Without these data points, your predictive risk modeling will fail

Abby Progin – Vice President – Product Management

October 7, 2020

When a new economic threat emerges, traditional risk models that focus on historical data and aggregated risk models lose their predictive value. Most credit unions have enhanced their risk assessments this year, but in many cases these strategies only include limited demographic variables, income data, account balances and ACH patterns.

As we approach the end of the road for deferment periods, the need for effective predictive risk assessment strategies gains urgencyFor credit unions, this call to action must work double duty, not only effectively minimizing losses, but also providing help and assistance to financially stressed members. However, significant uncertainties regarding COVID-19 make it difficult to project when and how the economy will recover.

Which combination of predictive data can help you identify areas of your portfolio that are most at risk, down to the member level, so you can achieve these two important objectives of protecting your cooperative from losses while also supporting members?

While the answer is multifaceted, the first step is to brainstorm insights into the financial health of your field of membership that can help you identify members who will experience financial distress when the government sponsored stimulus and protective actions end. Here are some ideas to get started:

  • Changes in credit utilization 
  • FICO score migration 
  • Identify members by income source and industries hardest hit 
  • Sudden absence of Payroll direct deposit
  • Market employment data

General, broad market data can also identify which members may be hit harder than others. For example, according to research conducted by Finicity in June 2020, households with incomes less than $50,000 have been hit harder by this recession in the following ways:

  • Half have lost their job or had their income reduced, compared to only 31% of households with more than $100,000 in income. 
  • Nearly three-quarters are having trouble keeping up with bills and payments, versus 57% of those with income between $50,000 and $100,000 and 54% of those with household income over $100,000. 
  • While 25% of lower earners have tried to tap credit less often during COVID, 21% have had to use it more often. Another 23% have not attempted to use credit because they assumed they wouldn’t qualify. 
  • Sixty-eight percent of those who earn less than $50,000 worry that the recession will damage their credit, while only 52% of people earning more than $100,000 income have that concern.

Identifying vulnerable members will allow a credit union to focus its resources on effective loss mitigation strategies. These include matching members who have the lowest propensity-to-pay with the best collectors and the right repayment strategies, more accurately identifying accounts that are likely to pay and enhancing risk mitigation efforts by concentrating your efforts on high yield inventory.

Predictive modeling is nothing new, and nearly all credit unions wish they could augment their traditional, historical data with new datasets, and leverage it with the power of data science to effectively quantify risk. However, they are limited by legacy technologies, expertise, and budget. On the other end of the asset spectrum, large financial institutions have built proprietary scoring engines that combine internal data with external factors. While this strategy produces improved predictability, it comes at a cost, requiring an enormous investment of time and specialized resources.

The Coronavirus recession is different from the Great Recession and all that came before it. However, one difference comes in the form of a blessing: technology in 2020 has made access to advanced analytics more affordable and relatively simple to execute.

The scales have tipped toward investing in a trusted data science vendor to help credit unions more effectively manage Coronavirus losses. AKUVO provides web-based solutionconstructed, powered, and maintained by utilizing the best in Data Science. This is more than just a 2020 bottom line decision. For some credit unions, it could determine whether the cooperative survives to provide their communities with access to credit for the next financial crisis. Let’s get through this together!