Teach First – a Failed Programme?

By Minsky Moment

Over the last few years, even as the US has become increasingly disillusioned with Teach for America, Britain has fallen in love with its Teach First programme. I’m not so sure this has been a good thing.

The idea behind Teach First is to narrow the socio-inequality gap in the British education system by raising the aspirations, and in turn attainment, of students from lower socio-economic backgrounds. It tries to achieve this by placing the best graduates in to the most challenging schools – the aim, here, is that smarter teachers will better motivate their students.

No one really disputes that an education gap divides rich and poor. And the most careful research finds a pervasive knock-on effect on social mobility. A 2006 OECD report found that Britain was the worst of its member states when it came to social mobility. But even though inequality is a clear and present problem, it hardly follows that Teach First is an effective solution. In fact the rationale behind Teach First is based on premises that are at worst disproven, and at best inconclusive. They come down to the following:

  1. The top graduates make the best teachers
  2. It is a bad thing that too few new graduates from other teaching programmes enter the toughest schools
  3. Six weeks is enough time to train teachers for tough environments

That notwithstanding, the vision and aim of the programme is commendable. But little has been asked about how much closer it is to achieving its aims, 12 years since its inception. So this post is devoted to debunking the premises behind Teach First, and then looking at some ways of reform.

The top graduates make the best teachers

Not only are over 80% of Teach First teachers from Russell Group institutions, but it has also become the largest graduate recruiter from Oxbridge in recent years. Many other employers with similar recruitment patterns are labeled as elitist – these statistics are morbidly ironic considering the cohort that Teach First claims to be serving are those most likely to be excluded from the top universities.

In a blog post Laura McInerney, a Teach First alumnus who is currently studying for her PhD, claims this elitism is a myth:

‘Academically you do need to be decent: a 2:1 or above (from any university). But on its own, that’s not enough. There are also 8 competencies that are tested over the full-day assessment centre all participants go through.’[1]

But then why are the students from a small minority of the best universities in the UK overwhelmingly more likely to have these competencies? Is it therefore unfair to argue that Teach First is singling out the best graduates? Hardly. We can put the statistics in the context of Teach First’s (somewhat boastful) evidence to parliament in 2012. It’s pretty clear that Teach First has an obsession with the top graduates, ignoring other competencies when describing the background of their teachers. Their submission reads:

‘Teach First has succeeded in making teaching a profession of choice for top graduates, recruiting from 149 universities including top selective institutions. For example, in 2010, 282 applications were received from Oxford graduates—almost 10% of the graduating class.’[2]

Of course if the best graduates necessarily made the best teachers, this would be fine. But the reality is blurred in the statistical vacuum of the word, “inconclusive.” Papers analysing this topic have found little to no correlation between academic attainment and teacher quality, apart from (possibly) in mathematics.

So the top achieving graduates from Russell Group Institutions are apparently coincidentally more likely to meet Teach First’s required attributes, even though they don’t necessarily make the best teachers. So who, then, do make the best teachers?

Too few new graduates from other teaching programmes enter the toughest schools

Before Teach First, purportedly only 1 in 10 graduates of teaching programmes chose to work in the toughest schools. This statistic was first published in a paper, by the now defunct Training and Development Agency for Schools (TDA) in 2010. Since then, it has been used as a battle cry for Teach First and its supporters. Teach First, by sending graduates in to the poorest schools, aims to rectify this problem.

Imagine a recent graduate from a medical school who wished one day to be a surgeon, being told to perform open heart surgery on their first day. Regardless of what interesting experiences the graduate gains, regardless of their ability to see things differently to someone who has been performing surgeries for twenty years, it is the patient that is the biggest loser in this procedure.

Or perhaps someone begins work at Sea World, and their immediate responsibilities are looking after the killer whales and great white sharks. Of course, there would be regular supervisions, a six-week crash course and a buddy to lean on when you’re not strong, in order to be your friend and help you carry on.

You get the point.

