Minter v. Wells Fargo Bank, N.A.

286 F.R.D. 273, 2012 WL 4903315

Opinion

MEMORANDUM OPINION

SUSAN K. GAUVEY, United States Magistrate Judge.

Before the Court is plaintiffs’ Motion to Compel defendant Wells Fargo Bank, N.A. and Wells Fargo Ventures LLC (collectively “Wells Fargo”) to respond to interrogatories seeking fees paid to or charged by Defendants in Class members’ mortgage loan transactions. (ECF No. 322). Briefing is complete. The Court held two hearings on July 26, 2012 and October 9, 2012. Between the two hearings, the parties worked admirably together to better understand the data systems and the needs of the case presentation and have narrowed the dispute significantly.

I. Background

A. The Underlying Action

Plaintiffs’ allege that Wells Fargo, working with Long and Foster, created Prosperity Mortgage, a sham Affiliated Business Arrangement, in order to generate unlawful fees and kickbacks. (ECF No. 263, 2). Prosperity Mortgage is a joint venture of Wells Fargo and Long and Foster. (Id). Customers seeking a loan from Long and Foster are referred to Prosperity, whose loan officers work out of Long and Foster’s office. (Id at 8). Plaintiffs allege that Prosperity has no functional capacity to provide these loans, however, and as a matter of practice refers loans to Wells Fargo. (Id). Unsuspecting customers pay fees disguised as legitimate charges to Prosperity. (Id at 10). These fees allegedly go directly to Wells Fargo, which then shares the fee with Long and Foster. (Id). Plaintiffs allege that the scheme generated unearned fees and kickbacks in violation of state and federal laws. (Id).

At issue here are the amounts of the various mortgage settlement fees paid by customers. (ECF No. 322-23, 1-2; ECF 322-17, 3). Specifically, plaintiffs’ request an itemized breakdown of particular fees, found in the “800” range of HUD-1 statements, including “application,” “undexxwriting,” “processing,” “documents,” and other charges. (Id).

B. First Round of Briefing

In their first round of briefing on this issue, the parties focused on two sources of information: (1) a series of Wells Fargo databases, available from 2000 onwards, that contained preliminary HUD-1 information, although with occasional gaps during specific time periods and various limitations in the presentation of the data, and/or, (2) backup tapes that held preliminary HUD-1 information from 2003 onwards. (ECF No. 322-23). Defendants estimated that it would cost approximately 6.7 million dollars to produce HUD-1 information from back-up tapes. (ECF 329-1, 7).

Both of these soxxrces were limited, however, in comparison to final HUD-1 statements, which are available throughout the time-period relevant to the case, and which contain final, rather than preliminary, HUD-1 information. (ECF 322-20, 6). The majority of these final HUD-ls are only available in hardcopy format. At the time of a July 26, 2012 hearing, the cost of converting these documents—viewed by all as the “gold standard” in terms of quality of evidence—into a usable electronic format was unclear. The parties therefore asked for additional time to further consider their positions on the fee data and to investigate the full cost of producing final HUD-1 loan files.

Since the July hearing, the parties have, through a series of meet and confers, commendably reduced the amount in dispute to produce the requested fee data from 6.7 million dollars to approximately $138,000. (ECF No. 432, 2). Specifically, plaintiffs have abandoned their request for information from backup tapes. (ECF No. 431, 2). Plaintiffs now seek the final HUD-ls that both parties agree is the most valuable information available. (ECF No. 431, 2). Problematically, however, Wells Fargo stores this final HUD-1 in a loan file along with several other documents. (ECF No. 431, 7). The amount in dispute is the cost of segregating and identifying the final HUD-1 loan document within certain loan files. (ECF No. 432,2).

C. Available Loan Files and Wells Fargo Practices

There are HUD-1 loan files available for 138.662 of the 143,494 members of the Minter Classes. (ECF No. 432, 3). Out of the 138.662 total, 64,112 have previously been imaged into electronic format by Wells Fargo, and 74,550 are in hard copy format. Wells Fargo has already produced the 64,112 imaged documents: 50,474 electronic loan files in August 2012, (ECF No. 431-5, 3), and 13,638 further files in October 2012. The remaining 74,550 loan files are currently stored in hard copy at Wells Fargo’s off-site storage facility, although Wells Fargo has begun imaging these documents in preparation for production. (ECF No. 431, 4; ECF No. 438, 3-^4). At the hearing, Wells Fargo was unable to report on the status of this imaging, however.

