SPECIAL MASTER ORDER #54
ROBERT D. MARIANI, District Judge.
THE BACKGROUND OF THIS ORDER IS AS FOLLOWS:
One of the claims presented by Plaintiff Consumer Financial Protection Bureau (the "Bureau") against Defendants (collectively referred to as "Navient") is that an incorrect code was reported by Navient to "National Credit Reporting Agencies" ("NCRAs" (also known as "credit bureaus")) for student loan borrowers whose educational loans had been discharged due to the student borrowers' total and permanent disability ("TPD"). Specifically, the Bureau contends that from 2012 to 2014, Navient improperly used special comment code "AL" when reporting to NCRAs that a student borrower's educational loan had been discharged due to TPD. The Bureau asserts that student borrowers' credit scores were adversely affected as a result of Navient's use of the AL special comment code, and that this adverse impact on credit scores caused the student borrowers to suffer various types of harms, including higher interest rates on mortgages, car loans, etc.
In support of its position, the Bureau has tendered the August 8, 2019 report of Michael A. Turner, Ph.D. To ascertain the impact on credit scores resulting from Navient's reporting of the AL code, Dr. Turner contracted with a credit reporting agency, TransUnion. He tasked TransUnion with running simulations on more than 19,000 student loan borrowers whose education loans were discharged for TPD and for whom Navient reported the AL code during the years 2012 through 2014. TransUnion was asked to report the credit scores with and without the AL code for affected student loan borrowers on an anonymous basis at the end of each quarter for the years in question using a computer-based credit scoring program known as VantageScore 3.0.1 According to Dr. Turner:
For each of [the] quarterly snapshots, two versions of credit scores (VantageScore 3.0) and attributes were returned. One version shows the credit file as it was at the time, with no changes to the credit file data (i.e., the borrowers had the "AL" code, as furnished by Navient). The second version was based on the same set of files with the Special Comment Code "AL" removed. It was replaced with the previous code that may have been displaying prior to the Special Comment Code "AL" being added. Account Status Code was displayed as "05", as of the date when the Special Comment Code "AL" was incorrectly added. This second version shows what the credit scores and credit profiles would have been if Navient had simply not added the Special Comment Code "AL" and had left other aspects of how the account was reporting unchanged (such as the Payment Rating Code).
(Turner Report, (Ex. 1 to Navient's letter of Aug. 27, 2019) at p. 36.)
Navient raised the sufficiency of the Bureau's production of information related to Dr. Turner's report during a conference call held on August 22, 2019. Consequently, the parties were directed to provide letter briefs on the issue and to address the matter during a subsequent conference call. (See Special Master Order #42.) By Special Master Order # 46, the Bureau was directed to produce by September 13, 2019 the TransUnion data received by Dr. Turner as well as copies of all written communications between Dr. Turner and TransUnion related to this litigation. By Special Master Order #49, dated September 20, 2019, Navient was directed to "report in writing, no later than September 25, 2019, whether the information received in connection with the VantageScore 3.0 simulation ran by TransUnion is sufficient to assess the propriety of the methodology employed by [Dr. Turner], . . . or whether additional information is needed or whether the matter is one that must be presented by a Motion in Limine challenging Dr. Turner's report." Navient's letter of September 25, 2019 contended that the TransUnion data produced in response to Special Master Order #46 was not adequate. Moreover, Navient's letter took issue with the Bureau's explanation of the nature of the work performed by TransUnion, pointing out that TransUnion had performed services over a period of at least six weeks and for which it had been paid at least $265,000. Contending that it was unable to replicate the work performed by TransUnion without receipt of significant additional information, Navient asked that the Bureau be ordered to:
[D]isclose . . . no later than October 9, 2019, the details of any analysis, manipulations, and/or simulations performed by and the supporting data/documentation used by TransUnion pursuant to the Research Data Services Agreement . . ., including, but not limited to, the following categories of information:
1. A copy of the specific VantageScore 3.0 iteration(s) that was used in TransUnion's simulations and modeling exercise. In addition, a description of the methodology and process that this VantageScore 3.0 iteration used to weight Metro 2 special comment code "AL."
2. All iterations/version(s) of VantageScore that were in use between January 1, 2012 and December 31, 2014. In the alternative, a description of the differences present in each such iteration from the version that was used for the simulations and modeling exercise. In addition, a description of the methodology and process that each of these VantageScore iterations used to weight Metro 2 special comment code AL, including any modifications made to the scoring model during the specified time period.
3. The inputs into the "quarterly archive files," including the underlying credit files and tradelines and TransUnion's treatment of certain Metro 2 special comment code AL.
4. The methodology used to process the "Input Files" received from Dr. Turner against the TransUnion credit archives, including what fields were used to match and identify the approximately 19,000 consumers to specific tradelines.
5. The manner in which TransUnion's "quarterly credit archives" were constructed, including whether each quarterly archive was a snapshot of a credit file on a certain date
or was an aggregation of a time period within that quarter, and whether the archive
incorporated information reported within the quarter, or information effective in that quarter.
