This assumes that investments made with premiums are precisely equal with the margin kept by the insurer. Considering the variety of insurers, insurance products, seasonal variation, etc. this is a pretty wild assumption.
That said, insurers do make a profit on their income and make a profit on their investments. They may tell you that they have a negative combined ratio, but it's a misrepresentation, mainly for lobbying reasons. They only include base premiums in that calculation, not all the extras they sell (alloy wheel insurance, key/fob insurance, no-claims protection, etc).
Again though, it's worth pointing out that smart contracts aren't going to get regulator approval any time soon... and will probably never have access to re-insurance markets. Customers are also going to have a hard time trusting smart-contract based systems them after recent events.
The value of optimising for decentralisation over optimising for disruption* seems negligible in this case.
If you can create a crowd-sourced insurance pool there are tons of real tangible benefits, mainly coming from the fact that the customers and "insurer" are economically aligned (unlike a conventional insurer) so the incremental value (and cost) of making it decentralised doesn't seem worth it to me.
(source/disclaimer: I founded a p2p insurance startup)
Edit: added asterix and list of genuine benefits
- Since the "insurer" isn't on the hook for paying it doesn't need whitelists of "permitted" tradespeople, or price lists.
- It can share information freely about crime hotspots, health issues, etc. It can act on the customers behalf in order to keep them safe, and a happy customer.
- Customers in a pool can work together to solve problems
- A pool of safe (and lucky) drivers might see premiums drop to 25% of market rate, which mainly covers re-insurance costs
- Knowing the repurcussions of your actions will affect you and your pool group fairly directly may change some types of risky behaviour
- Lower incidence of fraud
( * By disruption I mean the proper sense of the word. It's an entirely different business model rather than just being an innovation. )
I don't see customers caring about distribution, whereas it would make several areas more difficult to manage... fraud for one. It's cool and all, but it's a walled-enough-garden even without that constraint... a constraint that regulators, reinsurers, etc will find alien and confusing.