The students that Teach First often unleashes its graduates upon often come from challenging backgrounds beyond only socio-economic concerns. To assume that a recent graduate, that is often only three years older, will be adequately able to take on the sometimes-necessary roles of mentor, disciplinarian or translator is not sound logic. It places unreasonable pressure in the short term on people who otherwise could make excellent teachers in the long term.

Wouldn’t it however be wiser to get more experienced teachers with a proven and successful record in to these schools? As it turns out it would, according to practically every major study out there. The best example of this is the Talent Transfer Initiative (TTI) in America, which pays top quintile teachers to move to schools in low income areas for a period of two years. Over 90% of teachers completed the two-year programme, whilst 60% stayed beyond the end of the programme.[3] The results saw an average of between 4-10% increases in attainment in all subject areas – which is staggeringly good compared with most US education pilot studies. And Slate’s Dana Milbank has perhaps the most interesting bit:

‘These transfer teachers were far from the Teach for America archetype of a young, transient Ivy League grad. Their average age was 42, and they had an average of 12 years of experience in the classroom. They were also more likely than control group teachers to be African-American, to be homeowners, and to hold a master’s degree. In short, they were stable adults with deep ties to the cities in which they worked.’

Sounds relevant at all? Sadly, whilst the results have been promising and the report caused a stir, less than 10% of possible applicants took up the opportunity to participate in TTP.

The challenge shouldn’t be getting the best graduates in to the toughest schools in a sort of finding-yourself-on-ones-gap-yah zeal, but rather getting the best teachers in to the toughest schools – and this often means recruiting experience.

So given that teachers get better with time, how is the Teach First faring at retaining its staff?

Does Teach First retain its staff?

It’s important to first be clear on why this is an important battleground. Teachers who stay in schools for prolonged periods often achieve better results for disadvantaged students; as they tailor their pedagogic techniques to suit the needs of the class and there is less disruption in the child’s learning.

People often underestimate Teach First here; a high number of TF-ers not only qualify, but also stay on. In certain areas, these numbers match and even surpass the PGCE programme.[4]

We now know Teach First trainees (like all trainees) aren’t the best teachers, they merely have the potential to be so one day. With this in mind, there are two knock on effects of people dropping out early on any teaching programme. Firstly, it weeds out the ineffective teachers who were never suited for the job in any way. The extent of this, however, is impossible to gauge. Secondly, some teachers who had the potential to be effective are lost. The problem that applies to all dropouts is that we will never know which one they were.

So even if Teach First did have higher drop out rates, it could potentially and perhaps reasonably argue that its monitoring systems are better at weeding out the teachers who will never make the grade with greater effect than government schemes. To understand this though you would need to have a look at the reasons for Teach First-ers leaving. The results aren’t pretty:


The little table above shows that Teach First are five times as likely to leave teaching in the first five years due to reasons of “burn out” than their counterparts on other programmes.[5]

Teach First supporters argue that this is because of the nature of the schools that they have worked. Whilst this is probably fair, the significant differences in burnout could be decreased if TF weren’t focused at these schools. One former TF-er puts it more succinctly:

‘Teach First ‘ambassadors’ talk proudly about the difficulties they had to start with – the high workload, the lack of sleep, hostility from colleagues, daily mistakes – and how they fought through these and lived to teach another day.’

The issue of “burnout” is especially worrying when one takes in to account that rates of suicide and depression, amongst teachers, have been increasing at an alarming speed in recent years. The problem with such patterns is not only that people who could otherwise have made good teachers leave, but also, and most importantly, that the students lose out.

Furthermore, whilst Teach First (nigh on) matches its counterparts after the first two years, the long-term projections for teacher retention is far more worrying.


I am not going to delve too far in to the above as it speaks for itself. What I will say, is that considering the cost of persistently replacing teachers, the damage this can cause to a child’s education, and the fact that teachers often reach their peak after five years this is a statistic that poses serious questions not only about the sustainability of Teach First but also the allocation of government funds.

Where do we go from here?

It seems fairly clear, then, that for all its noble purpose Teach First has some severe structural flaws. In this light I wanted to outline a few specific areas for reform.

1.     Teach First should become a programme that trains teachers on the job by easing them in to the teaching profession, as opposed to putting them in the toughest situations.

