IRS RESPONSE TO RECOMMENDATION: The ASFR program prioritizes cases to ensure the tax law is applied fairly and equitably to all non-filers in conjunction with the principles outlined in Policy Statement 5-134 that provides that operations should be geared to produce the greatest revenue yield. Basing case selection on taxpayer populations where an original return is likely to be filed focuses enforcement on individuals who become compliant, while ignoring individuals who are not. The NTA’s recommendation does not consider successful collection for modules where taxpayers did not file, but were assessed under the ASFR process with no subsequent response by the taxpayer. The IRS authority to make assessments in the ASFR program should be used when necessary to assess individuals who will not file voluntarily. Selecting cases based on taxpayers who respond more often would enforce filing requirements and collection on a more compliant taxpayer population, while failing to enforce for taxpayer populations who are least compliant. This would be unfair and inequitable.

CORRECTIVE ACTION: N/A

TAS RESPONSE: The National Taxpayer Advocate is disappointed by the IRS’s reluctance to review annually where ASFR assessments have had the most success in getting taxpayers to file an original return and adjust the ASFR selection process to focus on similar types of cases.  As the IRS stated above, the purpose of the ASFR program is to promote filing compliance. This recommendation would focus the IRS’s ASFR authority on cases where this objective will most likely be achieved.

The National Taxpayer Advocate is not suggesting that the IRS would not attempt to promote filing compliance in other cases where an ASFR assessment has historically not generated an original return, but is rather suggesting that a different approach might be more successful.  For example, in cases where the IRS determines that ASFR assessments have typically not generated an original return, it can impose a different approach on these cases (i.e., sending a soft notice and making phone calls to the taxpayer, as is sometimes done by Field Collection and ACS, as explained above).  This approach will improve case resolution by focusing on a smaller number of cases and adding the element of in-person contact with taxpayers to solicit and secure tax returns.  If these personal contacts prove unsuccessful in securing tax returns, then the IRS should use its Substitute for Return authority (including Automated Substitute for Return) to make the assessment and move forward to collection.

ADOPTED, PARTIALLY ADOPTED or NOT ADOPTED: Not Adopted

OPEN or CLOSED: Closed

DUE DATE FOR ACTION (if left open): N/A

IRS RESPONSE TO RECOMMENDATION: ​IRS agrees it would be beneficial to include ASFR abatement reason codes to capture additional data for analysis and improvement of the ASFR program.  Additional reason codes would be useful in determining why returns are filed, such as when abatements are necessary to move tax liabilities to spouse SSNs for joint returns. However, any changes will be dependent on Information Technology (IT) resources and acceptance of a Unified Work Request (UWR) to perform the work.

Update: After review, it was determined reason codes are currently available that will allow for additional issues to be identified, such as “Schedule A” (076), “Business Income” (012), or “Education Credit” (035). IRM 5.18.1 will include additional instructions when it is updated in FY17. It should be noted that reason codes are a good tool to identify return issues, but their true purpose is to designate paragraphs for a resulting CP notice to the taxpayer. Employees will be required to use judgment in identifying the most appropriate reason codes to meet the taxpayers’ individual situations.

CORRECTIVE ACTION: The Collection Inventory Delivery and Selection, Non Filer and Inventory Analysis function will coordinate with IT stakeholders to determine whether additional reason codes can be created for ASFR modules. Coordination will occur in FY 2016, with a determination by October 2016.  A UWR will be input by December 2016 if IT resources are secured to perform the additional work.

TAS RESPONSE: The National Taxpayer Advocate is pleased that the IRS is willing to refine ASFR abatement reason codes.  More specific reason codes would allow the IRS to better understand why the ASFR liability was abated and to consider refinement of its ASFR selection criteria based upon that information.  The National Taxpayer Advocate understands limited resources is always a consideration, but urges the IRS to take a more analytical view to the commitment of resources for the refinement of abatement reason codes.  Specifically, investing in the abatement reason codes, which will allow the IRS to enhance its ASFR collection criteria, would involve a commitment of resources up front, but such costs would likely be offset by mitigating the cost of abating ASFR assessments.

ADOPTED, PARTIALLY ADOPTED or NOT ADOPTED: Not Adopted

OPEN or CLOSED: Closed

DUE DATE FOR ACTION (if left open): N/A

IRS RESPONSE TO RECOMMENDATION: Update: IMF CCNIP Modeling was implemented for TY 2016 modules in August of 2018. Due to systemic issues with Standardized IDRS Access (SIA), modeling could not be passed to the ASFR system at that time. Corrective programming was implemented in November 2018, but TY 2016 modules had already been referred to ASFR without scoring and could not be updated. Scoring is available and is being passed with all TY 2017 modules, but it will not be available on individual modules until TY 2017 calculations are implemented. Implementation is scheduled after the filing season blackout period has ended.

TY 2017 modules cannot be selected until May/June of 2019, when tax calculations are implemented for that year. A test & learn was scheduled with RAAS for June 2019, but was suspended due to the government shutdown. In the interim, IMF CCNIP and ASFR coordinated a plan to test implemented scoring for 2,000 modules. Inventory for that test will be based on filing frequency.

ASFR programming was updated in July 2018 to receive and display the projected balance due score based on prior filed returns. Due to the SIA issue mentioned above, ASFR could not receive scoring until November 2018. Once TY 2017 modules are selected and started, ASFR will monitor and compare filed returns with provided scoring for accuracy to determine if adjustments are needed. Results will be shared with Strategic Analysis and Monitoring (SAM).

