IRS Correspondence Exams: Doing Less with Less

0 Flares 0 Flares ×

I’m not a huge fan of the suggestion to “do more with less.” To me, it tastes like a corporate-sugar coat to otherwise clearly insipid advice: “be more efficient.” During the pandemic more than ever it should be clear that there are actual bandwidth limitations: sometimes you legitimately need more (time, resources, staff, etc.) to do more.

But that isn’t to say we should ignore inefficiencies, particularly when cuts (to budgets, staffing, etc.) all but force us to confront them. Recently Anna Gooch posted on a TIGTA report that covered some of those inefficiencies in EITC exams. Here, I will discuss another TIGTA report on the inefficiencies of non-EITC correspondence exams. As will be seen, the numbers are pretty shocking in terms of how much “less” the IRS has in terms of resources, but also in terms of how much less the IRS is doing with those resources. In a subsequent post I will drill into some more practitioner-oriented insights that can be gleaned from the report. For now, I will focus on what can be gleaned from the report on tax administration writ large.

read more…

The TIGTA report contains three recommendations to the IRS on how to improve the (non-EITC) correspondence exam process. To me, however, those recommendations are the least interesting and least consequential aspects of the report. They largely boil down to minor recommendations on how the IRS change their selection criteria taxpayers based on prior exams. I’ll touch a bit on that in my practitioner-focused later post.

But what I found more important, and more eye-opening, are the numbers that the TIGTA report lays out. The numbers show just how much less the IRS is working with (in terms of employees) and just how much less the IRS is bringing in (in terms of tax assessments). But they also suggest that the IRS is doing “less with less” at least in part because of inefficient resource allocations. Let’s take a look at some of the statistics…

Doing Less with Less: Massive Drop in IRS Compliance Personnel

One number that jumps out is the Compliance Personnel reductions the IRS has had to contend with. By the numbers, there are 15,000 fewer enforcement employees (a drop of over 25%) from 2010 to 2018, resulting in a 40% reduction in exams.

This mind-blowing statistic largely serves as the impetus for the TIGTA report. We should take as a given right now that the IRS is stretched extraordinarily thin. Accordingly, the focus shifts to the best use of these preciously limited resources. And there is certainly room for improvement.

Over a quarter of the IRS correspondence exams closed between 2017 and 2019 had “no impact on net dollars recommended.” This means that the audit resulted in either “no-change” (i.e. the return was correct) or “no-pay” (that is, the return was wrong but the taxpayer hasn’t made payments on it).

Assuming, as is historically accurate, that a large percentage of the “no-pay” taxpayers stay no-pay, you are looking at a sizeable waste of exam resources going towards returns that will never bring any money into the fisc. Yes, I know there are important reasons to audit beyond just bringing in money like encouraging voluntary compliance. But perhaps, given the resource allocation constraints alluded to above, we should focus our attention on auditing returns that do both: encourage compliance and bring in money. They just might exist…

Doing Less with Less: Trending Towards More Low-Income Exams

Looking at the trends from 2017 to 2019, the numbers are either baffling or appalling. In 2017 the very rich (total positive income (TPI) of $1 million or more) comprised 3% of all non-EITC Correspondence Exams. That same year, the merely well-to-do (TPI of $200K to $1 million) comprised 10% of those exams. The less-fortunate (TPI less than $200K) comprised 86% of those exams. By 2019 the less-fortunate comprised 91% of non-EITC correspondence exams while the very rich only comprised 1% and the well-to-do 8%.

From a pure collectability standpoint this doesn’t make much sense. As noted in the TIGTA report, “despite a lower level of collectability for lower income taxpayers, the relative percentage of non-EITC correspondence examinations [for lower income taxpayers] continues to increase.” But there are other reasons this surprised me.

I talk to my students about why it is our clients are at a significantly higher risk of audit than, say, the students themselves are likely to be when they presumably become upper-middle income lawyers. The answer generally hinges on the Earned Income Tax Credit. I generally bring up the Pro Publica articles here and here. I also recommend as further reading (that is to say, something that maybe one of my students would be interested in but doesn’t have the time to read) the excellent Lawrence Zelenak law review article here. The simple bullet points, as far as I’m concerned, is that low-income EITC returns get audited more because (1) the method of audit is inexpensive and (2) the EITC is sometimes conceptualized as a quasi-welfare-style benefit (that is, a cash transfer) which flares up a greater sense of cheater-detection. Of course, the PT team has also dug into these exact issues here. And the problem of EITC as being an improper payment here.   

