Opinion
19-11984-CTG Adv. 21-51093-CTG
10-29-2024
IN RE FRED'S, INC., er al., Debtors. v. THE TERMINIX INTERNATIONAL COMPANY LIMITED PARTNERSHIP, Appellee. FI LIQUIDATING TRUST, Appellant, Civ. No. 23-1233-RGA
Erika F. Johnson, Steve D. Adler, Bayard, P.A., Wilmington, DE; Joseph L. Steinfeld, Jr., Richard J. Reding, ASK LLP, St. Paul, MN; Edward E. Neiger, ASK LLP, New York, NY, attorneys for appellant, the FI Liquidating Trust. James Tobia, The Law Offices of James Tobia, LLC, Wilmington, DE; Roland Gary Jones, Jones & Associates, New York, NY, attorneys for appellee, The Terminix International Company Limited Partnership.
Chapter 11
Erika F. Johnson, Steve D. Adler, Bayard, P.A., Wilmington, DE; Joseph L. Steinfeld, Jr., Richard J. Reding, ASK LLP, St. Paul, MN; Edward E. Neiger, ASK LLP, New York, NY, attorneys for appellant, the FI Liquidating Trust.
James Tobia, The Law Offices of James Tobia, LLC, Wilmington, DE; Roland Gary Jones, Jones & Associates, New York, NY, attorneys for appellee, The Terminix International Company Limited Partnership.
MEMORANDUM OPINION
ANDREWS, UNITED STATES DISTRICT JUDGE:
This matter arises from the chapter 11 cases of Fred's, Inc. and certain affiliated debtors (collectively, the “Debtors”) and their confirmed plan of reorganization. On appeal is the Bankruptcy Court's Order Granting in Part and Denying in Part the Defendant's Motion for Summary Judgment and for Final Judgment, dated October 18, 2023 (D.I. 1-1) (the “Order”).
The docket of the chapter 11 cases, captioned In re Fred's, Inc., el al., Case No. 19-11984-CTG (Bankr. D. Del.), is cited herein as “B.D.L,” and the docket of the adversary proceeding, captioned FI Liquidating Trust v. The Terminix Int'l Co. Limited P 'ship. No. 21-51093-CTG (Bankr. D. Del.), is cited herein as “Adv. D.I.” The appendix (D.I. 11) filed in support of the Trust's opening brief is cited herein as “App. App'x.
The Debtors operated general merchandise and pharmacy stores in multiple states in the southeastern United States. On September 9, 2019 (“Petition Date”), the Debtors filed voluntary petitions under chapter II. On June 4, 2020, the Bankruptcy Court entered an order confirming the Debtors' chapter 11 plan (B.D.l. 1162) (the “Confirmation Order” and “Plan,” respectively). In accordance with the Plan and Confirmation Order, the Fl Liquidating Trust (“Trust”) was established and authorized to prosecute and settle certain causes of action under chapter 5 of the Bankruptcy Code, including the avoidance action that is the subject of this appeal. (B.D.l. 1109.)
Defendant, The Terminix International Company Limited Partnership, provides pest control services and protection against termites, rodents, and other pests in commercial and residential markets. (App. App'x at 34.) Defendant provided regular pest-control services and on-demand services to the Debtors prior to the bankruptcy filing. (77.) Defendant would send separate monthly invoices to the Debtors, whose payment terms were net 30 days. (Id. at 36.) There is no dispute that the Debtors almost always paid outside the 30-day term, typically by check, and often covering multiple invoices in a single payment. (App. App'x 36.)
It appears undisputed that the Debtors voluntarily changed their mode of payment in September 2018 to ACH payments, and that Defendant never required (or even discussed with the Debtors) any change to their mode of payment. (See App. App'x 76-82 at ¶ 16.)
