Decision Analytics

	Decision Analytics is the application of statistical techniques against customer-based data which, in turn, derive probabilities and allow agencies to make more educated decisions. Used for decades in other industries such as banking, finance and even sporting events, Decision Analytics is now being implemented in tax administration, with the aim of increasing taxpayer compliance, improving audit selection, reducing fraud and maximizing an agency’s resource utilization. The initial application of Decision Analytics has been in collections where, by determining a taxpayer’s probability of paying, cases can be organized, prioritized and assigned to appropriate treatment strategies to increase collections. Why call someone who would pay upon receipt of a letter? Why call someone if they will never pay without enforcement actions? By using models that score taxpayers based on truly predictive characteristics, appropriate collection actions can be taken immediately upon delinquency. These characteristics may or may not be obvious. Figure 1 shows the output from a predictive model, developed for collection purposes. A taxpayer’s probability of paying in full clearly varies by score, increasing as the score increases. This enables the score range to be divided into a number of “risk” categories, such as low-risk, medium-risk and high-risk, each of which has markedly different average probabilities of paying. Consequently, taxpayers in each risk category can now be treated, from the beginning, very differently from each other. Such differentiation in how taxpayers are collected is enabled by the development of predictive models. Further, a probability can be applied that provides managers with a high degree of confidence in the outcome (e.g., 90% of the time, a taxpayer with this score will pay in response to the letter). Some tax agencies have implemented analytical audit selection solutions to choose the best audit candidates (e.g., audits that will yield an assessment greater than $5000). Figure 2 presents an example of the output of such a model, superimposed with the distribution of actual audits. Clearly, the distribution of actual audits is sub-optimal, peaking in the middle of the score range, where the probability of a successful audit is in the 60% - 70% range, rather than peaking in the higher score ranges where the probabilities of a successful audit are also higher. Because revenue agencies have more collection cases and potential audits than people to work them, Decision Analytics helps prioritize the workload, enabling staff to always work the next best case. Massachusetts and South Carolina are two of RSI’s clients that have recently implemented both collection and audit Decision Analytic solutions.
Collections Distribution Figure 1: This figure shows the output from a predictive model, developed for collection purposes. A taxpayer’s probability of paying in full clearly varies by score, increasing as the score increases.
Analytical Taxpayer Compliance Management The application of Decision Analytics and Business Intelligence is ushering in a new approach to tax compliance that RSI calls Analytical Taxpayer Compliance Management (ATCM). ATCM utilizes Decision Analytics to drive interactions with taxpayers to maximize compliance. By identifying and understanding the differences in the taxpayer population, auditors and collectors can make decisions with a higher probability of outcome when reviewing work inventories. ATCM extends analytics from predictions through decisions to results. Analytics creates “actionable intelligence” which, when acted upon, produces measurable results. RSI’s approach is to integrate data sources into “portfolios” for every taxpayer, which ensures that “intelligence” reflects everything known about each taxpayer. Integrated data, plus actionable intelligence, enables the assignment of taxpayers to the right type of compliance action at the right time. ATCM selects cases for compliance action objectively. Selection reflects each taxpayer’s prior behavior, as measured by predictive behavior models. The probability of non-compliance, together with an agency’s tax policy, determines the compliance action to which the taxpayer is assigned. Combining statistical probability and tax policy ensures that assignment to actions doesn’t rely on the skills or prejudices of agency staff - it becomes entirely objective and, thus, fair. ATCM is about being more effective. At the staff level, effectiveness is achieved by focusing on the right cases. In Audit, this means focusing on candidates with high assessment probabilities. In Collections, collection agents would focus more on “medium-risk” cases (cases that are tractable but which require working) and thereby maximize the return on their efforts. In fact, any case inventory in an agency (such as refund fraud or abatements) can be examined to determine which cases require manual intervention. The days when “refunds less than $500 get automatically approved” are over. RSI believes that any “judgmental” rule embedded in business processes should be evaluated and replaced with statistically based decisions. Finally, by enabling all units to focus on non-compliance risk, the agency becomes more effective at improving voluntary compliance and increasing enforcement driven revenues.
Sales Tax Audits Figure 2: This figure presents an example of a model’s output, superimposed with the distribution of actual audits. Clearly, the distribution of actual audits is sub-optimal, peaking in the middle of the score range, where the probability of a successful audit is in the 60% - 70% range, rather than peaking in the higher score ranges where the probabilities of a successful audit are also higher.
ATCM Benefits ATCM produces significant benefits, such as: Increased revenues. Risk-based collection implementation can raise collections by tens and, in larger states, hundreds of millions of dollars. Fifteen to twenty percent increases in collection effectiveness is the benchmark. In Massachusetts, a lift greater than 20% was observed. South Carolina realized benefits in the millions of dollars, just months after implementation. When applied to audit selection, increases in assessments per auditor hour can exceed twenty percent. At one client, almost half of the cases under audit scored low to medium. After transitioning auditors to the high risk taxpayers, audit assessments increased dramatically. Clearly, risk-based collection and audit selection implementations produce substantial returns on investment! Increased employee productivity and morale. Decision Analytics enables better case selection assignment, allowing staff to focus on cases that produce results. Thus, auditor, collector and examiner resource utilization is maximized and productivity increased, which results in higher morale and job satisfaction. Improved customer service. When low-risk collection cases are allowed to self-cure, and when low-probability audit candidates are not selected for audit, compliant taxpayers are not bothered, resulting in improved customer service and increased customer satisfaction.
Performance Improvement Measures After Using Decision Analytics to Select Audits Figure 3: Significant performance improvements can be obtained by using Decision Analytics to select audit leads.
Revenue Solutions, Inc. Can Help Since 1996, our mission has been to help government agencies realize their business objectives by getting the most out of their information technology investments. In that time, over 25 federal, state and local government agencies have relied on RSI’s extensive expertise in both business and technical domains. Drawing on specialized business processing experience across multiple disciplines, as well as IT best practices, RSI is able to shorten and straighten the path to implementing change. RSI’s experts can assist your agency with the entire compliance management life cycle: from developing compliance plans through data aggregation and data warehouse creation, scoring, lead selection, case management in audit and collection systems, to benefits tracking and management reporting. RSI’s staff is skilled in all business and technical aspects of compliance business-process improvement. In particular, RSI’s staff includes extensive expertise in the development and application of predictive models, decision strategies and treatment scenarios – core components of any ATCM implementation – in audit and collection improvement projects.