Double for Nothing? Experimental Evidence on the Impact of an Unconditional Teacher Salary Increase on Student Performance in Indonesia
by Joppe de Ree, Karthik Muralidharan, Menno Pradhan, Halsey Rogers – #21806 (CH DEV ED LS PE)
Abstract:
How does a large unconditional increase in salary affect employee performance in the public sector? We present the first experimental evidence on this question to date in the context of a unique policy change in Indonesia that led to a permanent doubling of base teacher salaries. Using a large-scale randomized experiment across a representative sample of Indonesian schools that affected more than 3,000 teachers and 80,000 students, we find that the doubling of pay significantly improved teacher satisfaction with their income, reduced the incidence of teachers holding outside jobs, and reduced self-reported financial stress. Nevertheless, after two and three years, the doubling in pay led to no improvements in measures of teacher effort or student learning outcomes, suggesting that the salary increase was a transfer to teachers with no discernible impact on student outcomes. Thus, contrary to the predictions of various efficiency wage models of employee behavior (including gift-exchange, reciprocity, and reduced shirking), as well as those of a model where effort on pro-social tasks is a normal good with a positive income elasticity, we find that unconditional increases in salaries of incumbent teachers had no meaningful positive impact on student learning.
The long term effect of who becomes and stays a teacher is probably more important if you want academically capable teachers — probably even more so in places that, unlike Indonesia, do not have a large surplus of workers earning next to nothing.
How did they measure pupil outcomes? With standardised tests?
just how do you measure teacher productivity?
You can check out the paper and answer your questions from this link (which I think will be active for one week)
Basically 2-3 years is just not enough time to expect a change. And there are some weaknesses in the analysis. Here is a quote from the paper:
“Yet despite this improvement in teachers’ pay and satisfaction, teachers in treated schools did not score better on tests of teacher subject knowledge, and did not self-report any increase in measures of effort such as attendance, or the number of teaching hours. ‘
1. There’s no logical connection between increase pay and satisfaction and improvement in subject knowledge unless there were also advanced or refresher training courses provided or encouraged (possible if teachers no longer had to work two jobs and so had the time to do them). Even if this had started exactly at the same time as the pay increases, would an effect be seen so soon?
2. Why would there necessarily be an increase in effort, if teachers believed that they were now finally getting fairly paid for what they already did, after being underpaid for so long?
For these (and other) reasons, I would not expect a quick result from the current teachers, but a future improvement in the quality of people attracted to teaching.
An impressive study. They got to randomise over all the schools in Indonesia and essentially gave teachers a few years advantage over other teachers in when they got the higher pay.
For sure, you’d expect the main effect to be selection and improvements in the prior education of the teachers (since the prize is worth more, you get better people going into i and more effort to learn to get the prize), but still: to find an effect of zero is surprising.
A weakness of the study is that the intervention is not really to double the pay of some teachers whilst keeping the pay of others constant: all the teachers eventually got the pay increase, but only some got it earlier. This means that the effect on the incentives to remain in the education system (ie the importance of not losing the job) is similar for the teachers who got the increase earlier and later: those getting the pay increase earlier stood to lose a current job if they did poorly, whilst those getting the pay increase later stood to lose a future pay increase if they did poorly. So the authors’ assertion that their model runs counter to efficiency wage models doesnt quite have to hold. They will probably be told to weaken those assertions by the referees. Still, a great study. Well done to the authors.