The Trust filed an adversary proceeding against Defendant. The complaint sought the avoidance and recovery of $129,934.00 in transfers made by the Debtors to Defendant during the 90-day period prior to the Petition Date (the “Preference Period”). (App. App'x 1-15.) Defendant filed its Answer and Affirmative Defenses in which Defendant denied that the Transfers were preferential or, in the alternative, were protected from avoidance by the affirmative defenses available under § 547 of the Bankruptcy Code. (See Adv. D.I. 7.)
On March 1,2023, Defendant filed its motion for summary judgment (App. App'x 3039), together with a brief in support and an affidavit containing exhibits comprising the relevant billing and payment records. (App. App'x 40-105.) The Trust filed a response to Defendant's summary judgment motion. (App. App'x 106-143.) While the Trust's response to the summary judgment motion included its own affidavit and records, the Bankruptcy Court found that “the actual historical facts with respect to the timing of the payments is not disputed in any material way.” (App. App'x 187.) Defendant filed a reply. (App. App'x 144-157.)
The issue before the Bankruptcy Court was whether Defendant should be granted summary judgment on its affirmative defense that two payments it received during the Preference Period could not be avoided by the Trust. The total of the two payments was $129,934, which meant that putting a lot of effort into the litigation was not economically a good business decision. Nevertheless, after the Bankruptcy Court decided on summary judgment that $129,348 of the $129,934 could not be avoided because the invoices relating to the $129,348 were paid within the “one standard deviation range” of 20.61 to 179.39 days, the Trust appealed.
The Bankruptcy Court commented at the beginning of argument on Defendant's summary judgment motion that “in view of the stakes,” and for other reasons, it made sense to conduct the argument by Zoom. (App. App'x at 160.)
The parties' dispute over the remaining $586 was separately resolved, and it is not at issue on appeal. Thus, the appeal is from a final judgment, and I have jurisdiction over the appeal. 28 U.S.C. § 158(a).
The record in the Bankruptcy Court primarily consisted of the history of invoices and payments for the Preference Period and for the approximately two years preceding the preference period, referred to as the “Base Period.” A complicating factor was that, both before and during the Preference Period, the Debtors often paid multiple invoices in one transaction, and that there was, to put it mildly, a lot of variability in the invoiced amounts and the age of the invoices when they were paid. Defendant had put in the record before the Bankruptcy Court as “undisputed material facts” that:
Although often an issue of contention, here, the Defendant and the Trust have relied on the same time frame for the Base Period.
In the Preference Period, the Defendant's invoices were paid with an average of 128.50 days and weighted average of 115.08. During the Base Period, the Defendant's invoices were paid with an average of 100 days and weighted average of 104.05 days. The average age of each invoice during the Base Period exhibits a standard deviation of 79.39 days. This creates a one standard deviation range of 20.61 to 179.39 days. (App. App'x 36.)
The Trust disputed most of the “undisputed material facts.”
The Plaintiff [i.e., the Trust] agrees with the “Procedural Background” section of Terminix's statement of undisputed facts as well as the first four paragraphs under the section “Statement of Facts”. However, the Plaintiff does not agree that any of the remainder of Terminix's Statement of Undisputed Material Facts (the “Terminix Statement”) are actually undisputed. This includes nearly all portions of the Lino Affidavit referenced in the Terminix Statement. The Plaintiff expressly does not agree with the statistical calculations and conclusory statements in paragraphs 16-24 of the Lino Affidavit. The Plaintiff does not concur with the statistical calculations in Terminix's Appendices 3,3-A, 4, 4-A, or 5.(App. App'x at 112.) The Trust further explained:
[T]he reason why the standard deviation approach does not work is because Terminix uses significantly late invoices (in excess of 300 days) without any sort of weighting mechanism to prevent those late invoices from skewing the data. This is why Terminix ends up with a standard deviation range of 20.61 to 179.39 days because there are a relatively small number of invoices that were paid extremely late artificially “skews” the calculation. For the same reason that a “total range” approach is inappropriate under these facts, a standard deviation approach is as well.(App. App'x at 118.)
