6.1 As part of his work program consultations, the IGT received significant feedback from individual stakeholders or their representatives in relation to the ATO's risk assessment tools applied to this market segment. The main concerns related to the following areas:

  • delayed refunds arising from the ATO's Income Tax Refund Integrity Program (ITRIP); and
  • use of third party data in ATO compliance activities.

6.2 The above concerns are the subject of a current specific IGT review.326 Chapter 8 draws on this later review to make additional observations.

6.3 Although the ATO uses other risk assessment tools in this market segment, stakeholder concerns were largely limited to the above two topics. This may be due to a number of reasons including that there may be a lack of awareness, limited resources of stakeholders in this segment to engage with the IGT and/or the low level of adverse impact caused by the use of these tools.

6.4 This chapter seeks to provide additional information on other risk assessment tools used in the individuals market. This may promote greater awareness and facilitate further stakeholder concerns which, if sufficiently significant, may form the basis of a future specific IGT review.

6.5 In other jurisdictions, some revenue authorities have published details and results from the use of their risk assessment tools. For example, HMRC in the United Kingdom have published details about how they use their system, called 'Connect', to gather and make connections between large data holdings.327


6.6 The individual market segment includes 12.4 million individuals who lodge income tax returns. For many individual taxpayers, the lodgment of their income tax return is their only interaction with the ATO.328

6.7 Within the ATO, the Micro Enterprises and Individuals (ME&I) business line has main responsibility for administering the individual market segment. Other business lines may be involved in different parts of this market segment, depending on the circumstances. For example, high wealth individuals are administered by the Small and Medium Enterprises (SME) business line.

6.8 Within ME&I, the Individuals Compliance and Data Management (ICDM) stream is responsible for managing the compliance risks of individuals for this market segment.

ATO risk assessment tools used in the individuals segment

6.9 The ATO income tax risk assessment tools used by various areas affecting the individuals market segment are listed in the following table:

ME&I Serious Non-compliance
Agent rating tool Identity Crime Model
Data matching with third party data  
Expert business rules  
Pattern detection  
Risk assessment and profiling tool  

6.10 The ATO has through the course of this review supplied the descriptions for each of the above tools. These are outlined below.

Agent Rating Tool (ART)

6.11 The Agent Rating Tool compares returns prepared by a tax agent against those of a pre-determined peer (reference) group. The manner in which the peer groups are established has been validated on two occasions by representatives from Monash University.

6.12 A range of measures are determined for each agent and then compared against their reference group. For example, the Tax Agent Measure may be the median specified work related expenses (WRE) value for their client base. These include median client amounts in relation to motor vehicles expenses or other travel amounts.

6.13 The Reference Group Measure is the median Tax Agent Measure for the specified WRE label for the reference group. The median is used in preference to the average to avoid the influence of extreme scores and outliers.

6.14 For each agent performance measure, the difference between the agent's value and the Reference Group value is indicated by the Deviation Score. The Deviation Score indicates how much an agent differs from the reference group median for that performance measure.

6.15 For each agent, deviation scores are added together to give a Total Deviation Score for each set of performance measures (for example WRE, Rental, CGT and Offsets). Based on these assessments, a risk score is generated to identify which tax agents within a particular population are significantly different to their Reference Group. The product developed to support the presentation of this analysis is called the Agent Rating Tool.

6.16 A key element of the ART is the ability to produce a consolidated one page profile which serves as a tool for the ATO officer to conduct a conversation with a tax agent regarding the return preparation processes and standards.329

Data matching with third party data

6.17 The ATO receives data in respect of property transactions from state and territory title and revenue offices and continues to expand the use of data matching to identify the omission of capital gains. The ATO use standard data matching rules based on its understanding of the risk to identify individuals who make a gain from disposing property where the property is not covered by main residence exemption.

