Get the best
out of your models to improve decision making
17 December 2018
With increasing reliance on models by the insurance industry and beyond, the risk of “big data” causing “big losses” for insurers has never been greater.
Insurers use a wide range of models to make business decisions. Although these models are usually built well, there is often a disconnect between the model builders' detailed knowledge of the modelling, and how management uses the outputs to make decisions.
At one of our regular roundtables, we asked insurance Chief Risk Officers (CROs) what the main model risk issues were.
The number one issue was lack of clarity on the reasons behind the assumptions used in the model. Almost all participants also agreed that pricing models would benefit most from better model risk management, with one CRO commenting that “pricing models are of key importance to the business but can be a black box”.
Source: results from recent poll at LCP’s CRO roundtable
Model risk management
So how can insurers manage model risk and make the most out of the models across the business?
We have identified five key critical actions when developing your model risk response plan:
- Use a unique reliable data source to help ensure that the assumptions derived for models across the business are consistent.
- Have named owners for specific assumptions, regardless of the model they are used in to help boards have confidence in the veracity of assumptions and to improve consistency between different models.
- Make any margins explicit. This helps boards understand the strength of the underlying assumptions and also makes it easier to see differences between models, for example the board may want to set a stretch business plan, but to allow for specific margins in their reported reserves.
- Maintain an inventory of the material models across your business, including who is responsible for their structure and for signing off the assumptions.
- Implement reconciliations between models to check inputs and outputs make sense in the context of the other models used across the business.
These actions help ensure responsibilities are clearly defined so that the board can be confident that its decisions tie together appropriately and allow it to focus on adding value to the business.
By taking these simple steps, we have seen insurers improve their decision-making in a virtuous cycle where the board asks better questions of the modellers, which in turn drives improvements in the models to answer these questions.
By doing this consistently across the business, management gains more confidence in the models and asks better questions, and so the cycle continues.