6. The manner in which corrections or updates made to a consumer's credit file were incorporated into the "quarterly credit archives."
7. The relationship between TransUnion's proprietary remark code and the Metro 2 special comment code "AL" and whether TransUnion considers Special Comment Code "AL" to a be a field that needs to be reported each month it applies.
8. The specific modifications performed pursuant to the simulation, including what specific code(s) were substituted for Metro 2 special comment code AL — "Assigned to Government" in TransUnion's modification to produce the "Modified Observation Period Archive Dates," and whether these modifications were made only to Navient student loan tradelines.
9. Whether TransUnion ran any other simulations or modeling in connection with the work it performed for Dr. Turner in connection with this lawsuit.
10. The name(s) of the TransUnion personnel that performed the work pursuant to Dr. Turner's Research Data Services Agreement.
(Ex. 7 to Navient's September 25, 2019 letter.)
The Bureau responded to Navient's September 25th letter on October 3, 2019. The Bureau's letter included Dr. Turner's Declaration along with several other exhibits. The Bureau asserted that Dr. Turner's report and Declaration provided the information sought in paragraphs 4 through 10 of Navient's proposed order, and that neither Dr. Turner nor the Bureau were in possession, custody or control of the information sought in paragraphs 1 through 3 of the proposed order. Navient responded by letter dated October 7, 2019, contesting the assertions that the Bureau lacked the ability to produce the information sought in paragraphs 1 through 3 of the proposed order and that the Bureau had otherwise produced the information sought in paragraphs 4 through 10. Navient stated:
Defendants expected to receive all information provided to and relied upon by Dr. Turner in forming his opinions, including anonymized credit files. In his report, Dr. Turner explicitly stated that TransUnion "provided [] credit files in completely anonymized form," Turner Report, ¶ 57, and during the last status conference, CFPB counsel represented, "TransUnion pulled the historical versions of those borrowers' credit files" and "then provided Dr. Turner with those historical versions on a quarterly basis," Tr. at 10:24-11:3. Yet, this is not what Defendants received. Nor does the other information Defendants have received fill in the gaps. Dr. Turner admits that "my email communications with TransUnion, which I produced previously, present an incomplete picture of my communications with TransUnion." Turner Declaration, ¶ 9. He also admits that "how their data architecture is designed and how they [TransUnion] internally track a code" is "unique." Id. at ¶ 10.d. Moreover, the VantageScore 3.0 model has changed over time. And Dr. Turner apparently does not know what specific iteration of VantageScore 3.0 was used by TransUnion, or how it differs from the older versions in place during the relevant time period.
Without this information, there is no way to test whether the change in credit score that resulted from the simulations run by TransUnion is a reliable assessment of the impact of the "AL" code on each of the 19,509 borrowers between January 1, 2012 and December 31, 2014. Contrary to the CFPB's statements, Defendants cannot replicate the custom simulations performed by "[Dr. Turner's] assistants employed by TransUnion," CFPB Oct. 3 Letter at 4, because they lack the undisclosed data inputs used by TransUnion and other key information about the simulations and data manipulations performed by TransUnion (such as what code replaced the comment code "AL" in the modified credit files).
(Navient letter of October 7, 2019 at 2-3.)
Oral argument on the question of ordering the Bureau to produce the information sought in Navient's proposed order was held telephonically on October 9, 2019. During the course of the argument, Navient acknowledged that Dr. Turner's declaration had provided the information requested in paragraphs 4, 9 and 10 of the proposed order.2 Accordingly, what remains to be decided is whether the Bureau should be ordered to produce the other information sought in Navient's proposed order. The seven paragraphs of that order that remain in dispute will be addressed seriatim.
Paragraph 1 of Navient's Proposed Order: A copy of the specific VantageScore 3.0 iteration(s) that was used in TransUnion's simulations and modeling exercise. In addition, a description of the methodology and process that this VantageScore 3.0 iteration used to weight Metro 2 special comment code "AL."
The Bureau represents that neither it nor Dr. Turner is in possession of VantageScore 3.0, a credit scoring algorithm that is the intellectual property of TransUnion. Navient rejoins that the Bureau's expert, Dr. Turner, controlled the work performed by TransUnion and so the unavailability of VantageScore 3.0 is attributable to his failure to contract for its production. Dr. Turner, however, has asserted that VantageScore 3.0 is an accepted credit scoring model. It may be, therefore, that he was entitled to rely upon the results produced by this algorithm without receiving the algorithm itself. That question, however, is premature. The question is whether the Bureau must produce the algorithm employed by TransUnion under its contract with the Bureau's expert witness, Dr. Turner. Because the algorithm is not in the possession of the Bureau or Dr. Turner, they cannot be compelled to produce it. Navient, however, retains the right to challenge the admissibility of Dr. Turner's opinions due to the unavailability of the algorithm that produced the results upon which Dr. Turner relied.3
Paragraph 2 of Navient's Proposed Order: All iterations/version(s) of VantageScore that were in use between January 1, 2012 and December 31, 2014. In the alternative, a description of the differences present in each such iteration from the version that was used for the simulations and modeling exercise. In addition, a description of the methodology and process that each of these VantageScore iterations used to weight Metro 2 special comment code AL, including any modifications made to the scoring model during the specified time period.