2.     Teach First places too much emphasis on the impact of teachers within our education system. This has coincided with an increased emphasis from the DFE in the past two governments. Thus Teach First should be more careful of its language when talking about the impact of teaching. For instance, the 2010 Schools White Paper, while emphasising the importance of parents holding schools to account, does not once mention the input that parents can have in to a child’s attainment. Several studies have shown that increased parental engagement from an early age can drastically improve attainment. According to parents, the two biggest barriers to their participation are information and confidence, the latter being based around an inability to help a child as they get older and are set more challenging work. Wealthier parents circumnavigate this issue by hiring tutors. Clearly, then, there are other important factors that feed in to social-inequality besides teacher quality. Emphasising teaching to the exclusion of other areas, diminishes the responsibility but not the negative impact of other key actors.

3.     Teach First needs to either admit that it specifically targets the best graduates or address why Russell Group students are overwhelmingly more likely to meet their criteria. And what if it becomes clear that TF specifically targets the best graduates, even though there is a significant body of evidence showing little to no correlation between high performing graduates and outcomes for students? Then the government would have to justify giving over £70 million to a pointlessly elitist body.

4.     Personally, I am concerned by costs that go in to training each Teach First teacher that we know about and the lack of transparency on the hidden costs. It is disconcerting (and morally dubious) that we have some teachers who perform an equally vital role within our schools system who pay £9k fees for the pleasure of completing their government sanctioned courses, whilst on Teach First quite the opposite happens. According to government figures, if you’re a teacher on the PGCE programme teaching English, History, Biology, Geography, Music, and Design and Technology with a 2:i the amount they invest per teacher is around £12,000 pounds. If you’re on Teach First however, the cost to the government is £23,277 per trainee. There are many questions this raises:

  • Do the government actively value Teach First’s training programme more than they do their own?
  • Are these costs sustainable in the long term?
  • What does this say about how the government is currently managing its dialogue on an increase in teachers’ wages?
  • Considering the government’s own bursary system for PGCE is already linked to university grades in order to attract the top graduates, and we have ascertained that TF-ers aren’t necessarily the best people for these schools, what problem is Teach First actually solving?
  • Imagine a world where the government cuts funding to Teach First to zero and reinvests the money elsewhere…. If the PGCE were needs blind but rather gave out places on the basis of meeting Teach First’s competencies, would the “top graduates” still be interested?

5.     Lastly, (and most importantly of course) The Wilberforce Society should do a feasibility study of the TTI programme implemented in Britain, asking whether the government’s money would be better invested there.

The challenge for these schools (as with all) is: how can we give our students the best possible education? Considering all of the above, would it not be better for children from difficult backgrounds to find themselves taught by more experienced teachers, rather than recent graduates? If you agree with this basic premise, then Teach First isn’t the answer to the problem of social inequality in education.

[1] These special attributes that are four times as likely to be found in the Russell Group are; Interaction, Knowledge Resilience, Self Evaluation, Planning & Organising, Humility, Respect & Empathy, Leadership Problem Solving.

[2] In any case the claim of “recruiting from 149 universities,” is very misleading. There is no specific year. There is no country bias. But we do know that 80% of their entrants are Russell Group graduates.

[3] A retention rate that is far higher than Teach for America.

[4] Though let’s not get too happy clappy. Many people on the PGCE programme are funding themselves and these dropout rates are not too dissimilar to other Masters programmes where funding is often an issue. TF-ers, with the help of gov funding, do not have the same problem to contend with.

[5] We also know that those in leadership roles with less than five years experience in schools are also significantly more likely to leave due to burnout than those who had had more experience. Clearly, though, this is only important if you hold the unproven belief that TF-ers are often fast-tracked in to management roles.

BLOG WARS!!!!; or, T-Simps Smackdown Watch

Jonathon Hazell

‘T-Simps’, the nom de plume of a Cambridge undergraduate called Thomas Simpson, has a go at the paper Felix Nugee and I wrote and presented the day before yesterday on what we call ‘helicopter money’. I encourage you to skim over it if you haven’t already, before returning to luxuriate participate in its demolition discussion. Snark aside, it’s a great piece, and raises a lot of salient and controversial points. Highly recommended reading, and no doubt the start of an excellent blog. It’s just quite nice to have some conflict in the Cambridge blogosphere – or indeed to have a Cambridge blogosphere at all.