Beginning in early 2002, Wells Fargo’s practice has been to image and retain certain documents in incoming HUD-1 files. (ECF No. 438-5, 2). The imaged file is stored in an on-site image repository. (ECF No. 438-5, 2; Fritze Decl., ECF No. 438-1, ¶ 2). After the file is imaged and stored on-site, the hardcopy is sent to a third party vendor to be stored in an off-site facility. (Fritze Decl., ECF No. 438-1, ¶2). In November 2006, Wells began imaging and indexing the entire loan file before sending the file to storage. (Id,.). Wells Fargo will also periodically recall an older, un-imaged hard copy loan file from its off-site storage facility for its day to day operations; beginning in July 2007, Wells Fargo imaged and indexed these older loan files before they were returned to storage. (Id., ¶ 3-4; ECF No. 438-5, 2).

Important to the instant dispute, all loan files—including both those stored electronically and the hard copies in storage—contain an array of information, including preliminary HUD-ls, draft HUD-ls, final HUD-ls, unsigned HUD-ls, and other materials. (ECF No. 431, 7). The vital final HUD-1 document is therefore within, as plaintiffs describe, a “haystack of extraneous and duplicative documents.” (Id.). Accordingly, the 50,474 electronic files produced in August 2012 did not contain only the final HUD-1, but also an array of other related, but for this purpose unwanted, documents.¹ (Id., ECF No. 432-4, 1). In order to extract the final HUD-1, Wells Fargo offered after the production to “manually review both the 50,-474 electronically imaged documents already produced,” and the 13,638 due to be produced in October, but warned that the review “will take some time.” (ECF No. 432-4, 2).² Plaintiffs declined the offer as to the 50,474 (ECF No. 431, 7) and is relying on the efforts of its expert to identify only the final signed HUD-1 from these 50,474 files in his compilation of fee data for trial.

The two parties differ as to the ease by which Wells Fargo can isolate the final HUD-l document from the extraneous information also held in the loan file. Referring to Wells Fargo’s “Procedure Template” for document management, (ECF No. 439-4,11), plaintiffs contend that as part of the imaging and indexing process, ‘Wells Fargo’s regular procedure is to identify and move final HUD-ls into the ‘Final HUD-1 Settlement Statement Bucket.’” (ECF No. 439, 13). Plaintiffs “seek only the contents of that bucket, Wells Fargo’s regular HUD-1 business record.” (Id.).

In reply, Wells Fargo argues that plaintiffs oversimplify the process. While acknowledging that Wells Fargo has an index called ‘Final HUD,’ ” defendants note that “the index may include both the final signed HUD-1 and possibly non-final HUD-ls for each loan file.” (ECF No. 432-4,2). Accordingly, defendants argue, “it is not possible to ‘push a button’ and pull all the final signed HUD-ls for each loan file.” (Id.). Ultimately, defendants contend that, using their normal indexing procedures, they “have been unable to pull the final signed imaged HUD-ls without also getting extraneous documents.” (Id.).

Isolating a final HUD-1 is seemingly more challenging when imaging older, hardcopy HUD-ls, as compared to newer loan files that are imaged by Wells Fargo upon receipt. At the October 9, 2012 hearing, defendants noted that new loan files have barcodes that streamline the imaging process and allow for easier isolation of individual documents. At least in part due to the difficulties associated with imaging older hard copy loan documents, defendants claim that identifying and segregating an individual HUD-1 for the remaining hard copy loan files necessitates an additional manual or personal review beyond their normal procedures. (Id.).

Plaintiffs have agreed to bear the cost of identifying and segregating the final HUD-1 documents for the 50,474 electronic loan files that have already been produced. (EOF No. 431-6,1). Plaintiffs have also agreed to bear the cost of compiling and organizing the line-item data from the 800 lines of the final HUD-ls for all of the 138,662 loan files, estimated at a cost of $113,000. (ECF No. 431, 3). For its part, Wells Fargo agreed to make best efforts to isolate a final HUD-1 for the 13,638 loans produced in October. (ECF No. 432, 6). It has also volunteered to bear the cost of imaging into electronic format the entire loan file for the remaining 74,550 loans, estimated at a cost of approximately $750,000. (Id., 2).