This TIGTA report, however, adds an unfortunate wrinkle to my explanation. How do I explain that the trend is to audit relatively low-income taxpayers at a higher rate than high-income earners when (1) the method of audit is the same for both groups (correspondence exam) and (2) there are no “welfare-style benefits” at issue?

There may be some innocuous explanations for this trend. It is possible that high income returns raise fewer flags because they tend to have better tax return preparers. That is obviously just speculation on my part, but I do tend to see middle-income earners falling in the “donut hole” of tax preparation where they don’t qualify for free services and it economically doesn’t make sense to hire someone (competent).

Another, more testable explanation is that the types of issues that are likely to be subject to correspondence exams are, again, simply more likely to be found on low-and-moderate income taxpayer returns. Itemized deductions are highly audited and may not be an issue for high-income because of the AMT. Similarly, Schedule C expenses from sole-proprietors are highly audited (in fact, “Form 1040 Schedule C Issues” were the second most audited project code in the report. They can also be quite time consuming…). However, they may not be an issue for high-income because wealthy (sophisticated) taxpayers are more likely to have a partnership or other entity for their business income. I would think that the wealthier taxpayers are more likely to file the Schedules that come after C… specifically, Schedule D, Schedule E, and Schedule K-1. Those schedules do not appear in the list of top correspondence exam project codes, possibly because there just are few returns listing them or possibly because they don’t lend themselves to correspondence exams. It is also possible that these issue would require more training and time than the IRS invests in examiners, which is a paradigm shift in itself.

Doing Less with Less: Assessing Much, Much, Much Less Tax…

Let’s let the numbers speak for themselves: the value of correspondence exam assessments decreased 52% from 2015 – 2019.

Read that again, please, after your brain has pieced itself back together.

In 2015 the IRS had proposed assessments of $7.3 billion from correspondence exams. In 2019, the proposed assessment value declined to $3.5 billion. Yes, you may recall that there has been a 40% decrease in total correspondence exams from 2010 as compared to 2018, but that is over a longer time-period than 2015 – 2019 and can’t explain the full difference. I think this number is the best indication that there are some inefficiencies factoring into the IRS doing less with less.

Going Deeper into the TIGTA Report

The TIGTA report implies that there are two different ways the IRS could more efficiently run its exam resources. First, the IRS should emphasize certain sources of selected exam cases. Second, the IRS should emphasize certain types of issues. Presently, the IRS does not (by TIGTA’s estimation) emphasize the most efficient sources of exams or the most efficient issues.  

What correspondence exams take the least amount of time and result in the highest value assessments? Perhaps as no shock to those in the tax community it is far-and-away exams on “non-filers.” The average assessment for non-filer exams from 2017 – 2019 is a shocking $28,102. By contrast, the average assessment for Schedule C Issues was only $4,576 and also took significantly more time than non-filer exams did.

It is impossible to determine how much income the average non-filer would have to result in an assessment of $28,102. But for frame of reference in 2018 a single individual with no dependents would have to earn over $140,000 in wages to have that type of income tax due (granted they would need significantly less self-employment income to hit that same assessed tax number). Anecdotally, that amount of wage income is consistent with the tax-protestors I come into contact with. They are rarely low-income.

TIGTA does not specifically “recommend” more exams of nonfilers but does note that “Schedule C issues were less productive than other issues, such as nonfiler issues.” I imagine (and sincerely hope) that the IRS would be quick to examine more nonfiler issues, but that there is simply a smaller pipeline of nonfiler cases. Perhaps that is something increased information reporting from banks could have fixed were it properly focused on upper-income earners.

It should be noted that PT has written about the nonfiler issue multiple times in the past, including the IRS’s essential abandonment of the “Automated Substitute for Return Program.” There is nuance to the issue, including how much how is actually collected from nonfilers (here).  

