Nearly a full decade has passed since the housing bubble burst, which resulted in the global financial crisis, the most damaging economic downturn since the Great Depression. The extraordinary actions taken to stabilize financial markets were largely successful, yet memories of the pain have begun to fade.
Recently, FHFA posted a Request for Input (RFI) asking for opinions concerning options for changing the delivery of credit scores to the GSEs, underwriters and investors. At present, credit scores are delivered via the three credit bureaus (Equifax, Experian and TransUnion), using a model developed by the Fair Isaac Corporation about a dozen years ago known as Classic FICO. More recently, FICO has updated its model, FICO 9, and the three credit bureaus have collaborated to build their own model, VantageScore 3.0.
FHFA is considering various options to change the current system. These include 1) a single delivery of one of these two scores, 2) a requirement that both scores be used, 3) lender choice as to which score to use or 4) a waterfall through which a primary and secondary score is established.
Any such choice in the end is the outcome of a detailed cost-benefit analysis. While FHFA acknowledges this, they fall short of providing the necessary framework for reaching an informed conclusion. In my recent comment letter to FHFA Director Watt, I provide a detailed critique of these options. The major points are below:
- While FHFA states that new scores provide “only marginal benefits,” they go on to say that there are “compelling reasons” to change. It is not clear how that conclusion is reached.
- They ask market participants for cost estimates, and state that the GSEs will require 12-18 months to implement changes, but do not provide cost estimates for the Enterprises.
- They acknowledge that there are risks associated with the proposed changes but do not provide a structure for ameliorating these concerns, or a cost estimate of implementing it.
- They ignore the potential risks associated with the credit loosening implicit in the new VantageScore model that stem from more relaxed trade line histories and increases to systemic risks associated with its adoption by lightly-regulated nonbanks.
Over the past couple of years, we have seen successive risk layering implemented by the Enterprises related to higher loan-to-value (LTV) and debt-to-income (DTI) ratios, along with generous use of various sources of income. For underwriters, servicers, investors and policymakers to successfully navigate through this perilous market environment, accurate up-to-date data on loan performance at a micro level is an essential resource.
At present, delinquency data provided by Recursion Co reveals a trend increase in the share of loans with DTIs over 40 and 45. Overall, the level of stress in the market remains low, but delinquencies for FHA loans show a troubling rise, even when adjustments are made for recent natural disasters.
In the end, advocates of “lender choice” have failed to make a case for a change that would serve to boost leverage in the mortgage system at the peak of a market cycle. If there is one lesson to take away from the last crisis, it is this: Policies that begin by allowing more borrowers to obtain credit who cannot confidently sustain their positions in a downturn will inevitably end with the costs of a bust shouldered by the very groups that the policy is designed to serve.
Implementation of multiple versions of a credit model is likely to lead to added cost and complexity. If FHFA finds a need for change, the proper choice is a change to provision of FICO 9 as the single score, as this provides the opportunity for forecast benefits with minimal risks associated with credit loosening.