OPINION

RESTANI, Judge:

This matter comes before the court following the court's decision in Dongguan Sunrise Furniture Co. v. United States, 931 F.Supp.2d 1346, 1348 (CIT 2013) ("Dongguan III"), in which the court remanded Commerce's second redetermination in Wooden Bedroom Furniture From the People's Republic of China: Final Results and Final Rescission in Part, 75 Fed.Reg. 50,992, 50,992 (Dep't Commerce Aug. 18, 2010) ("Final Results"), to the U.S. Department of Commerce ("Commerce") to reconsider its four partial adverse facts available ("AFA") rates assigned to Fairmont's unreported sales of dressers, armoires, chests, and nightstands. For the reasons stated below, the court finds that Commerce's selected AFA rates are not supported by substantial evidence, and thus Commerce's third remand results are remanded.

BACKGROUND

The facts of this case have been documented in the court's previous opinions. See generally Dongguan III, 931 F.Supp.2d at 1348-49. The court presumes familiarity with those decisions but summarizes the facts as relevant to this opinion. In the Final Results, Plaintiffs Dongguan Sunrise Furniture Co., Ltd., Taicang Sunrise Wood Industry Co., Ltd., Taicang Fairmont Designs Furniture Co., Ltd., and Meizhou Sunrise Furniture Co., Ltd. (collectively "Fairmont" or "Plaintiff") received a rate of 43.23%, which was calculated based on a rate of approximately 34% for reported sales and a partial adverse facts available ("AFA") rate of 216.01% for unreported sales. Final Results, 75 Fed.Reg. at 50,997; Dongguan Sunrise Furniture Co. v. United States, 865 F.Supp.2d 1216, 1234 (CIT 2012) ("Dongguan I"). In Dongguan I, the court sustained Commerce's application of a partial AFA rate to calculate the overall dumping margin, but held that Commerce's selected AFA rate of 216.01% was not supported by substantial evidence. 865 F.Supp.2d at 1232-34. Commerce failed to demonstrate that the 216.01% rate, which was calculated in a new shipper review for a different entity during a different period of review ("POR"), was relevant and reliable for Fairmont. Id. at 1233.

On remand, Commerce grouped the unreported sales into four categories based on general product type: armoires, chests, nightstands, and dressers. Dongguan Sunrise Furniture Co. v. United States, 904 F.Supp.2d 1359, 1362 (CIT 2013) ("Dongguan II"). Commerce then determined an AFA margin for each of the four general product types by selecting the single highest CONNUM-specific1 margin below 216.01% from Fairmont's reported sales that fell within the corresponding general categories.2 Id. Fairmont received a rate of 39.41%, which included partial AFA rates of 182.15% for the unreported armoires, 215.51% for the unreported chests, 134.42% for the unreported nightstands, and 183.52% for the unreported dressers. Id.; Dongguan III, 931 F.Supp.2d at 1348. The court again remanded to Commerce, stating that Commerce had failed to demonstrate "a rational relationship between the AFA rates chosen and a reasonably accurate estimate of Fairmont's actual rate," because the AFA rates were based on minuscule percentages of Fairmont's actual sales. Dongguan II, 904 F.Supp.2d at 1363-64. The court also noted that the weighted-average margin for the reported sales, which constituted the vast majority of Fairmont's sales during the POR, indicated that Fairmont's actual rate was much lower than the selected AFA rates. Id. at 1364.

During the second remand proceedings, Commerce calculated partial AFA rates of 189% for the unreported armoires, 161% for the unreported chests, 140% for the unreported nightstands, and 161% for the unreported dressers, which resulted in an overall rate of 41.75%.3 Dongguan III, 931 F.Supp.2d at 1349. Commerce arrived at the partial AFA rates by selecting the single-highest CONNUM-specific margin below 216% where at least 0.04% of the total reported sales in that product category were dumped at or above the selected margin. Id. Once again, the court remanded to Commerce. Id. at 1356. The court noted that "Commerce ignored the majority of the reported and verified information" regarding the sales data for the four general product types at issue "and instead relied on an extremely small percentage of [those] sales." Id. The court held that because "Commerce declined to consider the very evidence it identified as most indicative of Fairmont's actual rate for the unreported sales, and given that the disregarded record evidence suggests a reasonably accurate estimate of Fairmont's actual rate would be much lower, Commerce's determinations are not supported by substantial evidence." Id.

