Skip to main content

Assumption And Directions

FHFA Market Risk Stress Testing

Last Updated:  5/29/2018

  1. The OAS and Volatility shocks as posted incorporate an adjustment to make them more representative of the correlations between these measures and interest rates during a stressful period.  The adjustment is calibrated to reflect a 75th percentile of correlation, or a more stress-period measure of correlation, in contrast to the 50th percentile, at which the correlations are more over-the-cycle measures. 

  2. FHFA was unable to obtain sufficient data to statistically determine a stressful shock to the OAS applicable to Private Label MBS.  FHFA did review available historical data and concluded that a shock to PLMBS OAS of 200 bps, applied in all scenarios, is reasonable.

  3. The list of assets to which the agency OAS shock applies includes the following:

    • Agency MBS, CMO, DUS, and HECM securities
    • Repos backed by TBA’s
    • AMA loans
    • State Housing Finance Agency bonds

  4. Student Loans—as the Bank deems most appropriate, the Bank should apply an OAS shock of either the flat 200 bps for all scenarios, or the scenario specific agency OAS shock.

  5. Assets not subjected to an OAS shock include:

    • Advances
    • Fed Funds
    • Agency non-mortgage backed assets (debt)
    • Term CD’s and all other money market assets not otherwise listed
    • Repos backed by Treasuries

  6. The market risk assessment should occur at least quarterly.  The shocked rates, volatilities and OAS posted are derived from applying the historical shocks to the current market data as of the end of the month, for the middle month, of each calendar quarter.  The shocks are then applicable to the portfolio data as of the same date.

  7. FHFA recognizes that some financial models may not accommodate cap volatilities.  In such circumstances, there is no option but to proceed to assess market risk without including shocks to the in-and-out-of-the-money cap volatilities.

  8. A measure of market risk under stressful macroeconomic conditions is reasonably approximated by an average of the five worst (largest loss in portfolio market value) outcomes from subjecting the portfolio to the set of historical scenarios.  Each of the five outcomes should be equally weighted in determining the average outcome.