The only remaining dispute is over the cost of imaging and segregating the final HUD-1 document from the approximately 300 pages of extraneous documents vsithin the 74,550 electronic loan files that Wells Fargo has agreed to produce. Both sides agree that this process will cost approximately $138,000. Wells Fargo has agreed to use its own systems to accomplish the task, and the process has already begun. (ECF No. 438, 3-4). The parties ask the Court to rule on which party should bear that cost.

II. Analysis

Both parties agree that this dispute implicates Federal Rules of Procedure 33 and 34. Wells Fargo asserts that it has already met and “far” exceeded its obligations under these rules, and therefore has no obligation to pay the cost of identifying and segregating final HUD-ls from imaged loan files. (ECF No. 432, 11). Plaintiffs, in turn, argue that “[w]hether analyzed under Rule 33 or 34,” they are entitled to final HUD-ls at defendants’ expense. (ECF No. 431,12).

As Rule 33 provides a clear resolution to the current dispute, the Court finds that an analysis of Rule 34 is unnecessary.

A. Federal Rule of Civil Procedure 33(d).

The general presumption “is that the producing party should bear the cost of responding to properly initiated discovery requests.” Thompson v. United States HUD, 219 F.R.D. 93, 97 (D.Md.2003); see also Oppenheimer Fund, Inc. v. Sanders, 437 U.S. 340, 358, 98 S.Ct. 2380, 57 L.Ed.2d 253 (1978). Rule 33(d), however, functions to shift some of the burden of discovery to the requesting party under certain circumstances. It provides that where:

the answer to an interrogatory may be determined by examining, auditing, compiling, abstracting, or summarizing a party’s business records (including electronically stored information), and if the burden of deriving or ascertaining the answer will be substantially the same for either party, the responding party may answer by ... specifying the records that must be reviewed, in sufficient detail to enable the interrogating party to locate and identify them as readily as the responding party could. Fed.R.Civ.P. 33(d).

The principle behind Rule 33(d) is that “the interrogating party should bear the burden of gathering information ... when the burden ... would be the same for either party.” 1 Discovery Proceedings in Federal Court § 15:12 (3d ed.); see also Compagnie Francaise d’Assurance Pour le Commerce Exterieur v. Phillips Petroleum Co., 105 F.R.D. 16, 44 (S.D.N.Y.1984). The Rule serves to “put the burden of extracting and collating the information on the party seeking it in cases in which it would be time-consuming and expensive to extract data from the records.” 8B Charles Alan Wright, Arthur R. Miller, Mary Kay Kane, Richard L. Marcus, Federal Practice and Procedure § 2178 (3d ed.); see also Securities and Exchange Comm’n v. Elfindepan, 206 F.R.D. 574, 576 (M.D.N.C.2002).

In the Fourth Circuit, district courts have adopted a two-part test when analyzing a Rule 33(d) dispute. First, the moving party on a motion to compel must “make a prima facie showing that the use of Rule 33(d) is somehow inadequate to the task of answering discovery.” Hillyard Enters. v. Warren Oil Co., NO. 5:02-CV-329, 2003 WL 25904133, 2003 U.S. Dist. LEXIS 27922 (E.D.N.C. Jan. 31, 2003) (internal citations and quotations omitted); see also Elfindepan, 206 F.R.D. at 576. If plaintiff makes such a showing, “the burden then shifts to the producing party to justify the use of Rule 33(d) instead of answering the interrogatories.” Elfindepan, 206 F.R.D. at 576.

A party seeking to answer interrogatories through Rule 33(d) business record production must make four prerequisite showings. See id,.; Hillyard at *2, 2003 U.S. Dist. LEXIS 27922 at *6; 8B Charles Alan Wright et al., Federal Practice and Procedure § 2178 (3d ed.). First, the responding party must affirm that the information sought is available in the specified records. Id. Second, the producing party must specify where the information will be found. Id. Third, the producing party must demonstrate that answering the interrogatory in the more traditional manner—in this case by actually producing the specific final HUD-1 documents requested—would impose a burden. Id. Finally, the responding party must demonstrate, as the rule requires, that the burden of acquiring or compiling the information is substantially the same for both parties. Id. This final prerequisite requires “a determination that the effort necessary for the interrogating party to obtain the information from these records would not be substantially greater than the burden on the responding party to do the same thing.” 8B Charles Alan Wright et al., Federal Practice and Procedure § 2178 (3d ed.).

