Compliance notes

The Treasury Department is out with a fact sheet on Biden’s proposed tax increase for his infrastructure bill.  It’s mainly corporate rate increases (taking back half of TCJA’s rate cut) and stiffening TCJA’s imposition of minimum taxes on foreign earnings, along with reviving some version of the book income minimum tax from 1990s. Another round of tax increases will be proposed for a third initiative. I assume that will have individual increases in it. I had hoped he would propose a healthy compliance element – e.g., rebuilding the IRS, which has been emaciated by budget starvation over the last decade, expanded information reporting, and similar. The fact sheet gives a modest nod in that direction (see pp. 15 – 16) mainly for C corporations. Since the proposal exclusively focused on C corps, that makes sense. But compliance problems are likely bigger for business income reported (or not reported) on individual returns (see below).

Biden’s separate budget proposal does include a 10% funding increase for the IRS (Business Insider story), on top of the increase in ARPA. Those one-time bumps are good but what is needed is an sustained (multiyear) series of increases to reverse years of underfunding.

My preference for compliance measures derives from two thoughts:

• The putative targets of Biden’s likely tax increases (i.e., high-income folks and businesses) are prime contributors to the tax gap and it makes sense to get the noncompliant to pay what they already owe before piling more on compliant taxpayers.
• Putting on my SALT hat, improved compliance provides an automatic fiscal dividend to state and local governments, strengthening their tax and fiscal systems. The national debate glosses over or completely ignores that effect, but it is a big deal.

This post describes and archives a couple of items that I have noticed over the last weeks on compliance issues:

• I discovered former IRS commissioner Charles Rossoti’s has a website which provides details on his proposals for expanded business information reporting.
• A recent NBER paper that analyzes noncompliance by the highest income filers, specifically undisclosed offshore accounts and underreported earnings of pass-through businesses.

Note: just as I was finalizing this post IRS Commissioner Rettig testified to Senate Finance that the tax gap is likely $1 trillion or more than 2X the official number (Politico story).

Rossoti website

Former IRS Commissioner Charles Rossoti (appointed by Clinton) has a website, Shrinkthetaxgap.com, that he and another former IRS executive (Fred L Forman who served under Rossoti, mainly focusing on IT issues) maintain. Its purpose is to promote their ideas for increasing tax compliance, as the website’s name implies.

Most of their ideas are described in a series of Tax Notes Federal articles – initially by Rossoti alone, then another with Furman, and finally ones with Larry Summers and Natasha Sarin. I had read the last of the articles, since it was made available outside of Tax Notes’ paywall and blogged about it [here and here for CBO estimate and Summers and Sarin’s response]. All the articles are available on the Shrinkthetaxgap.com website.

I am skeptical about the viability of collecting $1 trillion in more revenue over a decade just through more IRS funding, as Summers and Sarin have claimed. CBO’s estimates are more modest. However, the first two articles, which I had not read, focus mainly on Rossoti’s idea for expanded information reporting for sole proprietor and pass-through businesses. Expanded information would pump up the potential revenue, and that is mainly what the website focuses on. This earlier (2019) Summers and Sarin NBER article (p. 11) estimates $815 billion for expanded IRS funding and $350 billion more for expanded information reporting over 10 years. The information reporting piece seems credible but very difficult to get through Congress. (IRS might be able to implement some of the requirements without congressional authorization, but that would be dicey politically and take even longer.)

Rossoti and Forman refer to their information reporting idea as the 1099New proposal. Its essence would subject some subset of businesses to IRS information reporting by their financial institutions on bank deposits and withdrawals. These numbers would be plugged into tax returns and reconciled (!!). Tax gap statistics show that there is rampant underreporting of business revenue and income – i.e., by proprietors on Schedule C. Information reporting on the amount of that revenue that shows up in bank accounts seems likely to reduce underreporting. At least that is the core of the idea. Similar information on interest and dividends greatly reduces under reporting, of course.

They have spent a good deal of time and effort working out the details of how this could be implemented, as one would expect of former tax administrators. Much of the focus in the earlier articles is on implementation issues, largely from an IRS perspective. They have modified the proposal in response to comments. If something like this starts to gain legs (e.g., by attracting interest by the administration or influential members of Congress), I suspect a host of objections will be raised regarding its costs and practicality.

I have never run a business, represented businesses, or been a banker, but I would imagine the hue and cry about a proposal like this will be substantial – particularly the need to reconcile bank transactions on tax returns. Aside from imposing costs on financial institutions and software providers and expanding tax reporting significantly, the information would undoubtedly generate a lot of false positives that will need to be explained to the IRS.

I must say that I am not optimistic about provisions like these being implemented. Congress has occasionally enacted somewhat similar, albeit much less ambitious, provisions – e.g., the ACA required expanded reporting of credit card receipts. But my observation is the following pattern often occurs:

1. Congress enacts a provision with delayed implementation, a practical necessity, obviously.
2. The IRS starts issuing guidance and proposed regulations.
3. Affected businesses and individuals panic when details of requirements are revealed, and their accountants and lawyers tell them horror stories about how complying will affect their operations.
4. Trade associations and other interest groups lobby Congress to repeal or delay the provisions or to hopelessly water them down.
5. Congress does that thwarting matters and wasting a lot of IRS time and resources.

