Using Machine Learning to Predict Outcomes in Tax Law

23 Pages Posted: 20 Oct 2016 Last revised: 19 Dec 2017

See all articles by Benjamin Alarie

Benjamin Alarie

University of Toronto - Faculty of Law; Vector Institute for Artificial Intelligence

Anthony Niblett

University of Toronto - Faculty of Law; Vector Institute for Artificial Intelligence

Albert Yoon

University of Toronto Faculty of Law

Date Written: December 15, 2017

Abstract

Recent advances in artificial intelligence and machine learning have bolstered the predictive power of data analytics. Research tools based on these developments will soon be commonplace. For the past two years, the three of us have been working on a project called Blue J Legal. We started with a view to understanding how machine learning techniques can be used to better predict legal outcomes. In this paper, we report on our experiences so far. The paper is set out in four parts. In Part 1, we discuss the importance of prediction. In many fields, humans are outperformed by mechanical and algorithmic prediction. We explore this phenomenon and conclude that the legal field is no different. In Part 2, we discuss recent advances in machine learning that have generated powerful tools for prediction. These new methods outperform traditional statistical techniques in predicting outcomes. In Part 3, we describe the Blue J Legal project. We discuss how Blue J Legal is using these machine learning technologies to provide predictions in grey areas of tax law. We provide a number of examples to illustrate the strength of these predictions. In part 4, we discuss the broader possibilities for technologies such as those powering Blue J Legal. We foresee a world where information about legal rights and responsibilities is more affordable; where the informational asymmetries that lead to wasteful expenditure on litigation is reduced; and where regulators use these tools to create a more effective and efficient administration of government. A final section concludes.

Keywords: Machine Learning, Artificial Intelligence, Taxation, Tax Law

JEL Classification: C18, C19, H20, H83

Suggested Citation

Alarie, Benjamin and Niblett, Anthony and Yoon, Albert, Using Machine Learning to Predict Outcomes in Tax Law (December 15, 2017). Available at SSRN: https://ssrn.com/abstract=2855977 or http://dx.doi.org/10.2139/ssrn.2855977

Benjamin Alarie

University of Toronto - Faculty of Law ( email )

Jackman Law Building
78 Queen's Park
Toronto, Ontario M5S 2C5
Canada
416-946-8205 (Phone)
416-978-7899 (Fax)

HOME PAGE: http://www.benjaminalarie.com

Vector Institute for Artificial Intelligence ( email )

Anthony Niblett (Contact Author)

University of Toronto - Faculty of Law ( email )

78 and 84 Queen's Park
Toronto, Ontario M5S 2C5
Canada

Vector Institute for Artificial Intelligence ( email )

Albert Yoon

University of Toronto Faculty of Law ( email )

78 and 84 Queen's Park
Toronto, Ontario M5S 2C5
Canada

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
2,245
Abstract Views
7,788
Rank
12,439
PlumX Metrics