| |
LoanDynamics™ Model
The AD&Co Approach to Mortgage Credit Modeling
The risks inherent in credit-sensitive mortgages and related securities have the market clamoring to better understand delinquency, default, and loss severity, as well as prepayment on mortgage loans. The new LoanDynamics™ Model extends AD&Co’s current prepayment model subroutines to address the needs of those who are exposed to these risks. For issuers and investors, the new LoanDynamics™ Model produces required performance metrics such as CPR (prepayment), CDR (default), 60+ delinquency and loss severity, given loan characteristics and a scenario for interest rate and house price indices.

The launch of the LoanDynamics™ Model is the culmination of a 2-year company-wide development effort. As a starting point, we examined historical data, including roughly 8 million loans from 144 issuers from 1990 to 2006. Over the course of our analysis, we found that borrower behavior can be described realistically yet parsimoniously using 4 categories of payment status: current, delinquent, seriously delinquent and terminated
Our team built a set of economic models to explain the motivation for borrowers to transition from one payment status to the others given a wide variety of economic scenarios and loan characteristics. The economic model is calibrated to observed historical data in order to provide a probabilistic assessment of how a loan will behave in the future. The model has some flavor of quantum mechanics in that a single loan (like an electron) can exist in several distinct states over the course of its existence.

The resulting LoanDynamics™ Model goes beyond a traditional “2-state” competing risk model (which includes just prepayments and defaults), and can be used to account for the effects of 60+ delinquency on bond triggers. Our model also builds upon traditional transition models with 7 or 8 payment status categories by focusing on those payment status categories that have the greatest impact on investment performance. This simplification allows for greater emphasis on the dynamic aspects of the loan transitions.
The model is unified across credit sector and product type, and it relies upon observed loan characteristics (i.e., fields available in the typical servicing system file) to make its projections. As a result, users are not required to make potentially arbitrary judgments about whether a loan falls into jumbo, Alt-A, High LTV, or Subprime credit sectors, and users can apply the model to pools of loans containing a wide mix of underlying collateral.

We offer the LoanDynamicsTM Model as a C++ subroutine, which can be delivered either as a Windows dynamically linked library (DLL) or as a Unix shared object. The subroutine is ready for integration into customers’ internal systems, and is also embedded into sophisticated Excel and Quantrix interfaces. We are working with several other vendor partners to make the model available within their analytical systems. We are happy to help with integration and offer high-quality customer support on an ongoing basis to enable clients to take full advantage of these cutting-edge behavioral models.
The Credit Product line also currently includes the following tools:
HPI Generator - produces a Monte Carlo path of projected HPI at both a national and state level. While it is consistent with the AD&Co Interest Rate Model, the HPI generator can accommodate other firms’ interest rate models as well.
|
Implied Default Model - draws on AD&Co’s advanced OAS technology to compute the value of ABS tranches under a range of prepayment and default scenarios, relating bond prices to a distribution of defaults and losses. In the near future, the implied model will be integrated with the newly released LoanDynamics™ Model.
|
Risk Profiler™ is AD&Co.’s database-driven (MS SQL or Access) valuation solution that incorporates standard and advanced valuation techniques including OAS, prOAS, credit OAS (with the LoanDynamics™ Model) and derivative pricing. A thouroughly designed front-end exposes market data, valuation options, prepayment tunings and results. Multi-dimensional risk measures can be compiled for large portfolios of MBS, ARMs, CMOs (including credit impaired instruments) and standard rate derivatives to produce portfolio or strategy level summary reports. |
We are committed to further development and enhancement of the LoanDynamics™ Model and other credit valuation tools. Please contact Rob Landauer (rob@ad-co.com) for more information.
Credit in The Pipeline |
Six Degrees of Separation --
by Andrew Davidson , August 2007 |
| The Implied Expected Defaults of ABX 06-2 -- by Andrew Davidson, November 2006 |
| Active-Passive Vasicek -- by Alex Levin, November 2006 |
| A Framework for Market-Implied Defaults -- By Anne Ching, March 2006 |
| Modeling Home Prices as Dynamic Assets -- by Alex Levin, April 2006 |
| Data Availability and Its Implications for Behavioral Models -- By Kyle Lundstedt, PhD, September 2005 |
| Modeling Mortgage Risk: Definitional Issues -- By Kyle G. Lundstedt, Ph.D., April 2005 |
| The Competing Risks of Mortgage Prepayment and Default -- By Kyle G. Lundstedt, Ph.D., February 2005 |
| Forecasting Conditional Default Rates for Manufactured Housing ABS -- By Anne Ching, October 2003 |
|