Board Report: 2014-SR-B-015 September 30, 2014
During the slotting, review, and validation process, Board staff identified weak or missing data in two servicers' systems.20 Board staff determined that the data could not support the slotting of borrowers in specific categories. As a result, the two servicers could not complete the waterfall processes according to the categories defined by the Board and the OCC. To address these data reliability issues, Board staff directed these two servicers to move borrowers to higher categories or refrain from excluding borrowers from higher categories. Board staff explained that these solutions were a conservative approach that benefitted the borrowers affected by the data reliability issues by placing them into the highest possible waterfall category based on the available data. For example, one servicer reviewed a sample of its slotting results for 255 loans and identified 45 incorrectly slotted borrowers, or 18 percent of the sample. The servicer attributed the incorrect slotting to inaccurate system data. Board staff instructed this servicer to not slot any borrowers by manual intervention.
The guidance for the Reserve Bank dedicated teams' review of the slotting sought to ensure consistency to the extent possible. The servicers' internal audit functions sought to assess the completeness and accuracy of the servicers' waterfalls. In our opinion, consistency to the extent possible was a guiding principle for this exercise, although the scope of that principle was not clearly defined. Board staff directed servicers to move borrowers to higher categories or not exclude borrowers from higher categories because the Board (1) emphasized completing the payment agreement as quickly as possible and (2) sought to avoid repeating the file review process. As a result of the servicers' movement of borrowers into the highest possible category, some similarly situated borrowers may have received different payment amounts.
During the Reserve Bank teams' reviews of the servicer's internal audit validation of the slotting process, Board staff identified weak or missing data in the systems of two servicers—hereafter referred to as Servicer A and Servicer B. Because the data were not sufficiently reliable to support moving borrowers out of specific waterfall payment categories, Board staff instructed these two servicers to move borrowers to higher waterfall categories or refrain from excluding borrowers from higher categories. Generally, borrowers in the higher categories received a higher payment amount.21 Moving borrowers into a higher waterfall category, or keeping borrowers in a higher waterfall category, benefitted those borrowers since they would likely receive higher payment amounts. Servicers completed the slotting process before Board and OCC staff determined the actual payment amounts associated with each of the categories. Therefore, the precise payment amounts did not affect the Board's or the OCC's decisionmaking with respect to options for addressing the data reliability issues.
Each Reserve Bank dedicated team drafted a conclusion memorandum to summarize the results of its review of the slotting process at each mortgage servicer. The Reserve Banks shared those memorandums with the Board to evidence completion of the review. According to the Servicer A Reserve Bank conclusion memorandum, Servicer A self-identified issues with the reliability of its servicing system data fields used to populate certain waterfall categories. To compensate for the data reliability issues, the servicer sampled 255 loans and performed a manual review of the system query results to confirm appropriate waterfall placement. As a result of the manual review, Servicer A reslotted a total of 45 accounts across multiple waterfall categories from the sample of 255 loans (18 percent). Out of the 45 incorrectly slotted borrowers, 36 were incorrectly slotted in what the Reserve Bank dedicated team referred to as lower-tier categories, categories 7 through 11. Thus, based on the sample, the bulk of the servicer's slotting errors were concentrated within lower waterfall categories.
The Reserve Bank dedicated team communicated to the servicer that the reslotting of accounts based on the servicer's manual intervention would not be consistent with the expectations set forth in the IFR waterfall definitions and the terms of the payment agreement. According to the Reserve Bank conclusion memorandum, the payment agreement terms required the servicer to slot borrowers using an objective and systematic process rather than a manual process. Board officials noted that the IFR was a time-consuming, manual process involving loan file reviews at the individual-borrower level, while the payment agreement sought to provide timelier remediation using an objective, systematic process. Therefore, slotting borrowers based on a manual process did not align with the terms of the payment agreement. As a result, the servicer was not permitted to slot borrowers to lower categories based on a manual review. Instead, the servicer revised its waterfall to reslot the 45 accounts back into the originally identified higher payment categories, even though manual intervention identified that those results were not reliable.
