You may also be interested in:

Articles

Case Study: A Shift to Random Forecasting Improves Accuracy

  • By Nilly Essaides
  • Published: 7/13/2015

When Seattle-based healthcare insurer Premera Blue Cross faced the Affordable Care Act (ACA), it quickly realized it needed a new way of approaching its forecasting process.

According to Jon Gilbert, an internal consultant at Premera who largely focuses on innovation and decision analysis, the onset of the ACA meant going forward without some of the tools and underwriting assumptions that Premera typically had. Post-implementation, health insurance companies were required to accept all comers; before insurers could pick and choose their customers based on the perceived risk.

“We’re not selling an iPad or an iPhone; the cost can be very significant,” Gilbert said, during an AFP roundtable on financial planning and analysis (FP&A) in Seattle. “It can be an asymmetric payoff with an unlimited downside.”

One significant fact Premera faced was that it could no longer screen potential members through traditional medical underwriting. Conversely, the insurer now has many more potential buyers. “We didn’t know what our customers would look like and how many would want to buy,” Gilbert explained. The company also didn’t know early on who its competitors would be.

At Premera, so much of the cost of revenue depends on difficult-to-predict member healthcare consumption. Despite this, in the past, the range of uncertainty was narrower and company leadership and the board were accustomed to more predictable results. 

“When we looked at the shift under the ACA and the requisite amount of uncertainty, we clearly needed new tools,” Gilbert said. For a health insurer, the calculation for taking on a new customer is very different than a retailer selling to a customer, he explained. A single person may not be profitable, however, a pool of customers may be. Also, the insurer is tethered to each person for the length of the program. 

A range of possible outcomes

To help develop the tools, Premera approached consultancy Strategic Decisions Group (SDG), out of Palo Alto, Calif. SDG evaluated the company’s readiness to expand its toolset and confront the new challenges presented by the ACA.

“We used their probabilistic approach to what we now call ‘decision quality,’ and it’s been very useful for us,” Gilbert explained. “Under the deterministic approach, you come up with a single number. The stochastic approach comes up with a range of possible outcomes.” Additionally, the stochastic approach encourages a broadening of the range. 

But it’s not enough to determine the best alternative; Premera had to build a culture around it, and that meant getting other departments involved. “We had to ask:  ‘Who else needs to be involved in the process?’ Sales, as an example. We realized we may have to talk to more people,” Gilbert said.

SDG worked Premera through the process in 2013, for the company’s 2014 corporate projections. The knowledge the consultancy imparted on Premera helped it continue the process the following year. “Right now, we have made it standard work,” Gilbert said. “We do a primary corporate forecast annually, and update that several times within the year using the stochastic approach, relying on Monte Carlo simulation to come up with a range of possible outcomes. We’re starting to use simulations to run models and communicate recommendations afterwards.”

The process involves looking for the drivers and the risk factors, and then examining the range of possible values for these drivers. “You turn over every rock, and then see whether any of them correlate,” Gilbert said. Initially, Premera looked at 150 different variables.

The approach is not suitable to all problems; some can be easily resolved with one number. But it lends itself well to fuzzy, complex problems.

That does not necessarily mean avoiding risk, Gilbert noted. Sometimes there are good commercial opportunities on the left side of the curve. But if you realize the risk, you can take action to mitigate it, such as deploying reinsurance strategies or partnering with another company to limit the potential for bad outcomes. “Companies, like people, have utility and risk aversion curves,” he said. “It’s best to show the true underlying uncertainty and then work with executives to figure out whether there are mitigation strategies to move forward in profitable way.”

Culturally, Gilbert believes it is important to create a safe environment where people can voice their concerns about the narrowness of the ranges. The ranges Premera came up with for 2014 were subsequently found to be too tight. The insurer became more comfortable with widening the ranges around key input variables, such as membership. “They did not want to make the same mistake again,” he said.

Premera’s annual forecasting is now done stochastically. “We compare ranges against actuals to see early results,” Gilbert said. “It’s a rigorous way to outline the effects of potential strategies and communicate them in a comprehensive, yet straightforward, format.”

Overall, the probabilistic approach remains a very valuable tool for Premera. “We’re going through a great amount of change. There’s still tremendous amount of upheaval in our industry,” Gilbert added. “Our ability to think more probabilistically will benefit us in a lot of ways down the road.”

The Call for Speakers is OPEN
Think you have what it takes to lead an educational session at FinNext 2019? We KNOW you do. Just pick a topic you're passionate about and share it with your FP&A peers. This is your time to shine. 
Submit your Proposal

Copyright © 2018 Association for Financial Professionals, Inc.
All rights reserved.