6.18 The ATO also performs calculations to estimate the capital gains and make allowance for holding costs, such as stamp duty. The ATO compares this information to what has been declared in the return and contacts relevant taxpayers (after the notice of assessment has issued) where it identifies a capital gain has been potentially omitted or calculated incorrectly.330

Expert Business Rules

6.19 The Expert Business Rules use rules or parameters of known identified 'at risk taxpayer' behaviour across income tax return labels. These rules may use a range of ATO data holdings including corporate data holdings, compliance history information, Payment Summary Annual Reports, Annual Investment Income Reports and Department of Immigration Temporary Visa Holder data. Where these rules are triggered, a decision is taken to either monitor the behaviour or the individual income tax return is taken offline for review prior to issuing the assessment.331

Pattern Detection Model

6.20 The Pattern Detection Model is designed to detect patterns or commonality within individual income tax return data to detect 'high risk' behaviour. The model considers common variables amongst returns and determines whether a pattern exists. This nature of exploration and discovery distinguishes data mining approaches such as the pattern detection model from traditional risk identification models such as the expert business rules.332

Identity crime and network detection model

6.21 This model is also used in the individuals market segment but is owned and maintained by the Serious non-compliance (SNC) business line instead of ME&I. The identity crime and network detection model uses known attributes of identity fraud to detect high risk returns using pre-set thresholds. Networks linked to this high risk population are subsequently identified and their size, value and growth assessed. These models run across all lodgment channels.333

6.22 The expert business rules, the pattern detection model and the identity crime and network detection model are used in the ITRIP. The use of these rules and models is described in the IGT's, Review into the Australian Taxation Office's Compliance Approach to Individual Taxpayers — Income Tax Refund Integrity Program (ITRIP Review).334

Risk Assessment and Profiling Tool (RAPT)

6.23 The ATO's ME&I business line has commenced developing profiles and rules within RAPT for a number of key populations and risk areas including Pay As You Go Instalments (PAYGI) and the 'High Income Individuals' population. The ME&I business line currently use other risk models outside the RAPT. The intent, however, is for these models to be migrated in the RAPT over time.335

IGT observations

6.24 Although the IGT is conducting an in-depth review into the ATO's ITRIP and data-matching in separate reviews, it is useful to draw certain high level observations here about the ITRIP and data-matching from a risk management perspective.

Inputs and accuracy

6.25 Stakeholders had raised concerns that the risk models used in the ITRIP were not accurately detecting cases of incorrect or fraudulent returns. As part of his IGT ITRIP Review, the IGT has analysed the strike rates to determine the rate at which returns stopped and held for investigation were the subject of adjustments.

6.26 Furthermore, the IGT has also examined the average adjustment amounts to assist in determining the actual risk posed by returns stopped via the ITRIP models. However, the use of strike rates and average adjustments as a means to assess the accuracy of risk tools used in ITRIP is limited if there are no benchmarks against which the results can be judged.

6.27 For example, a random sampling of the risk population may indicate a certain strike rate if the returns were stopped at random. Ideally, risk assessment tools would yield strike rates above this benchmark. Chapter 8 further discusses the use of strike rates and random sampling.

6.28 Any changes in the strike rate and average adjustments may be attributable to changes in other variables, such as general community compliance levels and the behaviour of the risk population, rather than directly being attributable to the risk models themselves.

6.29 With respect to data-matching, the ATO receives data from a variety of sources. Each data-set has a different level of quality and usefulness. For example, legislative data, such as interest payment information received from financial institutions, is generally of a high quality since the type of information and its format is mandated by law. For example, the type and format of information is designed to allow the ATO to readily match the identity of the account holder with the identity of the taxpayer.

6.30 Other data sources, known as non-legislative data, may have a low quality and may not be a reliable source of information for the ATO. For example, some state-based property disposal data does not have sufficient identifiers to enable the ATO to match the vendor's identity with that of the taxpayer. This is because state-based property disposal data is mostly focussed on the identity of the purchaser, as they are usually the party liable for stamp duty.

6.31 When data cannot be relied upon with a high level of confidence, the ATO needs to take care in any subsequent compliance activity which is based on that information. The ATO notes that for pre-filling electronic tax returns, the ATO will only use information that has a high confidence level. Medium and low level confidence level data is disregarded.

6.32 The IGT also notes that the government has included additional funding in the 2013-14 Budget to enhance the quality of data provided to the ATO. This includes data in relation to real property. Such measures should increase the quality of data, and the level of confidence with which the ATO can rely on it.