The legal basis for Defendant's argument was the affirmative defense set forth in section 547(c)(2) of the Bankruptcy Code, commonly referred to as the “ordinary course of business defense.” 11 U.S.C. § 547(c)(2). Section 547(c)(2) provides that:
(c) The trustee may not avoid under this section a transfer
...
(2) to the extent that such transfer was in payment of a debt incurred by the debtor in the ordinary course of business or financial affairs of the debtor and the transferee, and such transfer was
(A) made in the ordinary course of business or financial affairs of the debtor and the transferee; or
(B) made according to ordinary business terms.11 U.S.C. § 547(c)(2). The Bankruptcy Court, citing the statutory language, considered the only disputed issue: whether the two transfers at issue were “made in the ordinary course of business or financial affairs of the debtor and the transferee.” (App. App'x at 190.) The “made according to ordinary business terms” prong of the analysis was not and is not at issue.
Prior to the Bankruptcy Abuse Prevention and Consumer Protection Act of 2005, the “made according to ordinary business terms” was a requirement that had to be met under § 547(c)(2). See In re Molded Acoustical Products, 18 F.3d 217, 219 (3d Cir. 1994). It is now merely an alternative.
The creditor bears the burden of proof of showing that a preferential transfer is not avoidable under the ordinary course of business defense. 11 U.S.C. § 547(g). The Bankruptcy Code does not define the phrase “in the ordinary course of business.” In re Hechinger Inv. Co. of Delaware, Inc., 489 F.3d 568, 576 (3d Cir. 2007). To be in the ordinary course of business, the transfers at issue should “conform to the norm established by the debtor and creditor in the period before, preferably well before, the preference period.” In re Molded Acoustical Products, 18 F.3d 217,223 (3d Cir. 1994) (discussing how to interpret “ordinary business terms” to give it meaning separate from what was required by other parts of the ordinary course of business defense) (quoting In re Tolona Pizza Prods. Corp., 3 F.3d 1029, 1032 (7th Cir. 1993)). Thus, the creditor is obligated to produce evidence of a “baseline” of dealings that establishes the contours of the ordinary course of business between the parties. Stanziale v. Superior Technical Resources, Inc. (In re Powerwave Technol., Inc.), 2017 WL 1373252, at *4 (Bankr. D. Del. Apr. 13, 2017). The creditor must then show that the activity in the preference period comported with the pattern established in the baseline period.
Section 547(c)(2)(A) has been the subject of two precedential Third Circuit cases. See In re Hechinger, 489 F.3d at 568; In re J.P. Fyfe, Inc. of Fla., 891 F.2d 66 (3d Cir. 1989). In Fyfe, the Court held that the ordinary course of business determination was a factual one subject to clearly erroneous review. Id. at 70. Fyfe's analysis suggests the importance of making the factual determination on the basis of a comparison of financial relations before and after the debtor was showing financial distress. In Hechinger, after a trial, the bankruptcy court considered changed credit terms, and the length of time the parties had engaged in the type of dealing at issue, the way the payments were made e.g., whether the subject transfers were in an amount more than usually paid or tendered in a manner different from previous payments and whether the creditor did anything to gain an advantage in light of the debtor's deteriorating financial condition. Id. at 578.
In Fyfe, the Bankruptcy Court found facts based on testimony. In the instant case, the appeal arises from a decision on summary judgment. So, review is de novo, not clearly erroneous.
Thus, bankruptcy courts examine a variety of factors when determining whether the preference period conforms to the baseline period norm, but of particular importance is the timing of invoices i.e. the amount of time elapsed between the creditor issuing an invoice and the date that the debtor pays that invoice. In re AE Liquidation, 2013 WL 3778141 at *5 (Bankr. D. Del. July 17, 2013) (citing Burtch v. Detroit Forming, Inc. (In re Archway Cookies), 435 B.R. 234, 241-42 (Bankr. D. Del. 2010) and Radnor Holdings Corp. v. PPT Consulting, LLC (In re Radnor Holdings Corp.), 2009 WL 2004226, *5 (Bankr. D. Del. July 9,2009)).
