My answer would be to just compare the NPV of investment income to the NPV of the underwriting P/L. In WRB case, the float is worth substantially more than the underwriting profits. In Lancashire's case, it's the opposite.

On a totally differing point- One thing I've had a bit of fun doing is valuing float in hindsight, specifically Berkshire's. If it's 1990 and you were all-knowing about what float would do over the next 23 years, what would you have valued it at? Since 1990 float has grown at 18% compounded annually. I don't know what actual investment returns from their float were, but two years ago Charlie said that Warren would sometimes earn 20-30% on float. Just say on average they earned 7% after tax. Float growing at 18% a year generating 7% investment returns, with a 10% discount rate, creates a valuation of **3.7 times nominal value of float**! That boggles my mind.