Algothrims in Service of Tax Compliance (10/30/15)

I missed this article when it came out earlier this month. Lynnley Browning, Computer Scientists Wield Artificial Intelligence to Battle Tax Evasion (NYT 10/9/15), here.  Lynnley Browning is an excellent observer of the tax avoidance / fraud scene.  Her articles are worth reading.

Readers of this blog are aware if the traditional means for identifying tax evasion -- such as civil audits, traditional informants (such as disgruntled special friends, even wifes), etc.  But now, algorithms can be deployed to large available data sets where, without the algorithms, it might be hard to perceive patterns and correlations that can identify the high potential for tax avoidance or evasion.

Here are some relevant excerpts:

“We see the tax code as a calculator,” said Jacob Rosen, a researcher at the Massachusetts Institute of Technology who focuses on the abstract representation of financial transactions and artificial intelligence techniques. “There are lots of extraordinarily smart people who take individual parts of the tax code and recombine them in complex transactions to construct something not intended by the law.”
A recent paper by Mr. Rosen and four other computer scientists — two others from M.I.T. and two at the Mitre Corporation, a nonprofit technology research and development organization — demonstrated how an algorithm could detect a certain type of known tax shelter used by partnerships. 

First, the researchers translated tax regulations governing partnerships, a growing source of tax trickery, into source code. Then they rendered the transactions underpinning a questionable shelter known as “installment-sale bogus optional basis,” or Ibob, as a series of codes. The Ibob shelter artificially inflates the basis value of an asset on a tax return to wipe out taxable gains when that asset is sold. While some of Ibob’s individual transactions are perfectly legal, the collective result is a bogus deduction. 

Next, the researchers mapped out in code the tangle of entities that make up typical partnerships. The results flagged specific combinations of transactions and partnership structures that were likely to produce the Ibob dodge. 

Large corporations attract most of the attention when it comes to tax avoidance and tax evasion, but partnerships, which have separate tax rules, are a growing source of worry for the authorities. Commonly used by hedge funds, private equity funds, real estate outfits and oil and gas concerns, partnerships are far less likely to be audited than corporations. A Government Accountability Office r eport from 2010 said that the I.R.S. knew of one million “networks” involving partnerships and similar entities, adding that “the I.R.S. also knows that many questionable tax shelters and abusive transactions rely on the links among commonly owned entities in a network.” 

Rooting out fraud in corporate tax returns takes place largely through data mining, in which the I.R.S. collects pre-existing data from filed tax returns and analyzes them for patterns. The data goes into a database within the agency’s Office of Tax Shelter Analysis, created in 2000 in the wake of a crackdown on mass-market tax shelters sold by accounting firms, law firms and banks. The data-analytical approach depends upon already having some kind of smoking gun, such as a suspicious deduction on a return.
By contrast, the artificial intelligence approach does not require pre-existing evidence. Instead, it focuses on rule mining, in which individual tax code regulations are lined up against one another to ascertain if they can be used collectively to create a sophisticated tax dodge. 

Rule mining takes advantage of a surprising feature of tax shelters: While their inner workings are convoluted and complex, their general aim at the highest level is usually simple and clear — to lower tax bills by improperly generating bogus losses, deductions, offsets and credits that minus the shelters would not exist. 

“It’s incredibly difficult to have a computer algorithm that duplicates the enormous creativity of taxpayers, but it’s very promising,” said Robert A. Green, a tax professor at Cornell Law School who read the M.I.T./Mitre paper.

The ibob shelter was identified as follows in a 2010 GAO report, here.

IRS needs an agency-wide approach for addressing tax evasion among the at least 1 million networks of businesses and related entities

Why Area Is Important 

At least 1 million networks involving partnerships, trusts, corporations, and similar entities existed in the United States in tax year 2008. These networks can serve a variety of legitimate business purposes. However, transactions made among related entities within networks also can be used in tax evasion schemes to hide taxable income or shift expenses. Such schemes—such as the one described in the text box below—result in lost tax revenue and are difficult for the Internal Revenue Service (IRS) to identify, due to data limitations. 

IRS recognizes the risk from network-related tax evasion and is developing new tools and programs to better identify such evasion. These IRS efforts are in various stages of development, but their potential effectiveness in terms of cost savings or added revenue, is not known. However, GAO has identified the need for additional efforts to strengthen enforcement in the networks area and to assess progress.

What GAO Found 

IRS knows that many questionable tax shelters and abusive transactions rely on the links among commonly owned entities in a network, but it does not have estimates of the associated revenue loss in part because data do not exist on the population of networks. IRS generally addresses network-related tax evasion through its examination (audit) programs. These programs traditionally involve identifying a single return from a single tax year and routing the return to the IRS division that specializes in auditing that type of return. From a single return, examiners may branch out to review other entities if information on the original return appears suspicious. However, this traditional approach does not align well with how network tax evasion schemes work. Such schemes can cross multiple IRS divisions or require time and expertise that IRS may not have allocated at the start of an examination. A case of network tax evasion also may not be evident without looking at multiple tax years. 

Network Scheme Example: Installment Sale Bogus Optional Basis Transaction (iBOB)

An iBOB is an example of a network-related tax evasion scheme that shows how networks pose enforcement challenges for IRS. In an iBOB, a taxpayer uses multiple entities, all owned or controlled by the taxpayer, to artificially adjust the basis of an asset to evade capital gains taxes. The scheme can involve multiple transactions and take place over many tax years, making it difficult for IRS to detect. A short video illustrating the iBOB is available at http://www.gao.gov/products/GAO-10-968.

 IRS is developing programs and tools that more directly address network tax evasion. One, called Global High Wealth Industry, selects certain high-income individuals and examines their network of entities as a whole to look for tax evasion. Another, yK-1 (sic?), is a computerized visualization tool that shows the links between entities in a network. These efforts show promise. They represent new analytical approaches, have upper-management support, and cut across divisions and database boundaries. However, there are opportunities for more progress. For example, IRS has no agencywide strategy or goals for coordinating its network efforts. A strategy would include assessing of IRS's network tools and determining the value of incorporating more data into its network programs and tools—neither of which IRS has done. Without a strategy and assessments, IRS risks duplicating efforts and managers will not have information about the effectiveness of the new programs and tools that could inform resource allocation decisions. 

Actions Needed 

GAO recommended in its September 2010 report that IRS create an agency-wide strategy with goals to coordinate and plan its enforcement efforts on network tax evasion. The strategy should include assessing the effectiveness of network analysis tools to ensure that resources are being devoted to those that provide the largest return on investment; determining whether to increase access to IRS data or collect new data for network analysis; developing network analysis tools on a specific time schedule; and deciding how to manage network efforts across IRS. IRS should ensure that its staff understand the network tools and establish formal ways for users to interact with tool programmers and analysts to ensure that the network tools are easy to use and achieve goals. IRS agreed with GAO's recommendations and said it would make plans to take actions on them but it is too early to determine IRS's progress. 

Estimates are not available on the potential for increased tax revenues because IRS has not measured the potential impact of its network efforts on reducing tax noncompliance due to data limitations, but these efforts have significant potential, based on the number of networks that exist.