The purpose of the IRPAC Tax Gap Subgroup is to help the IRS improve its estimates of the tax gap. This is the report for 2009, which is the third year in the panel’s three-year term.
In pursuing its mission, the panel conducted two meetings with IRS staff from the Office of Research, Analysis, and Statistics (RAS) this year – a telephone conference call on November 10, 2008, and a conference call January 15, 2009.
In addition to the conference calls, one of the panel’s members, Marsha Blumenthal, participated in a panel discussion entitled “Is There a Gap in the Tax Gap Estimates?” at the IRS Research Conference on July 8, 2009. In her comments, Prof. Blumenthal summarized the findings from this panel’s 2008 report.
Most of the panel’s deliberations during this year were centered on reviewing and providing constructive comments on potential estimation methodologies for the estate non-filing tax gap and individual income tax underreporting gap. Prior to each conference call, IRS provided panel members with white papers detailing the potential estimation methodologies so that members would have sufficient time to review and formulate questions. During the conference calls, the IRS researcher responsible for preparing the white paper presented the methodology to the panel. In addition to providing a number of verbal comments during the conference call, panel members provided extensive written comments on the individual income tax underreporting gap methodology.
A. Estate Tax Non-filing Gap Estimation
The November conference call was devoted to discussing a document prepared by staff from RAS, “Estate Tax Filing Noncompliance.” This document updated the methodology used by the IRS to generate the current estimate of the estate tax non-filer gap. The methodology relies on an external panel survey of older households and the wealth held by that segment of the population. IRS uses this data to estimate wealth adjusted mortality curves in order to estimate the number of estates with a filing requirement. Comparing these predictions to the actual number of filers in the Statistics of Income (SOI) data provides an estimated number of non-filers and the sum of the estimated taxes due for each predicted non-filer gives a prediction of the non-filing tax gap.
Based on that telephone call and subsequent conversations, the Tax Gap Subgroup recommended the following actions concerning the estate tax non-filing gap estimation methodology:
- IRS should account for estates that file, but do not have a filing requirement.
- IRS should account for married couples where both spouses die in the study period because those households may have different filing patterns.
- In estimating the estate tax non-filing gap, it is necessary to estimate household wealth and allocate wealth among spouses. The estimates of allocation of wealth within married couples should be informed by knowledge of residency in a community property state.
- The following recommendations concern the use of the Health and Retirement Study (HRS) to estimate wealth across various demographics;
- IRS should consider benchmarking the HRS wealth distribution to the Survey of Consumer Finances (SCF) to account for the truncated right hand tail of the wealth distribution observed in the HRS.
- IRS should determine whether the HRS sample captures people over the target age that are living in households over the target age (e.g., elderly parents living with their adult children).
- IRS should consider using multiple years of HRS data to smooth the wealth estimates.
- IRS should compare the wealth values reported to HRS to estate values reported for tax purposes.
- IRS should consider the performing the following sensitivity analysis. Instead of estimating mortality by level of wealth (bin) across demographic groupings (cell), the IRS should investigate applying the mortality model directly to the individuals in the HRS study over the filing threshold. IRS should allow the binning of wealth to vary by cell. The $675,000 filing threshold should be lower for those individuals who have already used some of their unified credit for gift taxes; IRS should investigate what effect this fact may have on the estimates. IRS should estimate confidence intervals for both the number of non-filers and the dollar value of noncompliance.
- IRS should discuss the link between tax planning and estate tax filing behavior. How does tax planning differ by age, marital status, etc.? Can one identify tax planning by looking at the trajectory of wealth over time? Does the HRS capture information on common tax planning techniques (e.g., Family Limited Partnerships (FLPs), closely held businesses, and valuation discounts)?
- Filing compliance is likely different (higher) for high value estates than low value estates. Therefore, IRS should account for increases in the filing threshold when projecting estimates from one year to another.
B. Individual Income Tax Underreporting Gap Estimation
The purpose of the January 2009 teleconference was to discuss a proposed new methodology for estimating the individual income tax underreporting gap. Prior to the conference call, the IRS provided a copy of a white paper applying this methodology for Tax Year 2004 entitled “Tax Year 2004 Individual Income Tax Underreporting Gap: Description of Methodology and Line Item Estimates.” This white paper described four major changes in the underreporting gap estimation methodology from what was used to produce the official estimates for TY2001. The white paper also suggested an approach to estimate the individual underreporting gap for tax years where National Research Program (NRP) reporting compliance data do not exist. The four changes are: (1) line-item estimates of undetected misreported income using the technique of detection controlled estimation (DCE) (Erard, Brian and Jonathan Feinstein, Adjustment of Income Tax Underreporting Using Detection Controlled Estimation. Final report under contract order number TIRNO-05D-00050 0001, November 15, 2007.), (2) implementation of a tax calculator to compute total income tax (and self-employment tax) liability for TY 2001 and TY 2004 using both reported and corrected income and offset amounts, (3) use of SOI’s Complete Report File (CRF) to provide a data bridge to estimate tax underreporting in the absence of NRP data, and (4) imputation of TY 2001 DCE-based estimates of misreported income specifically to TY 2004. Following the meeting, members of the panel provided written comments on the white paper to the Chairman who then forwarded the comments to the IRS.
The Tax Gap Subgroup expressed concerns with the “black box” nature of the DCE methodology used to estimate the amount of income that is undetected by IRS examiners during the audit. Although the goal of the white paper was to describe a proposed methodology for estimating the underreporting gap, the DCE methodology is a key component and deserves more explanation than was given in the white paper. This is especially important since the new DCE methodology developed under contract to the IRS includes large changes in the corrections for undetected income for several line items. The panel was concerned about understanding the reasons for large DCE corrections for some line items where there is information reporting and where the correction was significantly larger than in prior estimates.
The Tax Gap Subgroup recommends that the IRS conduct additional research on the DCE methodology and provide more detail on the DCE methodology and the underlying equations in subsequent tax gap reports. We believe that this additional research is needed to increase the transparency and credibility of using the DCE methodology to measure the tax gap and will provide useful insights on areas for improved data collection in future random audit studies conducted by the IRS' National Research Program Office.
The Subgroup also recommends that IRS provide estimates of the extent to which changes in methodology affect changes in the estimated tax gap. Ideally, IRS should be able to provide estimates that separate out the sources of changes in the tax gap into changes in methodology, changes in tax law, changes in the underlying income distribution, and changes in voluntary compliance. Finally, the panel provided IRS with some technical comments on the imputation methodology.