In the remand proceedings currently challenged before the court, Commerce calculated new partial AFA rates for each of the four types of unreported sales.4 Final Results of Third Redetermination Pursuant to Ct. Order, ECF No. 193-1, 2 ("Third Remand Results"). Commerce based these partial AFA rates on the weighted-average dumping margins of the 15% of reported sales with the highest dumping margins within each of the four general product categories. Id. at 12-13. This resulted in an overall dumping margin of 44.64%. Id. at 34. Fairmont contends that the partial AFA rates are not supported by substantial evidence. Pl. Fairmont Cmts. on Third Remand Results, ECF No. 204, 1-10 ("Pl.'s Cmts."). Defendant-Intervenors continue to argue that 216.01% was the appropriate AFA rate, but otherwise do not object to the Third Remand Results. AFMC's Cmts. Concerning Commerce's Final Results of Third Redetermination Pursuant to Ct. Order, ECF No. 195, 1-2. Defendant argues that the partial AFA rates are in compliance with the court's remand order in Dongguan III and supported by substantial evidence. Def.'s Resp. to Fairmont's and AFMC's Remand Cmts., ECF No. 213, 2-11 ("Def.'s Resp.").

DISCUSSION

Fairmont makes several arguments in support of its contention that the partial AFA rates are not supported by substantial evidence. First, Fairmont notes that the selected partial AFA rates are impermissibly excessive because these rates are very close to or even higher than the rates before the court in Dongguan II and Dongguan III. Pl.'s Cmts. 1-2. Thus, the rates have the same flaw of unconnectedness to Fairmont's true commercial behavior, as the court previously found. See id. Fairmont also argues that Commerce acted unreasonably in comparing the volume of sales relied upon by Commerce to calculate the AFA rates to the volume of unreported sales in determining what constitutes a significant portion of the available evidence. Id. at 3. Fairmont further contends that Commerce impermissibly focused on the quantity of reported sales used to calculate the AFA rates without properly considering whether those sales were representative of Fairmont's commercial reality with respect to its unreported sales.5 Id. at 3-8. Finally, Fairmont argues that the AFA rates are far higher than necessary to deter future non-compliance. Id. at 8-10.

Commerce provides several justifications to support the partial AFA rates. First, Commerce defends its conclusion that the 15% of sales with the highest dumping margins for the four categories constitute a meaningful portion of the available data by noting that this volume of sales is several times larger than the volume of unreported sales and by comparing it to the 15% of sales used to support the AFA rate upheld in Lifestyle Enterprise, Inc. v. United States, 896 F.Supp.2d 1297, 1301-02 (CIT 2013), aff'd in part, rev'd in part on other grounds, 751 F.3d 1371 (Fed. Cir.2014). Third Remand Results at 12-14. Commerce also reasons that because many of Fairmont's transactions deviated greatly from the overall dumping margin, the 15% of sales relied upon reflect Fairmont's commercial reality, even though the AFA rates appear to be high when compared to the overall dumping margin. Id. at 13.6 Commerce further states that AFA rates based on the sales with the highest margins are necessary to ensure that sales with high dumping margins are reported in the future. Id. at 15-16.

When a party has failed to act to the best of its ability, Commerce, "in reaching the applicable determination under this subtitle, may use an inference that is adverse to the interests of that party in selecting from among the facts otherwise available." 19 U.S.C. § 1677e(b). Commerce's chosen AFA rates must be "a reasonably accurate estimate of the respondent's actual rate, albeit with some built-in increase intended as a deterrent to non-compliance." Gallant Ocean (Thail.) Co. v. United States, 602 F.3d 1319, 1323 (Fed.Cir.2010). The built-in increase, however, cannot go beyond the amount sufficient to deter respondents from future non-compliance. See id. at 1324. "Substantial evidence requires Commerce to show some relationship between the AFA rate and the actual dumping margin." Id. at 1325. In any case, there is no finding that respondents were willfully avoiding reporting sales. Rather, the scope of the antidumping order was misunderstood by Fairmont, or its employees were not properly instructed in how to identify in-scope sales. See Dongguan I, 865 F.Supp.2d at 1229-30. Thus, Commerce did not use total AFA but used AFA for the unreported sales only, presumably to encourage greater care on the part of respondents, not to deter intentional uncooperative conduct. As explained below, the partial AFA rates still are not supported by substantial evidence because they are not reflective of Fairmont's commercial reality and are higher than necessary to encourage careful compliance.

