When Less is More: Experimental Evidence on Information Delivery During India's Demonetization

Abstract

How should policymakers disseminate information: by broadcasting it widely (e.g., via mass media), or letting word spread from a small number of initially informed "seed" individuals? While conventional wisdom suggests delivering information more widely is better, we show theoretically and experimentally that this may not hold when people need to ask questions to fully comprehend the information they were given. In a field experiment during the chaotic 2016 Indian demonetization, we varied how information about demonetization's official rules was delivered to villages on two dimensions: how many were initially informed (broadcasting versus seeding) and whether the identity of the initially informed was publicly disclosed (common knowledge). The quality of information aggregation is measured in three ways: the volume of conversations about demonetization, the level of knowledge about demonetization rules, and choice quality in a strongly incentivized decision dependent on understanding the rules. Our results are consistent with four predictions of a model in which people need others' help to make the best use of announced information, but worry about signaling inability or unwillingness to correctly process the information they have access to. First, if who is informed is not publicized, broadcasting improves all three outcomes relative to seeding. Second, under seeding, publicizing who is informed improves all three outcomes. Third, when broadcasting, publicizing who is informed hurts along all three dimensions. Finally, when who is informed is made public, telling more individuals (broadcasting relative to seeding) is worse along all three dimensions.

Messages on COVID-19 Prevention in India Increased Symptoms Reporting and Adherence to Preventive Behaviors Among 25 Million Recipients with Similar Effects on Non-recipient Members of Their Communities

“…This interpretation explains both the large drop in conversations for the non-Jio sample, who also report far fewer inperson interactions, and the small drop in conversations in the Jio sample, for whom there is no drop in in-person interactions. The finding that mass messaging doesn't necessarily spark conversations on a complicated topic is also consistent with other recent empirical work in India (Banerjee et al, 2019). Overall it does not appear likely that it is through conversations about COVID-19 that the intervention led to diffusion: it seems more likely to have come from direct observations and imitation of behavior.…”

When Celebrities Speak: A Nationwide Twitter Experiment Promoting Vaccination In Indonesia

“…Before proceding to our empirical analysis, we begin by discussing a simple framework to think through how individuals make the decision to pass on information. The framework is standard, developed in Chandrasekhar, Golub, and Yang (2018) and also previously applied in Banerjee, Breza, Chandrasekhar, and Golub (2018b). In our framework, individuals pass on information for two reasons.…”

Can Network Theory-based Targeting Increase Technology Adoption?

“…6 For example, promising results in Banerjee et al (2018) imply that households know who is central in their village, and this type of information may be easily elicited from a random sample of people. Kim et al (2015) use a related elicitation mechanism based on friends-of-friends, and conduct an experiment to distribute public health coupons in a sample of 32 villages.…”