Without further ado, let’s take a look at Thomas’s arguments.  From the top ….

(1) Negative interest rates can be a thing

Well, duh. Central banks can cut interest rates to negative infinity and beyond if they so choose. They have not and never will, because it would be pointless. This is because when interest rates are negative on any given security (say the bank balances which UK banks hold with the Bank of England), there’s almost no reason not to just withdraw its value into physical cash, which has a strictly better interest rate of zero. I say almost because there are costs to storing all that cash – think Fort Knox, bank vaults etc. Thomas rightly points out that Sweden has introduced negative interest rates. And the ECB might do the same. But no one – or almost no one – seems to think this would be more than marginally effective. Because once the cost of holding bank balances with the BoE at negative interest rates exceeds the costs of holding all that money in physical form, banks will withdraw it all. It’s that simple. In fact this is why when we are at the zero lower bound we call it a ‘liquidity trap’. Central banks are trapped by the fact that they can provide unlimited liquidity (i.e. money) without changing interest rates at all. Outsourcing to Barry Ritholtz on Sweden:

“Governor Stefan Ingves was quick to point out that benefits from negative interest rates did not seem to be observable. He did not get into the costs associated with negative interest rates, but he did affirm that it was unlikely that Sweden would use them again.”

The zero lower bound on interest rates is most definitely a problem.[1] And this is why we need to reform. Or as Gotham City’s Caped Crusader once put it:


2)        It’s not about the state of financial intermediaries. It’s about expectations of future nominal income

I’m confused about this point on a number of levels. Firstly, we have absolutely no problem with nominal GDP level targeting (NGDPLT). In fact, I think (and Felix agrees) that NGDPLT would be a wonderful initiative. It is, however, a different argument – one that we’d fully support, but well outside the scope of our proposal (I seem to remember Felix making this point in our seminar). Most of the people who support NGDPLT (e.g. Michael WoodfordDavid Beckworth [2]) are very keen on seeing the two policies as complements, as are we.

But that notwithstanding, it is simply not true to claim that you could fix a broken financial system solely by stabilising future expectations of nominal income, which is I think the main line of Thomas’ and others’ arguments. It requires a world that is fundamentally at odds with what happened during the financial crisis. It is a world in which banks were willing to borrow from other banks, because they were not afraid that whatever collateral they received might be worthless in a few days time. It is a world in which banks were willing to lend to other banks because they were not worried about the prospect of them going bust. It is a world in which hot money didn’t take flight from the banking system and leave them without funds to lend at all. It is not the world we’ve been living in for the last five years. Instead, it is a world with a financial system that chose to stop lending because of future expectations, and was not forced to stop lending because of current financial panic. When Northern Rock had depositors queuing outside its cash machines in 2007, when RBS was scrabbling around to fund itself after buying the basket case that was ABN Amro in 2008, whatever the Bank of England might have done to the future economy was irrelevant – they simply did not have the balance sheets in place to lend out to the real economy. And so Felix and I remain sure that the financial system was sufficiently bust over the crisis that, we need other ways to target the real economy, which don’t rely on merely stabilising nominal income expectations and trusting to the health of the banking sector.

3)        Can QE or forward guidance cause bubbles?

Admittedly I’m not entirely convinced of this point myself, and lots of evidence points against the UK currently being in a bubble. But I’m certainly concerned about the possibility – and who wouldn’t be? We know from the last five years that the small tail risk chance of a bubble is worth weighing, because the consequences are so catastrophic. But Thomas claims to be confused about it, so I’ll try and point him in the right direction. The gist of the battleground here is this: when QE and forward guidance happens, what happens to long-term interest rates? And if long-term interest rates fall, why would that cause a bubble?