That is what occurred with the expanded reporting of credit card receipts that Congress enacted in the ACA and then repealed in 2011 (Accounting Today article). In summary, it takes a long time and effort to implement these provisions, the benefits are often unclear or not obvious, and Congress is susceptible to listening to problems that occur at the margins and then, repealing the whole thing.

All that said, I think exploring the Rossoti and Forman proposal would be a good place to start rather than straightway raising rates. Something along those lines seems necessary to get at one of the problems identified in the second item I noticed.

Top income noncompliance

A new NBER paper, John Guyton, Patrick Langetieg, Daniel Reck, Max Risch, and Gabriel Zucman, Tax Evasion at The Top of the Income Distribution: Theory and Evidence (March 21, 2021), is out on noncompliance by taxpayers with the highest incomes. The first two authors are IRS economists, the other three academics.

The authors use IRS microdata to produce their estimates – combining data from (1) the IRS’s National Research Program or NRP, the series of random audits the Service uses to statistically measure noncompliance, (2) direct compliance audits, and (3) the quasi-amnesty (“offshore voluntary disclosure program” or VDP) offered before implementing FATCA, the 2009 legislation cracking down on nondisclosure of foreign bank accounts. The main take away is that noncompliance is larger than previously estimated because of offshore accounts and pass-through entities (PTEs) and this noncompliance is concentrated in the top income filers.

The NRP is an IRS program of intensive audits of a randomly selected, stratified sample of returns (oversampling high-income filers) to measure noncompliance based on filer characteristics and income type and amounts. (As an aside, the authors note the obvious about the stratification – it is done based on reported income. That misses filers who underreport their incomes, keeping them out of higher strata.) The results are adjusted statistically for known factors that distort the numbers, such as the visibility of income, difficulty and quality of auditing, and so forth, coming up with the Service’s best estimate of noncompliance, the “tax gap.” The NRP results are mainly used to target IRS compliance audits.

The authors take NRP adjusted results as their starting point (more or less – they replicate the formal IRS efforts themselves) and attempt to determine how accurate they are for the highest income filers.  Their premise is that lack of information on offshore accounts (pre-FATCA) led to underreporting of that income, and the difficulty for NRP auditors to track PTE income and expenses would lead to under detection. Both types of income are concentrated at the top and the standard adjustments to the NRP do not take this into account.

They next compare the results of operational audits with the NRP data as corrected. That is, with the audits IRS conducts to collect tax. These audits found more unpaid tax by the top filers than NRP’s corrected estimates suggested they would, even though only about 10% of these filers were audited. (That was before the decline in audit rates caused by reduced IRS funding, by the way.) Some of this effect is caused by the audits moving filers into the top 0.1% (i.e., adding to their reported incomes).

They, then, focused specifically on two potential areas of undercounting in the NRP and operational audit data: (1) income from offshore accounts and (2) income from PTEs.

Offshore accounts. For offshore accounts, they took returns from filers using the VDP amnesty or who started to disclose their foreign bank accounts after implementation of FATCA and matched it to the adjusted or corrected NRP returns. They found that NRP had detected only 7% of these filers. As they say (p. 18), “This finding suggests that NRP random audits seldom detected concealed offshore wealth.” The income was highly concentrated at the top with almost 7% of the top 0.1% in the potential evaders group. They find the concentration is even higher for “hidden wealth” and that the distribution of the accounts by size (i.e., a few large accounts represent most of the value; the median value is less than 10% of the average!). They suspect that most of the FBAR wealth assigned to the bottom 90% should be reassigned to top strata (p. 20). In the authors’ words, “The striking and non-obvious result of our computations is that even under very conservative assumptions about offshore evasion, taking this form of noncompliance into account implies large adjustments to detected evasion at the top.” (p. 25). Figure 4(c) (on page 19) shows how dramatically the distribution is skewed to the top.

A caveat is that FATCA likely has diminished these effects by reducing under reporting of income from offshore accounts. It’s hard to know how much. I think a consensus is that there is still a lot of underreporting. The Brockman and related cases tend to support that idea.

PTE income. NRP likely fails to detect much noncompliance from underreporting of PTE income – i.e., income reported on K-1s. K-1 information reporting is not equivalent to that on W-2s or 1099s for the simple reason that most PTEs issuing K-1s are not really third parties; they are reporting to their owners – essentially to themselves for closely-held PTEs (most of them). W-2s are issued usually by independent employers and 1099s by financial institutions.  The same tendency to underreport Schedule C income likely exists for K-1s issued by closely held PTEs (at least that would be my supposition). NRP audits show a lot of underreporting of Schedule C income. The authors note that less than 4% of NRP audits go upstream to the K-1s. The last NRP audit of PTEs was done in 1982. Thus, one would naturally conclude that there is likely a lot of underreporting (or evasion) by PTE owners and most of that income is concentrated at the top. The tax gap numbers, because they are based on adjustment of the NRP, likely miss much of that PTE income.