The information sought in paragraph 2 of Navient's proposed order is not in the possession, custody or control of the Bureau or its expert. The questions raised by the information sought by Navient may bear upon the reasonableness of Dr. Turner's reliance upon VantageScore 3.0, but the information does not fall within the ambit of Federal Rule of Civil Procedure 26(a)(2)(B).4 Accordingly, the Bureau will not be compelled to produce the information sought in paragraph 2 of Navient's proposed order.
Paragraph 3 of Navient's Proposed Order: The inputs into the "quarterly archive files," including the underlying credit files and tradelines and TransUnion's treatment of certain Metro 2 special comment code AL.
Navient strenuously argues that it needs access to the credit files of the more than 19,000 student loan borrowers for whom TransUnion ran credit scores pursuant to Dr. Turner's directions. Navient's desire for the credit files is understandable, as the credit files were a basic input into the VantageScore 3.0 simulations. I understand that TransUnion produced credit scores for these borrowers by removing the AL code and the resulting score was compared against the borrowers' scores with the AL code present. Navient would need the files to test TransUnion's results. But once again, the files were not produced by TransUnion to Dr. Turner. Instead, Dr. Turner received certain anonymized credit file data, and he did produce that data to Navient. Accordingly, neither the Bureau nor Dr. Turner can be compelled to produce the credit files. It may be, however, that Navient could pursue discovery from TransUnion to obtain the credit files, and if it needs leave to do so, leave will be granted.5
Paragraph 5 of Navient's Proposed Order: The manner in which TransUnion's "quarterly credit archives" were constructed, including whether each quarterly archive was a snapshot of a credit file on a certain date or was an aggregation of a time period within that quarter, and whether the archive incorporated information reported within the quarter, or information effective in that quarter.
Dr. Turner's Declaration addresses this question in a confusing manner. On the one hand, he suggests that a simulated score was produced if the AL code "was present during the applicable period." (Turner Decl. at p. 6.) On the other hand, he states that "[i]f Navient had not yet reported the AL code for a borrower or had already removed the AL code for that borrower, then a quarterly snapshot was not provided for that borrower because the AL code would not have been present." It is thus not clear that a simulated score was produced only if the AL code was present on the last day of the quarter in question, or was produced if the AL code was present at any time during that quarter. Accordingly, the Bureau will be required to provide a more coherent explanation of how the "quarterly credit archives" were constructed.
Paragraph 6 of Navient's Proposed Order: The manner in which corrections or updates made to a consumer's credit file were incorporated into the "quarterly credit archives."
This requested information, which is evidently related to the issue of whether TransUnion took a "snapshot" of a credit file as it existed on the last day of a particular quarter or compiled an aggregate of the data appearing in the credit file during that quarter, does not appear to have been addressed by Dr. Turner's Declaration. Accordingly, the Bureau will be directed to provide a forthright explanation for how the "quarterly credit archives" were constructed.6
Paragraph 7 of Navient's Proposed Order: The relationship between TransUnion's proprietary remark code and the Metro 2 special comment code "AL" and whether TransUnion considers Special Comment Code "AL" to a be a field that needs to be reported each month it applies.
This particular request does not appear to have been addressed in the Turner Declaration. Accordingly, the Bureau will be directed to provide the information sought in paragraph 7 of Navient's proposed order.
Paragraph 8 of Navient's Proposed Order: The specific modifications performed pursuant to the simulation, including what specific code(s) were substituted for Metro 2 special comment code AL — "Assigned to Government" in TransUnion's modification to produce the "Modified Observation Period Archive Dates," and whether these modifications were made only to Navient student loan tradelines.
Dr. Turner's Declaration on this point appears to indicate that no code was substituted for Metro 2 special comment code AL when simulated scores were produced. Dr. Turner's Declaration, however, is not pellucid on this point. If no modifications were made, he should say so. If no code was substituted for special comment code AL, he should say so directly. Accordingly, the Bureau will be directed to provide a response to paragraph 8 of Navient's proposed order.
NOW, THEREFORE, THIS 15th DAY OF OCTOBER, 2019, IT IS HEREBY ORDERED THAT:
1. Navient's request that the Bureau be compelled to provide the information sought in the proposed order attached as Exhibit 7 to Navient's letter of September 25, 2019 is GRANTED IN PART.
2. Within seven (7) days from the date of this Order, the Bureau shall provide the information requested in paragraphs 5 through 8 of Navient's proposed order.
3. In all other respects, Navient's request is DENIED.
