Nudging at Scale : Evidence from a Government Text Messaging Campaign during School Shutdowns in Punjab, Pakistan

Text and voice messages have emerged as a low-cost and popular tool for nudging recipients to change behavior. This paper presents findings from a randomized controlled trial designed to evaluate the impact of an information campaign using text and voice messages implemented in Punjab, Pakistan during the COVID-19-induced school closures. This campaign sought to increase study time and provide academic support while schools were closed and to encourage reenrollment when they opened, to reduce the number of dropouts. The campaign targeted girls enrolled in grades 5 to 7. Messages were sent out by a government institution, and the campaign lasted from October 2020 until November 2021, when schools had permanently re-opened. Households were randomized across three treatment groups and a control group that did not receive any messages. The first treatment group received gender-specific messages that explicitly referenced daughters in their households, and the second treatment group received gender-neutral messages. A third group was cross randomized across the first two treatment arms and received academic support messages (practice math problems and solutions). The results show that the messages increased reenrollment by 6.0 percentage points approximately three months after the intervention finished. Gender neutral messages (+8.9 percentage points) showed larger effect size on enrolment than gender-specific messages (+ 4.3 percentage points), although the difference is not statistically significant. The message program also increased learning outcomes by 0.2 standard deviation for Urdu and 0.2 standard deviation for math. The paper finds a small positive effect on the intensive margin of remote learning and an (equivalent) small negative effect on the intensive margin of outside tutoring. In line with similar studies on pandemic remediation efforts, the paper finds no effect of the academic support intervention on learnin g. The findings suggest that increased school enrollment played a role in supporting the observed increase in learning outcomes.