This piece was published in the American Banker — click here to read on the publication website.
Recently, the Federal Housing Finance Agency, acting in its capacity as conservator of Freddie Mac and Fannie Mae, made some modest changes in the pricing of mortgage risk. These modest changes in pricing have created a firestorm of reactions. Even The Wall Street Journal’s editorial board weighed in decrying the cross-subsidies and income redistribution effects of these changes. Most often cited is the flattening of the credit score curve. (Low credit score borrowers now pay upfront roughly 1.25 percentage points more than high credit score borrowers compared to 2.0 percentage points before.)
Having spent a lifetime and then some analyzing and pricing mortgage risk, I was amused and chalked up the intensity of the reaction to Washington politics and vested interests protecting profits. But it is useful to peel back the rhetoric and discuss a few fundamental policy issues. When viewed through this lens, FHFA’s recently announced changes to loan-level price adjustments (LLPAs) are consistent with safety and soundness and the charter purposes of the government-sponsored enterprises.
I will focus on three key questions.
First, do the GSEs need to earn the same return on equity on each loan they purchase to remain profitable and present a low risk to the taxpayers?
Of course not. The GSEs offered one flat guarantee fee which applied to all borrowers for many years. LLPAs only began in 2008 when the GSEs wanted additional revenue to bolster earnings. The issue of profitability and risk relate to the entire company, and the questions we should be asking are what are the profits, and what is the overall leverage in the firm relative to risk? In 2008, The GSEs were charging too little (0.15 percentage points on their outstanding loan portfolios) and held only 0.45 percentage points of capital against credit risk.
In fact, risk-based pricing often encourages the loosening of credit standards and was used to justify expansion into lower credit quality loans because the GSEs thought they were earning higher risk-adjusted returns on Alt-A and subprime products. So flatter pricing accompanied by tight credit policies may be the profitable, lower-risk strategy. The FHFA’s modest change of flattening the credit curve does not create a risk to taxpayers and may very well be beneficial.
Second, should financial firms set pricing to earn the same return on regulatory capital for each asset?
Of course not. Regulatory capital is just one measure of economic risk to shield firms from unexpected losses. There are many other considerations on how firms price, including leverage ratios, demand elasticities, competition, etc. Would JPMorgan Chase use the Basel accords as the sole basis for all its pricing decisions? There is no evidence that the new FHFA pricing deviates from its recently enacted risk-based capital framework, but even if it did, that would not be a cause for concern.
Finally, what additional considerations beyond credit risk need to enter into GSE pricing decisions?
First, prepayment risk. It always baffles me that LLPAs factor in credit risk but not prepayment risk. Prepayment risk is the risk that borrowers pays off their mortgage earlier or later than expected. An early prepayment leads to an overall lower principal and interest income stream over the life of the loan, reducing that loan’s overall profitability. Because lower-credit-score individuals don’t prepay as quickly as higher-credit-score individuals, the net impact of lower-credit individuals on a company’s returns and performance is complicated. Second, the GSE charters direct them to consider setting lower return targets for low- and moderate-income families. Third, as often has been noted, pricing affects what loans go to the Federal Housing Administration vs. the GSEs. Policymakers may want to consider this. When I headed FHA, I vehemently argued against risk-based pricing at FHA because it would unnecessarily expand FHA’s footprint.
Former acting FHFA Director Ed DeMarco wanted pricing to consider state foreclosure laws and to charge more for states with longer foreclosure timelines. Others have wanted GSEs to price the earthquake risk from “the big one” that will occur one day. Now, some are looking to price in climate risk. All these have or will face political pushback as inappropriate factors for the GSEs to consider — as they were chartered by Congress to create a liquid national secondary market, not to set policy on these matters. So blind adherence to pricing purely on risk may not be a desirable policy objective or consistent with the charters of the GSEs.
It seems that FHFA pricing changes were well considered. Whether by intent or not, the flattening of the credit score curve is very consistent with incorporating prepayment risk into the pricing framework.
That said, I would encourage one change to the new pricing scheme. The additional LLPA for debt-to-income ratios greater than 40% is an unwelcome complication in the pricing of mortgages. Debt-to-income ratios are difficult to measure, so the difference between a DTI of 39% and 41% is insignificant; creating a DTI cliff at 40% will cause lots of work finding more income to get below the 40% threshold. Debt-to-income also has a weak correlation with risk partly because it is not well measured. And the mismeasurement of true (permanent) income is not the same across protected classes, as was documented in Milton Friedman’s 1957 paper, “The Permanent Income Hypothesis.” So, I would encourage the elimination of this fee.
In conclusion, FHFA’s recently announced changes to LLPAs are consistent with safety and soundness and the charter purposes of the GSEs. Eventually, explicit regulatory guidance on what factors to consider when setting prices would be useful for, if and when the GSEs emerge from conservatorship and are free to set their own prices.
Edward L. Golding is Executive Director of the MIT Golub Center for Finance and Policy. He ran the Federal Housing Administration from 2015 to 2017.