Data Overhaul Promotes More Accurate RMBS Ratings and Faster, More Efficient Transactions
A three-year, groundbreaking project to modernize data standards for private-label, residential mortgage-backed securities (RMBS) is nearly complete. Spearheaded by SitusAMC Senior Director Julia Curran, the effort involves updating a massive set of data with unified definitions, enabling any party in the industry to more accurately evaluate the quality of a loan asset. The new standards, currently open for comment until mid-November, also pave the way for faster, more efficient securitizations.
“The quantity of data needed to actually understand a mortgage loan in today’s world is massive,” Curran said. “This data standard allows everyone to communicate seamlessly – nothing gets lost in translation.” We recently spoke with Curran, who leads SitusAMC’s securitization efforts, about the data project and its pervasive implications for the industry.
What inspired this project?
Back in 2009, the American Securitization Forum (ASF) created a data set with specific definitions, which has been used by agencies such as Standard & Poor’s, Moody’s and others to rate private-label RMBS. These ratings give investors comfort as to the quality of loan assets. There’s a difference between that data and the standardized data set used by the government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac, because they offer standardized loan products. In the private-label market, these loans don’t fit the box of the GSEs, and there are multiple guideline requirements, so the private-label market needs its own data file. Unfortunately, ASF disbanded shortly after the creation of that data-set tape. There was no entity remaining to keep this data standard up to date.
What happened over the last dozen years that required the data set to be revised?
A lot of new products entered the market, especially non-qualified mortgage (non-QM) products, which may use creative and non-standard methods to qualify loans. For example, things like bank-statement loans, in which borrowers can be qualified based on bank statements; or CPA-letter loans, which allow for accountant certification of self-employment income. These new loans all had different types of documents, different levels of documentation, and different time periods for the documentation. In addition, the industry developed something called the TILA (Truth in Lending Act) designation – it could be a QM loan, non-QM loan, a safe harbor QM loan – which spoke to the risk of a specific loan. This information just didn’t have a home in that ASF tape.
How did the industry deal with the proliferation of data?
Each of the third-party review (TPR) firms created their own extra data sets, adding fields requested by the rating agencies. Unfortunately, there was no single definition for each of those new fields. So, what SitusAMC might consider cash-out, one of the other TPRs might define differently, because “cash out” could be comprised of a bunch of different combinations of numbers. That happened across many fields. We eventually added 100 fields to the ASF tape with no definitional consistency behind them. Oftentimes a securitization will have loans in it reviewed by more than one TPR and it becomes a mess to put those securitizations together, when each firm had a different format, and unique definitions. There was a lot of back and forth and wasted time.
How did you get involved in this project?
I’m co-chair of the private-label securitization (PLS) development work group at MISMO®, The Mortgage Industry Standards Maintenance Organization. MISMO develops the standards for the mortgage industry, which are used by every type of entity involved in creating mortgages, and required by most regulators, housing agencies and the GSEs. I reached out to MISMO and Geran Combs, Managing Director IT & MISMO at Actualize Consulting, who is an expert on data structure. We collaborated on the project together.
What did the work involve?
We took the universe of the ASF tape to understand all the data points associated with it and also took all of the supplemental fields the rating agencies had been requesting to fill the data gap. After aggregating that large data set, we went to each rating agency and asked for input on what fields they still needed, which ones they no longer use, and additional fields we should add. Then we just started at the top and went down data point by data point. We incorporated the structure of MISMO in the hierarchy, and started mapping. We had input from the originators, RMBS issuers and rating agencies to develop the data set. It took us three years to update the definitions. We also kept the linkage back to the ASF tape, so users would be able to track back to a file they have been using for years, helping to smooth the transition.
What’s the outcome and why is this standard important?
The standards mitigate risks related to not being completely clear on what has been originated. Issuers and investors have clearer insight on the quality of the assets. In addition, a single standard helps the industry work more efficiently. SitusAMC relies heavily on technology, and on repeatable processes. If we know the definition of a field, we can code that into our systems, so the answers and data we produce will be consistent, and help eliminate human error. From a cost perspective, it means the time to market is faster.
The project is out for open comment until the middle of November, then we must aggregate the comments and make changes. It will likely be ready for implementation by the second quarter of 2022.