Charting The High-Tech Evolution Of Loan Portfolio Analysis

12017_knowledge_is_power Charting The High-Tech Evolution Of Loan Portfolio Analysis REQUIRED READING: In February, 49 states signed a deal with the five largest banks to settle legal action concerning foreclosures and defaults. As part of the terms of this agreement, the banks will spend $25 billion to help underwater borrowers refinance and to compensate those foreclosed on improperly.

While the deal only applies to the largest banks, all servicers should be prepared to deploy more sophisticated, defensible portfolio management practices that focus heavily on collateral risk management. Servicers will need to evaluate their portfolios and make difficult risk-based decisions using the best information available to determine the best course of action.Â

Unfortunately, traditional portfolio-review methods typically have not incorporated vital property- and loan-level data, which limits the ability to create current and accurate risk models and modification policies. To counter these limitations, servicers are turning to sophisticated analytic solutions that can be overlaid on top of deep property data.

By incorporating a handful of simple best practices into portfolio analysis, servicers are able to realize two critical advantages:

  • Build more accurate loss severity, prepayment, default and pricing models; and
  • Identify problems with individual loans more quickly, allowing for more appropriate, immediate action.

The evolution in best practices for portfolio analysis begins with integrating post-origination data and then dives deep into local pricing trends using new valuation solutions and strategies designed to better reflect today's market trends and realities.

The biggest mistake servicers often make is to ignore the wealth of post-origination intelligence available to the market. Changes in borrower behavior, such as the addition of new liens or changes in occupancy or local market conditions, significantly impact the more significant driver of loss risk: combined loan-to-value (CLTV). CLTV is considered one of the most significant risk factors, because a homeowner's willingness to stay in a home, negotiate modifications or enter into default is often based on whether the homeowner owes more on the property than it is currently worth.Â

It is also important to consider the ‘C’ in CLTV. Servicers must understand the ‘combined’ loan-to-value to make the right decision. If they are only considering the LTV on the loan they secure and not all others on the property, they will again be basing critical decisions on incomplete intelligence and opening themselves up to additional risk.Â

Furthermore, risk models must account for wide variances in distressed-property values. Calculating accurate property valuations is the key factor in evaluating distressed properties and developing effective modification, default or refinance strategies. Every property, even ones with similar characteristics, will have a different valuation based on the stage of default, liquidation strategy and location of the property.

Most servicers have not been able to factor in how the stage of default or liquidation strategy impacts property values. Servicers can now use specialized automated valuation models to provide a clear view of the discounted value that can be expected at different stages of default, as well as the impact of locations on the price.

An analysis of more than 117,000 distressed properties in Los Angeles County shows that value clearly drops as a property works its way through the default process. Specifically, discounts average 2.5% for short sales, 6.6% for properties that go to real estate owned (REO) liquidation, and 14.6% for properties that go to an auction sale.Â

Any analysis to understand general property valuation trends must be done at the neighborhood level. Factoring in the level of default is only half the story. Evaluating trends – both pricing and discount – at a nationwide, metropolitan or even county level is a mistake. There are significant variations at the ZIP code and neighborhood levels that must be considered in order to make truly accurate risk decisions.

Even the worst counties have pockets of recovery, and the best counties have plenty of problem areas. Servicers should use ZIP code or neighborhood-level evaluations on any number of key metrics to identify these pockets within a larger geographic area. These metrics include the following:

  • Average home price;
  • Number of notice of default;
  • Number of notice of trustee sale;
  • Number of bank REOs;
  • Number of government REOs;
  • Total REOs;
  • Number of foreclosure auction sales;
  • Number of short sales;
  • Number of REO liquidation sales; and
  • Number of non-distressed sales.

Finally, it is important to take advantage of additional loan-level data to predict likely default candidates. In addition to CLTV, there are two other primary indicators of default. Based on data collected during a recent portfolio review initiative for a larger national lender, two of the best default predictors are occupancy status and the presence of second liens.

Research shows that non-owner-occupied properties have a much higher incidence of delinquency (50%) than owner-occupied properties (31%). When occupancy and the presence of a second lien are combined, the results are more dramatic: Owner-occupied properties without a second lien have a default rate of 22%, compared to 66% for non-owner-occupied properties with a second lien.

Ultimately, the accuracy of loan- and portfolio-level decisions depends on the underlying intelligence used in these decisions. The more local, more current and more nuanced property and market data a servicer can access and deploy, the more intelligent and informed the resulting risk decision will be. Utilizing this type of intelligence across an entire portfolio significantly improves the efficiency of the portfolio review and, most importantly, allows servicers to make better modification decisions and lower their overall default risk.Â

Randy Wussler is vice president of product management and marketing for San Diego-based DataQuick. He can be reached at (858) 597-3100.


Please enter your comment!
Please enter your name here