Resumen

BEVILACQUA, Solon  y  MIRANDA, Edivan do Socorro Fonseca de. Default in Paying Taxes: Intervention with Nudges and Social Standards. Gest. polít. pública [online]. 2024, vol.33, n.1, pp.141-162.  Epub 12-Ago-2024. ISSN 1405-1079.  https://doi.org/10.60583/gypp.v33i1.8191.

In this research, a group of charging messages for the payment of taxes was investigated. Altogether 12 variations of billing messages (social norms, simplification, disclosure, previous engagement, reminders and previous choices) were evaluated, and their effectiveness was tested. The messages were transmitted to defaulting microentrepreneurs in four Brazilian states. From a database containing information about defaulting micro entrepreneurs, 250 thousand text messages were sent making charges. The data were obtained from the Secretaria Especial da Micro e Pequena Empresa (Sempe). Tests were used to analyse the difference between means and Logistic Regression was used in sequence. The Random Forest, Logistic Regression and Naïve Bayes widgets were used to indicate the robustness of the Machine Learning predictive model. The research findings indicated that the formats “simplification”, “previous choices” and “alert”, employees, did not have an effect in combating default. However, when aligned with social norms, messages in the form of “past options” and “reminders” increase the payment of debts. The widgets used indicated an excellent fit to the machine learning model. The Random Forest tool attested with superiority that the model is robust and suitable for the predictive function. The results of the research provide a contribution to public policies when they present an effective action to reduce defaults in the payment of taxes. The use of messages in the social norm format can be adapted to other situations, constituting a suggestion for future research.

Palabras llave : default; nudges; social norms; entrepreneur; machine learning.

        · resumen en Español     · texto en Español     · Español ( pdf )