Mortgage Fees and Incentive Compatibility

In my last post I considered the new mortgage fee schedule that is due to go into effect next week, and made two points.

First, despite some outlandish claims online, it is not the case that borrowers with better credit scores will pay higher fees than those with worse scores. The advantage to having a stronger credit rating has been diminished for some values of the loan-to-value ratio, but it has not been erased or reversed. And second, conditional on credit score, the new schedule has significantly lower fees at the highest loan-to-value ratios. This suggested to me that some people might try to game the system by borrowing more than they need to, and then making a large principal payment after the loan has closed. I wondered whether the FHFA administration had taken this incentive effect into account when designing the schedule.

These two points can be seen in the table below, which shows the new fee structure.

The lowest fees are shown in green and the highest in red. It is aparent that at any given loan-to-value-ratio, higher credit scores are associated with lower fees. But it is also clear that at some credit scores the fees are quite a bit lower if one makes a very small down payment. For instance, at a credit score of 730, fees are 40 percent lower if one puts down less than 5 percent of the property value, as compared to a standard 20 percent.

The FHFA Director Sandra Thompson has now released a statement that speaks to these issues. On the issue of gaming the system, she states:

The new framework does not provide incentives for a borrower to make a lower down payment to benefit from lower fees. Borrowers making a down payment smaller than 20 percent of the home’s value typically pay mortgage insurance premiums, so these must be added to the fees charged by the Enterprises when considering a borrower’s total costs.

This is true in the absence of strategic prepayment. However, for those buying expensive homes and easily able to afford 20 percent down, it might make sense to put down less than 5 percent, purchase the mandated insurance, and make a large principal payment once the loan has closed. The insurance could then be cancelled. See, for example, the comment here (and my reply). I don’t see how the inventives embedded in the new fee structure protect against this.

If there is indeed a strong enough incentive for strategic loan choices along these lines, then there will be a shift in the distribution of loans across cells in the above table. Higher fee cells will end up being depopulated as lower fee cells attract borrowers. The net effect will be lower fees flowing to the government sponsored enterprises, which are already undercapitalized. Since greater safety and soundness of the GSEs is an explicit goal of the new fee structure, this might be a cause for concern.

There is another claim in Director Thompson’s statement that is worth considering:

Some mistakenly assume that the prior pricing framework was somehow perfectly calibrated to risk—despite many years passing since that framework was reviewed comprehensively. The fees associated with a borrower’s credit score and down payment will now be better aligned with the expected long-term financial performance of those mortgages relative to their risks.

Since the changes in fee structure are substantial, this suggests that the existing schedule had become wildly miscalibrated to risk. Perhaps that is the case. But it would be useful to see some evidence on this point, and some explanation of the process by which an initially well-calibrated fee structure fell so far out of balance.