We study how text messages incorporating behavioral insights can be used as a tool to affect civil servant performance when state capacity is weak. By experimentally varying the content of a messaging campaign targeted to civil servants implementing a school maintenance program in Peru, we test the effectiveness of reminders and treatments making salient either monitoring, social norms, the possibility of public disclosure of noncompliance, or audit risk. All messaging treatments improve compliance by similar magnitudes, increasing the probability of submitting a key expense report by an average of 3.9 percentage points over a base of 74%. The inability of this large-scale experiment to detect differential impacts by treatment arm is consistent with timely reminders being the main driver of increased compliance. We explore generalizability across time and populations in two supplemental experiments, confirming the promise of such campaigns to improve civil servant performance when the state lacks enforcement capacity.
Nota bibliográficaFunding Information:
We thank Ernesto Dal Bo, Kitt Carpenter, Claudio Ferraz, Fred Finan, Alvaro Forteza, Federico Gutierrez, Daniel Hicks, Rachid Laajaj, Mounu Prem, Santiago Saavedra, Juan Vargas, Andrew Zeitlin, and seminar participants at Universidad del Pacífico, Universidad del Rosario, Vanderbilt University and participants at the Advances with Field Experiments, Association for Public Policy Analysis and Management, Midwest International Economic Development Conference, North East Universities Development Consortium Conference, Pacific Conference for Development Economics, Peruvian Economic Association, Southern Economic Association, and RIDGE/LACEA-PEG conferences for insightful comments. Field implementation would not have been possible without the support of the MineduLAB and PRONIED teams at the Ministry of Education and the CUNA MAS team at the Ministry of Social Inclusion. We especially thank Jorge Mesinas, Fabiola Caceres, Luis Baiocchi, Dennis Velasco, Rosie Fontinier, Milagros Barrantes, Jorge Fernandez, Karina Tecse, Marcos Bravo, Liz Espinoza, Susan Fiestas, and Martin Blas. Juliana Aragon, Maria Paula Medina, and Cesar Huaroto provided excellent research assistance. Maldonado acknowledges financial support from Colombia Cientifica – Alianza EFI 60185, contract FP44842-220-2018, funded by The World Bank through the Scientific Ecosystems, managed by the Colombian Ministry of Science, Technology, and Innovation (MINCIENCIAS). IRB approval was secured with Vanderbilt University (Study 151334). The experiments were registered with the AEA RCT registry under ID AEARCTR-0003722.
© 2022 Elsevier B.V.