Description:
The Government of Colombia imposes a variety of taxes that must be paid by individual wage earners, called in their entirety "social protection contributions". Since 2007 individual payments have been collected using an on-line mechanism. In order to improve compliance, the Government used a controlled field experiment in which various "pop-up messages" were sent to individuals when making their on-line payments, as behavioral "nudges". We examine the impact of these nudges on individual reporting behavior. We find mixed evidence that these messages increased compliance rates relative to a control group that received a so-called "neutral" message. However, we also demonstrate that the use as the control group of individuals receiving a so-called "neutral" message creates considerable bias; that is, the receipt of any message of any type clearly influences behavior. Instead, we show that the appropriate control group should be individuals who receive no message at all. When this control group is used, we find that self-employed individuals generally increase their contributions; individuals who are making declarations on behalf of all employees in their company are less likely to respond to messages in a systematic way.