Post-Lockdown Data: Still Fit for Purpose?
The transition from a full lockdown life to something that will feel more normal has begun. While there is huge uncertainty about what lies before us, we can recall lessons from our past that may help us in this next phase. In this article, FTI Consulting’s insurance expert, Peter Kelly, shares his insights.
Uninformed Decisions: Follow the Data
We know that in social and economic disruptions of a systemic scale, the data we capture for our business decisions is altered and/or degraded. The data changes in either absolute measurement, accuracy or predictive power.
Insurers and others in financial services should be proactive by learning from the past and acting now to build tools to measure and detect this degradation and avoid making naïve and uninformed decisions.
Only 12 years ago many industrial countries experienced a financial crisis which led to widespread loss of wealth, business failures and unemployment. In many countries, there was a response on the part of consumers to de-leverage. In the USA, the household savings rate doubled over three years from 3.6% in 2007 to 7.1% in 20101.
Over a five-year period, US consumer debt as a percentage of GDP fell from 99% in 2007 to 84% in 20122. One result of this widespread de-leveraging was an overall across-the-board improvement in credit scores. This created a problem for insurers who use credit for predictive models.
Since the data change was an overall shift of the credit curve, an improvement in someone’s credit score wasn’t necessarily any movement at all in their relative place in the shifted curve.
When the de-leverage shift occurred, the predictive models could not recalibrate quickly enough and as a result, insurers were caught off guard. The emerging bias (and under-pricing) took years to observe and correct.
1: Personal savings rate in the US from 1960 to 2019 – Statista, 2020.
2: US Bureau of Economic Analysis, 2019
June 10, 2020
Forensic & Litigation Consulting
System.Collections.Generic.List`1[FTI.Foundation.Person.Models.IJobTitle] ?? string.Empty