Dr. Cliff Rossi
In a remarkable display of the speed of change in philosophy over FHA policy between the Trump and Obama Administrations, the new Administration on Inauguration Day suspended a recently announced 25bps reduction in FHA mortgage insurance premiums (MIP). FHA premiums which should be set based on statistically valid pricing models are instead overly influenced by policy. This is no way to run a portfolio where taxpayers face a contingent liability on future defaults on the $1 trillion plus FHA Mutual Mortgage Insurance (MMI) Fund. Providing FHA with the tools and analysis to effectively price and manage the MMI Fund helps ensure the long-term financial viability of this federally guaranteed loan portfolio.
Each November, an external analysis is conducted of the current and prospective health of the MMI Fund. It is based on a sophisticated set of statistical models that take into consideration each FHA borrower’s risk profile, collateral, property, loan and macroeconomic characteristics over a variety of different scenarios on house prices, interest rates and unemployment rates. This analysis produces an estimate of the Fund’s economic value from which the Fund’s capital ratio is calculated and compared to the statutory minimum of 2 percent. In 2016 this capital ratio was estimated to be 2.32 percent1. In recent years, the capital ratio had been well below the 2 percent threshold, an indication that the Fund was in jeopardy of requiring an infusion from the US Treasury. With the ratio above 2 percent, the outgoing Administration made the decision to lower premiums, arguing that doing so would help offset the costs to borrowers from rising interest rates. An unintended consequence of this action is to crowd out private capital. Private mortgage insurance companies (PMIs) participate in the market for high loan-to-value (LTV) borrowers; an important segment of the market for the FHA. By lowering premiums to high LTV borrowers it reduces the market presence of the PMIs. For years, policymakers have espoused a desire to reduce the federal government’s footprint in housing. Reducing FHA premiums runs counter to this strategy. Clearly FHA must balance the interests of borrowers and taxpayers; however, its ability to fulfill its mission to provide low- and moderate-income borrowers with access to mortgage credit depends first and foremost on maintaining the actuarial soundness of the MMI Fund.
Today the FHA essentially flies blind in its ability to understand the real-time risks of the MMI Fund, with the annual actuarial review providing an infrequent benchmark of limited value in portfolio risk management. It does not currently possess the technical or analytical resources to manage a portfolio of the size, the risk and complexity of the MMI Fund. By comparison, both Fannie Mae and Freddie Mac use rigorous proprietary models to price their credit guarantee fees and manage their credit risk exposure. The FHA’s lack of analytic capability has translated into an overreliance on the annual actuarial report where a host of assumptions and model-related issues can have a dramatic effect on the assessment of the Fund’s financial health. To see this consider that the latest actuarial report results showed the economic value of the Fund more than doubled from $17.04 billion to $35.27 billion between 2015 and 2016, helping propel the Fund’s capital ratio above 2 percent and thus giving policymakers more comfort that they had room to reduce premiums2. However, this logic rests on the assumption that the actuarial report models and assumptions have been validated. All regulated financial institutions are subject to strict model validation standards for analytical models used to manage their businesses. Banks that do not provide evidence that their models have undergone external validation reviews are in violation of model risk regulatory guidance. While the actuarial report presents voluminous information on the details of their models, insight into how the models compare against actual performance is scant other than a brief letter by a distinguished Canadian professor specializing in non-mortgage finance issues stating that the models had been reviewed.
To gain a better sense of why validation is important, consider first the way the actuarial report models compute economic value. Using a standard technique called Monte Carlo simulation, the model generates 100 different economic scenarios that it then averages to come up with its estimate of expected economic value. It also produces a range of sensitivities in economic value looking at various percentiles of the distribution of economic value, such as the 90th percentile lowest outcome. Using 100 paths to generate views of average and extreme stress outcomes for the MMI Fund is insufficient to generate statistically reliable estimates of credit losses. Further, the limited number of paths begs the question of whether the tail of the distribution has been accurately measured. The impact of using so few paths could lead to vast underestimation of the amount of losses the Fund could experience during a stress event, thus providing policymakers with a false sense of security that the Fund is safe. For example, the 10th worst path produces an economic value estimate of $25.8 billion in 2016 which is only 27% lower than the average estimate of economic value3. It isn’t clear from the analysis why we should be comfortable with this result. The FHA must enhance its pricing and portfolio management capabilities. Specifically, that means acquiring technical and analytical resources to develop or use tools similar to those commonly used by large financial institutions or those featured in the actuarial report. MIP premiums could be generated using a similar framework as that used in the actuarial report with greater attention given to model validation and key assumptions to improve transparency and ensure confidence in the analysis. The resources to deliver this capability could come from restructuring HUD’s Policy Development & Research Division which today conducts a variety of academic-style research reports on housing. This seems to be a luxury at a time when the FHA has no ability to conduct pricing and portfolio analysis consistent with major market participants.