LAPS
Loan Analytics and Pricing System
A complete model, valuation, tracking and reporting system for lending, loan servicing and securitization.
Table of Contents
Introduction
LAPS (Loan Analytics and Pricing System) was designed and developed by a group of mortgage operation, investment, trading and research professionals who have had ample successful experiences using model-driven analytical systems for making critical business decisions.
LAPS provides a complete yet a simple solution for compliance in loss projection and reserve calculation, i.e. CECL.
Modeling Philosophy
LAPS is designed on these firm beliefs:
- All models are biased
Models need to be constantly monitored, validated, and adjusted if necessary. Adjustments need to be saved, documented and validated as well. - All model systems are limited and incomplete
Businesses cannot be dictated by any model system. Model systems are tools that need to be constantly improved.
LAPS is different from most other model systems because it was developed not only to implement a statistical model, but also to be capable of dealing with potential changes and to be a complete analytical tool. It is designed to balance complexity and simplicity, efficiency and flexibility. While most models fail to work as the environment experiences dramatic changes, LAPS is built to last.
Model Structure
LAPS combines the most comprehensive data input, the latest modeling techniques and a well-thought-out implementation design into a system that provides not only a wide range of mortgage analytical solutions, but also a tool to implement a new model, manage existing models and deal with compliance. The unique advantages of LAPS can be summarized as follows:
- Default and prepayment models
- Estimation (Core Model) – One month transition probabilities from month t to t+1.
- Simulation – Likelihood of a loan being in a particular state at month t+T.
- Severity Model
Loss components, i.e. home price change, interest carrying, MI reimbursement and other costs are modeled separately with real data. Regional distressed real estate inventory and sales trend are incorporated into the severity model.
Scenario Manager
As a key integral part of the LAPS system, the Scenario Manager gives user the power to manage and define their own economic scenarios at desired geographic level. Statistics of regional economic data are readily available for constructing meaningful scenarios. The example scenarios are as follows:
Cash Flow
Cash Flow and Pricing Engine
LAPS has a built-in cash flow and pricing engine that takes the loan level projections from the models, aggregates them and generates prices at loan/group/pool level as specified by the user. A yield table is illustrated as follows:
Loan level cash flow and pricing information include CPR, CDR and severity etc. are calculated on monthly basis.
Validation and Recalibration
Model validation could be just as complicated, if not more, as model development if it is to be implemented properly and to run continuously. It becomes more challenging for transition probability models because of the number of models that need to be tracked.
LAPS automates the validation process which makes it easier for developers, users and auditors to work together on a standardized platform.
LAPS allows users to select a starting date for back-testing as long as the data is populated as far back. Validation results will be displayed in a standard report. They can also be output to data files for further analysis.
Validation at transition level enables users to identify the problem areas and therefore make reasonable adjustments.
Validation Chart: Example 1 – VPR (Voluntary Prepaymemt Rate) projection
A typical validation chart will include a projected CPR curve (red) and a realized CPR curve (blue) as of the starting date for back-test, and a projected CDR curve (yellow) as of today.
Validation Chart: Example 2 – Validation for each transition probability model
Automated Recalibration
LAPS is designed to automatically recalibrate every model based on the results of validation. Users can control this feature as well as choose a time period, i.e. past 6 months, past 12 months, etc., that a model is to be recalibrated to. It gives users a handy option to deal with model shortcomings and a changing market environment.
After recalibration is done, the calibration information can be saved and documented in LAPS. If a recalibrated model is used for any business decision making, this recalibrated model would have to be tracked and validated in the future. It is critical for model risk management. With LAPS, the process would be more systematic.
Reporting
LAPS can output very complete validation and projection information at both the loan level and the pool level for users to conduct more detailed analysis. The output can also be easily input to other third party analytical systems. The output will include:
- Historical transition probability, delinquency/default/loss/prepayment rates at pool level. Historical cash flow at both loan level and pool level.
- Projected transition probability at both loan level and pool levels.
- Projected delinquency rate up to any state (D30, D60, …, D180, …) at both levels.
- Projected CDR, CPR, cumulative default and loss, severity at both levels.
- Projected Basel CDR (PD) and Basel severity (LGD) at both levels.
- Projected cash flow and pricing information at both levels.