To alleviate poverty, is the underlying goal of all conditional cash transfer (CCT) programmes. This relatively new form of social protection provides an immediate additional income to the poor in exchange for fulfilling specific behavioral conditions, that are often focused on improving human capital through access to health and education (Doetinchem et al. 1). Linden and Shastry examine the efficacy of one such program – the conditional distribution of grain, part of the National Programme of Nutritional Support to Primary Education (NP – NSPE) introduced by the Government of India in 1995, by observing the behavior of teachers implementing the programme in Mumbai, India (129). The primary objective of the programme is “to boost universalization of primary education by increasing enrolment, retention and attendance and simultaneously impacting on nutrition of student in primary classes” (National Programme 1). This explains that the NSPE programme has dual objectives – one, to improve student attendance and two, to improve nutrition among students. In this memo, I argue that the teachers manipulating students’ attendance do not qualify as corrupt bureaucrats or agents, as they make no private gains. I then use a simple workhorse model to summarize the governance problem that inflicts this decentralized program implementation. Finally, I conclude with policy recommendations to constrain the behavior of teachers and improve the efficacy of the policy intervention. 

Banerjee et al. define corruption as “the breaking of a rule by a bureaucrat (or an elected official) for private gain” (6). Interpersonal interviews with teachers suspected of manipulating select student attendance records reveal that they “were responding to the extreme poverty in which their students lived” (Linden and Shastry 129). Furthermore, the findings suggest that “there were no reports of teachers demanding kickbacks” (Linden and Shastry 129). This establishes that the underlying intention of teachers was compassionate in nature and in adherence with meeting the nutrition goals of the NSPE programme. The intention was not to illicit bribes for private gain. 

The key governance question in the implementation of CCT programs remains – Does decentralizing the implementation to local agents help in targeting the poor? Often, CCT programs decentralize “the task of allocating transfers among potential beneficiaries to local agents” with the aim “to improve targeting and reduce the cost of gathering necessary information” (Linden and Shastry 128). To closely examine this case of decentralization, and focus on the incentives the teachers face, I follow a simple workhorse model to understand the teachers’ decision to manipulate students’ attendance records. The following variables can be identified – the local teachers receive a wage from the government; the probability that the teachers get caught for manipulation and receive an outside wage (v) is p (Ravanilla 5). However, as the gains are not private, we cannot account for a bribe, and given that the manipulation takes place on compassionate grounds, the likelihood of a dishonesty cost is low. Instead, the teachers incur a cost in terms of the time (t) it takes to manipulate the attendance records at the end of the month. In this scenario, if the teachers are undetected, they receive the wage minus the time cost (t). If the teachers are detected with probability p, and are fired, they receive an outside wage v (Ravanilla 5). We know that local agents “will be corrupt only when it pays to be corrupt” – when (1- p)[w – t] + pv > w; that is, when the returns to manipulating attendance net of the time cost is higher than the returns to earning the wage net of the opportunity cost (Ravanilla 6-7). 

To constrain the behavior of teachers, and to improve the efficacy of the program, I propose that attendance rosters be monitored by an external NGO on a weekly basis and this monitoring practice be transparently introduced to the school and the teachers. We know that monitoring reduces the propensity of teachers to manipulate and increases the probability of being caught. Hence, introducing such a measure will increase the cost in terms of time (t) – as it will take the teachers longer to manipulate the data on a weekly basis, and they will not be able to accurately estimate the students who are more likely to reach the 80% attendance rate at the end of the month. This will also increase the probability of them getting caught (p). Second, to reduce the general incentives for teachers to inflate attendance, I propose separating the performance indicators of teachers from those of the school. The performance of teachers should be evaluated against childrens’ learning outcomes and reputation of the teacher’s integrity. Such an incentive will help teachers focus on the quality of teaching and will deter them from engaging in manipulation practices that may affect their reputation. Third, it would help to change the behavioral condition for receiving grains from a singular condition of students’ school attendance rate to a combination of conditions that include learning outcomes, participation in extracurricular activities, attendance rate, classroom engagement as well as student nutrition records. This will make it difficult for teachers to manipulate any one record and will drastically increase the cost in terms of the time it takes to manipulate all records. Finally, the success of this program should not rely solely on improving the student attendance rate to above 80%, but instead should focus on improving students’ learning outcomes and their nutritional status, simultaneously. 

The full article with the works cited can be viewed below.