Stopping Identity Fraud in UI Before it Happens: Innovating with Behavioral Science

In our current series on preventing improper payments, we’ve discussed why confusion, stress, and self-deception can contribute to UI claims issues. Below, we explore premeditated fraud in UI claims, specifically identity fraud. Whether you work in fraud, claims, or other departments, we hope the thought experiment and COM-B tool can help unlock your own ideas—and help prevent fraud in your state workforce agency.

Picture yourself as a behavioral scientist working with law enforcement – a criminal profiler of sorts. Your new assignment is to help prevent identity fraud in state unemployment systems by “getting in the heads of” would-be criminals and diverting them before they commit a crime.

You are handed a stack of documents. You read how identity fraud cases in UI, especially during the pandemic, have been hard to close and have cost billions annually. You read about how new technological and procedural efforts are helping to catch fraudsters after the fact, such as by validating UI claims data in multiple systems and enhancing claims audit procedures. However, fewer solutions stop fraudsters from acting in the first place.

To get in the heads of fraudsters, you read this case example:

The case gives you a specific focus: what low-cost, easy-to-implement solutions can you invent to stop people like Brenda? How do you even begin?

Behavioral science gives us some options. To understand any behavior, a good place to start is by making a map. A framework favored by behavioral scientists is the “COM-B system,” or the Capability, Opportunity, and Motivation Behavioral System. COM-B is based on a successful method used by crime investigators; you may recognize it better as “means, motive, and opportunity.” (See here for more information.)

The COM-B system aims to map a decision against the three eponymous criteria (capability, opportunity, and motivation), which are each further divided into psychological and physical capability, social and physical opportunity, and reflective and automatic motivation.

The model is helpful as a brainstorming tool, as it pushes us to consider psychological and environmental factors that lead to a decision. Other models work similarly but are less expansive. For example, the traditional rational decision model (a go-to for many economists and criminologists) prompts us to only think in terms of costs and benefits and thus leaves out a lot of useful context.

To map Brenda’s decision using the COM-B tool, we can follow these four steps.

1. First, define the questions. Following each of the categories of the model, we create targeted questions to think through the context, such as in the example below.

2. Next, answer the questions. We respond to these questions by brainstorming “enablers.” These enablers offer a full picture of what is going on inside Brenda’s head and in her environment that may be leading her to commit identity fraud. Below we add some examples to our chart.

3. Then, brainstorm solutions. We can determine solutions that address each of these enablers. While not all solutions may be worthwhile, we will have a menu of options that directly or indirectly target the enablers of fraud. We add a few possible solutions to the chart below. Notice many of these solutions can be implemented as informational warnings when filing claims.

4. Last, choose a solution to implement and test. The above solutions are low-cost and likely to have a modest impact on fraud (based on existing behavioral science research). We cannot, however, take for granted that any solution gets the results that we expect. For example, recent research shows that while some of the tips on fraud protection are effective (like shredding documents), some are not (like checking your credit history regularly). Trust, but verify!