This situation is somewhat unique as Wells Fargo has already agreed to identify and produce a final HUD-1 to plaintiffs. They argue only that plaintiffs should cover Wells Fargo’s costs associated with this specific production. For its analysis, the Court will discuss whether a production of the entire imaged loan file—not simply the final HUD-1—would meet the requirements of Rule 33(d). If such a production comports with 33(d), plaintiffs should properly bear any additional cost of segregating and identifying the final HUD-1, as defendants have satisfied their discovery obligation. If such a production is improper under 33(d), however, defendants must bear the cost of identifying the final HUD-1.

As to the threshold 33(d) showing, both sides agree that imaging and segregating the final HUD-1 from the loan file is a relatively costly process. It will cost Wells Fargo an estimated $138,000 to complete. (ECF No. 431, 10). If plaintiffs were to attempt to isolate the HUD-1 independently, they would require the assistance of a third party vendor. (ECF No. 431, 7). They have already contracted with a vendor to both isolate a final HUD-1 and compile the relevant line item fee data for the electronic loan files already produced by defendants. (ECF No. 431, 9). The process will cost approximately $113,000. (Id.). As noted at the October 9, 2012 hearing, however, identifying and segregating a final HUD-1 would be more complicated for the 74,550 documents, as the stored hard copy loan files are approximately 300 pages, whereas the electronic loan files already produced were far smaller, primarily consisting only of various versions of the HUD-ls, including preliminary HUD-ls and duplicate final HUD-ls (ECF No. 431, 7).

The Court finds this is a sufficient showing to pass the first step of the 33(d) test. A demonstration that specific, relevant information in defendants’ production can only be extracted with substantial difficulty is sufficient to pass the 33(d) threshold. See Hillyard at *2, 2003 U.S. Dist. LEXIS 27922 at *6 (finding that first step is passed where “the information sought would be too difficult to extract” from an archive of information offered by plaintiff); Elfindepan, 206 F.R.D. at 576.

Turning to the second step, the first prerequisite is easily disposed of in the present case. Neither side disputes that available final HUD-ls will be found in the loan files. (ECF No. 431, 7; ECF no. 432, 9).

Plaintiffs also do not argue that defendants have failed to specify which records contain the information sought by the interrogatory. The specificity requirement in 33(d)(1) is designed to limit the practice of responding by “directing the interrogating party to a mass of business records or by offering to make all ... records available.” Fed.R.Civ.P. 33 advisory committee’s note. In order to satisfy 33(d), a party must therefore identify the particular document or documents within a production that will answer the interrogatory. Elfindepan, 206 F.R.D. at 576. A document dump of thousands of documents will not suffice. Graske v. Auto-Owners Ins. Co., 647 F.Supp.2d 1105, 1108 (D.Neb.2009) (production of 7,000 documents in 7 files not sufficient under 33(d)). If the records produced are voluminous, the responding party may need to produce an index to guide their opponent to responsive documents. See United States ex rel. England v. Los Angeles, 235 F.R.D. 675, 681 (E.D.Cal.2006).

Defendants’ production passes this test. Both parties agree that the final HUD-1 contains the information that responds to plaintiffs’ interrogatory, and that it is contained within the electronic loan file that will be created. (ECF No. 432, 7). Each of the 74,550 loan files will be individually produced and labeled. This is not a case where the responding party will produce an “undifferentiated mass of business records” without any guidance as to where relevant documents may be or what they might look like. Matthews v. USAir, 882 F.Supp. 274, 275 (N.D.N.Y.1995). By directing plaintiffs to individually organized files that each contain a final HUD-1 document, defendants have met their “duty to specify, by category and location, the records from which answers to interrogatories can be derived.” Fed. R.Civ.P. 33 advisory committee’s note. However, Wells Fargo has not produced an index to the voluminous individual loan files; it is the locating of the final HUD-1 which is the work at issue here.

The third prerequisite ensures that the 33(d) option may be used “only if the interrogatories require a party to engage in burdensome or expensive research into its own business records in order to give an answer.” Jay E. Grenig, Jeffrey S. Kinsler, Handbook of Federal Civil Discovery and Disclosure § 8:24 (3d ed.). This burden is beyond the task of simply referring to ordinary business records: there must be “some burden involved in compiling or extracting the requested information, above and beyond the simple task of referring to the records in order to answer the interrogatories.” Pascale v. G.D. Searle & Co., 90 F.R.D. 55, 60 (D.R.I.1981); 8B Charles Alan Wright et al„ Federal Practice and Procedure § 2178 (3d ed.).