Both the servicer and its internal audit function agreed that no alternative data systems existed with data capable to slot borrowers more accurately into the waterfall. According to the Reserve Bank conclusion memorandum, the definitions set forth in the waterfall guidance state that all accounts are to begin at the top of the waterfall (category 1a) and then flow down into lower waterfall categories only when servicing system data evidence that they do not meet the definitions of each sequential category. Under Board staff's guidance, the Reserve Bank dedicated team communicated to the servicer that the data available in the servicing system were not sufficiently reliable to slot borrowers below category 7, and consequently, no accounts should be allowed to flow below that point in waterfall. Based on this directive, Servicer A truncated its waterfall at category 7, and approximately 83,000 borrowers who had been slotted into lower categories were moved to category 7, resulting in those borrowers being eligible to receive higher payment amounts.
During the review of Servicer B's slotting, the Reserve Bank dedicated team identified two instances in which the measures the servicer took "to compensate for weak or missing system data did not yield the most favorable outcome to the borrower." These instances also prevented movement of certain borrowers out of higher payment categories.
First, the Reserve Bank dedicated team recognized that the data used to place borrowers into the bankruptcy category, category 3, were weak or missing. Based on prior supervisory work performed during the IFR, the Reserve Bank dedicated team was aware that the servicer's system indicator for active bankruptcy was unreliable. After discussions among the Reserve Bank dedicated team, the servicer, and Board staff, the servicer agreed to purchase Public Access to Court Electronic Records (PACER) data and rely solely on this external data source to place borrowers into category 3.
The servicer extracted dates from PACER to identify borrowers in active bankruptcy at the time the foreclosure action occurred and determine whether the borrowers should be excluded from the category due to a court-granted motion for relief from stay on the borrower's mortgage loan, which would have permitted the foreclosure to proceed. Due to time constraints, however, the servicer was only able to extract the necessary PACER motion for relief from stay data for approximately 14,000 of 26,000 borrowers identified as having been in active bankruptcy. About 72 percent of those 14,000 were excluded from category 3 based on the motion of relief from stay data and moved to a lower category. Servicer B requested an extension to complete this process for the remaining borrowers. Board and OCC staff did not grant the extension, however, because it would have caused delays completing the overall payment agreement.
Granting an extension would have caused a delay because Board staff needed to know the total number of borrowers in each waterfall category before the Board and the OCC could finalize the payment amounts for each category. As such, the final waterfall payment amounts could not be determined until all the servicers submitted their completed waterfalls. Therefore, Servicer B slotted the remaining balance of approximately 12,000 borrowers for which it did not have time to extract PACER motion for relief from stay data in category 3, the bankruptcy category. Reserve Bank examiners were comfortable that the final number of loans in the bankruptcy category reflected a conservative approach to identifying borrowers in category 3 for this servicer based on the decision not to exclude borrowers.
The second instance of weak or missing data at Servicer B related to the modification status of borrowers' mortgages. In certain instances, Servicer B did not retain or could not determine the modification status of the borrower and slotted those borrowers into category 8.22 However, Board staff instructed the servicer that if it could not determine modification status, those borrowers should default to the highest applicable waterfall category, which was category 7. As a result of this directive, approximately 60,000 Servicer B borrowers were moved to category 7 from 8. Board officials considered the actions taken by the servicer to be the most conservative approach because the actions resulted in placing borrowers in the highest possible waterfall payment category given the available data.
The Reserve Bank dedicated teams used the Supervisory Expectations as a project plan to monitor the servicers' implementation of the terms of the payment agreement. According to the overview section of the document, the Supervisory Expectations was "designed to outline the requirements of the exam teams and to ensure consistency, to the extent possible." We believe that consistency of results to the extent possible was a guiding principle for this exercise, although the Board's expectations for applying this principle were not clearly defined. Additionally, the scope of the document states that "the Dedicated Teams will direct the Servicers to have their waterfall independently reviewed by internal audit or independent compliance control function to assess the accuracy and completeness of the waterfall."
We attribute the approaches taken to mitigate the data reliability issues to two contributing causes: (1) the Board had a sense of urgency to complete the process as quickly as possible and (2) the Board conveyed broad objectives as guidance to the Reserve Bank dedicated teams that did not consider how to address the impact of data reliability issues on the results of the slotting exercise.
First, the Board and the OCC emphasized speed in the design and execution of the payment agreement. Under the IFR, the precursor to the payment agreement, independent consultants conducted file reviews for about 18 months, but no money had been provided to borrowers. In response to the slow progress of the IFR, the payment agreement emphasized providing remediation to borrowers as quickly as possible. The payment agreement term sheets stipulated a narrow window of 45 days for completing the Reserve Bank dedicated team reviews of the servicer internal validation testing for the execution of the slotting process. Given the sense of urgency and narrow time frame, Board staff explained that the payment agreement was designed by Board and OCC officials to accurately slot borrowers using an objective, systematic process rather than a manual process.
