Using AI to Characterize Financing Between Related Parties

Abstract

The debt-versus-equity question is largely relegated to the common law. The central inquiry is the “extent to which the transaction complies with arm’s length standards and normal business practice.” The courts have established a list of recurring factors to determine whether a transfer should be characterized as a debt interest between the contributing party (the lender) and the receiving business (the borrower) or whether the transfer resulted in an equity interest between the contributing party (the holder) and the receiving corporation (the issuer).

We show how it is possible to identify patterns within a data set of past debt-versus-equity cases by applying machine learning algorithms. By comparing two recent decisions involving related parties, one with a debt outcome and the other an equity outcome, we illustrate the relative significance of various factors as they pertain to related parties. The presence of formal indicia of debt such as a loan document is not significant in relation to other factors that address the substance of the transaction as opposed to the way it was papered. The existence of enforcement rights is slightly more significant, but how the parties behaved in terms of any repayment is the most significant out of the three.

Machine-learning-powered systems can allow lawyers to make more confident and efficient predictions based on all the relevant information. And while there remains some anxiety about the disruptive potential of artificial intelligence for the legal field, it is important to recognize that machine learning is not a replacement for the judgment of human lawyers. Instead, it is a powerful new tool to augment their professional knowledge and instincts.

Keywords: Tax, Artificial Intelligence, Machine Learning, Tax Avoidance, Debt, Equity, Financing

JEL Classification: H2