Concerns about declining student achievements have become central in education debates, with teacher quality drawing attention as a key driver. This column examines how wages affect the quality of the teaching workforce by studying how uniform salary increases alter the composition of teachers who remain in the profession. Exploiting a French teacher salary reform, the author shows that uniformly increasing salaries improves workforce quality by disproportionately retaining better teachers.
Declining student achievement has become central to education policy debates across OECD countries. Growing concerns point to a decline in teacher quality, given teachers’ central role in shaping student outcomes (Chetty et al. 2014a, 2014b). This has prompted governments to re-examine what they can do to keep their most effective teachers in the profession. Teacher pay has therefore become a focus of policy debate. In most OECD countries, salaries have stagnated in real terms while outside opportunities have grown more attractive. In France, for instance, the wage gap between teaching and the occupations that teachers commonly transition to after exiting teaching has doubled over the past decade (Figure 1). A growing literature shows that higher wages improve retention on average (e.g. Falch 2011, Hendricks 2014), while performance-linked pay can retain the most effective teachers (e.g. Muralidharan and Sundararaman 2011). But performance pay is politically difficult to implement, and the key question regarding uniform raises remains unanswered: which teachers are more likely to stay when wages increase?
In my job market paper (Tartova 2025), I shed light on this question using quasi-experimental wage variation from a 2014 French reform that sharply increased salaries in highly disadvantaged schools and only modestly in slightly less disadvantaged ones.
I show that high-productivity middle-school teachers in either maths or French are 2.5 times more likely to be retained in response to wage increases than low-productivity teachers. Uniform raises therefore improve average teaching quality through selective retention.
A discrete-choice labour supply framework explains this pattern and allows me to quantify the relative cost-effectiveness of uniform and targeted bonuses for high-value-added teachers for achieving the same average quality gains. It also shows that rising outside wages produces disproportionately large losses in teacher quality. Together, these results highlight the importance of wage competitiveness for retaining effective teachers and inform broader ongoing debates about public-sector workforce quality.
Figure 1 The wage gap between outside-option and teaching wages is widening in France
Notes: The figure shows the widening of the average annual wage gap in France (in thousands of euros). The Average outside option wage is the weighted-average wage of non-teachers in occupations identified as outside options for teachers, using teacher exits from the profession. Weights account for both the average share of teachers exiting to a given occupation over the period 2009-2021 and the share of workers in an occupation in a given cell of age, sex, commuting zone and year. The Teacher wage varies per position on the pay scale and year. Both teaching wages and outside option wages are CPI-deflated to 2024 thousands of euros.
Source: Tartova (2025), based on data from Bases Relais (DEPP), BTS-Postes, and teacher wage data compiled by Lucas Chancel, 2009/2010-2020/2021.
The puzzle
Whether uniform salary increases improve the average quality of the workforce through retention depends on which teachers are at the margin of leaving. If these ‘marginal teachers’ are disproportionately more productive, uniform raises would increase average workforce quality. If the reverse is true, they risk becoming a costly intervention with limited, or even negative effects on learning outcomes.
This is not a priori obvious. Teachers are more likely to be marginal if they have better outside options, and less likely to be marginal if they have stronger pro-social motivations. Because the distribution of these traits across high- and low-productivity teachers is unknown, theory does not predict which group should respond more to a wage increase.
For instance, if the average high-productivity teacher both earns more outside teaching and has stronger pro-social motives, it is unclear whether they lie closer to being indifferent between staying or leaving, relative to low-productivity ones.
High-productivity teachers respond more to wages
I exploit the 2014 French reform by comparing exit probabilities between teachers in highly disadvantaged schools and those in slightly less disadvantaged ones, using a difference-in-differences design based on teachers’ pre-reform school assignments.
The (relative) salary increase succeeded in retaining more teachers in the profession: exit probabilities fell by 27% more for teachers in highly disadvantaged schools relative to those in slightly less disadvantaged schools. This implies a fairly large exit elasticity: a 1% wage increase – about €350 per year based on the average wage just before the reform (in 2024 euros) – reduces the probability of exit for teachers receiving the bonus by as much as 8%.
Most importantly, using an unbiased measure of teacher value-added (Tartova 2023), namely, the causal impact of a teacher on their students’ test scores, I find much larger effects for high-productivity teachers (Figure 2). Among maths and French teachers with above-median value-added, exit rates decline by 65% more in disproportionately treated schools, and the effect persists over time. Among below-median teachers, exit rates fall by only 26.5%, on average. In other words, high-productivity teachers are 2.5 times more responsive to wages than low-productivity teachers. These differences are not driven by teacher characteristics or local labour market trends.
This result has important implications for policy: even when all teachers receive the same nominal increase, high-value-added teachers are disproportionately retained, raising the average quality of those who remain.
