Tax Compliance by the Very Wealthy: Red Flags of Risk

Abstract

The issue of detection of tax evasion and avoidance is no small challenge for tax authorities. Oftentimes detection depends on extensive auditing. The ambiguity and difficulty of the task has led tax authorities to devise risk management strategies so that they can dedicate their limited resources to detection exercises that are likely to uncover instances of evasion or aggressive avoidance. The process of identifying a set of red flags that alerts tax officers to aggressive tax planning is the focus of attention in this chapter. We illustrate how a statistical methodology can be deployed with subjective estimates of risk made by expert analysts and more objective data in the form of dollars at risk. 1 In 1996, the Australian Taxation Office (ATO) established a High Wealth Individuals (HWI) Taskforce to enhance compliance management for HWls. It did this in a way no other tax authority in the world has done before to our knowledge. Traditionally, tax authorities treat individual persons and corporate entities (partnerships, trusts, corporations) as separate taxpayers in their case management. The innovation of the HWI Taskforce was to treat the individual return of a HWI and the entities they control as a single case. In the first year of operation, 180 HWis received a questionnaire about the groups of entities they controlled, or from which they received income. These were formalised in subsequent years into expanded returns, called Current Year Data Collection (CYDC) returns. In 1997 and 1998, 285 HWis (142 rated as medium overall risk and 143 high overall risk) were required to complete expanded returns for their 2371 and 2599 associated companies, trusts, partnerships, or individuals. The level of these individuals' wealth, and the amount of tax they paid, to qualify for this level of scrutiny is confidential. However, all those completing expanded returns are extremely wealthy.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 67 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.