Patricia Mosser: Principles–and a solution–for GSE reform
April 26, 2016
Replacing the housing securitization government-sponsored enterprises (GSEs)—Fannie Mae and Freddie Mac—is the biggest financial reform issue in the United States. It is a glaring gap compared with the extensive financial reforms in other parts of the financial system and seven-and-a-half years (and counting) of temporary government conservatorship. There have been many proposals to replace the activities of Fannie and Freddie, including a mutualized securitization utility that some of my colleagues and I outlined in two papers (Dechario et al. 2010; Mosser, Tracy, and Wright 2013). But more important than the details of our proposal to replace the GSEs are several key design principles—and a few hard-learned lessons—which need to guide reform.
Principles for reform
First, any replacement for Fannie and Freddie should be structured to reduce systemwide risk by insuring that those who make decisions about housing risk appropriately price and bear those risks. Because securitization is atomistic, this incentive alignment needs to happen across the entire housing securitization chain (i.e., borrowers, lenders, securitizers, investors, and the government). The entities responsible for covering losses must have a stake in which loans to securitize, the risk monitoring of loans, and the amount of credit protection (capital) needed. The failure of private-label securitization in the financial crisis was partly because of the failure to align the incentives of lenders, securitizers, and investors in this dimension. Any proposal to replace GSE-type securitization should not be structured with the same flaws that destroyed the private-label market. Instead, it must ensure that the securitization entity has skin in the game and can monitor, price, and enforce credit standards.
One implication of this principle is that there needs to be greater capacity to absorb losses for core mortgages consistent with capital requirements for other financial institutions. Fannie Mae and Freddie Mac had skin in the game, just not enough. Another implication is that borrowers, lenders, and securitizers—all of whom influence lending decisions—should provide sufficient information to understand and monitor mortgage risks and have skin in the game. In other words, we need greater disclosure, meaningful household down payments (10 to 20 percent), close monitoring of credit quality, and risk retention.
Second, the housing finance system structure must be more robust and provide access to mortgage credit through the housing cycle. While this was an ostensible goal for the GSEs, they clearly could not do so without a government takeover. Moreover, this robustness cannot be achieved by merely spreading out or diversifying mortgage credit risk across many investors or financial companies. As the crisis proved, all boats rise and fall (or fail) together in a systemic real estate downturn.
Third, governments own the tail risk in housing finance. Across decades, countries and political regimes, and different types of financial systems, governments step in when house prices collapse and precipitate a financial crisis. While the form of government bailout can vary, there is no question that governments “reinsure” their housing finance systems in extremis (Laeven and Valencia 2010). Therefore, it is not credible (and not time-consistent policy) for the government to claim that it will not bail out the housing finance system. Moreover, if the government claims it will not provide a backstop, it is likely to provide incentives for more risk taking and leverage in the financial system through an unpriced implicit guarantee, making a housing-related boom-bust cycle more likely. Consistent with the first principle above, the government’s catastrophic reinsurance needs to be credible, explicit, and priced ex ante.
While governments must manage and price tail risk, they should not price and manage housing credit risk on a day-to-day basis. Setting standards and regulatory minimums is a government regulatory responsibility. Daily risk monitoring, repricing mortgage credit risk, and making monthly adjustments to issuer-specific fees are not. The history of government agencies and corporations has many examples of the private-sector actors "gaming” government entities, minimizing the costs and risks to the private sector and maximizing the risk taken by the government. In short, the government should price the (remote) tail risk it does face (already a difficult task) but not price normal mortgage credit risk.
Fourth, the current high degree of standardization of mortgages (particularly for fixed-rate loans) and the associated market structures (particularly the agency mortgage-backed securities, or MBS, market) are worth preserving if possible in a system consistent with the other design principles. Of course, the evolution of agency securitization reflects, in part, US homebuyers’ long-standing preference for long-term fixed-rate, fully amortizing loans with no prepayment penalties. Because such mortgages are a poor fit for most lender balance sheets, securitization and its associated market structures will be important for financing home mortgages.