Figure 2 High-productivity teachers decrease their exit more in response to wages
a) High teacher value-added
b) Low teacher value-added
Notes: The figures plot the estimated year-by-year effects of the 2014 reform on teachers’ probability of leaving the profession. Effects are estimated using an event-study difference-in-differences design that compares teachers who were working in REP+ schools in 2013/14 (and therefore eligible for the larger wage increase) to teachers in REP schools (eligible for the smaller increase). The sample includes all Math and French teachers in these schools for whom teacher value-added (TVA) can be measured. Exit is defined as the last year a teacher appears in the administrative records. Each panel shows results under five specifications, ranging from simple comparisons with no controls to versions including teacher characteristics, school fixed effects, and local-labor-market-by-year fixed effects, where local labor market is either approximated by an academy (educational region) or commuting zone. All estimates control for teachers’ pre-reform characteristics (gender, qualification, experience, age, and subject taught), and standard errors are clustered by teacher and by school-year. Pre-reform differences are normalized to zero because the analysis focuses on teachers employed in 2013/14, before the reform was announced. Panel (a) shows results for teachers with above-median TVA (“high productivity”), and Panel (b) for those with below-median TVA (“low productivity”), where medians are computed within local labor market. Ninety-five per cent confidence intervals are shown.
Source: Tartova (2025), based on data from Bases Relais (DEPP), Scolarité (DEPP), Sysca (DEPP), 2013/2014-2019/2020.
Why are better teachers more responsive to wages?
To interpret these results, I develop a discrete-choice framework in which high- and low-productivity teachers choose each year between staying in teaching and leaving for an outside option. Utility from each option reflects wages, non-pecuniary preferences (e.g. pro-social motives), and idiosyncratic tastes. Because average exit rates are relatively low – as supported by the data – both groups generally prefer teaching.
The framework shows that high-productivity teachers are more responsive to wages if:
- Their average preference for outside options is closer to the indifference threshold between staying in teaching and leaving. This average would be closer to indifference the higher the wage gap is between teaching wages and outside wages, and farther from indifference the stronger pro-social motives are. It follows that for the average high-productivity teacher to be closer to indifference, it must be that they either have better outside options or are less pro-social. The former is consistent with findings that high-TVA teachers tend to have better outside wages once they exit (Chingos and West 2012).
- Their preferences are less dispersed, so a given wage increase moves a larger share of them from “about to leave” to “deciding to stay”.
A micro-foundation consistent with both conditions is that teachers with better outside options devote more attention to understanding those opportunities, which reduces uncertainty around their choices.
In my paper, I uncover empirical evidence which suggests that teachers with wider wage gaps – driven by their better outside option – are those more likely to be retained following a salary increase.
What do these results imply for policy?
Alternative teacher wage policies
The labour supply framework allows me to evaluate the effects of counterfactual wage policies for the change in quality implied by selective retention.
A uniform €1,800 wage increase raises aggregate teacher quality by 1.5% of a standard deviation over five years, driven by the heterogeneous exit elasticities. Targeting raises exclusively to high-value-added teachers yields the same quality improvement at just one-fourth of the fiscal cost, because this reduces exit only among high-productivity teachers.
Rising outside wages
By contrast, a €1,800 increase in outside wages lowers teacher quality by 5% of a standard deviation – more than triple the positive effect of raising teacher wages by the same amount. This asymmetry arises because most teachers lie far from indifference in steady state: teacher wage increases shift a thin margin (Figure 3, panel a), while higher outside wages move many more teachers toward the exit threshold (Figure 3, panel b).
Figure 3 Asymmetric effects of teacher versus outside wage increases on aggregate teaching quality
a) Increase in teacher wage
b) Increase in outside wage
Notes: The figure shows how changes in teacher wages and outside wages affect the share of teachers who are close to leaving the profession. The curves represent the distribution of teachers’ relative preferences for switching to an outside option versus staying in teaching, separately for high- and low-value-added teachers. Relative preferences depend on the wage gap between the outside option and teaching (Dw), pro-social motivation (q) and idiosyncratic preferences (e). The vertical line marks the point at which a teacher is indifferent between the two options. Teachers to the right of this line are those who prefer the outside option. Panel (a) illustrates that increasing teacher wages shifts both distributions slightly to the left, reducing the share of teachers near the exit margin. Because most teachers are already far from that threshold, the improvement in overall teacher quality is modest. Panel (b) shows the opposite experiment: an equivalent increase in outside wages. This shifts the distributions to the right and moves many more teachers—especially high-productivity teachers—toward the exit threshold. As a result, the decline in aggregate teacher quality is much larger than the gain produced by raising teacher wages by the same amount.
Source: Tartova (2025), underlying framework parameters are estimated using the exit elasticities to the wage for high- and low-productivity teachers, and an estimate of the 2013 wage gap obtained by teacher wages (Lucas Chancel data) and the average feasible outside option wage (BTS-Postes data).
Conclusion
As governments confront decreasing student achievement and teachers play a big part in their educational attainment, understanding who stays and who leaves when wages change is crucial.
In my job market paper (Tartova 2025), I highlight that uniform salary increases can raise teacher quality by disproportionately retaining high-productivity teachers, even though they are far less cost-effective than targeted wage increases. I also show that rising outside wages have disproportionately negative effects on average teacher quality. As wage gaps between teaching and outside occupations continue to grow in many countries, this is an important insight for policy makers that target average teaching quality.
Source : VOXeu