Here, the Bankruptcy Court noted there was a “long course of dealing” between the parties. (App. App'x 192.) The Bankruptcy Court did not otherwise discuss the factors listed in Hechinger-likely because the summary judgment briefing reflects no dispute that, apart from the consistency of the parties' transactions, these other factors did not weigh against Defendant's ordinary course of business defense. See App. App'x 40-105 (Terminix's summary judgment motion) at 54-56 (arguing years' long history involving the same services; consistent amount and form of tender; no unusual collection practices; no evidence Terminix knew of or took advantage of debtors' deteriorating financial condition); App. App'x 106-143 (Trust's opposition to the summary judgment motion) (failing to address Hechinger factors and arguing only that “Terminix uses questionable statistical methods to obscure what is a foundational fact in this litigation: prior to the preference period, the debtors paid approximately 80% of Terminix's invoices in 80 days or fewer. In the preference period, the debtors paid nearly 80% of Terminix's invoices in more than 100 days ... No amount of statistical sleight-of-hand can cover up what is a fundamental and drastic change in the course of dealing between the parties”).
Accordingly, the Bankruptcy Court's analysis focused on the consistency or similarity of the transactions and whether Defendant had met its burden of showing that the activity in the Preference Period conformed with the pattern established in the preference period.
“Payments made during the preference period do not have to possess a rigid similarity to each past transaction; however, there must be some consistency.” In re Powerwave, 2017 WL 1373252 at *4 (internal quotations omitted). Small deviations in timing will not preclude a finding of ordinariness, while greater deviations in the timing of payments are more likely to overcome an assertion of ordinariness. See Archway Cookies, 435 B.R. at 243. Overall, this inquiry is a fact-specific inquiry. Goldstein v. Starnet Capital Grp., LLC (In re Universal Mktg, Inc.), 481 B.R. 318, 327 (Bankr. E.D. Pa. 2012). “There is no single formula the court must use when deciding whether preferential transfers were made in the ordinary course of business.” In re Powerwave, 2017 WL 1373252 at *4 (internal quotations omitted). “A variety of mathematical processes that include the range, averages, and weighted percentages may be appropriate.” Id. (citing In re Moltech Power Sys., 327 B.R. 675, 681 (Bankr. N.D. Fla. 2005)). (“[T]he court may use any or all of these mathematical methods as tools by which to determine what the ordinary course of business between the parties actually was.”) (internal citations omitted). Various methodologies are discussed in Powerwave. 2017 WL 1373252, at *5-7.
Here, the parties focused on four methodologies, which the Bankruptcy Court characterized as the “range of payment” approach, the “average lateness” test,” the “bucketing” approach, and the “standard deviation” approach. (App. App'x 192-193.) The Bankruptcy Court, in its oral ruling (App. App'x 185-196), rejected the first three while finding that the standard deviation approach “makes the most sense in the circumstances of this case.” (App. App'x at 193.) Specifically, “where the parties have a fairly long course of dealing and a robust set of dealings that can be compared,” and where “there is a large data set with multiple points of data, it seems useful to rely on a test that accounts for the variability of the values.” (Id.) Once the Bankruptcy Court decided on the standard deviation approach, the rest was mathematical. The Bankruptcy Court concluded that “a range within the base period of 20.61 to 179.39 days [] are within 1 standard deviation of the mean.” (App. App'x 069.) All the invoices (except one for $586) were somewhere in that range. That meant that all but $586 of the transfers were unavoidable as those transfers fell within the ordinary course of business defense. (Id. at 070.)
On appeal, the Trust makes two main arguments. First, that the Bankruptcy Court “incorrectly applied] the standard deviation approach.” (D.I. 10 at 12.) Second, that the Bankruptcy Court erred “in its application of the standard deviation approach.” (Id. at 20.) Despite the two arguments sounding pretty similar, they are actually different. The first argument is that the standard deviation methodology is an “inappropriate statistical method” to analyze data that is not normally distributed. The second argument is that the range of 26.61 to 179.39 “is not mathematically correct.”
