Predicting and preventing tax fraud with analytics

The Swedish Tax Agency uses SAS® to identify risk profiles, pinpoint debtors and increase revenue

The Swedish Tax Agency has two important objectives: to ensure that taxes are correctly reported and that the resulting tax debt is then paid. Using new analytics solutions, the agency can now identify risk profiles more easily, thereby preventing and stopping tax fraud and neglectful tax returns.

Like many other public sector organizations, the Swedish Tax Agency is facing a future of diminishing resources. "This increases the need for automated solutions and requirements for better resource management through well founded risk analysis," says Andreas Voxberg, Analyst/Section Coordinator at the Swedish Tax Agency's Analysis Unit.

We have high tax compliance in Sweden, both in terms of reporting and paying it," explains Voxberg's colleague, Analyst and Statistician Joacim Danielsson. "This is underpinned in part by good system solutions where, for instance, employers pay their employees' preliminary tax and submit a statement of income to the Swedish Tax Agency. Even so, the aim is still to ensure there are as few discrepancies as possible."

In Sweden, the tax discrepancy, defined as the tax that should be reported but isn't, is estimated to be roughly 10 percent of the theoretically correct tax. One of the Swedish Tax Agency's goals is to half the tax discrepancy, and one of the Analysis Unit's tasks is to identify anything that could jeopardize meeting this goal.

We need to identify the non-payers as early as possible. Using predictive models, we can identify risk profiles and indicators so that we can focus where the risk of non-payment is greatest.

Joacim Danielsson
Assistant Project Manager, Analysis Unit, Swedish Tax Agency

To reduce the percentage of tax errors, the agency has endeavored to make life easier for tax payers wherever possible, gradually simplifying its tax returns. It has also introduced various control measures and conducted information campaigns.

"We have published information about some of our controls online, so that individuals and companies can run checks themselves before submitting their returns and avoid being flagged in our control system. This saves time for them as well as for the agency," Voxberg explains. The process of identifying tax returns with the highest risk of errors takes place within the framework of the Swedish Tax Agency's selection project, for which Voxberg and Danielsson are project manager and assistant project manager, respectively. All tax returns are scrutinized. They undergo an automated check, and some are also selected for a more in-depth examination according to certain criteria which may vary from year to year.

"All income and tax returns, from individuals and companies alike, pass through an auditing net, which is set up by the Swedish Tax Agency's selection project. It is the project's task to identify risks at return level, based on the operational plan and the various activities therein. The risk analysis at a more general level is dealt with by the agency's Analysis Unit, and is used as a basis for decisions on what measures should be implemented to deal with the risk groups," says Voxberg.

To produce as reliable risk profiles as possible, the agency uses analysis solutions such as data mining and statistical analysis. According to Voxberg, this is not simply a means of "rating" tax payers, rather it is a model for pinpointing high-risk behavior.

Thanks to the agency's new analysis solutions, it has been able to broaden its risk examination from the component to the general level. Voxberg and Danielsson regard the move from company level to network level as the next fascinating challenge. Putting the risk into a broader context – a network – is a way of building a more substantial picture.

The aim is to further complicate tax fraud. When it comes to the risk of non-payment of tax, this has become an increasingly important area, particularly in the wake of the credit crunch. Tax payment compliance is at 99.7 percent, which means that only 0.3 percent is not paid.

It's more about maintaining that level," says Voxberg, while noting that even 0.3 percent of Sweden's total tax revenue still equates to billions of kronor. "We need to identify the non-payers as early as possible. Using predictive models, we can identify risk profiles and indicators so that we can focus where the risk of non-payment is greatest," says Danielsson.

Along with two master's students from Copenhagen Business School, the Swedish Tax Agency is currently going through the extensive research literature, so that, if possible, it can improve its credit risk models with further indicators of inadequate solvency at both a company and a macroeconomic level. Danielsson can see many benefits in the improved risk analysis, in addition to increased efficiency for the Swedish Tax Agency:

It is good if we can find any debtors and reach out to them early so we can help out, and they can pay before the matter is referred to the enforcement authorities. Obviously it's also a plus for society in general if we have a stronger tax foundation to fund welfare initiatives, for instance."


Swedish Tax Agency required an automated solution for predicting and preventing tax fraud.


SAS® Fraud and Improper Payments


Early detection of fraud and the ability to reach debtors more quickly has increased tax revenue in Sweden.

The results illustrated in this article are specific to the particular situations, business models, data input, and computing environments described herein. Each SAS customer’s experience is unique based on business and technical variables and all statements must be considered non-typical. Actual savings, results, and performance characteristics will vary depending on individual customer configurations and conditions. SAS does not guarantee or represent that every customer will achieve similar results. The only warranties for SAS products and services are those that are set forth in the express warranty statements in the written agreement for such products and services. Nothing herein should be construed as constituting an additional warranty. Customers have shared their successes with SAS as part of an agreed-upon contractual exchange or project success summarization following a successful implementation of SAS software. Brand and product names are trademarks of their respective companies.

Back to Top