Rising interest rates, the end of steadily rising home prices – the potential for borrower default has never been greater. But for many institutions modeling the risk of loan default is treated as an afterthought or is crudely approximated by limited, ad-hoc measures.
LPS Applied Analytics announces its Loan Default Model™, the latest tool in a suite of software applications and models developed for MBS professionals that characterize and project the propensity toward prepayment and default for mortgage backed securities of all types.
Overview
The Default Model is complimentary to the LPS Applied Analytics Prepayment Model™ and MBS Calculator Library. In addition to standard collateral description inputs, initial populations of several delinquency cohorts (0, 30, 60, 90+ days late and Foreclosure), original LTV, MSA (metropolitan statistical area) and LPS Applied Analytics Default Score are used.
A future HPI (housing price index) scenario – in the form of a vector of projected monthly home prices over the term of the collateral for each relevant MSA – forms the primary driver for default projections. The model has been extensively fit against historical loan data calculated using actual HPI for a broad range of MSAs.
The model projects prepayment rates, the rates of loan delinquencies and defaults, the dollar balance lost each month due to defaulted loans (severity) and the loan populations across a range of delinquency cohorts.
Modeling Approach
The Default Model implements a behavior based approach to modeling economic activity. Its fundamental drivers are not abstract macro-economic conditions, but instead are factors that are directly experienced by individual mortgage holders.
There are many variables that influence default propensity – enough to make simultaneous aggregations by each of them a practical impossibility. To address this, the Default Model uses the LPS Applied Analytics Default Score™ – a metric based on time-independent attributes – as an input to compress the dimensionality of the problem. The score captures the combined impact of several loan attributes recorded at origination and represents them in modeling default propensity.
A set of downward transition functions (one pertaining to each of the factors that drive the model) and a state-transition matrix implement the model’s methodology.
Characteristics
The Default Model combines detailed loan data, projections of future interest rates and home prices in the relevant MSAs, with LPS Applied Analytics’ proprietary algorithms to generate results. Input collateral must be bucketed by MSA, Original LTV, Origination Date, and Loan Type.
Inputs to the model include:
- Original LTV
- FIS Applied Analytics Default Score – proprietary measure representing the collective effects of several loan characteristics on the propensity to default
- FIS Applied Analytics Prepayment Score – proprietary measure representing the collective effects of several loan characteristics on the propensity to refinance or sell
- Loan Age
- Projected WAC
- HPI Scenario – future housing price movements for the locations relating to the loans being analyzed
- Projected Mortgage Rates
The model outputs monthly projections for the remaining life of the subject collateral (max 360 months) for the following:
- Delinquency Cohorts – proportion of loans projected to be delinquent by less than 30 days, between 30 and 59 days, between 60 and 89 days, and 90 days or more
- Foreclosure – proportion projected to be in default
- Prepayment – proportion projected to prepay
- Default – percent of loans projected to default
- Loss (Severity) – percent of the total outstanding principal balance lost due to projected defaults
Architecture
The LPS Applied Analytics model is open – allowing full access to the factors that drive output projections. Users can review and modify scores, interest rates, fixed cost assumptions or HPI vectors using a simple text editor. Plus, LPS Applied Analytics makes it easy to integrate the model into internal programs or 3rd party systems.
Client Service
Comprehensive customer service – ranging from technical implementation to tactical consulting – is part of your license agreement and provided at no additional cost. LPS Applied Analytics professionals have years of experience working for and with major Wall Street firms, institutional investors, hedge funds and sophisticated private investors. They understand far more than theory and formulae – they get what you need to do right from the start.
For more information contact LPS Applied Analytics at 1.415.989.9800.