Which gets to the second issue: the source of return examination selections. Some sources are more productive (in terms of assessed dollars) than others. The IRS has multiple different methods of selecting returns for correspondence exam. There are the “Compliance Data Environment (“CDE”) filters. There are the Dependent Database (“DDB”) and Discretionary Exam Business Rules (“DEBR”) filters. And there are also “referrals” from other, non-Exam IRS functions. Guess which brings in the highest average assessment?  

Far and away, it is from referrals. And it isn’t even close.

Referrals brought in an average assessment of $16,941 as compared to $3,132 from DDB/DEBR and $4,195 from CDE. Oh, and referrals take significantly less time, too: 0.74 hours as opposed to 2.09 from DDB/DEBR and 1.79 from CDE.

On seeing these discrepancies my first thought was that referrals must comprise a small amount of correspondence exams. But that is not so. In fact, referrals are the second most common source of case selection: 33% of the total cases (244,269) from 2017 – 2019 were referrals. The most common source (CDE) was 353,064.

Conclusion: Opportunities for Doing (A Little) More with Less

From the outside looking in it seems clear that the IRS should (1) focus more on non-filer issues and (2) rely more on referrals for case selection. But the outside looking in usually distorts the true picture, so I will reserve at least some of my judgment on the grounds that there are likely many factors I am not privy to.

That said, the IRS does not, in my opinion, have a great record when it comes to efficient use of its resources. Let me close with an example from a recent conference.

During a presentation from the IRS, the speaker said that Collection Statutory Expiration Dates (CSEDs) are listed on taxpayer transcripts, and that she’d “show us where.” CSEDs are an issue of frequent concern to tax practitioners, particularly since the IRS often (systemically) makes mistakes in its calculations. Keith has written (here) about these mistakes and how unhelpful the IRS sometimes is when you ask for a CSED (here). Tantalized at the words of this IRS presenter, I picked up my pen, ready to learn something new and amazed that I hadn’t already known that the CSED was on the account transcript…

So my expression was not unlike a child tearing the wrapping paper off a box of new socks when I saw the presenter’s slide. Yes, the CSED is on the transcript (along with the ASED and RSED). Only not the transcript that is made available to practitioners pulling them online. The proposed solution was to call the IRS for the CSED or make a FOIA request.

Let me repeat that: instead of giving practitioners the information upfront, the IRS adds an extra step that (1) ties up an IRS employee and (2) wastes the taxpayer/practitioner’s time. Why? If the IRS is willing to give out their calculation of the CSED, why not do it upfront? What I hear from the IRS when they explain their decision is: “Well, we could give you this information and save everyone time and improve customer service but… what if instead we just didn’t?”

Perhaps there is something behind the scenes that I just don’t appreciate (yes, IRS technology is not great). But to an outsider it certainly appears sometimes that the IRS shoots itself in the foot to make bad situations worse -in audit selection no less than in customer service decisions. Hopefully their resource constraints may force a re-evaluation of some of these decisions, so that they may “do more with less.”

Caleb Smith About Caleb Smith

Caleb Smith is Associate Clinical Professor and the Director of the Ronald M. Mankoff Tax Clinic at the University of Minnesota Law School. Caleb has worked at Low-Income Taxpayer Clinics on both coasts and the Midwest, most recently completing a fellowship at Harvard Law School's Federal Tax Clinic. Prior to law school Caleb was the Tax Program Manager at Minnesota's largest Volunteer Income Tax Assistance organization, where he continues to remain engaged as an instructor and volunteer today.


  1. The “doing less with less” part may have something to do with the brain-drain the IRS is experiencing. I know of many (in the dozens) of experienced auditors, Appeals Officers, and Chief Counsel attorneys, who either retired or left the Service after experiencing frustrations with management or budget cuts. More recently, the work-from-home transition has further eroded efficiency. Noticeably so.

  2. Mr. Burton Smoliar says

    It seems to me that the most relevant statistic is the percentage of people in each category that are audited rather than the percentage of audits in each category. Low income people are more numerous, so the statistic used has a biased impact.

