Fighting Improper Payments

IBM-logo-promoLast week I visited Capitol Hill to discuss with lawmakers how better data can help improve both operational and program efficiency. Over the next few blogs, I will discuss a few of these highlights. One of the most interesting conversations I had that day involved a topic on which, believe it or not, members from both sides of the isle seemed to agree. Nonetheless, even with this agreement in principal, this issue remains one of the largest solvable threats to the federal government’s fiscal responsibility. This threat is improper payments.

In a world of tight budgets and pushes for continued fiscal austerity, it makes sense to take a look at an area that if solved could save the federal government over $100 billion annually – improper payments. Although progress has been made in plugging this fiscal hole in the operation of the federal government, there is still a long way to go. OMB itself declared that improper payments are the “biggest government problem,” with an estimated $108 billion in government-wide improper payments made in 2012.

According to the federal government, “improper payments” occur when:
• Funds go to the wrong recipient
• The right recipient receives the incorrect amount of funds (including overpayments and underpayments) documentation is not available to support a payment
• The recipient uses funds in an improper manner

Fortunately, there seems to be increased agreement amongst lawmakers and government executives alike on the need to measure improper payments and reduce its waste and fraud. In evaluating improper payments, there are two approaches that should be employed, in parallel, to solve this problem:

Operational effectiveness: Traditionally, organizations have relied on “internal controls” to fight operational errors and fraud. There has been much attention, intellectual capital, and progress in using traditional controls to root out improper payments. Tight internal controls are necessary in any organization to ensure a base level of compliance. There are fairly mature methods in implementing and testing these controls. Unfortunately, internal controls are not sufficient to prevent all improper payments. Internal controls are the equivalent of locking your doors while you are away from home. This might deter criminals, but seasoned, determined criminals are always looking for new advanced ways to break in.

Advanced Analytics: Traditional internal controls are not sufficient to solve the problem or improper payments. One government official stated to me that “we are in an analytics arms race against those wishing to defraud the government and its people”. Meaning that those looking to defraud the government are using data and analytics to find new ways of commiting fraud all the while going undetected. And the only way to fight this is to apply advanced analytics as a part of the solution. Applying advanced analytics to the problem of improper payments utilizes large quantities of inter-agency data to look for predictive, non-obvious patterns in data to estimate a probability of fraud or improper payment and deliver this probability to decision-makers for real-time improper payment prevention and investigation.

Big Data as an Enabler
In addition to analytics providing a solid foundation to mitigate the problem of improper payments, recent advancements in big data technology further empower advanced analytic techniques. Until recently, the traditional approach to applying analytics to problems such as improper payments involved business users determining what questions to ask and IT stakeholders structuring the data to answer the questions. Now, with the new big data paradigm, IT delivers the platform to enable creative discovery such that business users can explore what questions to ask. It is this exploratory analysis that empowers business users to find and operationalize the analytics to root out improper payments.

Whereas the traditional approach uses structured and repeatable analytics such as queries on data at rest, the big data approach involves iterative, exploratory and, autonomic analytics on data at rest and in motion enabling insight to drive the answers — not Just IT. Because of these shifts and challenges in the analytics landscape, public sector stakeholders are empowered to think and work differently.

Effectively Utilizing Enterprise Data Models
Effective organizations conduct analytics with data that spans multiple environments. While in most cases, the data is already valuable on its own, the real transformative effect of the data comes when these data is combined with other data to get a 360° programmatic view. Many federal regulatory authorities are looking to analyze the whole body of data, structured and unstructured, to provide better evidence-based service with better outcomes at lower costs.

Unfortunately, at this point in fraud and claims analysis, information is coming in faster than organizations’ infrastructure can typically handle. Therefore, organizations typically conduct post-payment fraud analysis in batches, long after the ability to actually recoup the losses. Being able to capture and analyze that information as it comes in is important to reduce improper payments. Nonetheless, the influx of information makes it progressively more difficult to distinguish between essential data and clutter – a fact relished by perpetrators of fraud.

Building enterprise big data models that span organizational boundaries and are empowered by advanced analytics are helping government organizations that are hardest hit by improper payments. For example, the IRS has used data shared by the Social Security Administration and Medicare to stop identity theft leading to improper payments of tax refunds.

The Role of Advanced Analytics
To assist government organizations better understand their payment streams and root our improper payments, our team at IBM has developed an innovative advanced analytics approach to manage improper payments through prevention and early intervention, even in the context of a changing program environment. At a high level, this approach employs an organization / program-specific scoring system to target likely error-prone and potential improper payments, thus allowing for early detection prior to disbursement—without impeding access and/or unduly delaying payment delivery to the proper recipients.
Benefits of this approach include:
• Identifying and resolving errors prior to payment being made
• Identifying errors that are not obvious in the application data
• Allowing investigative resources to focus on the most likely errors

Furthermore, this provides investigative resources with a description of what to look for. IBM’s approach integrates risk scoring with data from various source systems and deploys the solution on the program’s infrastructure as a part of the payment stream. Therefore, this system can be built into existing rules, with triaging at the transactional level and could use the same system for future integrations into other native systems so that improper payments could be identified on-line or in real-time.

About Brian Murrow
As a leader in IBM’s Business Analytics and Optimization team, Brian’s business analytics experience includes designing and implementing capital adequacy models, decision support systems, actuarial reviews, economic impact assessments, new financial product assessment models, performance measurement programs, business operations reporting processes and systems, opinion research analytics, and portfolio optimization.