Stress testing: looking backwards to go forwards

Companies: Conning

Daniel Finn, a director at Conning, discusses how the firm's historical stress testing and scenario analysis is helping to change how the insurance industry understands risk

What are common mistakes firms make when undertaking scenario analysis and stress testing?

Daniel Finn, director, ConningFor years, many US life insurance companies have been doing the "New York Seven" – a set of interest rate scenarios first prescribed by the state's regulator – but they do not go beyond this. As a result, firms are constructing portfolios based on the same stress test outcomes – they are making similar choices in their investments to avoid the same potential negative scenario outcomes. This homogeneity in risk approach creates its own risks.

For property and casualty companies, most have been performing stochastic analysis and now they want to expand on that. However, firms then go and build a whole new system for stress testing, and then they wind up with significant disconnects between the way the two systems operate, and the meaningfulness of their outputs. These firms need a solution that enhances their stochastic analysis and gives them another way of thinking about risk holistically.  

How should senior management and boards analyse the outcomes of historical scenarios to better understand the risks the firm faces?

Often when teams create scenarios to perform stress testing with, they struggle to explain what that model is composed of. Using a historic scenario gives the team a story to talk about with these stakeholders. They can say, "You remember what this was like and here's how bad it would be today."

The team can then show how the stochastic and the deterministic make sense with each other. Senior management are very smart people but they are not spending every working day thinking about interest rate modelling. By doing both historical and stochastic modelling, and seeing how they relate, it can be easier to convey the meaning of the outcomes of the modelling to these stakeholders. It helps firms to have a more constructive discussion about the risk/reward trade-off.

How can firms effectively communicate the outcomes of scenario and stress testing work to improve management decision-making and risk-taking?

Using both stochastic and the historical approaches also helps firms ensure they are balancing a more long-term view with what may have happened in the past quarter in their models. Firms don't want to be too reactive to what happened in the last quarter, but they don't want to miss anything, either. Senior management wants a strategy that makes sense and will last for a while – they need to understand if they are structuring the organisation correctly for the long term. So, having the right balance between the reactive and the stable is a good way to get senior management's head around this decision-making versus risk-taking component.

How is the regulators' approach to scenario analysis and stress testing evolving?

Regulators are becoming much more open to what companies are doing. A number of our clients who are using these stress scenarios have presented them to the regulators – who said they like them. The regulators said they understood what these firms are doing to enhance analysis and to help senior management understand the risks and opportunities. The regulators are asking if they can share this approach with other companies they oversee.

How will the availability of data be different or better for future "big events" that could impact firms?

We run into data challenges for historic events such as the US Great Depression, where there is only some interest rate and equity data. Data challenges can also pop up when looking at other jurisdictions – there can be an issue around the credibility of the data some governments produce. Research now helps augment and correct data in these cases. Ten years ago, this wasn't possible – today we have more data and sources, and we have the horsepower to work with the data.

If you could make one key prediction for the industry for 2018, what would it be?

Firms really like this "best of both" approach, and I think that is going to lead to people wanting to do a lot more. Firms are looking at how they can learn enough from these historical events to create scenarios for future events, such as a hard Brexit. This is the goal of risk management – to be prepared for the event before it happens, rather than be reactive afterwards.