Securitization of standardized mortgages has several benefits for borrowers, lenders, and investors. For borrowers, standardized fixed-rate mortgages provide some protection from interest rate fluctuations and simplify comparison shopping across competing lenders. Because of their long history in the United States, they are consumer protection products in that their properties and risks are well understood by borrowers. Moreover, one reason that the current agency MBS market is so much more efficient than private securitization markets is because of the relatively standardized, homogenous underwriting process and the straight-through processing embedded in the GSE securitization structure.
The large size and deep liquidity of the agency MBS market mean that investors in agency MBS require lower yields because the bonds can be sold with lower transactions cost and without a long, uncertain wait for a buyer. These lower yields contribute to lower borrowing costs for households and to less risk (and thus lower costs) to lenders making loans. Similarly, the current agency MBS market is a hedging vehicle not only for investors but also for lenders who hedge their pipeline of originated mortgages and interest rate locks offered to borrowers. In addition, lenders use the MBS markets to hedge risks of all types of mortgages—even those not intended or eligible for GSE-type securitization—which likely lowers the cost of those mortgages as well.
Last, standardized securitization is an infrastructure or a utility. It is a high-volume, low-margin business, as demonstrated by the relatively flat profitability of the Fannie Mae and Freddie Mac securitization business lines over time. In other words, standardized securitization has relatively large fixed costs and low marginal costs (the characteristics of a natural monopoly), which suggests that there will be one or very few firms. An important implication of this principle is that there are no efficiencies or benefits for conducting nonstandard securitizations within such a utility. Securitization of idiosyncratic or bespoke loans—such as those included in commercial real estate or multifamily residential MBS—is the opposite of a utility business. Margin costs are high. Every security—compositionally and structurally—is different, and most involve few individual loans, so there is no “law of large numbers” to diversify idiosyncratic risk. In short, any replacement for Fannie and Freddie should not include such securitizations.
What do we advocate?
My coauthors and I previously proposed a mutualized financial market utility (FMU) to replace Fannie and Freddie securitization. The proposed FMU is a regulated private firm mutually owned by lenders, focused exclusively on securitization of standardized residential mortgages.1 The utility pools mortgages into pass-through MBS and provides a credit guarantee. It prices and manages credit risk of the mortgage pools it securitizes, subject to safety and soundness (capital) regulation, and it is required to buy catastrophic reinsurance from the government. It allows mortgage credit markets to function (or restart) even in times of market and economic distress by managing its capital waterfall on a vintage basis. Each vintage of mortgage risk (supporting multiple MBS) is separately capitalized and separately reinsured by the government, allowing for new capital (via a new vintage) to support origination of mortgage credit at the bottom of the cycle without additional government intervention.
How is this structure consistent with the principles above?
The mutualized ownership and loss-sharing structure in our proposed utility are designed to address the first principle of incentive alignment. Our FMU creates incentives to adopt and maintain common underwriting and securitization standards and, more importantly, for owners to closely monitor the credit quality of each other’s lending. Based on this monitoring, the utility can adjust pricing and guarantee fees of individual lender and member loans.
Appropriate incentive alignment also limits how much credit exposure the FMU can sell to other investors. This is important during housing booms, when it is easy for the utility to sell off all its credit risk. But without skin in the game, utility members would collectively ease credit standards, making the boom (and subsequent bust) worse. To avoid this outcome, the FMU must retain a portion of the mutual loss pool to insure appropriate risk management and ongoing monitoring of mortgage credit quality. Importantly, this also insures that the FMU has sufficient capital to fund new vintages, even in periods of market stress when external financing may not be available.
Moreover, characteristics of mutuals that may be seen as disadvantages for some types of financial companies are in fact advantages for our FMU proposal because they align the utility’s incentives with other actors in the securitization chain. Creating a utility with a narrow focus, an emphasis on cost control and efficiency, low risk taking, low return on equity, and low incentives for entrepreneurship and “mission creep”—all of which are characteristics of mutuals—is well aligned with the interests of owner/lenders and investors; with the government’s interest in insuring its backstop is sufficiently remote; and to offset monopolistic rent seeking. In practice, mutual ownership structures work well when they solve a principal agent problem—that is, when their business provides a service closely aligned to the needs of their owners and when owners are relatively homogeneous and knowledgeable (Hansmann 1999). All these characteristics fit our FMU.