6.33 In relation to data matching, concerns regarding proportionality differ according to the specific data matching program. For example, adjustments of low amounts of income may be reasonable for one taxpayer in their circumstances but for other taxpayers with lower incomes, such amounts and associated compliance costs may represent a much higher proportion of their income. Therefore, the actual adjustment amounts, the associated compliance costs as well as the circumstances of the target population, are important factors in understanding proportionality.

6.34 In relation to ITRIP, stakeholders have raised concerns that the ATO compliance approach with respect to refunds was not proportionate to the risk posed. An important element of risk is the distinction between fraudulent claims (fraud) and incorrect claims resulting from errors and mistakes (over-claiming).

6.35 The ATO regards fraud as a higher risk compared to over-claiming. This distinction is consistent with the ATO's Compliance Model in Figure 3 in Chapter 2. It is important, therefore, that the ATO's compliance approach is sufficiently differentiated so that those taxpayers who over-claim due to mistakes are not subjected to ATO activity more appropriate for fraud cases.

6.36 In the IGT's ITRIP Review336, it has been observed that the ATO often bundled fraud and over-claiming concerns in the same communication. For example, reason codes given to tax agents to explain why their clients' tax returns were being held included 'Reason code 1: Potentially fraudulent and/or overstated claims'.

6.37 Furthermore, since the ITRIP stopped refunds as part of pre-issue compliance activity, stakeholders raised the question of whether cases of potentially limited over-claiming that are not fraud are more appropriately addressed as a post-issue activity, that is, after the refund is issued, to reflect the lower level of risk.

6.38 Lack of differentiation of different risk such as the one in the ITRIP example above, may increase taxpayer and tax agent perceptions of unfairness in the way the ATO is treating taxpayers and may negatively affect levels of voluntary compliance.

6.39 As indicated in the ATO Compliance Model, the ATO attempts to distinguish between taxpayer attitudes to compliance, ranging from 'willing to do the right thing' through to those who have decided not to comply. The model also indicates, that in light of different taxpayer attitudes, the ATO should also respond in such a way as to positively influence taxpayer behaviour.337 This distinction is reinforced elsewhere in ATO publications. For example, in the ATO's Second Report of the Cash Economy Taskforce, it is asserted:

The community expects fairness and individual treatment. The ATO needs to recognise and differentiate between those trying to do the right thing and those who intentionally disregard their taxation obligations. This will require the ATO to be firm, but also fair, in bringing to account those who are not meeting their obligations.

... Importantly, the ATO needs to be sure that those already in the system have full knowledge of their obligations and have been given every opportunity to comply. The ATO must be sure that previous good behaviour, or a history of poor behaviour, is acknowledged and taken into consideration in current dealings. The use of stronger enforcement measures on an industry or individual taxpayer will be supported by evidence that lesser measures have proved unsuccessful.338

6.40 The extract above highlights the importance of taxpayer perceptions of fairness. In the context of the ITRIP, this means the ATO action should also be proportionate to the risk posed by the taxpayer. The ATO Compliance Model needs to be viewed in its historical context. It was originally a model for guiding ATO decision making in relation to sanctions for confirmed non-compliance. The approach in more recent times requires a more holistic consideration of the taxpayer experience and relationship.

326 Ibid. See also Inspector-General of Taxation, Review into the Australian Taxation Office's Compliance Approach to Individual Taxpayers - Income Tax Refund Integrity Program.

327 HM Revenue & Custom, Closing in on tax evasion - HMRC's approach (December 2012).

328 ATO, above n 33, p 18.

329 ATO communication to IGT, 3 June 2013.

330 ATO communication to IGT, 4 June 2013.

331 ATO, above n 329.

332 Ibid.

333 Australian Taxation Office, Risk Treatment Plan - Income Tax Refund Integrity, p. 8. (from Enterprise Risk Manager).

334 Inspector-General of Taxation, Review into the Australian Taxation Office's Compliance Approach to Individual Taxpayers - Income Tax Refund Integrity Program.

335 ATO, above n 329.

336 IGT, above n 334, para [5.30].

337 Australian Taxation Office, Introduction to the Compliance Model (25 March 2009).

338 ATO, above n 52, p 58.