Since entering conservatorship 10 years ago, Fannie Mae and Freddie Mac, the government-sponsored enterprises (GSEs), have not been subject to regulations that set the amount of capital they need to maintain. Nevertheless, their conservator, the Federal Housing Finance Agency (FHFA), is assessing capital requirements for the GSEs if they are released from conservatorship.
The FHFA proposed a new capital framework in June 2017, the Conservatorship Capital Framework (CCF), and initiated a critical conversation about the right capital standards in a postconservatorship US housing finance system.
At a recent Sunset Seminar hosted by the Urban Institute’s Housing Finance Policy Center, experts discussed how well the CCF aligns capital with risk and how much borrowers with low-risk profiles are likely to subsidize borrowers with high-risk profiles, a process known as cross-subsidization.
The panel generally agreed the CCF would be an improvement over what was in place before the 2008 housing crisis. Taxpayers would be better protected, and most homeowners would not see a major change in their mortgage rates. But some disagreed on how much the proposal would lead to low-risk borrowers cross-subsidizing high-risk borrowers.
Who benefits from cross-subsidization?
Capital requirements significantly affect the pricing of guarantee fees and hence mortgage rates. Depending on how these requirements are set, they can encourage cross-subsidization.
Currently, based on expected loss probabilities, risky GSE borrowers with high loan-to-value (LTV) ratios and low credit scores likely pay lower mortgage rates than their risk profiles warrant and are subsidized by lower-risk GSE borrowers paying higher rates than their risk profiles warrant. This outcome is partially a product of capital standards that do not fully align with risk.
The panelists differed in their views of how the CCF would alter the market’s current cross-subsidization.
Andrew Rippert, CEO of the global mortgage group at Arch Capital Group, argued that the CCF would preserve the significant cross-subsidies in the system, with borrowers with credit scores above 740 paying greater amounts to compensate for borrowers with scores below 680.
Mark Zandi, chief economist at Moody’s, noted that the capital requirements set out in the CCF are close to those implicitly used by the GSEs today. He quantified the current cross-subsidy from low-risk to high-risk borrowers in the GSE market at $4.1 billion a year.
In a dissenting view, Urban Institute nonresident fellow Ed Golding presented work done jointly with senior researcher Jun Zhu using a different measure of cross-subsidies to show that the CCF likely creates little or possibly a “reverse” cross-subsidization.
Golding and Zhu argue that expected losses over a long period are the wrong measure. Rather, they compared the required capital with the losses the GSEs suffered following the Great Recession and showed that in times of stress, original credit scores matter less.
For example, while a borrower with a low credit score might be a higher risk during normal times, a person with a 720 credit score might be just as likely to lose his or her job as someone with a 620 credit score during a severe recession.
They also noted the failure to consider the impact of prepayment risk: high-credit-score homeowners are more likely to prepay their mortgages. To measure cross-subsidization in the mortgage market, it is important to consider the pricing of both prepayment risk and credit risk.
Golding maintained that there might be little cross-subsidization in the current system and that in some instances, the CCF might encourage reverse cross-subsidization, with high-risk homeowners subsidizing low-risk ones.
The CCF needs high-quality data to succeed
The panelists agreed that high-quality data are critical to accurate modeling of mortgage performance and risk and for accurately assessing the level of cross-subsidization in the market. Credit scores, LTV ratios, debt-to-income ratios, and other credit variables, along with performance outcomes, are often not available to researchers at the loan level.
The panelists also agreed that the quality of available mortgage data is inconsistent. Data generated from mortgages issued before 2008 are contaminated by low-documentation and predatory lending.
But since the crisis, underwriting quality has improved. Which data are used to create the new capital requirements will have significant implications for how the GSEs determine borrower risk.
An important start on a complicated topic
Achieving perfect capital requirements for the GSEs might not be possible, according to some panelists. Although risk management remains a priority, federal law also requires the GSEs to support low-income borrowers and underserved markets, like manufactured housing. These responsibilities require added risk.
The plan for the GSEs postconservatorship must also mesh with plans for others in the housing finance system, such as the Federal Housing Administration, the US Department of Veterans Affairs, and the US Department of Agriculture. The panelists agreed, however, that we can’t wait for comprehensive housing finance reform to happen before we make decisions about capital requirements.