Our proprietary models are used throughout our services and are available for client use. Additionally, our team develops custom analytics and tools tailored to specific needs including enhanced risk management and surveillance systems, pricing and valuation tools, management reporting modules, and data infrastructures.
Our experience also facilitates the best “buy” versus “build” evaluations to help you make the most appropriate decision on models and analytics. With our team’s experience in both building systems and integrating models and analytics, regardless of the path, we can provide a solution.
Our data has been organized to be the engine for the development of our econometric models. The Five Bridges’ edge in constructing prepayment, default, and loss models is a deep understanding of the influence of local economic activity on the borrower’s ability to repay a mortgage loan. Loan performance is based on geographic location as well as borrower, property, and loan characteristics. Moreover, the rapid decline in the quality and integrity of underwriting which preceded the current crisis, and the rapid deterioration of local economies subsequently have heightened the importance of modeling economies to supplement traditional borrower metrics.
We have developed individual models for each mortgage loan type, product type, and origination year in each of the 363 metropolitan areas (“CBSA”). These models incorporate baseline econometric forecasts of payroll employment, the household unemployment rate, per capita personal income, and median existing house price (from the Federal Housing Finance Agency) for each one of the 363 metropolitan CBSAs.
By developing such detailed models, we are able to capture the propensity of borrowers pay, prepay or default with more accuracy. Additionally, our modeling infrastructure was built with the flexibility to permit the imposition of additive or multiplicative stresses on any exogenous or endogenous variable. Given the importance of updated or current borrower and loan metrics to surveillance and valuation, loan-to-value ratios are updated on every loan in the database on a monthly basis using our metropolitan area-based monthly home price indexes.
Lastly and importantly, these models were not developed in response to the current financial crisis, but rather were in development in 2004 with subsequent revisions in 2005, 2007, and 2009. As a result, we possess out-of-sample results which allow us to make intelligent enhancements to our process quickly over time.
