Because it is so much fun, I've put together a simple model relating Radian's delinquency inventory to reserves. It should tell you nothing other than I know how to use a spreadsheet and make imaginative assumptions. Nonetheless, you can judge the reasonableness of the assumptions and copy the model into your own spreadsheet to test the implications of your own expectations.
Here's a little bit of the little non-obvious reasoning I put into it my assumptions:
-Claims paid at Q4 actual justified by severity already at 100% though loan size differences are admittedly unknown.
-12% mod rate is consistent with Radian point of view that most of mod benefit has been realized.
-21% total rescission rate reflecting decreasing proportion of poorly underwritten loans overall but bucket over 12 months behind is still dominated by those loans.
-1-in-50 natural cure rate for loans over 12 months behind reflect only debtor windfalls and reformed strategic defaulters.