The idea of an absolute contradiction between insurance float as a resource and as a liability seems poorly calibrated; a more nuanced classification would be more accurate.
The loss ratio of an insurance company determines a conversion score of this liability into an investment resource, much more than a binary opposition.
For example, a company with a stable loss ratio under 50% over a decade classifies this float with a potential for investment of 0.9 out of 1, a very high level.
Conversely, a 90% ratio places the same float at a risk score of 0.8 out of 1, mainly making it a pure liability, which is not the same situation.