  3. Some good stuff here, Caleb! I especially like the idea of more transparency in CSED calculations. “Doing Less With Less” would make a good title for a law review article! Here are some thoughts. First, you say , “we should focus our attention on auditing returns that do both: encourage compliance and bring in money. They just might exist.” That statement would benefit from more concrete suggestions of how to accomplish identifying returns that “bring in money.” Exam has historically NOT attempted to evaluate the collection potential of taxpayers. Are you suggesting that the IRS should now create and then implement some kind of collection potential screen? How would that look? How would that increase efficiency? Second, you then abandon this idea when discussing the idea of shifting exam resources from Schedule C exams to more non-filer exams. Yet why would you believe non-filers are inherently more able to pay than Schedule C filers? I would have the opposite intuition. Similarly, you abandon the idea when you discuss which source brings in the biggest “assessments.” So if you DO further work here (whether in future blogs or a law review) I think your work would benefit from giving more thought to the relationship between Exam and Collection. That is a central feature of tax administration that has not, I think, been as well studied as it might be.

  4. Thanks all for the comments. A couple points I’d want to quickly hash out, both of which deserve more time (to Bryan’s point) than I can give at the moment:

    First, to Burton. It is correct that there are more low-income filers than high income. But that doesn’t explain the drop. There are actually more high income earners and high income returns filed in 2019 than there were in 2017, but fewer audits of them both as a raw number AND as a percentage of total audits done. Yes, the IRS did fewer audits overall, but it cut down the RATE of audit on high income at significantly greater level than it did for low income. That leads to my quick second point, responding to Bryan…

    I don’t (necessarily) think Exam employees should make collectibility determinations in their audits -they are already advised not to do so. But properly understood, I’m not really talking about exam employees in audit at all. I’m talking about an earlier stage of exam SELECTION which (I imagine) is mostly the product of a 1970s era IRS computer somewhere and not an employee (kidding, somewhat). One simple idea would be to write into that selection criteria “higher income = higher selection rate” rather than what appears to be the opposite decision.

    So much more could be written on this, I think, on the nonfiler issue alone… I concede that it isn’t a simple task to know who will actually end up paying. My bet (born out a bit by the TIGTA stats) is that a large amount of nonfilers make significant income, and thus have significant collection potential. But actually GETTING that money after assessment could, indeed, end up being extremely costly since my bet is also that these are the taxpayers most resistance to voluntarily paying.

    So much food for thought…

  5. Bob Probasco says

    Caleb, great post. With respect to CSED on transcripts – and I fully agree with you – my impression is that we’re facing a *particularly* bad case of IRS inertia. It sounds as though it took a *lot* of pushing (presumably by Taxpayer Advocacy Service) just to get the crumb of the limited information we have for CSED on our transcripts. The IRS finally added, for a subsequent adjustment after the original return was filed, the CSED for that adjustment amount shows up on the line underneath.

    Of course:
    (1) It’s not labeled as such.
    (2) Adjustments to refundable credits are broken out separately, and it doesn’t show up for those portions of the total adjustment – which may be the entire adjustment.
    (3) Subsequent events that have the effect of prohibiting collection activities and thereby extending the CSED – e.g., OIC, CDP hearing request – even if they are shown on the transcript (and I’m not sure they always are), do not show the revised CSED.

    And – you see this coming don’t you? – that opens up a diagnosis of this as an *educational* issue. That is, the IRS just needs to explain to taxpayers and tax practitioners how to determine the CSED from all that information already available. Tell them how to come up with the right answer, instead of change the programming to provide that answer (already calculated and readily available) to them.

    Yes, that doesn’t make sense to me, either. But I think it’s the most likely response by the IRS, simply because of the silo effect. The IRS groups that have to do more work because that information is not readily available on the transcripts we can access are not the same groups that would have to make the programming changes.


Comment Policy: While we all have years of experience as practitioners and attorneys, and while Keith and Les have taught for many years, we think our work is better when we generate input from others. That is one of the reasons we solicit guest posts (and also because of the time it takes to write what we think are high quality posts). Involvement from others makes our site better. That is why we have kept our site open to comments.

If you want to make a public comment, you must identify yourself (using your first and last name) and register by including your email. If you do not, we will remove your comment. In a comment, if you disagree with or intend to criticize someone (such as the poster, another commenter, a party or counsel in a case), you must do so in a respectful manner. We reserve the right to delete comments. If your comment is obnoxious, mean-spirited or violates our sense of decency we will remove the comment. While you have the right to say what you want, you do not have the right to say what you want on our blog.

Speak Your Mind