Would fully-compliant Sarbox actually have helped avoid trouble?

    With the recent five year anniversary of the passage of Sarbanes-Oxley, now is an appropriate time for mortgage companies to review the significance of “Sarbox” for the mortgage industry.  In an information industry such as the mortgage industry, adherence to Sarbox is often a complex and challenging ordeal where the costs are apparent and the benefits elusive.  With all the recent turmoil in the Subprime and Alt-A secondary markets and with GAAP pushing forward a new mark-to-market accounting standard, SFAS 157, mortgage bankers are not only being asked to implement new mark-to-market methodologies in a manner compliant with Sarbox, but the market turmoil significantly challenges their asset valuation methodologies and their price discovery procedures.  A careful review of the asset valuation methodologies, assumption discovery procedures, price discovery procedures, business control procedures supporting the valuation methods, and the valuation software model validations will be a critical focus of mortgage companies in this market confusion and illiquidity.  In the post-Sarbox environment, the solutions for implementing these challenges are daunting, but implementing effective procedures will help your company in avoiding potentially significant troubles.

Asset Valuation Methodologies – Disaster Recovery Procedures Round II:

    Pricing mortgage assets involves assumptions about the underlying market mechanics and market liquidity of the numerous markets that exist for distinct mortgage assets.  For example, pricing an agency eligible, newly created whole loan involves a completely different valuation methodology and completely different price discovery procedures than pricing a NIM (Net Interest Margin) bond class.   Given the turmoil in the Subprime and Alt-A secondary markets, companies must review their asset valuation methodologies as to whether their stated methodologies are robust enough to adequately function in tumultuous markets.   The recent liquidity meltdown scenario is the kind of disaster scenario that asset valuation methodologies must contemplate.   Wouldn’t American Home have had a much smaller loss if their computer servers were severely damaged than what happened to the market acceptance of their loan products in their warehouse?  Nevertheless, American Home probably had sound policies and internal systems in place to address severely damage computers servers, but unfortunately for them didn’t have a backup system or disaster recovery procedure for selling or holding their warehouse loan inventory when their take-out bids declined significantly.  If they had fully addressed this potential scenario in advance of the situation, they might have avoided bankruptcy.

Business Control Procedures implementing the Valuation Methodologies:

    With SFAS 157 scheduled to be adopted by November 2007, mortgage companies are going to have to rethink all their methods and procedures for establishing Fair Value for each of their mortgage assets.  Each asset will be assigned a Fair Value class as Level 1, Level 2, or Level 3.  Level 1 being a quoted price.  Level 2 being established from the quoted price of a similar asset.  And Level 3 fair value established from an assumption discovery process.

Assumption Discovery Procedures

    For mortgage assets, the three most important assumptions are prepayment assumptions, default/FCL assumptions and relative yield or spread assumptions.    The disclosure standards for residual interests in 10-K filings has been to provide single prepayment speeds, FCL loss assumptions and discount rates (yields).   However, in order to establish a defendable and justifiable fair value of a mortgage asset, single prepayment assumptions, e.g. 12% CPR, are inadequate and inconsistent with market pricing practices.   Moreover, the market participants make realistic and complex assumptions about the FCL behavior including age dependent FCL frequency curves, geographically dependent FCL timing periods, and loan product, LTV, and real estate market dependent loss severity assumptions.  Any fair value pricing methodology and assumption discovery procedures needs to be consistent with market pricing practices and not minimal 10-K disclosure standards or historical behavior.
The real fair value question: What will happen to Subprime Voluntary Prepayments?
With all the dislocation in the subprime and Alt-A origination markets, nearly every mortgage analyst is developing a new view of how voluntary prepayments of these borrowers will perform in the future.  Utilizing historical loan product or company specific prepayment speeds will not be consistent with market pricing practices and will not be appropriate for fair value disclosures.  Actual buyers and sellers of these assets will base their pricing on their anticipated future prepayment and FCL behaviors.  Mortgage companies must require that their independent third-party pricing providers supply not only a thorough disclosure of the prepayment and FCL projected vectors, but also their procedures and credentials in asserting that they are consistent with market consensus vectors.   Regression-based correlation prepayment and FCL models will offer only limited utility with the completely transformed subprime and Alt-A mortgage origination markets.

Price Discovery Procedures

            One of the most sweeping change anticipated with the adoption of SFAS 157 fair value standard will be the reduced acceptance of marking to model for illiquid mortgage price discovery.  Most analysts expect an increased need for independent third-party market valuations/appraisals for illiquid mortgages such as MSRs, seasoned whole loans, residual interests, and subordinated bond classes.  Many firms’ current price discovery procedures employ a mark-to-model procedure and occasionally verification of their modeled prices with third-party marks.  However, implementing SFAS 157 in a more fully-compliant manner will cause more mortgage companies to more frequently obtain third-party justifications of their model prices to insure that they are representative “Exit prices”.

Model Validation:

The cornerstone of any SFAS 157 Fair Value disclosure or Sarbox compliance of financial risk control is the validation of the software model employed in these mission critical asset valuations.  Clearly longstanding and wide-spread market acceptance of the software is the first order validation of a pricing model.  Transparency of the underlying mechanics of the model (i.e. an “Open-Box” not a “Black-Box”) and thorough user documentation are the next level of validation.  In addition, the industry standard for model validation – OCC Directive 2000-16 highlights the general model validation procedures as:

However, let’s review what else makes sound model validation procedures as identified in the OCC directive.  

Validating Model Inputs

For mortgage pricing models, there are several categories of model inputs:

To validate the Loan Data Attributes in a mortgage pricing model, the central focus should be on:

Validating Model Processing Components

 

Validating Model Reports

Conclusions