I hold that the Bankruptcy Court should not have granted summary judgment based on the record it had before it. I base this on the Trust's first argument.
The Bankruptcy Court had before it only the mostly disputed “undisputed material facts.” It had an affidavit of Charlene Lino, the “Director of Transaction Services of Terminix” (App. App'x 76-82) (the “Lino Affidavit”), which provided some background information about the relationship of the creditor and the debtor. (App. App'x at 77-82.) The Lino Affidavit cited two spreadsheets, labelled “App. 3” (referencing the Base Period) and “App. 4” (referencing the Preference Period). (App. App'x 79, 86-88, 92-93.) There were two other spreadsheets, labelled as “App. 3-A” and “App. 4-A”. The Lino Affidavit did not mention them. Their origin was thus a bit mysterious. App. 3-A included at the end six pieces of information, as follows: “Batching Average, 93.26; Min Range, 52.50; Max Range 152.22; Standard Deviation, 39.53; Stdev Min 53.73; Stdev Max 132.79”. (App. App'x 91.) The spreadsheet did not expressly explain these numbers. They played no part in the Bankruptcy Court's ruling.
Ms. Lino, it turned out, had nothing to do with preparing App. 3-A. It was generated by some of Defendant's lawyers. (D.I. 23 at 2.) It seems odd that they were submitted in sequence with the other exhibits to her affidavit, as though she had something to do with preparing them. I did not specifically ask who prepared App. 4-A (D.I. 22), but it seems certain that Ms. Lino had nothing to do with it either.
To the extent the Lino Affidavit authenticated the tables drawn from Terminix's records, the Bankruptcy Court could rely upon them. I do not think the Bankruptcy Court could rely upon unsworn statements of fact and disputed statements of fact. One such unsworn statement of fact was that one standard deviation for the Base Period was 26.61 to 179.39. To boot, the Trust disputed this assertion, albeit without any evidence of his own, both factually and methodologically. I expect the range could have been factually supported; it simply wasn't. Whether applying the standard deviation to the data set from the “Base Period” is a methodologically sound use of standard deviation, however, was an issue the Trust raised in enough detail to preserve the issue. The Bankruptcy Court acknowledged the argument as the “two separate curves” argument but did not accept it. (App. App'x 194.) On appeal, the Trust provides a more developed argument, complete with citation to some articles questioning the applicability of standard deviation to non-normal data sets. The articles were not cited to the Bankruptcy Court. But the Trust made the argument, and, under the circumstances, if it was going to be rejected, it needed to be rejected after Defendant provided some evidence that its use was valid. Defendant did not do that. There was no expert opinion on the appropriateness of using standard deviation with a non-normal data distribution (what the Trust refers to as “skewness”). “There is extensive literature describing the problems associated with applying the Cohen's d test to data that are not normally distributed or that are lacking equal variances.” Stupp Corp. v. United States, 5 F.4th 1341, 1358 (Fed. Cir. 2021). I do not think this statement is directly applicable to the standard deviation calculations upon which the Bankruptcy Court relied, but I think it suggests that the analogous arguments that the Trust was making needed to be resolved based on evidence. They weren't. Probably because of the “stakes,” see n. 3, lawyers, not statisticians, were responsible for all the “statistical” analyses that were presented to the Bankruptcy Court. I have considered the various articles cited by the Trust, but I think to resolve the dispute about whether standard deviation was an appropriate tool to apply to the Base Period data set in App. 3, an expert was needed. Thus, I cannot resolve it. I think it ought to be resolved, in the first instance, by the Bankruptcy Court, not by me.
1 put the numbers from App. 3 into an Excel spreadsheet, used one of its two standard deviation functions, and came up with a similar standard deviation. The mean of the 51 entries in App. 3 is in fact exactly 100, so the “one standard deviation range” Terminix asserted is roughly accurate.
See n. 8. On appeal, I asked for an explanation of how the 26.61 to 179.39 range was calculated, and where there was any explanation of that in the record. (D.I. 22.) The letter I got in response gave an explanation, but it did not cite anything to suggest that the explanation was already in the record.
