The effect of transparency initiatives depends on how citizens interpret and process the information they receive. We conceptualize these initiatives as signals and test this idea in a survey experiment conducted in Buenos Aires, Argentina. The signaling model incorporates multiple receivers through two dimensions that capture the heterogeneity of respondents: “belief” types differ in their priors (baseline trust in government) and “attention” types differ in the information to which they pay attention to (familiarity with transparency initiative). With multiple receivers, treatment effects depend crucially on respondents’ initial beliefs and information: Bayesian updating implies that receivers with no trust or full trust in government should not respond to the treatments, as additional information does not shift extreme priors; the same holds for receivers already familiar with the transparency initiative, whose beliefs should have adjusted prior to the experiment. We find that the initiative works as a partially informative signal that improves the government’s reputation. The effect is entirely driven by respondents who are vaguely familiar with the initiative. Those already familiar appear to have been impacted before our experiment was carried out. Unexpectedly, those unfamiliar remain unaffected; this latter finding suggests that they do not know because they do not care.

La presentación de la siguiente conferencia se realiza en idioma español.