As noted above, plaintiffs do not dispute that Wells Fargo’s cost estimate of $1.85 a file to image and extract the final HUD-1 documents, or approximately $138,000 in total for the remaining 74,550 loan files, is a substantial amount. (ECF No. 432, 7). Defendants estimate that it will take approximately 5 minutes a loan file to accomplish the task, in addition to the time devoted to imaging. (Id.). Plaintiffs do argue, however, that the imaging, indexing and identifying of the remaining HUD-ls is possible with “no out-of-pocket expenses,” (ECF No. 431, 9), because Wells Fargo would perform the task using “existing personnel and equipment consistent with its normal imaging procedures.” (ECF No. 431, 9).

Defendants counter that “but for plaintiffs’ motion to compel # 6, Wells Fargo would likely never need to image [or index] the vast majority of the 74,550 hard copy loan files.” (ECF No. 431, 9). They further contend that the process will result in a significant loss of resources that would otherwise be devoted to regular business activities. (ECF No. 438, 9-10). Finally, as noted above, Wells Fargo argues that the process necessary to produce a distinct, final HUD-1 document demands efforts beyond their normal process of simply indexing and categorizing loan files. (Id.). From Wells Fargo’s description of the procedures utilized in its ordinary business treatment of pre-2002 loan files, it does attempt to put the final HUD-1 into a separate bucket, but as the facts here indicate that defendants’ task of identifying and segregating the final HUD-1 necessitates the use of significant resources, and requires an effort beyond the simple time and cost of referring to more easily accessed records, the Court finds that Wells Fargo has met this prerequisite.

There is a clear dispute in briefing regarding the final prerequisite: a showing that the burden of acquiring or compiling the information is substantially the same for both parties. Citing Wilkinson v. Greater Dayton Reg’l Transit Auth., No. 3:11cv00247, 2012 WL 3527871, 2012 U.S. Dist. LEXIS 114025 (S.D.Ohio Aug. 14, 2012), defendants maintain that the burden of identifying and segregating the final HUD-ls is the same whether borne by Wells Fargo or plaintiffs, because the extraction process does not require specialized knowledge or arcane information. (ECF No. 432, 9). Wells Fargo further contends that plaintiffs’ knowledge of mortgage banking litigation will enable them to identify the HUD-1 as readily as Wells Fargo could. (ECF No. 432, 10). Finally, defendants note that “to lighten the burden on plaintiffs, Wells Fargo has offered its own staff to do the task, and merely asks plaintiffs to cover the costs.” (ECF No. 432,10).³

Plaintiffs make three primary arguments in reply. First, they note that Wells Fargo uses specialized software (“Kofax”) to both image the loan files and identify final HUD-ls. (ECF No. 439, 10-11). Plaintiffs argue that this software, along with the “finely-honed procedures for doing this work,” gives defendants an advantage over plaintiffs in accomplishing the task. (ECF No. 439, 11). Second, plaintiffs contend that Wells Fargo’s familiarity with the process makes the task comparatively easier for them. (Id.) Finally, they argue that Wilkinson is inapposite because the information there could only be extracted through a series of detailed mathematical calculations that could be accomplished with equal ease by either party. (ECF No. 439,12).

Plaintiffs have the stronger side of this particular dispute. Turning first to the relevant case law, Wilkinson does not provide substantial support for defendants. There, plaintiffs sought information relating to the “average length of time it takes GDRTA Human Resources to approve or deny an application for FMLA leave.” Wilkinson at *8, 2012 U.S. Dist. LEXIS 114025 at *26. In reply, defendant provided numerous FMLA files on each employee who requested FMLA leave, and instructed plaintiffs to independently make the calculations regarding the length of time it took to approve or deny leave. Id. Defendants had no specialized software process, experience, or expertise that gave them an advantage in accomplishing this task. Id. Wilkinson was concerned with creating a new set of information using tools equally available to either party. In contrast, the present case involves identifying and segregating already existing information in defendants’ possession, using defendants’ own software and methodology.⁴ (Id.).