Unlike the offshore account estimates, the authors do not have hard data to observe the level of underreporting. That requires them to make assumptions or to extrapolate from underreporting of other types of business income. Their baseline assumption is that PTE underreporting falls between that for Schedule C income and by C corps, both of which are measured in the NRP. They use that assumption to estimate how much PTE income is underreported. The effects are quite large at the top (bigger than the offshore accounts effect). They move the top incomes from below average underreporting in the NRP numbers (1.7 percentage points below) to well above average (1.2 percentage points higher). The sensitivity tests (e.g., disallowing 20% of reported losses since disallowance of losses is often a result of compliance audits) suggest their base assumptions are conservative.

The authors combine the two effects (“sophisticated evasion” is their term) to show the overall effects. It is easiest to just quote their description of the three effects they find:

First, although our benchmark series feature only slightly more evasion on aggregate than in the standard IRS methodology, our proposed adjustments have large effects at the top of the income distribution. Specifically, our adjustments increase the aggregate income underreporting gap by a factor of 1.1, but by a factor of 1.3 for the top 1% and 1.8 for the top 0.1%. Second, in our benchmark series, we find clear evidence that under-reported income rises with income until at least the 99.95th percentile of the true income distribution. Third, across all our specifications, we find that accounting for unreported income increases the top 1% income share significantly in 2006–2013.

Tax Evasion at The Top of the Income Distribution: Theory and Evidence, p. 34.

The last section of the paper develops theoretical economic models of both taxpayer and tax administrator behavior to help explain “sophisticated evasion” and tax administrator’s allocation of audit resources between high and low-income filers. I will say it is nearly all Greek to me (and not because they use Greek letters in their equations), so I won’t try to describe or evaluate their models or what they predict. The models of tax administrator behavior are, as far, as I can tell ambiguous. That is, they may or may be a basis for the increased auditing of lower-income filers (e.g., EITC audits) that has been occurring.

Some of my general observations:

• The PTE numbers make a good case for expanded information reporting, perhaps along the lines proposed by Rossoti. The IRS will also need a lot of money and IT infrastructure to make good use of that type of data, though. As a reading of Rossoit’s proposal makes clear, information reporting on business income will be complex partially because of the issue of expense deductions (not the case for W-2 and 1099 income). That requires focusing on net flows, not a gross amount.
• The offshore accounts information is troubling because that transparently is outright cheating and borderline criminal. I don’t know if FATCA can be strengthen. It got the Swiss to finally knuckle under and loosen their bank secrecy rules. But I assume pure tax havens (island nation version) will never compel their financial institutions to report. Their “business plan” is, even if covertly, premised on enabling illegal evasion. They are essentially modern day “pirate ship sovereigns.” I know that the CARES Act finally required disclosure of LLC ownership information, which should help but may just drive tax miscreants from using Delaware and other state LLCs to offshore tax haven LLCs (without similar reporting) at slightly higher legal cost.
• I’m surprised that the authors use the term “evasion” since to me that implies knowledge of bad intent. Do not know what that is about exactly. It fits with many offshore accounts, I guess.
• The authors touch on the implications of their estimates for measures of income inequality. Their adjustments make an ugly picture uglier. For background see this post by Timothy Taylor, US Income Inequality, According to CBO (October 4, 2020), which reviews the latest edition of the annual CBO report on the topic. Taylor notes, quoting CBO, that business income is a large contributor: “As a share of income among households in the top 1 percent, business income rose from 11 percent in 1979 to 23 percent in 2017.” The NBER paper reveals (assuming its estimates are accurate) those numbers at the very top are understated because of noncompliance, making inequality worse. The distribution curve does not uniformly move up, but more so at its right tail. (Their estimates may say nothing about growth over time, though. The fact that business income has grown so much at the top may suggest they do.) As an aside, how underreported income affects the statistical measures of income inequality has been an issue in the back-and-forth debate over just how much worse inequality has become in the last 40 year or so (horrendous or not so bad). A debate that has made it all the way into the Strib opinion page (courtesy of D.J. Tice highlighting the possibility (wish?) that things may not be as bad as everyone thought, based on a 2018 paper). See the back and forth on the issue here Saez and Zucman (p. 26) (Saez and Piketty were the original protagonists) and David Splinter (pp. 2 – 4), a coauthor of the revisionist study (Autin and Splinter) finding much lower growth in inequality. Whatever one thinks about who is correct in that debate on how the data were or should be adjusted, neither reflect the big adjustments at the top that the authors find (note that Zucman is an author of the paper and a partisan in the debate with Splinter).

After I started working through the paper (it’s 78 pages of academic economicsese and charts and tables), the general media picked up on it (see Ingraham’s WaPo story and Hiltzik LA Times column for easier to read accounts and graphics of some of the paper’s data as well as additional commentary by the authors in the WaPo article). For another take on this study (actually more of a rift on his ideas for improving collections from the top income filers) see Steve Rosenthal’s post at TPC.