On Complexity and Algorithms in Tax Administration (9/8/21)

The Procedurally Taxing Blog has a good post this morning, Bob Probasco (Guest Blogger), Complications from Extensions and Unprocessible Returns (Procedurally Taxing Blog 9/8/21), here.  It delves into the interaction of  (i) the Beard test for a valid return (Beard v. Commissioner, 82 T.C. 766 (1984), affd. 793 F.2d 139 (6th Cir. 1986)), (ii) the differences between a Beard-valid return and a processible return, and (iii) the arcana of interest computations based on those differences and the interaction of § 7508A disaster relief.  Bob does a great job weaving through that stuff, so I won’t even try to summarize it here (too great a risk of showing my ignorance).  Rather, I found Bob’s conclusion in this dance through tax arcana should resonate with students of the tax law and tax procedure:

Two Final Thoughts

            First, I think the PMTA is clearly right under the Code.  That the IRS originally reached the wrong conclusion, in part, is a testament to the complex interactions of the different provisions and the need for close, attentive reading.  I double-checked and triple-checked when I worked my way through the PMTA.  This was actually a relatively mild instance of a common problem with tax.  Code provisions are written in a very odd manner.  They’re not intended to be understandable by the general public; they’re written for experts and software companies, and sometimes difficult even for them.  I suspect that people who work through some of these complicated interpretations would fall into two groups: (a) those who really enjoy the challenging puzzle; and (b) those who experience “the pain upstairs that makes [their] eyeballs ache”.  My bet is that (b) includes not only the general public and most law school students but also a fair number of tax practitioners.  Which group are you in?

            Second, that (relative) complexity leads to mistakes.  I assume that these interest computations have to be done by algorithm.  There are simply too many returns affected to have manual review and intervention for more than a small percentage.  An algorithm is feasible but will require re-programming every time there’s a section 7508A determination, with changes from year to year.  Even when the law is clear, there is a lot of opportunity for mistakes to creep in.  (We’re seen some of those recently in other contexts, e.g., stimulus payments and advanced Child Tax Credit.)  Whether by algorithm or manual intervention, particularly given the change from the original conclusions, there’s a good chance that some refunds were issued that are not consistent with the correct interpretation and may not have included enough interest.  Is the IRS proactively correcting such errors?  If the numbers are big enough, it might be worth re-calculating the interest you received—it’s easier than you may think—and filing a claim for additional interest if appropriate. 

 JAT Comments:

1.  Bob recommends that persons looking for overpayment interest might want to pay attention to these complexities.  He asserts -- "it's easier than you may think."  My experience as a long-time tax controversy/litigator lawyer is that interest calculations are not my cup of tea.  Where interest calculations were important for some reason, I passed the work to professionals, usually accountants, who knew what they were doing.  I cover this issue at the conclusion of the Interest Chapter in my Federal Tax Procedure Book at Student Edition, p.  229, here, and Practitioner Edition p. 317, here, under a section titled "Who Really Does This Type of Work?"

2. In preparing this blog entry, I focused on word arcana in the opening paragraph.  In the past, for some reason lost in the smog of time, I had used the word “arcania” which I intended to  mean the same as arcana.  I find, however, that arcania is not the right word.  (E.g., Urban Dictionary definition here.)  That was my error.  I have corrected my malapropism in my blog entries and in the working draft of the 2022 Federal Tax Procedure Book.