Policy and Complex Systems
data should allow researchers to design and analyze more accurate and valid models . As a result , tax policymakers will have more accurate information upon which to design legislation and regulations to maximize the revenue generation and social welfare goals of tax policy . This will bring policy on tax compliance from the chaotic towards the ideal political decision making field in the Stacey Matrix .
The principal purpose of this paper is to derive an expected value measure of the tax underreporting rate given only tax authority enforcement data . The method derived here measures only the relative frequency of tax underreporting , not its magnitude . In other words , it only approximates how many returns contain net underreporting , not how much tax ( in monetary units ) is underreported . Although the derivation should be applicable to any tax system that meets the assumptions listed in § 2 , the paper makes specific reference to the United States tax system since those are the data both studied previously by Phillips and readily available for analysis .
To be clear , this paper seeks only to derive a method for approximating the expected value of the underreporting rate from enforcement data . It does not seek to accurately measure the true underreporting rate for any given taxable period . The paper first outlines it assumptions , and then provides notation sufficient to explore the principal purpose . It then presents preliminary claims and proofs before making the final derivation . The paper concludes with a brief summary .
II - Assumptions
This paper makes the following
assumptions about the tax system under consideration :
1 . Taxpayers assess their income tax liability and include this assessment on returns they file with the tax authority .
2 . The tax authority does not directly monitor taxpayer activities nor does it audit every assessment for accuracy . Instead , the tax authority selects a small sample of returns to audit . The paper refers to this as a “ self-report / audit system ” ( Kotowski , Weisbach , & Zeckhauser , 2014 ).
3 . The tax authority ’ s sample selection criteria for return examinations ( audits ) are at least marginally more efficient at predicting which returns contain underreported tax than if the authority selected returns only at random .
4 . The authority maintains statistical data that allows one to calculate the proportion of all filed returns that experience audit for any given tax period ( the audit rate ).
5 . These same data allow one to calculate the proportion of audited returns that , according to the tax authority ’ s final audit determinations , contain underreported tax ( the audit success rate ).
III - Notation
For self-report / audit systems , a tax
return in the population of all filed returns R has one of four event types . These types relate to whether the tax authority audits a return and whether a return contains underreported tax . In type A , a return experiences audit . In à , a return does not experience audit . In type U , a return contains underreported tax . In Ũ , a return does not contain underreported tax .
A return can also exist in one of four outcome states . In state AU , a return experiences audit and contains underreported tax . In AŨ , a return experiences audit and does not contain underreported tax . In ÃU , a return does not experience audit and
7