© 2024 Arizent. All rights reserved.

Synthesizing and Analyzing ABS Data: Keys to Sound Portfolio Management

With the worsening economic outlook, structured finance market participants have learned that it is necessary to read the fine print when it comes to managing their portfolios.

This has made data and technology providers an increasingly vital presence on the loan-level analytics front - especially with the slew of loan modifications expected in the coming months and deteriorating consumer credit conditions.

However, synthesizing the copious amounts of data is a challenge in itself. This is why service providers have refined their current models as well as unveiled a host of new platforms to help ABS collateral managers organize data more efficiently.

Indeed, portfolio managers need a single way to integrate different data sources and definitions into their risk controls and portfolio management, said Douglas Long, executive vice president of business strategy at Principia Partners, a software platform provider for managing structured finance operations and portfolios.

He noted that there is increased focus by data providers to expand their coverage by geography and asset class. "In the past, they have focused on key markets or asset types, but now every data vendor is redoubling their efforts to get broader coverage, which is key to transparency," Long said.

Principia provides standard interfaces to some of the main data providers and is currently working on new interfaces to its platform, Long said. "We have done a lot of work to standardize the use and reporting of data, bringing in the different forms our clients are working with," he said. "This facilitates the consistent deployment of data across the entire operation from portfolio management through to the risk oversight and accounting functions."

Recently, 1010data and Equifax teamed up to provide a statistical matching solution that allows customers to anonymously link current and historical borrower credit information to loan-level data. Leveraging 1010data's linking technology, the solution allows customers to match a borrower to a loan to monitor how consumer credit attributes and scores change over time, said Greg Munves, vice president at 1010data.

1010data allows clients to calculate loss severities, constant prepayment rates (CPRs) and cumulative default rates (CDRs) as well as carry out deal surveillance by providing an analytics platform through the Web that manages third-party data and allows users to upload and analyze their own information in conjunction with third-party datasets. "Managers don't have to worry about building massive IT infrastructure and updating data; they can focus on analyzing the data instead," Munves said.

An especially useful feature of 1010data's platform, given the most recent economic events, is the ability to identify modifications in pools of mortgage loans well beyond what is reported by servicers and trusts. "Investors can go through raw loan-level data and figure out where a loan mod may have occurred, and how that differs from what servicers and trusts report," Munves said.

Furthermore, with more than $1 trillion outstanding and conforming deals still coming through the pipeline, there remains an interest in agency and servicing data, he said.

But what's important is not only increasing transparency, data providers say, but also accessing and consistently implementing the right data.

Setting the Right Standards

More consistent data will create a level playing ground for investors, said Barrett Burns, president and chief executive officer at VantageScore Solutions. "Putting standards in place will be one major step toward putting credibility back into the market," he said, crediting the American Securitization Forum's Project RESTART for drawing attention to the need for monthly loan-level data across the market.

"The key is being able to compare like with like," Long said. "You could always get information, but getting it in a readily useable form that you can easily disseminate to all the interested parties was a challenge," he said. Long noted that historically, data providers have often been focused on one particular geographical area, location or product type - ABS or CMBS, for example. Although vendors still have a regional focus, there is more of a push for a global reach right now across the markets, he said.

VantageScore is also making steps toward standardization. The platform, a generic credit scoring model launched in 2006 and implemented by the three national credit reporting companies (Equifax, Experian and TransUnion), uses "characteristic leveling," the term for creating common definitions for data, according to Burns. Although lenders provide each of the bureaus with data at different times, an unavoidable cause of variation in the credit scoring process, these bureaus have also had different definitions of this data. This has created confusion and variation among credit scores, Burns said. "Now lenders using the program can interpret the data consistently, and consumers have less confusion about their scores," he said.

In September 2008, VantageScore announced a collaboration with Fitch Ratings to incorporate VantageScore into Fitch's ResiLogic 2.1 - its quantitative model for assessing credit risk on both an individual loan and pool level for residential mortgage loans. Fitch now has a FICO-based model and a VantageScore-based model for rating portfolios.

Burns said he expects to announce a similar initiative in the near future.

The company used an anonymous sample 7.5 million consumers in the June 2003-June 2005 time frame to build the scoring model, and just last month announced the completion of an annual revalidation - with a similar number of consumers within the June 2006-June 2008 time window. The program was initially developed in late 2006, during a volatile time period with consumer debt rising, new products popping up in the mortgage sector and a ramp-up of subprime and Alt-A. This has provided the platform with relevant initial data. But a revalidation is important, Burns said. "We want to make sure the platform remains highly predictive."

Not a Perfect Score

However, models are not without their limitations, a harsh reality of the recent credit crisis. While an accurate scoring model is needed to predict consumer creditworthiness, it does not factor in the what-ifs like unemployment, for example, Burns said. "The market must understand the limitations of models and data."

Ultimately, it has to be in the hands of the managers, Munves said. "They are the ones committing capital and taking risk. We make sure they have access to all the available information they need. We provide a platform to slice and dice the data however they want," he said.

Despite the revolving door of new managers entering and exiting the market, business has remained steady. While there are fewer major dealers, there has been a large pickup in hedge fund business, Munves said.

1010data manages loan-level mortgage data for more than 75 firms on the Street, including large and medium-sized dealers, hedge fund managers, asset managers and rating agencies, among others.

(c) 2009 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.

http://www.structuredfinancenews.com http://www.sourcemedia.com/

For reprint and licensing requests for this article, click here.
ABS
MORE FROM ASSET SECURITIZATION REPORT