I'm not particularly brilliant at understanding how patent language is constructed and after taking a look at the two patents for "Providing Termination Benefits for Employees" aka severance insurance, I'm not any more enlightened on how this insurance product is really structured. No great surprise there!
It appears that the software uses a three dimensional risk isolation approach and limits the coverage in any given cell (specific group of employees as determined by wage, tenure and job description) to a calculated number of employees (as a percentage of the total number of employees in that cell). This would appear to be the controlling strategy for dealing with adverse selection. Under the approach used in the two patents, it would appear that a corporation couldn't elect to fire just those employees who would be the most expensive to provide severance to, because they would only be covered for a certain number/percentage of employees who fit that description, e.g., those who earned over $150,000.