Robotic Process Automation and Tax Administration: Examining Theoretical Applications for Efficiency, Accuracy, Compliance, and Administrative Cost Reduction

Abstract

This paper examines robotic process automation's documented outcomes across six national tax
administrations, drawing on case evidence compiled in the OECD's Tax Administration 2022 report
and independently verified through the administrations' own published factsheets. The evidence
reveals a striking, task-dependent range in reported efficiency gains: the Israel Tax Authority's
automation of multi-employer tax liability flagging, a structured, rule-based task, achieved a 97
percent reduction in average time per case and a 99.6 percent reduction in cost, while the Australian
Taxation Office's application of RPA within its judgment-intensive audit process achieved a
considerably more modest 5 percent reduction in manual actions. Intermediate cases include the
Canada Revenue Agency, which automated eight processes and saved more than 34,000 staff hours,
equivalent to approximately twenty-five additional full-time employees; the United States Internal
Revenue Service, which completed 1,500 contract modifications in 72 hours using automation, a
task manual processing would have required approximately one year to complete; and the
Netherlands Tax Administration, which achieved four hours of daily time savings per employee
through invoice-processing automation. The OECD further reports that more than half of surveyed
tax administrations were using or planning to use RPA as of 2022. Interpreted through the
technology-organization-environment framework and this paper's companion research on
algorithmic audit selection, the paper argues that RPA returns are systematically larger for
structured, rule-based tasks and considerably more modest for tasks embedding genuine
professional judgment, with direct implications for how tax administrations should prioritize
automation investment.