The rate changes aren't supposed to ensure that higher risk borrowers make their payments. The rate changes are supposed to shift the costs of insuring mortgages such that people with lower risk mortgages pay more (than they otherwise would have) and people with higher risk mortgages pay less (than they otherwise would have).
It's conceptually comparable to raising the auto insurance premiums of 40 year olds with no accidents from $1,000 a year to $1,100 a year so that you can lower the auto insurance premiums of 18 year olds with 3 accidents from $4,000 a year to $3,000 a year. Ostensibly both groups will, respectively, create roughly the same amount of claim liability as before. And the total premiums collected will, ostensibly, remain roughly the same. But the first group now bears a larger relative share of the insurance cost. If the premiums were fairly set before - i.e., they accurately reflected the risks of the respective groups - then the changes amount to forcing the former group to subsidize insurance costs for the latter group.
I left an important if in this previous post because I had no idea whether the previous rates accurately reflected the risks of the respective groups. I'm not an actuary and at that point I'd never looked at analysis of default rates comparing them, e.g., across different credit scores.
So I tried to find such analysis and found the latest revision (at least I think it's the latest revision) of
an FHFA working paper analyzing mortgage default risk. It deals in broader strokes than the LLPA matrices do. The latest matrix breaks down into 9 credit score bands and 9 LTV bands. But the general picture painted by the analysis - a flatter default risk curve - comports more with the new modified matrix than with the previous matrix, at least when it comes to credit score differences.
Whereas previously someone with a credit score of 630 was (in a lot of LTV ranges) paying 13X as much as someone with a credit score of 750, now they'll be paying around 3X as much. Are the knew rates fairer than the old ones? I don't know, again I'm not an actuary and I don't have the reams of data I'd need to make a fair assessment; but it seems they might be.
I do strongly suspect they're fairer in one way though. Previously the benefit of having higher credit scores stopped at 740, which isn't really that high a score. Above that the rates stayed the same. Now you continue to get the benefit of lowering rates up to 780, which still isn't a really high score. But people at 780 and above are generally better off than or the same as they were before. What the rate adjustments really did is flatten the curve in the middle. People with high credit scores tend to be better off, people with low credit scores tend to be better off, and people in the middle tend to be worse off. But, again, that may well accurately reflect the relative risks. You'll still be paying a lot less if you have a mid-700s credit score rather than a mid-600s.
