That presumes someone takes time to evaluate the claim. Unless you have data already available that the counterfactual is true, it’ll take a while before a competitor appears to test out your claim and existing incumbents might just take your word for it. Additionally, the success of the competitor to eat your padded profits is not controlled solely by the truth.
The only thing worse for an insurance company than deliberately inaccurate statistics is accidentally inaccurate statistics. Anyone who wanted to make a profit selling solar panel insurance would independently seek to determine the failure rate of solar panels. Even if they believed their competitors were 100% honest, it's a value that is likely to change over time and they'ed want a better level of detail (rates for different manufacturers, detailed timing, weather effects, etc) than would be publicly available. Yes, if they just assumed it was accurate they might get lucky, but insurance is the business of not relying on luck.
I think you’re assuming that the cost of paying someone to continually track this and maintain everything is 0. It’s not and therefore the savings would have to justify this cost.
And it costs money for a bakery to bake fresh bread, but it's nevertheless a very safe assumption that they will. Statistics are to insurance what bread is to bakeries.