Update: This topic continues to evolve. We are continuing to look at using models to improve our case selection so that the cases referred to ASFR are more likely to result in a tax liability instead of a refund if/when the taxpayer files their delinquent return. However, the Tax Cuts and Jobs Act (TCJA) of 2017 made changes to the tax law (increasing the standard deduction, limiting the schedule A deductions, and eliminating personal exemptions) that appear to have rendered TAS’s recommendation less fruitful than it may have been when it was originally made.

IRS is currently working on additional scoring for the Case Creation Nonfiler Identification Process (CCNIP) and ASFR cases. The ASFR program continues to refine the selection process and began coordination to include additional modeling for case selection in FY 2014. Tax law prevents IRS from including certain exemptions, deductions, and credits that may only be claimed by the taxpayer on a filed return. However, future modeling will be used to select cases that are more likely to result in a tax liability instead of a refund if these credits were claimed on a filed return. Bringing taxpayers into compliance through improved selection criteria will help to close the filing tax gap and improve future filing compliance.

We considered using scoring for the CCNIP process and ASFR. Based on the analysis we conducted we have determined not to incorporate it for case selections at this time.

We developed and deployed in IMF CCNIP three scores for nonfiler cases which predict:
(1) the likelihood a taxpayer will file
(2) the likelihood there would be a balance due
(3) the predicted balance due.

Because of limitations in the ASFR system, the “predicted balance due score” is the only score employed by ASFR, and would be the predictive factor that directly links to this recommendation and corrective action. We analyzed the scores for Tax Year (TY) 2017 selections. We reviewed a random sample of 3,000 closed TY2017 ASFR cases where taxpayers responded by filing their return. Case selection was based on the ASFR proposed liability. We compared the proposed liability with the predicted balance due score and with the actual liabilities as shown on the return the taxpayer filed.

We found that using this scoring would have resulted in the non-selection of cases that should be selected. The “predicted balance due model” scoring estimated that 1,842 of the 3,000 would file a return that was below ASFR selection criteria. Upon review, of the 1,842 cases the model predicted would be below the ASFR criteria:

• actual filing results showed that 932 of the 1,842 taxpayers reported a liability above ASFR selection criteria;
• using predicted balance due model scoring for case selection would have excluded 147 High Income Nonfiler cases and 267 Federal Employee Return Delinquency Initiative (FERDI) cases from selection, among others, where taxpayers’ actual return filing showed they met ASFR criteria;
• scoring based on prior year filing was similarly problematic; while this predicted 61.4% of the population would be below ASFR criteria, the actual taxpayer responses showed that only 30.3% were below criteria.

  Further, current scoring does not include the TCJA changes (which increased standard deductions, limited Schedule A deductions, and eliminated personal exemptions). We determined that these tax law changes made the models less relevant. We plan to continue our scoring review in FY21 to identify potential improvements and to compare future selections, incorporating TCJA reporting changes into the models.

CORRECTIVE ACTION: The Collection Inventory Delivery and Selection, Non Filer and Inventory Analysis function will continue to coordinate with the Strategic Analysis and Modeling (SAM) group and IT stakeholders to pursue additional modeling and scoring for Nonfiler case selection. UWRs were submitted in FY 2015 to include placeholders. Implementation is planned for FY 2017. Testing and implementation is dependent on resources available for ASFR inventory.

Update: The Collection Inventory Delivery and Selection, Non Filer and Inventory Analysis function will coordinate with IT stakeholders to determine whether additional reason codes can be created for ASFR modules. Coordination will occur in FY 2016, with a determination by October 2016.  A UWR will be input by December 2016 if IT resources are secured to perform the additional work.

ASFR systemic programming updates will include a placeholder for scoring with a May, 2017 implementation date. Although the placeholder will be added, implementation of scoring by IMF CCNIP and ASFR cannot be completed at this time. Resource reductions in FY16 and FY17 have prevented any pilots that would enable full implementation for modeling and scoring. Modeling work continues but it has not been implemented.

Update: The ASFR scoring field was implemented in May, 2017, and systemic identifiers for conducting pilots were added. However, resources were not provided in the FY18 work plan to test modeling for ASFR selections. Scoring/modeling has not been implemented by IMF CCNIP due to IT resource issues. IMF CCNIP scoring requires JAVA programming that IT cannot accommodate at this time.

The Collection Inventory Delivery and Selection, Non Filer and Inventory Analysis function will coordinate with IT stakeholders to determine whether additional reason codes can be created for ASFR modules. Coordination will occur in FY 2016, with a determination by October 2016.  A UWR will be input by December 2016 if IT resources are secured to perform the additional work.

Update: ASFR programming was updated in July 2018 to receive and display the projected balance due score based on prior filed returns. Due to the SIA issue mentioned above, ASFR could not receive scoring until November 2018. Once TY 2017 modules are selected and started, ASFR will monitor and compare filed returns with provided scoring for accuracy to determine if adjustments are needed. Results will be shared with Strategic Analysis and Monitoring (SAM).

TAS RESPONSE: The National Taxpayer Advocate is pleased that the IRS is willing to consider third party information as part of its ASFR selection criteria.  Using this information will enhance the IRS’s ability to select cases for ASFR where a liability actually exists, rather than making an assessment on an account that will likely result in abatement, thereby wasting IRS resources that could be better used elsewhere.  Again, the National Taxpayer Advocate urges the IRS to consider how the up-front investment of adjustingits selection criteria to consider third party information would result in a more efficient and effective program long term.

ADOPTED, PARTIALLY ADOPTED or NOT ADOPTED: Partially Adopted

OPEN or CLOSED: Closed

DUE DATE FOR ACTION (if left open): N/A