rewrite this title Lemonade vs. Root – Revisiting Insurtech Stocks – Nanalyze

The most valuable segment of the $9.5 trillion global insurance industry represents money that people pay for a benefit that only gets realized when they die. Life insurance represents 41% of the global insurance industry followed by property & casualty (P&C) at 35%, then health insurance at 24%. Underwriting policies across all these segments is about statistics and probabilities. Consequently, no industry out there lends itself to AI more than the insurance industry.

Nonetheless, a few obvious problems arise when looking at emerging “insuretechs” claiming to disrupt these trillion-dollar spaces. Large insurance companies have loads of historical data, so emerging contenders need a demonstrated data advantage to thrive. Anyone can grow revenues by issuing insurance policies with cheaper premiums. It’s only down the road when the problems start to manifest themselves. In the insurance industry, surprises are never good ones. Today, we’re going to revisit a handful of insurtech stocks we’ve been covering over the years. One of the most popular dabbles in the largest subsegment of P&C insurance – auto.

Any insurance company can grow rapidly if it gets careless about underwriting – Warren Buffett

The Root of the Problem

In 2024, auto insurance claimed 42% of the P&C market – the largest segment ahead of property coverage (36%) and liability insurance (20%). We find auto insurance a questionable thesis because it could, itself, be disrupted by autonomous driving. Cars drive themselves at much safer rates than humans already, and they’re the most dangerous they’ll ever be today. It’s truly a self-improving ecosystem as software, hardware, bandwidth, and data improve simultaneously. How will this impact the auto insurance industry? It’s something we looked at in a past piece on How Technology Will Affect Big Insurance Companies. We concluded that a concentrated bet on auto insurance seems like one with more risk than reward.

Root (ROOT) is the largest auto insurtech in the country by premium with a market cap of $1.5 billion. That’s down 79% from the $7 billion valuation they achieved after taking the traditional IPO route in late 2020. Despite the dramatic decline in valuation, revenues have been ticking steadily upwards, giving it a current simple valuation ratio (SVR) of around one. Compare that to Lemonade at six, or our catalog average of 7.5, and it seems that Root shares are significantly undervalued. But it’s not quite apples to apples.

A prerequisite to investing in insurance stocks is that you need to learn key metrics that are specific to the insurance industry. One such metric is “loss ratio” which shows whether an insurance company is keeping more money than they’re giving away. For example, if a company receives $100 million in revenues from underwriting policies, and pays out $50 million in claims, during a single quarter then that’s a quarterly loss ratio of 50%. A loss ratio greater than 100% is a big problem because it means they’re not able to sell profitable insurance policies. Below you can see the accident period loss ratio for Root plotted out over time.

Industry averages are currently around 59% so Root is operating as expected with a “floor” appearing to emerge around that range. They claim to be using data and AI to avoid underwriting the 20% of drivers who are responsible for 80% of the claims (great idea), but their average loss ratio says otherwise. It’s improving over time, but nothing spectacular. Since insurance is a game of statistics and probabilities, what’s an event that might increase losses?

Weather related drama perhaps, and that’s why it’s important for any insurance company to be geographically diversified. Here we see Texas being the state Root is having the most success in – 20.5% of gross premiums last quarter.

In their 10-Q filing you’ll find additional information including other insurance industry metrics such as expense ratio and LAE ratio. Understanding each seems like a prerequisite to understanding Root’s progress, but is that a road we want to go down?

Three years ago, we said, “we’re liking the company, but not loving it enough to participate in the IPO.” Then last year we noted the company was down 98% since their IPO making them too small to be of interest. Now, they’re making somewhat of a comeback, but we’re worried about their moat. AI is now a major topic of discussion in just about every boardroom. How hard is it for insurance companies to replicate the technology on offer, then steal all those customers back with a bigger marketing budget? Certainly, large insurance companies have similar data – telematics is becoming pervasive – and we also have the threat of vehicle autonomy looming. That’s something our next company ought to be concerned about too.

Making Lemonade Out of Metromile

Metromile debuted their SPAC with an intuitive bull thesis. Paying for car insurance by the mile is the ultimate business model, especially when you can discriminately underwrite only the best drivers. That business model wouldn’t fly very well in the State of California where 58% of their revenues came from, and that was a showstopper for us. “Not being able to use someone’s driving habits or personal information as input to making underwriting decisions leaves Metromile up a creek. ” Following their $1.3 billion SPAC debut, Metromile was acquired by Lemonade (LMND) at a 89% discount to the original valuation.

