Alternative data in underwriting

Underwriting long relied on two sources: the insured's statements and claims history. Alternative data designates everything that comes from elsewhere: information published on a company's website and social media, customer reviews, job postings revealing its actual activity, aerial and satellite imagery, sensor and connected device data, browsing data, transaction data, public registers, technical information observable from outside about its IT systems. They make it possible to check statements, complete a file without a questionnaire, detect changes in risk during the contract and price segments where history is lacking. Their use nonetheless raises serious questions. Their quality and freshness are uneven, and a misattribution can wrongly penalize an insured. Some are personal data, subject to the data protection regulation and its purpose limitation and minimization principles. And several jurisdictions have regulated their use in insurance, precisely because seemingly neutral variables can reproduce prohibited discrimination. The underwriter must therefore document their source, their relevance to the risk and their effects on different groups of insureds.