Who knew Forbearance Analysis was so complicated?

We did and it is frustrating how such an important data element is not readily available and even when available, not always accurate. There is also no consistency.

In the governing documents, forbearance is not listed under the required information of the monthly statement to certificate holders. Most Trustees have added modification data to their remittance statements and three have added extensive modification data to their loan level files. Unfortunately, the data is not always accurate and/or complete.

In our attempt to help our clients, we are providing our Forbearance Analysis spreadsheet. It is probably the best source of forbearance data available. There are 2 sheets:

Forbearance Outstanding – we tried our best to take all the sources available and sum up the total outstanding forbearance by deal and how it is distributed among delinquency buckets. We also show the actual amount of forbearance recovered over the past year, which is derived from the monthly loss data provided by CoreLogic. Although we can see recovery from the CoreLogic data as a negative loss, the CoreLogic data does not provide us with a denominator to use to calculate a recovery rate. The prior cumulative loss cannot be used as it can be forgiveness, forbearance or something else. The best we can do is across all deals is to show the estimated amount of recovery. This will give you an idea of how much is being recovered versus how much is outstanding. It is not much on that basis. The column “Webbs Hill Comments” lists our opinion on the quality of the data source.
Forbearance Recovery – we only did this analysis on data provided by Wells Fargo and Bank of New York Mellon. We believe the Wells Fargo data is accurate and the Bank of New York Mellon data is fairly accurate. This analysis is dependent on having an accurate outstanding forbearance immediately prior to payoff or liquidation (a denominator), so we can estimate a recovery rate. The recovery rates are high for current and up to 60-day delinquent loans, but drops low when the loan is seriously delinquent. We can only assume that these recovery rate apply across the board. There will always be errors made by servicers, but forbearance is actually getting recovered.

We will do our best to improve this analysis as we make more discoveries.

Why can’t investors know for sure how much forbearance is outstanding?

Generally, forbearance and forgiveness looks the same at the modification effective date. Whether forgiveness or forbearance is provided by the servicer or a combination of the two, a loss equal to the debt forgiven or deferred is passed through to the trust at the time of modification. Most of the unrecognized forgiveness/forbearance has been corrected.
Wells Fargo, Bank of New York Mellon, and US Bank provide extensive modification fields on their monthly loan level data including separate fields for forgiveness and forbearance. Deutsche Bank does not. Although extensive data is provided, it is not always accurate (see item #3).
HAMP had a “PRA incented forgiveness program” whereby forbearance is eligible for forgiveness over a three-year period provided the borrower is not in default such that the equivalent of three full monthly payments are due and unpaid on the last day of any month, on each of the first, second, and third anniversary. Generally, if the borrower does not go delinquent 90+ days, the servicer reduces the eligible forbearance by one-third each year. In many cases, the HAMP modification provides for both forbearance eligible for forgiveness and forbearance not eligible for forbearance. The US Treasury HAMP data clearly shows this breakdown. We believe
Wells Fargo correctly updates the outstanding forbearance after a forgiveness event
Bank of New York Mellon mostly but not always updates the outstanding forbearance after a forgiveness event
US Bank never updates the outstanding forbearance after a forgiveness event and continues to report the original amount of forbearance.
Ocwen generally reports forbearance correctly but only for HAMP mods. Oddly, REALPortal does not track forbearance provided by Ocwen’s proprietary modifications.
Nationstar Master Servicing provides forbearance data but like Bank of New York Mellon, it is mostly accurate but not always. No way to know when it is right or wrong.