The combined companies now trade at a market cap of around $4 billion. (Compare that to the valuation Lemonade saw in their 2020 traditional IPO of about $3 billion.) The merger makes sense as Lemonade simply bolted on auto insurance to offer alongside their rental and pet insurance offerings. It’s anybody’s guess as to the success of this acquisition, and for that we need to know what percentage of Lemonade’s revenues come from auto insurance policies. Before going down that path, there’s a bigger concern we raised several years ago. Lemonade was sporting a consistent gross loss ratio in the 90s which – as we mentioned before – is far less than ideal. Two years later, has this changed?

The Combined Ratio

The short answer is kind of. Lemonade provides investor letters with rich loss ratio data plotted over time. Auto insurance seems to have some of the worst ratios averaging about 88% while pet insurance is the most consistent around 70%. For each segment there will be an industry average that fluctuates over time and serves as a benchmark as to whether these numbers are good or bad.

Any investment in Lemonade is a bet on two very niche segments – homeowner’s insurance and pet insurance which accounted for 48% and 32% of last quarter’s total revenues respectively. Auto insurance made up just 14%.

Another key metric for insurance companies is the “combined ratio” which takes “loss ratio” and adds another metric, “expense ratio,” which is simply expenses divided by “premiums written.” Root provides this metric in their financials – 94% last quarter and 100% the same quarter the year prior. That means profitability is now in sight if the number continues lower (or if they make it up on income from investing their float). Contrast this with Lemonade which had a combined ratio last quarter of 170% (69 loss ratio plus 101% expense ratio). More on this in a bit.

There are likely numerous ways to massage this data, but it all comes down to this. What happens if these companies stop throwing all that money into sales and marketing? With a net retention rate around the mid-80s, they always need to replenish customers and keep spending on sales and marketing. All insurance companies do. It’s one of the industries in which the most money is spent on sales and marketing.

Managing the combined ratio is delicate. You can easily expand premiums by loosening the quality of accounts you’re willing to underwrite, but then your loss ratio goes up. Instead, you can spend more on advertising to attract better clients but then your expense ratio goes up. Or you can actually run your business with a combined ratio at 100% or more and make it up on investment income. Neither Root nor Lemonade are bringing in much investment income on whatever float it is they’re investing so this isn’t an option. Lemonade is complicated by the use of reinsurance, a model they’re moving further from as the firm looks to put more risk on their balance sheet. Basically, they can increase revenues simply by reinsuring (or ceding) fewer policies. The below example shows how this works:

This will also decrease their expense ratio. For example, if they didn’t reinsure any policies their combined ratio would have been 114% last quarter. Still too high, but taking on more of this risk implies they expect to see loss ratios improve over time. Both companies do, of course, but they face an uphill battle against well capitalized insurance companies that can simply emulate the tech on offer with better data. Perhaps we should invest in insuretech companies that work alongside insurance companies, not against them.

No Guide For Guidewire

Probably the best thing about GuideWire (GWRE) is that you don’t need a secret decoder ring to understand their financial health. This $21 billion SaaS company offers a platform for P&C insurers, the second-largest category of insurance, and one that’s more resilient and durable in the face of global economic turmoil. Several years ago, we highlighted Guidewire as a more reasonable way to play the growth of insurance technology while noting the absence of retention rates which are usually provided for software-as-a-service companies. Guidewire may not be providing these metrics yet (aside from a visual depiction) but their revenue growth has accelerated – 22% last year and 16% expected for this year with annual recurring revenues (ARR) now over $1 billion.

An SVR of 15, compared to 7 several years ago, means investors may be discovering their potential. Share prices appear volatile over the past few years so don’t chase it. Set a valuation target and stick with it. Below we’ve plotted their SVR over time, and the average is about 14. That’s always a good place to start, and it’s how we set targets ourselves – just take the average over the past five quarters (or more if you prefer and have the data).

We’re reaching the word limit for this piece and haven’t really dug into Guidewire much. Seems like business as usual, but what about their competitor, CCC Intelligent Solutions (CCCS)? Several years ago we dismissed them because of concerns around the impending arrival of autonomy. Do you think this is a justifiable concern? Please let us know in the comments.

Conclusion

It’s hard to see any case where an AI algorithm wouldn’t be able to do a better job than a human at forecasting the likelihood of future events based on vast complex data sets. If an insurance company isn’t looking at AI right now, they’re going to be in trouble. Those that are will find the troves of internal data exhaust from doing business over decades probably gives them an advantage over others.

Rather than compete with insurance companies, perhaps it’s best to align with them and take a cut of the benefits. Otherwise, these “direct to consumer” insuretechs can easily be displaced by marketing campaigns. It’s why the insurance industry has some of the highest marketing budgets around. Nothing is more American than insurance and pharmaceutical commercials.