Curb Your AI Enthusiasm: Just Look at the Insurtech Carnage


Shares in “insurtech” startups, which look to upend traditional insurance through the use of automation, data and AI, have enjoyed a small bounce lately thanks to the mania unleashed by OpenAI’s ChatGPT. But they remain far below their levels a few years ago, when many went public.

Is this a good opportunity for investors to jump ship, or rather the start of a rebound that will reward AI faith?

A takeover proposal for car-insurance startup Root illustrates the dilemma. The $280 million offer from Embedded Insurance, a company run by serial insurtech entrepreneur James Hall, is much better than Root’s $85 million valuation before the bid, but also orders of magnitude below the almost $7 billion at which the company listed in 2020. Root has so far remained silent.

Other startups that went public at about the same time as Root include Lemonade, Hippo, Oscar Health, Clover Health and Bright Health. They all trade at a fraction of what they were worth then.

In theory, few industries should be easier to shake up than insurance, which scores very low on customer satisfaction and is run by big, boring corporations set in their own ways. Data and statistics have been at its very core ever since British mathematician James Dodson created the Equitable Life Assurance Society in the 18th century.

As a result, venture capitalists love it: According to reinsurer Gallagher Re, they poured $49 billion into insurtech between 2014 and 2022. Funding fell back last year, but mostly because 2021 was wild.

Unlike moonshots that promise air taxis or space tourism, these startups have sensible pitches. Root, which is backed by online used-car retailer Carvana, saw that the riskiness of drivers tends to be assessed through broad-brush variables such as age and occupation, leaving a gap for using telematics to assess individualized behavior, including factors such as gentle turning or not answering texts while driving.

Similarly, Hippo employs “smart home” kits to constantly monitor risks of fire, water and other damages. Lemonade uses AI and techniques from behavioral economics to fast-track all types of insurance.

Yet the revolters have failed to storm the gates.

Root’s business has slowed lately, a red flag for investors looking for fast-growing companies with potential economies of scale. Even more concerning is that new technologies haven’t yet proven that they are better at pricing risks: Old-school insurers still have lower claims relative to premiums than insurtech firms, though the record of the latter group is improving.

The digital age has never shown much sign of threatening insurance giants. They haven’t seen their market share eroded or even had to adapt their businesses much. Price-comparison websites may have injected some extra competition, but people haven’t fundamentally changed how they buy insurance, even as they embraced online apps to hail rides, get food delivered or go on vacation. That suggests even the takeover value of insurtech stocks may be low.

As it turns out, incumbents had good reasons to be conservative. “20 years ago we thought distribution costs would be lower using the internet, but this is absolutely not the case: Paying Google is more expensive than paying an agent,” said Frédéric de Courtois, group deputy chief executive of French insurer Axa.

New AI players face an additional problem: Growing aggressively means attracting riskier clients, and so they struggle to achieve scale. Without enough data, they can’t gain the edge required to compete in a very price-sensitive market.

These lessons have broader applications. As happened with personal computers and then the internet, AI could greatly benefit its suppliers—software and chip behemoths, whose stocks have gained the most in the recent rally. It could also open up some new markets and, yes, upend how established players perform specific tasks. For insurers, it is opening new avenues for identifying fraud, for example.

Investors who see revolution everywhere, however, could end up putting their own portfolios in the dock.



Source link