Optimal Design of Mobile Monitoring
This project is a follow-up project of Improving Services Delivery: An Assessment of Absenteeism Data.
Teacher absenteeism remains a serious challenge in Uganda, with estimated rates of absence as high as 27 per cent. Evidence from other countries suggests that a monitoring scheme, combined with bonus payments, could reduce absenteeism, and improve education performance. However, it is unclear what form this scheme should take. Automated monitoring via cameras or punch cards is expensive and has been subject to sabotage (Banerjee et al. 2007). Local monitoring by head teachers is cheaper and has the potential added value of providing useful information to government for planning purposes. However there is risk of collusion, especially when financial rewards are attached (Chen et al. 2001).Monitoring by beneficiaries, in this case parents could be attractive, although it is an open question whether parents are able and willing to take on this role.
This paper reports on a study assessing the efficacy of alternative forms of local monitoring, both as a means to improve teacher attendance, and to collect good quality monitoring reports. The study is taking place in 180 rural government schools in six districts in Uganda. At the start of the study, 40 schools were randomly assigned to a control group. Of the remaining 140 schools, 90 were randomly allocated to one of four local monitoring designs, which are the focus of this note. These designs differ along two dimensions: (i) the identity of the monitor and (ii) the stakes attached to the report. In some schools, head teachers have been asked to undertake the monitoring and reporting, whereas in other schools, monitoring and reporting are the responsibility of parents on the school management committee. At the start of the study, all of the nominated monitors were trained to report teacher attendance via SMS on a mobile phone, using a platform developed and managed by Makarere School of Computing and Informatics Technology. The stakes attached to the report vary as follows. In some schools a monthly bonus of UShs 60,000 (about USD23, or 30 per cent of a monthly salary) is paid to any teacher marked as present in every ‘qualifying report’ that month, whereas in other schools there is no financial bonus attached and qualifying reports are simply collated and re-broadcast via SMS. For the head teacher’s monitoring report to qualify s/he must report every day, while parents need only report once per week. Head teachers thus face a stricter monitoring protocol than parents.
Combining these treatments we obtain four local monitoring designs which are assessed against two criteria: cost effectiveness in inducing higher teacher attendance, and quality of reporting, both in terms of the frequency and reliability of reports. The key findings are as follows:
- First, local monitoring improves teacher attendance but only when the head teacher is responsible for monitoring and there are financial incentives for teachers at stake. Under this scheme, teacher presence is 11 percentage points higher than in the control group. Importantly, this design is also the most cost-effective since bonus payments were no larger than when parents were responsible for monitoring.
- Second, although monitors report frequently they all understate teacher absenteeism and parents more so than head teachers. Head teachers report an average of 2.5 times per week and parents, encouraged to select a single reporting day, report on average once per week. The understatement of teacher absenteeism is substantial across all four schemes: the actual presence rate, as measured by independent surprise visits, is 14 percentage points lower than the reported presence rate. Parents generate significantly less reliable reports than head teachers. There is no evidence that reporting is less reliable when bonuses are paid.
Further tests suggest that report quality is wanting because all monitors have a tendency to falsely report absent teachers as present. This is exacerbated under the parent led schemes because parents opt to report on days when more teachers are present. The stricter reporting protocol under the head teacher led scheme discourages head teachers from doing the same.
In sum, local monitoring schemes can improve teacher presence at relatively low cost. However, report quality may be a challenge, with reports systematically understating teacher absenteeism.