This next bit is quite tedious, but it’s Thomas’ argument as best I can make it. There’s supposedly an effect that says interest rates should go down. Some people call this the ‘liquidity effect’. The idea here is that QE or forward guidance lowers the path of future short-term interest rates, and this makes long rates fall. There are also a couple of effects that putatively act in the opposite direction. These are called the ‘inflation’ and ‘income’ effects. These reflect the fact that in the long run, if the economy picks up because of the stimulus, inflation and growth should rise, so the central bank will eventually raise interest rates.

Now I have no idea which of these effects dominate. I have no idea whether any of them are correct. They could both all be powerful or all be weak. We could have totally the wrong model of the economy and financial markets and how monetary policy works (and the evidence seems to be pretty clear that we do). This Scott Sumner bloke who Thomas brings up doesn’t have a clue either. And armchair theorising won’t get you there. You have to actually look at interest rates to figure out what’s going on. I took the precaution of actually doing this:


So the bulk of QE took place in the October 2011 to July 2012 phase – right next to that nice big cliff in the middle. We had a bit in November 2009, just before the slightly smaller valley to the left. The rest took place in the flattish 2009-2010 phase. But you might say that just eyeballing interest rates isn’t particularly persuasive. Fortunately some very smart people have already done much more careful analysis. Paul Tucker, BoE deputy governor is probably the best the UK has on this front. And he like others finds that unconventional monetary policy has acted to lower long-term interest rates. Which, you know, accords well with just about every piece of serious econometric work on the subject.

So why is this a problem? The issue is that when long term interest rates fall, financial intermediaries who need to make some kind of fixed return on their investments (think pension funds who need to provide steady returns to people drawing down pensions) take on riskier propositions that pay a higher yield. In the parlance, they ‘reach for yield’. And this could create problems further down the road when either (a) these risks all blow up in their face or (b) they all unwind the risky investments when interest rates normalise. In fact there’s a very good case to be made (and Jeremy Stein of the Fed makes it) that this was in large part a reason behind the erratic movements in interest rates over the summer. Now I hasten to add that we don’t have a very good idea of whether or not this ‘reach for yield’ thing matters at all. But one lesson we’ve learnt about finance over the last five years is that trying to poke institutions into overleveraging themselves is something that we should do our best to avoid if we can.

For the record, Thomas’ point about targeting foreign currency is an excellent one. But I digress ….

3) Income inequality

Not even sure how to argue this point. We have a choice between (a) a way of stimulating the recovery by raising the incomes of the few and hoping they trickle down; or (b) raising the incomes of the many. The latter is preferable on this basis. Period. Sure, QE may have raised asset prices because it raises current and future profits. But this is basically the point Felix and I made – poor people don’t own assets so they don’t benefit from this.

Whew. Almost there. I need to sleep.

4) Will helicopter money work at all?

This is a very important point – and one we communicated very badly in the paper and also when we presented it, as Thomas rightly points out. The point made is this; we hope that by design helicopter money would function like a temporary tax cut. So what if consumers simply saved the money or paid down debts, instead of using it to boost spending? We have two answers to that. Firstly, recoveries after financial crises are generally slow because consumers are paying down debts – or ‘repairing balance sheets’ in the jargon. One of the better aspects of helicopter money is that if consumers chose to use it to get rid of these debts, it would accelerate the process of balance sheet repair, and speed the UK along the path towards recovery. That aside, though, if consumers are busy saving or paying down debts with their helicopter-dropped money, the idiot-proof solution is to simply give them even more money. Which is the crux of our explanation.

I guess that’s pretty much everything. A series of very important criticisms from Thomas, but ones which Felix and I think our policy is broadly robust to. People who have made it this far in either of our posts may have also realised that the debate is something of a proxy argument over what Thomas calls ‘market monetarism’. I happen to take a much glummer view of that particular school of thought than he does (having once been a devout myself), but this is an argument for future blogging wars.

[1] There are some racier proposals to introduce negative interest rates on money itself, especially by Miles Kimball and Willem Buiter. This is a different thing entirely.

[2] Respectively, the man famous for working out all of the interesting ideas of the obscure school of thought known as ‘market monetarism’ about a decade before it existed; and the most financially literate and arguably most incisive of the aforementioned market monetarists.