The Court agrees that defendants’ expertise and familiarity with this technology gives it an advantage over plaintiffs in identifying and segregating final HUD-ls. Even if defendants must go above and beyond their everyday procedures to identify and produce final HUD-ls, they will use the same technology, staff and expertise used in the regular course of business. Defendants acknowledged at the October 9, 2012 hearing that they could accomplish the task more efficiently than plaintiffs. Plaintiffs do not have ready access to the documents, do not own or license the key software utilized in the extraction process, and do not have the specialized skills to operate this software. As plaintiffs note, the fact that defendants have offered to use their own technology and staff to “reduce the burden” (ECF No. 438, 8) on plaintiffs implicitly recognizes that defendants’ systems are best suited for the task. (ECF No. 439,11).

Defendants could have presented a stronger argument here if they “provide[d] some combination of technical support, information on application software, or other assistance” to allow plaintiffs to accomplish the task with the same ease as defendants. Fed.R.Civ.P. 33 advisory committee’s note. Defendants have made no such offer. This is understandable, given the complexities of software licensing, the disruption of Wells Fargo’s ongoing business operations, confidentiality concerns, among others, if plaintiffs were permitted to do the extraction utilizing Wells Fargo’s systems.

Defendants have offered to image and extract the final HUD-ls themselves, but only if the full cost is borne by plaintiffs. Essentially, Wells Fargo is offering to level the playing field, but only at the expense of plaintiffs. Rule 33 does not endorse Wells Fargo’s view. Quite the contrary, the Advisory Notes state that a party’s inability or unwillingness to provide access to its systems “may mean that it must derive or ascertain and provide the answer itself rather than invoke Rule 33(d).” Id. This is the case here. Accordingly, the Court finds Wells Fargo improperly relies on Rule 33(d) and must bear the cost of imaging and producing final HUD-ls.

III. Scheduling

Also in dispute is the schedule for production of final HUD-ls. In the course of briefing regarding the dispute over costs, Wells Fargo revealed that it currently estimates that it will take between 5 and 7.5 months to retrieve, image, and index the 74,550 hard copy loan files and then locate and export the final HUD-1 Settlement Statement. (Fritze Deck, ECF No. 432-1, ¶ 11). This was the first time plaintiffs had been notified of this prolonged a time-frame. (ECF No. 439, 5).⁵ After production, Plaintiffs estimate that it will take their expert 2-3 months to extract and verify fee data from the final HUD-ls, (ECF No. 439, 6), although defendant notes that the expert mentioned a “30 business day” turnaround, which is more accurately 6 weeks. (ECF No. 438, 4; ECF No. 431-4, 7). In either case, under defendants’ new, clarified timeframe it could be June 2013 before the data is fully prepared.

According to plaintiffs, this new timeframe was wholly unanticipated. A March 2012 declaration by Ms. Fritze estimated that it would take approximately 3-6 months for Wells Fargo to image and produce the 90,000 hard copy loans it then believed were in storage. (Fritze Deck ECF No. 329-3, ¶ 7). Plaintiffs thus expressed disbelief as to the new 5-7.5 month estimate to image approximately 15,000 fewer documents. (ECF No. 439, 5). They were also alarmed by Ms. Fritze’s admission that only 3 scanners out of 80-100 were being used for the process, although unspecified additional scanners would be used in the future. (Fritze Deck ECF No. 329-3, ¶ 9). Defendants offered no explanation in briefing for this unanticipated, long timeframe. They do note, however, that the timeframe is based on certain limitations, including (1) the scanning speed of Wells Fargo’s scanners, (2) the need to devote scanners to other projects, and (3) independent demands on the time of the staff who will review the loans to extract final HUD-ls. (ECF No. 438, 3).

At the October 9 hearing, defendants indicated that they are currently receiving 1,000 loan files a day from storage.⁶ They also explained that the loan scanning process began in mid to late September. They have 80-100 high speed scanners which can image loan files. (Fritze Decl. ECF No. 432-1 ¶ 9). Each machine can image approximately 2,750 pages per hour, meaning that a single machine can image an average 300-page loan file in under 7 minutes, or more than 9 per hour. (Id. at ¶ 9; ECF No. 439, 4). As noted above, Wells Fargo estimates that it will take 5 minutes per a file to review and identify the final HUD-1. (Fritze Decl. ECF No. 432-1 ¶ 7).

