DP8869 Can Institutions Be Reformed from Within? Evidence from a Randomized Experiment with the Rajasthan Police
|Author(s):||Abhijit Banerjee, Raghabendra Chattopadhyay, Esther Duflo, Daniel Keniston, Nina Singh|
|Publication Date:||March 2012|
|Keyword(s):||community participation, crime, India, institutional reform, police, police reform, randomized trial|
|JEL(s):||H11, H76, K42, O22|
|Programme Areas:||Development Economics|
|Link to this Page:||cepr.org/active/publications/discussion_papers/dp.php?dpno=8869|
Institutions in developing countries, particularly those inherited from the colonial period, are often thought to be subject to strong inertia. This study presents the results of a unique randomized trial testing whether these institutions can be reformed through incremental administrative change. The police department of the state of Rajasthan, India collaborated with researchers at US and Indian universities to design and implement four interventions to improve police performance and the public?s perception of the police in 162 police stations (covering over one-fifth of the State?s police stations and personnel): (1) placing community observers in police stations; (2) a freeze on transfers of police staff; (3) in-service training to update skills; and (4) weekly duty rotation with a guaranteed day off per week. These reforms were evaluated using data collected through two rounds of surveys including police interviews, decoy visits to police stations, and a large-scale public opinion and crime victimization survey--the first of its kind in India. The results illustrate that two of the reform interventions, the freeze on transfers and the training, improved police effectiveness and public and crime victims? satisfaction. The decoy visits also led to an improvement in police performance. The other reforms showed no robust effects. This may be due to constraints on local implementation: The three successful interventions did not require the sustained cooperation of the communities or the local authorities (the station heads) and they were robustly implemented throughout the project. In contrast, the two unsuccessful interventions, which required local implementation, were not systematically implemented.