Teachers are central for learning, but if they are civil servants, their management and hiring can be affected by the political cycle. Using an administrative school-level panel data set across India, we show that teacher transfers and the hiring of new teachers increase significantly after State Assembly elections. The identification relies on the staggered and pre-determined timing of elections across states. The restructuring can be harmful; test scores are up to 0.15 standard deviations lower for children whose schooling coincides with the post-election phase. We conduct various checks to establish the link between the two findings.
Keywords: Teacher turnover, Political cycle, Education, India
JEL Codes: D73, H75, I21, J24
Acknowledgements: We would like to thank the staff at NUEPA for providing us with the DISE database and for helpful discussions; in particular Professor Arun Mehta, Naveen Bhatia and Professor Pranati Panda. We would like to thank Dr Suman Bhattacharjea from the ASER Centre for the ASER data and for many illuminating discussions. Earlier versions of this study have benefited from feedback from participants at the Indian Statistical Institute Annual Conference 2013, NEUDC 2014, Trinity College Dublin and the RES Conference 2015. We would like to thank Amrita Saha and Aditya Das for research assistance. The authors acknowledge financial support from the British Academy (R56B). We are fully responsible for any remaining errors.
It is well acknowledged that teachers are a key input in the educational production function (see e.g. Glewwe et al., 2014). There is also growing recognition of the importance of the quality of management of schools and other public services within Economics (see e.g. Bloom et al., 2015). The relevance of management can extend to school systems as a whole, and can also be influenced by external factors, such as the political process. While the role of political factors in the provision and management of education is recognised, there is limited rigorous quantitative evidence for developing countries, in particular with respect to teachers (see e.g. Kingdon et al., 2014 for a systematic review). In contexts where teachers are civil servants, political processes and changes can affect the management of personnel.
In this study, we demonstrate that the electoral cycle affects the transfer and recruitment of Indian public primary school teachers. In addition, we show that the electoral cycle affects learning. Further analysis indicates that the two effects are connected, implying that the political cycles in the management of teachers can have performance implications. Our data source for teachers is an India wide administrative school records database (District Information System for Education, DISE). For the analysis on learning we rely on child-level data from the Annual Status of Education Report (ASER), given that the DISE does not include reliable indicators on learning.
Regular primary school teachers in India are civil servants, generally on permanent contracts. They are hired by Indian states and core recruitment decisions are made at this level (see e.g. Ramachandran et al., 2008 and Sharma and Ramachandran, 2009). Teachers can also be directly involved in the political process for instance due to their role in staffing election booths. (1)
Our study focuses on State Assembly Elections; the timing of which is pre-determined and staggered across states, and can thus be considered exogenous. We find increases in teacher turnover, the number of teachers and new hires after the elections in relation to other years, but no evidence of electoral cycles in reported days spent on non-teaching assignments. The findings can be compatible with bottlenecks created by a rule banning transfers in the pre-election period, but also with increased administrative and political momentum of the incoming government.
The Election Commission of India's Model Code of Conduct imposes a ban on the transfers of all government employees, who are connected with election duties in the run-up to elections (from the announcement of the elections). It also bans the appointment and promotion in Government/Public undertakings during the period. (2) Its aim is to reduce the capacity of politicians to influence the electorate during elections. Singh (2012) provides a useful overview on its content and development, including how political parties perceive to be constrained by it. Anecdotal evidence suggests that the ban is relevant for how teachers are managed. According to Jha et al. (2008), over the 2002-2005 period, the "Imposition of model code of conduct for assembly elections had also delayed teacher recruitment in Bihar and Haryana" (p.332).
To the authors' knowledge, the effects of electoral cycles on teacher recruitment have not been studied rigorously previously. A few studies have been conducted on bureaucrats in India. In an unpublished study, Iyer and Mani (2007) find evidence of an increase in bureaucrat transfers around election years, influenced by the incoming government. Other studies on Indian bureaucrats include Iyer and Mani (2012), which finds that political interests play a role in the transfers of Indian bureaucrats and a study by Bertrand et al. (2015) on the determinants of the effectiveness of Indian bureaucrats. On the other hand, there is a large literature on electoral cycles in public sector resources (see e.g. Nordhaus, 1975, Khemani, 2004).
After establishing an electoral cycle in the management of teachers, we proceed to test whether the electoral cycle also affects learning. We find that fourth grade pupils perform better in Reading and Mathematics tests if their schooling does not coincide with the post-election phase that is characterised by higher teacher turnover and new hiring. The role of the electoral cycle in learning itself has been little researched.
As evidence of a connection between the effects of the electoral cycle on learning and on the reorganisation of teachers, we show that 1) the timing of the two effects coincides and 2) the negative effects of the post-election period on learning are stronger in districts (3) that have a higher degree of excess teacher turnover in the post-election period. Further, we estimate models to rule out some potential competing explanations for the learning effects. Over the period studied, there is little connection between the electoral cycle and various types of reported crime or communal unrest, which might intensify during election periods and disrupt schooling. Furthermore, the electoral cycle has either no effect on learning (Reading), or a much milder effect (Mathematics) on pupils in private schools, indicating that the causes of the cycles lie within the public sector. Finally, we find no connection between the electoral cycle and pupil composition, when the latter is measured as the tendency of fourth grade pupils to be enrolled in a private school.
We therefore propose that our findings on learning are compatible with our findings on the teacher reorganisation process. This process incorporates the increased turnover of teachers, but the entire process cannot be captured with a single indicator in our data. (4) It is likely that such reorganisation can be disruptive and reduce effective teaching time, or the quality of teaching.
At a broader level, the results on the electoral cycles in teachers and learning can be considered symptomatic of impairments in the management of these services. Given that learning in private schools is largely unaffected by the electoral cycles, our findings also provide a new angle to the literature on the relative effectiveness of private versus public schooling (see e.g. Muralidharan and Sundararaman, 2015 and Singh, 2015).
We begin with a description of the data set used and the summary statistics (Section 2). The analysis on teachers, including a discussion of the identification and estimation, is presented in Section 3. Section 4 focuses on the effects of the electoral cycle on learning and Section 5 concludes.
Our data source on teachers is an administrative school records database, the District Information System for Education (DISE), managed by the National University of Educational Planning and Administration (NUEPA), Delhi. From the year 2005 onwards, the database has full, or nearly full coverage of government administered primary schools in India. The data are reported on an annual basis and form a panel dataset of schools. The database includes a rich set of variables on school resources, management and pupils. For each school, it also includes a teacher level file with information on each teacher and key characteristics. These include name, age, caste, gender, date of birth, starting point of career as a teacher and indicators on educational qualifications. (5) There are no other comparable India-wide, annual data sources on schools. In most estimations we use a panel data set of schools for seven years between 2005-2011.
In terms of timing, the year 2005 refers to the academic year 2005-06 (6), and the data are collected in the Autumn of 2005, and similarly for the other years as well. To focus on a unified group of teachers, the sample is restricted to lower primary schools, which in most states spans grades 1-5. (7) Most of our analysis focuses on public sector schools, with a few robustness checks for private schools. For schools that include both lower and upper primary schools, the variables in this study relate only to lower primary students, and teachers who teach such students. The ASER survey, which is the source for the learning data is described in more detail in Section 4.
Our key outcomes of interest are the numbers of teachers, whether teachers leave a school in a particular year (transfer) and the numbers of new teachers hired per year in a district. We also analyse the effects on the number of days...