Scheduling is an important element of this motion to compel because the liability trials related to this matter are scheduled for March and May 2013. While the trials are bifurcated into liability and damages phases, plaintiffs rightly argue that it is preferable to complete discovery prior to either trial. (ECF No. 439, 8). The Court has not yet ruled on the timing or method of damages determination. One suggestion on the table is use of same jury in both the liability and damages phases. As plaintiffs note, it would present logistical problems to reassemble the jury for the damages trial several months after the completion of the liability phase, when the information may become available under Wells Fargo’s schedule. (ECF No. 439, 9). As Wells Fargo chose to initially devote only 3% of their available scanner resources to this endeavor, they clearly do not feel the same—if any—sense of urgency regarding this production.

The Court has done basic calculations based on Wells Fargo’s imaging capabilities, which the Court presented at the hearing. It seems that approximately 16-23 scanners and 12-15 full time staff are required to complete the process of imaging the loan files and identifying final HUD-ls in 2-3 months. Defendants affirmed this calculation to the best of their knowledge at the hearing.

It is clear that the fast approaching trial date necessitates expediency. The use of up to 23 scanners (and likely less, as some unknown number of files have already been imaged of the 74,450) of 80-100 available is not an undue burden on defendants, particularly in light of its previous calculations regarding the timeframe of production. The Court therefore orders that Wells Fargo produce the relevant final HUD-1 documents on or before December 15, 2012. Plaintiffs had originally asked for a December 1, 2012 deadline for complete production, but given Wells Fargo’s agreement that production could be made on a rolling basis, plaintiffs appropriately acceded to a later completion production deadline of December 15. The Court notes that as the process began in mid-September, this date falls within the time-frame (3-6 months) originally given by Ms. Fritze for a considerably larger production.

Accordingly, the motion to compel is GRANTED. Because of the complexity of these issues, the Court shall issue a separate Order incorporating the rulings herein.

1 . While the 50,474 files initially produced contained several versions of the HUD-1 and some extraneous documents, it is important to note that these files are significantly more streamlined than the 74,550 complete loan files in storage, which contain approximately 300 pages of loan information. (ECF No. 439, 4).

2 . Wells Fargo appears to have been responsible for this over-production, complicating further analysis, as plaintiffs sought only the final HUD-1 in their interrogatories.

3 . As indicated earlier, plaintiffs have tasked their expert with identifying and segregating the signed final HUD-ls from the 50,474 partial imaged loan files as well as compiling fee data from the final HUD-ls. Wells Fargo does not argue that this expert could derive or ascertain the final HUD-ls as easily as Wells Fargo for the remaining 74,550 loan files. That is probably for several reasons. First, as plaintiffs point out, their expert’s extraction of final HUD-ls is not from the entire loan file (estimated to be 300 pages on average) but from a streamlined file that consists primarily of various versions of the HUD-1 document. Since the expert apparently identifies the final HUD-1 from the fact of a signature, that technique would undoubtedly be over-inclusive if the whole loan file is searched as there are likely many signed documents, such as applications, in a complete loan file. Moreover, tasking the plaintiffs’ expert with extraction of the final HUD-ls would not relieve Wells Fargo from imaging the loan files to allow this further electronic review for final HUD-ls. In short, there appears to be no reasonable, efficient method of obtaining this key fee data without utilization of Wells Fargo’s existing systems and personnel.

4 . The Court notes that the parties' current agreement follows the principles of 33(d) and is in accord with Wilkinson to the extent that plaintiffs have agreed to pay the cost of identifying and compiling the relevant information from the final HUD-ls. It seems that neither party has any specialized software or skills that would provide an advantage in identifying and compiling this information. Under this level playing field, it is appropriate for plaintiff to bear the burden of identifying and compiling the data from the HUD-ls provided to them by defendants.

5 . Plaintiffs acknowledge, however, that defendants noted that they would like to make "certain edits” regarding "timing issues” in a draft Stipulation that contained a November 30 production deadline. (ECF No. 439, 5). These issues were never expanded on because parties reached a stalemate over the issue of costs. (ECF No. 438, 3).

6 . Originally, Wells Fargo said that its business arrangement with Iron Mountain storage facility allowed it to demand production of only 1000 files a day—for all litigation and business needs. At the October 9 hearing, Wells Fargo reported that it had increased its allowance to 2000 files per day—1000 for this litigation and 1000 for other purposes. According to the Court’s calculations—discussed in open court—it would appear that even if this restriction of file production were absolute, the job could be completed by December 15. Moreover, the Court is highly skeptical given Wells Fargo’s size, that Iron Mountain would refuse a request for a higher production of files (at a per file charge) from a reliable customer.