The issue of teacher pay is controversial worldwide. Some argue that higher pay is necessary if we are to increase the status of teachers, and thus recruit, retain and motivate more high-ability individuals in the profession. However, others claim that higher pay across the board is not an efficient way to improve outcomes. First, higher salaries will not change the incentives to raise outcomes efficiently, which traditionally have been weak. Second, it has proven very difficult to separate who will become a good teacher based on their observable characteristics. And if it is difficult to separate good teachers from bad ones at the selection stage, we may not necessarily get better teachers for more money. In a new IFS Working Paper (W14/03), 'Estimating the Effect of Teacher Pay on Pupil Attainment using Boundary Discontinuities', economists Ellen Greaves and Luke Sibieta analyse the issue in the English context. Since national pay scales have been a defining feature of the teacher labour market for some time, salaries are basically the same all over the country. The exception is that there are three London weightings. This means that there are three different boundaries where the salaries for equivalent teachers vary depending on which side of the boundary the schools are situated. The authors focus on the boundary between ‘fringe London’ and the rest of England and Wales. By using a so-called ‘regression-discontinuity design’, they are able to compare like for like. The idea is that teachers in schools close to the fringe London boundary are likely to live in similar neighbourhoods and be at the same point in the pay scale, but still get paid differently simply because they work at schools on different sides of the boundary. This means that the authors can separate causation from correlation. The authors find no impact whatsoever on KS2 test scores in mathematics and English where salaries are about 5% (or £1,000) higher. The authors therefore argue that there is little evidence that higher salaries induce higher-quality teachers to sort into the communities in which they are paid more. This could be either because teachers are not sensitive to these salary differences, or because schools might find it difficult to separate high-quality teachers from low-quality teachers. In conclusion, then, authors argue that merely offering higher salaries across-the-board to teachers is unlikely to be an effective strategy for raising pupil performance, and that performance-related pay holds more promise. These are interesting results that are not entirely consistent with previous research on the effects of national pay scales by Caroline Propper and Jack Britton from 2012, which finds that if teacher salaries are low relative to salaries in other sectors in the area, pupils’ test scores will suffer. The authors reason that if wages in other sectors are higher than in teaching, it is less likely that high-ability individuals will go into teaching – which explains why pupil performance is lower in these contexts. The differences might depend on various issues. For example, it cannot be ruled out that stronger methodology explains Greaves and Sibieta’s results. Another difference is that these authors analyse primary school teachers, whereas Propper and Britton analyse secondary-school teachers who might respond differently to pay differences. Most importantly, however, the papers may very well identify different effects. Since they focus on the impact of differences between teacher wages and wages in other sectors, Propper and Britton are focusing on occupational choices – with more high-ability individuals going into teaching in areas where alternative occupations do not pay more – whereas Greaves and Sibieta’s results probably can tell us more about sorting across communities of existing teachers. (Of course, it may also simply be that the 5% higher salary is not sufficient to induce any sorting.) However, it is also notable that Propper and Britton do not find any positive effects of teachers being paid more relative to other sectors. This indicates that teacher salaries could actually be lowered in some areas without harming pupil achievement. In other words, it makes more sense to differentiate teacher wages, and pay some more than others, rather than increase them across the board. Similarly, Greaves and Sibieta’s research indicates that the way to increase teacher quality is not merely to raise salaries for all teachers. Instead, it is probably better to enable larger salary differentials – which means entirely abolishing national pay scales – so that we can reward productive teachers more than we reward non-productive ones. Gabriel H. Sahlgren This comment piece first appeared as the Editor's Pick in the third trial issue of the CMRE Research Digest (June 2014). The piece reviews an IFS Working Paper (W14/03), 'Estimating the Effect of Teacher Pay on Pupil Attainment using Boundary Discontinuities', by economists Ellen Greaves and Luke Sibieta, which may be downloaded here. You can download free copies of back issues of the CMRE Monthly Research Digest here. to subscribe.