Attorney for Defendant: “I don't have extensive argument about that, but we did - you know, courts do use standard deviation. I mean, I went to law school, so I wouldn't understand statistics, so now I have to explain statistics. So, you know, it - you know, it's an analysis that's used a lot and has a lot of support.” (App. App'x 167-168.) Attorney for the Trust: “[G]ranted, I also went to law school in large part to avoid doing math. But from my limited understanding of statistics, when you have a bimodal distribution, a standard deviation approach just isn't mathematically the right way of looking at it.” (App. App'x at 171.) The Bankruptcy Court: “I'm not a mathematician, either.” (Id. at 172.)
There is a second related issue. Why should standard deviation, to the exclusion of anything else, provide an up or down answer to the question of whether the ordinary course of business defense has been proved? The Bankruptcy Court preferred it to other approaches because it resolved the dispute in an objective manner. The standard deviation test is different than the other suggested approaches, which are “sufficiently malleable and manipulable that they can be engineered to obtain a desired result.” (App. App'x 193). Even if calculating standard deviation usually cannot “be engineered to obtain a desired result,” that does not mean that the result that is obtained resolves the issue of whether a payment is or is not in the ordinary course of business. Put differently, why are payments that fall within one standard deviation of the mean in the ordinary course, but payments that fall outside one standard deviation of the mean not in the ordinary course? I am not implying that standard deviation should not be considered. But it seems to replace consideration of the statutory term “ordinary course of business” with a bright-line statistical test. At most, a statistical test is a tool, not an answer.
1 say “usually” because there are two different formulas for standard deviation in the Excel program. I got different (albeit not significantly different) results from the two formulas using the same data set. I believe a statistician would know which formula was proper to use with a particular data set. Lawyers might not.
I agree with the Bankruptcy Court that, to some extent, application of any single methodology feels like “painting targets around the arrows.” (App. App'x 163; see id. at 192.) I think considering standard deviation alone, both generally and under the circumstances of this case, does not provide a full picture of payment consistency. For example, in this case, there was, as I have said, great variability not only in the timing of payments, but in the amounts of the invoices being paid and the amounts of the payments. During the Base Period, there were eleven invoices for less than $1,000. The other forty were all between about $25,000 and about $53,000. The twenty-seven transfers paying the fifty-one invoices included three for less than $1,000, and another five for less than $5,000. The five largest transfers were all in excess of $60,000. During the Preference Period, there were ten invoices. Five of them were for less than $1,000. The other five ranged from about $18,000 to about $37,000. (App. App'x 93.) At best, what the Bankruptcy Court's math determined only related to one factor the length of time it took to pay an invoice. For example, the inclusion of low-dollar invoices and low-dollar payments carried no weight in the standard deviation analysis. To the extent a standard deviation approach should ever be applied to a data set like this one data that are not normally distributed or that are lacking equal variances the result should be considered in conjunction with that of other statistical approaches to paint a full picture.
The Trust's second argument, that the 20.61 to 179.39 range is “mathematically incorrect,” is either a variation on the first argument or it is an argument that has been forfeited. The basis for the argument is the assertion that a one standard deviation range ought to cover 68% or so of the data, that the standard deviation range of 20.61 to 179.39 covers 88% of the data, and that it must therefore be incorrectly calculated. I think this is just a variation on the “skewness” and “non-normal distribution” arguments that the Trust did make. But, to the extent that it is a separate argument, it was not made to the Bankruptcy Court, and thus I find it is forfeited. Danny Kresky Enters. Corp. v. Magid, 716 F.2d 206,214 (3d Cir. 1983) (“We will not ordinarily consider issues which the parties have failed to raise below.”).
In conclusion, the Bankruptcy Court's grant of summary judgment will be REVERSED and REMANDED. The standard deviation analysis provided an insufficient basis for the determination that Defendant proved its ordinary course of business defense.
A separate order will issue.