Second, Board staff provided broad objectives to the Reserve Bank dedicated teams that did not specify detailed expectations and procedures. Specifically, the Supervisory Expectations document includes four objectives. The second objective outlined that "the process should also document and assess known system or data gaps/deficiencies and major assumptions associated with completing the Waterfall. In addition, it should document whether and how system gaps/deficiencies were addressed in preparing the Waterfall." Board staff sought to provide the dedicated teams with general guidance and provide specific responses to data reliability issues on a case-by-case basis. While the document provided instructions, it did not provide specific guidance concerning which approaches the Board would consider appropriate to resolve data reliability issues. Feedback on specific approaches was provided during frequent interaction between Board staff and Reserve Bank dedicated teams.
The Supervisory Expectations document provided detailed procedural steps to complete the review of categories 1 and 2 but not for the lower categories. The detailed procedures included several steps to be completed specifically to understand and assess the servicer's slotting processes for categories 1 and 2. For categories 3 through 11, however, procedures for reviewing the slotting are not specifically outlined; instead, a set of general instructions were provided. These instructions for the Reserve Bank dedicated teams included reviewing the servicer systems and developing an understanding of the process by which system queries were built. According to the Supervisory Expectations, the review may also include understanding which servicer systems were used in the slotting process and which data fields were accessed. The procedures to review categories 3 through 11 rely on analysis of automated system queries of the servicers' data. Moreover, the Supervisory Expectations document did not identify a threshold for when data systems would not be considered sufficiently reliable to support the slotting exercise. According to Board officials, a threshold was not used because decisions were made on a case-by-case basis based on discussions involving the Board and Reserve Bank staff.
Because of the actions taken to mitigate data reliability issues and move borrowers into higher categories or refrain from excluding borrowers from higher categories, similarly situated borrowers may have received different payment amounts across the servicers. For example, per Board staff's instructions, Servicer A truncated its waterfall due to data that were not sufficiently reliable. As a result, approximately 83,000 borrowers originally slotted into lower-tier categories were slotted into a higher category and were eligible to receive a payment of $2,000 to $6,000. However, if those borrowers had not moved to the higher category, they would have received $300 to $800. Among the other 12 servicers, approximately 2 million borrowers were slotted into the lower-tier categories and were eligible to receive a payment of $300 to $800.
Because of the approach to resolving data reliability issues, borrowers at Servicer A and Servicer B may have received different payment amounts compared to borrowers at the other servicers. For example, at Servicer B, approximately 12,000 borrowers were included in the bankruptcy category (category 3) when they may have been appropriately excluded. Borrowers slotted in the bankruptcy category were eligible to receive a payment of $3,750 to $62,500. Servicer B's borrower population slotted into the bankruptcy category accounted for 56 percent of the total borrowers in category 3. Among the other 12 servicers, approximately 13,000 borrowers were slotted into category 3.
Board staff explained that it selected the actions taken to mitigate data reliability issues on a case-by-case basis to slot borrowers into the waterfall as accurately as possible. This approach sought to foster consistency by acknowledging the limitations associated with the available data. The Board and the OCC published the payment distribution plan after the servicers finalized their waterfalls. As such, the servicers and the Reserve Bank dedicated team members were not aware of the payment distribution amounts for the categories during the slotting process.
Data integrity issues impacted the reliability and consistency of the slotting results for two servicers. These data issues resulted in those servicers moving borrowers into the next-highest category or not excluding them from a higher category. As a result of these actions, some similarly situated borrowers at the other servicers may have been slotted into dissimilar categories and received different payment amounts.
We recommend that the Directors of BS&R, DCCA, and the Legal Division
The Director of BS&R, the Director of DCCA, and the Board's General Counsel agreed with our recommendation. In their consolidated response to recommendation 3, the Board officials indicated that they plan to leverage information learned throughout the IFR process and the implementation of the payment agreement when planning for similar complex enforcement actions in the future. This effort would include addressing the potential impact of any data reliability issues as warranted.
In our opinion, the actions described by the Division Directors and the Board's General Counsel are responsive to our recommendation. We plan to follow up on the Board's actions to ensure that the recommendation is fully addressed.
