costs were also substantial, coming to £11.8 million across 21 schools.⁵

Effective heads don’t parachute in with a bag of tricks. Instead, they build a school’s capacity to improve without depending on them. They develop middle leaders, strengthen curriculum, make behaviour routines predictable, improve the quality of professional conversations, protect time, clarify accountability and make it easier for teachers to be their best.

The worst version of the heroic headteacher model does the opposite; centralising decision-making, accelerating only high profile change, frightening dissent into silence, chasing short-term indicators and mistaking compliance for strong culture. A school may become more orderly, but not necessarily a functional place for teachers to work.

With teachers, the temptation is to imagine classroom effectiveness can be moved without changing the routines, relationships and curriculum that shaped it. With heads, the temptation is to imagine institutional effectiveness can be imported without developing the people, habits and safeguards that sustain it. A strong head in a weak governance structure, with brittle middle leadership, poor trust and exhausted staff, may be able to impose change but this is not the same as building a culture in which teachers and students can thrive.

Towards a surplus model of school leadership

The influence of headteachers, though profound, is indirect. Leaders shape what’s prioritised, tolerated, measured, ignored and allowed to drift. A headteacher can’t improve teaching through force of will. They can, however, create the conditions in which excellence becomes more likely.

A deficit model of school improvement locates failure in individuals. Because teachers can’t be trusted, students don’t care and middle leaders aren’t holding the line, staff need tighter systems to make them comply. When things go wrong the default assumption is that someone is to blame and the solution is tighter monitoring, greater prescription and less professional trust.

A surplus model begins from a different premise. It assumes, until there is good evidence otherwise, that people are acting in good faith, working hard and trying to do the right thing. When things go wrong, it seeks to identify and remove the barriers to success. This means recognising that weak practice often emerges from unclear expectations, poor feedback loops, conflicting incentives, missing knowledge, unmanageable workload or convoluted routines.

Students arriving late to lessons offer a simple example. Deficit thinking reaches quickly for character: students are lazy, teachers are inconsistent, sanctions aren’t tough enough. A surplus model looks first at the design of the school day. Is movement time realistic? Are corridors congested? Do entry routines vary from classroom to classroom? Is the policy clear enough to be enacted day to day? The solution may still involve sanctions, but it begins with design rather than blame.

Deficit leadership is an attempt to find simple answers to complex problems, whereas a surplus model asks leader to take responsibility for the systems they oversee.

These models obviously contain elements of caricature. No school operates an entirely deficit or surplus model but you make recognise elements of these approaches to leadership in schools you’ve worked in. Importantly, a surplus model is not a call for blind trust. Trust without precision becomes drift; accountability without trust becomes theatre. Teachers are most likely to improve when they feel trusted, but they also need to know they are accountable. The aim is intelligent accountability: clear enough to prevent mutation, humane enough to avoid compliance theatre, and expert enough to distinguish looking good from being good.

Effective school leadership is to build routines in which being good is easier than looking good. That means turning behaviour follow-up into something staff can trust, reducing the number of priorities, protecting departmental time, ensuring assessment meetings to look at students’ work rather than just at spreadsheets, and insisting that policies are either maintained or abandoned rather than endlessly announced and ignored.

Every school trains its staff but does it train them into better habits or teach them how to survive bad ones? A surplus model of leadership is focussed on creating a culture where better teaching is more likely. This still leaves room for teachers to take responsibility for what they can control but acknowledges the limits of what is within their sphere of influence. Teachers still need secure subject knowledge, disciplined habits, clear boundaries, and to focus on the intellectual preparation required to teach effectively but it’s much easier to develop these when not having to act as a warlord in a failed state.

Students in the most disadvantaged schools absolutely need the best possible teaching but this is only possible when they are aligned around the principles that make such teaching possible. Where a well-led school multiplies professional development, a school which is poorly led absorbs it.

The belief that school improvement is only possible when great teachers work with great school leaders is unhelpful. What we need is great teaching and great school leadership. Both are best thought of as the product of systems rather than of individuals.

Moving people who have proved themselves successful in one setting into harder schools may still be worth doing, but this transfer is not what a sustainable improvement strategy.

1. Of the many attempt to transfer the ‘best’ teachers between schools the most well known is probably the US Talent Transfer Initiative, which offered high-performing teachers a $20,000 incentive, paid over two years, to transfer into low-achieving schools. Kraft et al. use this programme to test whether teacher effectiveness is portable across very different school contexts. The broader policy impulse also appears in differentiated-pay schemes and incentives for hard-to-staff schools, as well as in English school improvement through system leadership, academy sponsorship, National Leaders of Education and the “superhead” tradition, where the emphasis has often been on moving leadership expertise rather than classroom teachers.

2. See Rivkin, Hanushek & Kain (1998) “Teachers, Schools, and Academic Achievement”. Hanushek reviews a wider body of value-added evidence and discusses the economic significance of variation of teacher effectiveness in his 2011 paper, “The Economic Value of Higher Teacher Quality”. For evidence on longer-term outcomes, see Chetty et al. (2014) “Measuring the Impacts of Teachers II: Teacher Value-Added and Student Outcomes in Adulthood” - the authors use school district and tax records for over one million children and find associations between high value-added teachers and later educational and economic outcomes.

3. For useful cautions on the interpretation of value-added estimates, see Koedel, Mihaly & Rockoff (2015) “Value-Added Modeling: A Review,” which reviews the strengths and limitations of teacher value-added models; Bacher-Hicks & Koedel (2023) “Estimation and Interpretation of Teacher Value Added in Research Applications,”; and Raudenbush (2004) “What Are Value-Added Models Estimating and What Does This Imply for Statistical Practice?” On teacher-school match effects, see Jackson (2013) “Match Quality, Worker Productivity, and Worker Mobility: Direct Evidence from Teachers” which argues that teacher productivity partly reflects the match between teacher and school rather than only the teacher’s fixed quality.

4. Discriminatory assessment is designed to separate students, usually producing a spread of outcomes that allows ranking, grading or selection. This is appropriate for public examinations such as GCSEs and A levels, where the purpose is to discriminate between candidates. Mastery assessment has a different purpose: it asks whether students have learned the intended curriculum and should therefore reveal gaps, misconceptions and next steps. If a class has been taught something well, a well-designed mastery assessment should not be disappointed when most students succeed. I’ve written about this distinction here.

5. Some superhead cases have also exposed more basic governance risks. Greg Wallace, (no, not that one!) the former executive headteacher of the Best Start Federation in Hackney, had a DfE teaching ban overturned by the High Court after allegations of financial misconduct, while Jean Else, formerly of Whalley Range High School for Girls, apologised to a professional conduct committee for failing to follow procedures during her tenure. These examples need handling carefully, but they show how rescue narratives can make institutions too willing to suspend ordinary safeguards because someone appears to be getting results.