Doctorate – Estimated tax amount based on Benford's law

Tipo de evento: 
Data e hora: 
14/01/2021 - 13:00 to 16:00


Archibald De Araújo Silva

Doctorate – Estimated tax amount based on Benford's law

Advisor: Profa. Dra. Maria Aparecida Gouvea

Comission: Profs. Drs. José Roberto Ferreira Savóia, Eric Bacconi Gonçalves and Leandro Campi Prearo

Link YouTube: 


There are three types of tax within the State's competence: Tax on Transactions Related to the Circulation of Goods and Provision of Interstate and Intercity Transport and Communication Services (ICMS), Property Tax on Motor Vehicles (IPVA) and Transmission Tax Cause Mortis and Donation (ITCMD). According to the tax revenue report of the State of São Paulo, in 2019, ICMS represented 83.0% (R$ 144,037.5 million of reais) of the total amount of the State's annual tax revenues (Sefaz, 2020). From this amount, 25% is passed on to São Paulo City Hall and the other 75% is used for paying expenses and investments done by State government. This provides a clear dimension of the importance of ICMS, so that the State can continue operating. Therefore, caring for this collection is a matter of survival for the State itself. It is in this context that the battle against tax evasion is inserted, draining significant financial resources that could be directed to meet the needs of citizens, reducing the quantity and quality of public services offered, notably to most needy population. The technological advancement of computers has enabled large amounts of data to be stored and processed more quickly. This new reality has affected the work of internal, external and governmental auditing. According to Appelbaum, Kogan and Vasarhelyi (2017), companies' business processes are being substantially accelerated and external audit procedures need to keep track of this new reality, implementing new predictive, prescriptive and preventive analytical models that allow almost instantaneous auditing. The auditor needs instruments that help him to quickly assess a large number of tax documents in digital form, assisting him to detect any signs of tax and accounting irregularities. One of these instruments is the use of Benford's LB-Law in digital audits. The LB is based on the assumption that the first, the second and the other digits of a certain data set follow the Benford probability distribution and deviations from the LB, from certain values, could indicate possible manipulations in the data set. In this study, we used the traditional LB first and second digit Z, Chi-Square and MAD compliance tests, in addition to the ExcessMAD test, as metrics to gauge the degree of LB compliance of electronic invoices issued and received by companies declared void by SEFAZ / SP, in the period 01/01/2017 to 12/31/2019. Furthermore, it was investigated whether these companies have similarities, because of this conformity / non-conformity and of their registers, commercial and tax characteristics. In addition, in this work, the study by Bradley and Schulzke (2016) was expanded, by formulating the ExcessMAD test for the first and second digits, identifying the effect that small samples have on the type I error of this test. Based on these results, and on the optimization method proposed by Carreira and Silva (2016), which estimates the minimum number of digits that should be added to a data set to make it conform to the LB, the values ​​of commercial operations and ICMS were eventually handled by null companies. This study aims to contribute to improving the efficiency level of government digital audit, notably in the process of selecting taxpayers, which can prioritize those companies that present a higher level of manipulation of their commercial operations.

 *Abstract provided by the author



Voltar para a página de eventos