State of Texas Recovers $400 million Through Predictive Analytics
All states face compliance regulations when it comes to taxes. Tax gaps usually arise between the tax owed and the amount collected, which is the thrust behind tax audits: to reclaim that missing money. The state of Texas took a proactive approach to reducing its tax gap by implementing SPSS predictive analytics technology to aid in the process.
The state realized a need to leverage data in order to improve the audit selection process, and so initiated what it calls the Advanced Database System project, which utilizes SPSS predictive analytics. “Historically, the [state] agency had relied upon simple selection criteria for choosing which taxpayers should be audited in a given year,” explains Daniele Micci-Barreca, a principal with Elite Analytics, serving the state of Texas in an analytics consulting role. “They wanted to be able to leverage the information in the warehouse in order to make better decisions and increase the percentage of productive audits,” he adds.
Two primary data-driven criteria determine audit selection for the state. One, Micci-Barreca says, is based on the total reported taxes. The other is based on prior productive audits, meaning audits that generate a tax assessment above a certain threshold. “The rest of the audits were selected based on the field auditor’s own experience and intuition,” Micci-Barreca points out.
The use of analytics helps the state predict models for audit selection and prioritize models for non-tax filer discovery. “While non-filer discovery was done prior to the deployment of the data warehouse, the new technology enabled the agency to pursue this activity on a much larger scale and with a higher degree of efficiency,” Micci-Barreca asserts. “With the data warehouse, the agency can cross-match millions of records, and with predictive analytics, they can focus on the most promising leads.”
As a result of implementing predictive analytics, the state of Texas has recovered $400 million in unpaid taxes over a five-year period. During this timeframe, the ADS helped identify thousands of businesses that were operating in the state without complying with the tax obligations, Micci-Barreca explains. “However,” he adds, “this [$400 million] figure does not include additional benefits that came from the adoption of better selection methods for field audits, which is the primary source of tax dollars recovery for the agency. Last year, total tax assessment for sales tax audits alone accounted for nearly $150 million in Texas.”
The state’s most significant challenge in implementing ADS was not technical but cultural. “Processes like audit selection rely on well-established methods, and it isn’t trivial to convince field offices to follow the recommendation derived from some complicated mathematical formula,” Micci-Barreca notes. It took time for the field selectors to become confident in the tool.
For organizations considering data mining and/or predictive analytics technologies, Micci-Barreca advises, “The ROI potential for this type of project is phenomenal: tax administration agencies are only capturing a fraction of the so-called tax gap, and the answer to improve compliance with better non-filer discovery, audit selection collection strategies is in the mountains of data that often simply accumulate in legacy systems. While the technology behind predictive analytics can be very complex, once these models are deployed they can integrate seamlessly with existing operations and drive better decision making.”