There has been significant progress in recent years in getting kids into school, but what’s the point if the teachers don’t show up for work? In general, the poorer the country, the higher the level of absenteeism. The explanations are both obvious (wages are so low, teachers need to look for second jobs, or funnel their students into private tuition) and less apparent (demoralized teachers with no money for textbooks and huge class sizes; weak governance means no-one holds absentees to account, and good attendance is not rewarded).
The first priority should be to pay teachers properly, invest in education and build a real public service ethos, as has been achieved in countries such as Sri Lanka. But what other policies could contribute to reducing teacher absence?
A new paper from the Poverty Action Lab at MIT surveys some of the attempts to measure the effectiveness of efforts to reduce absenteeism, and comes up with some interesting success stories. Cautionary note: because PAL likes to conduct randomised control trials, it gravitated towards non-state providers willing to experiment with introducing new techniques in some schools and not others, however, the lessons might be useful for reforming state services too.
One way is to improve the accountability of teachers to their local communities. An ICS Africa programme in Kenya, the Extra Teacher Programme, tested a range of different approaches to improving education, including hiring locally accountable teachers, tracking students by ability level, and training the local school committees that oversaw these teachers. Locally accountable teachers had much lower absence rates than centrally hired teachers and pupils’ test scores improved.
Unfortunately, the Extra Teacher programme also paid the extra teachers a fraction of the government salary, (something the MIT researchers seem unhealthily delighted by), but the overarching point is that you can raise teacher attendance through local accountability. This is born out by the Citizen Report Card programme in Uganda, which informed communities about the quality of services (in this case health care services) and facilitated a “shared action plan” to monitor providers, including drawing up a “community contract” between local communities and health providers. The result was better quality health care, attributed to increased provider effort. Although provider absence was not the focus of the programme, the programme resulted in a 10 percentage point improvement in attendance.
More draconian methods also seem to work. The most successful project reported in terms of overall impact in reducing absenteeism took place in rural Udaipur in India, where Seva Mandir, an Indian NGO, runs informal 
schools to help students not reached by ordinary government schools. Each school has only one teacher who instructs about 20 students in basic Hindi and maths. Similar to other schools around the world, teacher absenteeism was high: the teacher absence rate was 44 percent. Because these were NGO schools and teachers, it was an ideal setting to test how teachers responded to incentives. Each teacher in the programme was given a camera with a tamper proof date and time stamp and was instructed to take a picture with students at the beginning and end of each school day. Teachers were paid for the number of days that they attended as recorded by the cameras, giving them a clear incentive to attend school. To test the effectiveness of the programme, Seva Mandir randomly assigned half of the teachers to the camera programme, while the rest were supervised and paid the normal way as a control group. Unannounced, random checks measured the true attendance of each group.
Ordinarily, teachers were paid a salary of Rs. 1,000 (about $22) per month, for 21 days of teaching. In the camera schools, each teacher was guaranteed a base pay of Rs. 500. Teachers were rewarded with Rs. 50 for each valid day taught, so if they turned up every day, they could significantly increase their admittedly paltry wage.
The cameras worked. Attendance increased from 58 percent in the control group to 79 percent in the group with cameras. Overall, this translates into 34 more days of instruction per student per year. Attendance increased for teachers with both relatively high and low attendance records. Teachers liked the programme because it gave them some control over their own income. Some initially resisted its inflexibility – invalid pictures meant no pay even if a teacher was present. But teachers grew to better understand the programme and technical problems became rare. Because the cameras were so successful, Seva Mandir has continued to use them long after the experiment has ended. From October 2006 to September 2007, attendance was 15 percentage points higher in classrooms with cameras, compared to those without cameras; attendance was still as high as before in the school with cameras, but it had some what improved in the other schools.
Not surprisingly, students in the camera schools learned more. They were 62 per cent more likely to be admitted to regular government schools. Seven percent more girls were able to take a test that required being able to write.
Finally, teachers respond to their pupils. A Girls Scholarship programme in Kenya, also run by ICS Africa, offered scholarships to sixth-grade girls who scored highest on tests, and showed an unintended (though positive) impact of incentives. The girls eligible for the scholarship showed academic improvement—as did girls who were unlikely to win and boys, who could not win. Teachers— who received no incentives themselves from the programme—responded to students’ increased motivation to learn and were absent 4.8 percentage points less than teachers in control schools. In other words, teachers responded to students’ increased motivation to learn.
At this point, the PAL researchers start to show the limits of the RCT number-crunching mindset and start to resemble that Monty Python sketch of the banker struggling to understand why someone should collect money for charity: ‘Is an appreciated or contented teacher less likely to be absent? If so, a programme carefully designed around a teacher’s situation could be successful without being financially expensive. It is not yet known exactly how teachers should be encouraged to want to teach. Researchers in J-PAL’s network are continuing to experiment with making teaching rewarding.’ Good luck with that, guys.