The FMUs vintage structure and government reinsurance are designed to address robustness. Vintages compartmentalize credit loss pools and capital so that losses on specific vintages of MBS are covered by the capital available for that pool. Excess profits or capital (as defined by the regulator) from older vintages may be used as capital for newer vintages but not vice versa. To ensure robustness, “new” capital may not be shifted to cover past losses. Securitization can restart with a clean slate with new mortgages from lenders and fresh capital and guarantee fees that can be used only for current and future vintages.
Government catastrophic reinsurance is purchased by vintage, so the government does not reinsure idiosyncratic tail losses on individual MBSs but on extraordinarily large and widespread tail losses on an entire vintage. The government’s involvement naturally unwinds as the balance of the affected vintage pays down. Applying government reinsurance only after the FMU’s failure is not robust (because mortgage credit cannot be restarted) and risks repeating the conservatorship experience. Finally, by compartmentalizing risk (and tail risk insurance) in vintages, failure of the entire FMU is much less likely.
The FMU structure also reinforces secondary market liquidity and standardization of mortgages and is most likely to be used for fixed-rate loans. Our structure is consistent with recent efforts to create a single to-be-announced securities (TBA) market for delivery of mortgage pools and efforts by the Federal Housing Finance Agency on modernization and a new single securitization platform. In addition, mutualizing credit risk could facilitate creating larger and more diversified mortgage pools within the TBA market, which could improve liquidity, reduce market fragmentation, and lower mortgage costs.
Risks and Trade-offs
Like any proposal, our FMU is not without risks and challenges. Government reinsurance is notoriously hard to price. However, the previous price was zero (or negative because the government was subsidizing Fannie and Freddie’s risk-taking behavior across several dimensions), so even an imperfectly calculated positive price would be a step in the right direction. Our proposal considers other key issues, such as retained portfolios (not allowed except for working out delinquencies), the amount of loss-absorbing capacity needed, and affordable housing programs (completely outside the FMU, perhaps funded by an addition to the guarantee fee).
Creating the mutual structure is challenging and reflects the large number of stakeholders in housing finance. From a governance perspective, the FMU would need to balance the needs of large and small members. This can be done several ways, including correspondent relationships akin to the arrangements used for other types of FMUs (e.g., clearinghouses). Even so, direct membership is likely to include small and large lenders, requiring a governance structure that encourages broad access for small lenders, but also risk-management oversight with sophisticated risk monitoring consistent with larger lenders.2 Creating a robust mutual structure will involve consultation with mortgage lenders, borrowers, and investors; new legislation; and importantly, a knowledgeable, sophisticated regulator with the power to write rules that can carefully balance the needs of the various stakeholders. Like many other proposals, ours likely needs a long lead time and creation of administrative “bridges” to new corporate structures.3
- Standardized mortgages need not necessarily be conforming, but those with common property, loan, and borrower characteristics. In practice and certainly at inception, most securitized mortgages are likely to be fixed rate.↩
- See Mosser, Tracy, and Wright (2013) for several proposed governance structures.↩
- Jim Millstein, “Jim Millstein: An Administrative Plan to Restructure and Reform the GSEs,” Urban Institute, last updated March 29, 2016, http://www.urban.org/policy-centers/housing-finance-policy-center/projec....↩
Dechario, Toni, Patricia Mosser, Joseph Tracy, James Vickery, and Joshua Wright. 2010. “A Private Lender Cooperative Model for Residential Mortgage Finance.” In The American Mortgage System: Crisis and Reform, edited by Susan M. Wachter and Marvin M. Smith, 286–304. Philadelphia: University of Pennsylvania Press. (A preliminary version is available as a staff report from the Federal Reserve Bank of New York.)
Hansmann, Henry. 1999. “Cooperative Firms in Theory and Practice.” Finnish Journal of Business Economics 4: 387–403.
Laeven, Luc, and Fabian Valencia. 2010. Resolution of Banking Crises: The Good, the Bad, and the Ugly. Working paper WP/10/146. Washington, DC: International Monetary Fund.
Mosser, Patricia C., Joseph Tracy, and Joshua Wright. 2013. The Capital Structure and Governance of a Mortgage Securitization Utility. Staff Report 644. New York: Federal Reserve Bank of New York.