1. Commercial Reality

Although Commerce has now relied on a greater number of sales in calculating the partial AFA rates, the record continues to show that these rates are not reflective of Fairmont's commercial reality. The court in Dongguan II stated that "[a] calculated rate of 34% for Fairmont's reported sales suggests that rates ranging from 134% to over 215% are not reflective of Fairmont's commercial reality, especially when there is no indication that Fairmont failed to report certain sales for strategic reasons." 904 F.Supp.2d at 1364. Similarly, the court in Dongguan III stated that the "record evidence suggests a reasonably accurate estimate of Fairmont's actual rate would be much lower" than the AFA rates ranging from 140% to 189% calculated by Commerce. 931 F.Supp.2d at 1349, 1356. As Fairmont rightfully points out, these same concerns apply in force when two of the four rates are even higher than the AFA rates at issue in Dongguan III, the other two rates are not substantially lower, and the overall dumping margin has increased as a result of the change in AFA rates.

The fact that Commerce relied on the 15% of reported sales with the highest margins in each of the four broad categories to support these rates, rather than, as previously, picking a specific margin representing a smaller percentage of sales, does not change the court's prior analysis of whether the partial AFA rates are reflective of Fairmont's actual rates. First, the court rejects Commerce's justification for relying on 15% of sales by comparing this volume to the volume of unreported sales. This logic seemingly would allow Commerce to rely on less evidence as the volume of unreported sales declines and consequently the level of compliance increases. The court will not endorse such dubious logic. Similarly, Commerce's reliance on Lifestyle to support its contention that using the 15% of sales with the highest margins is supported by substantial evidence is misplaced. The respondent against whom the AFA rate was applied in Lifestyle did not challenge the AFA rate calculated by Commerce using this methodology. Lifestyle, 896 F.Supp.2d at 1302. Rather, a domestic party argued that the rate should have been even higher, a claim which the court held lacked legal support. Id. Lifestyle did not hold that 15% of some body of sales will always constitute substantial evidence for an AFA rate.

Ultimately, the AFA rates must be supported by substantial evidence on the record and reflective of Fairmont's actual rates. See Gallant Ocean, 602 F.3d at 1325. Here, the transactions relied upon by Commerce to calculate the AFA rates had per-unit U.S. sales prices far lower than the U.S. prices for the unreported sales. Pl.'s Cmts. 4-5; Fairmont's Cmts. on Commerce's Initial Remand Results at 6, TRCD 8 at bar code 3169420-01 (Dec. 20, 2013), ECF No. 212-1 (June 5, 2014) (confidential). Although Commerce determined that these low U.S. prices were not the result of an erroneous calculation methodology, it did not address the issue of whether the sales it used were representative of the unreported sales. See Third Remand Results at 28-30. Commerce does argue that because Fairmont experienced a wide range of margins and made transactions at or above the selected AFA rates, the rates are reflective of Fairmont's commercial reality. Id. at 13-15. Although the record does show individual transactions at margins at or above the selected AFA rates, Fairmont's overall dumping margin is a fraction of the partial AFA rates, which indicates that the vast majority of Fairmont's sales are at margins far lower than the rates relied upon by Commerce in selecting the AFA rates. Furthermore, there is no evidence indicating that Fairmont failed to report the unreported sales in an attempt to hide sales with high margins. Dongguan II, 904 F.Supp.2d at 1364. In short, nothing in the record indicates that the unreported sales were made at average margins remotely close to the ones relied upon by Commerce in calculating the partial AFA rates, and the fact that some individual transactions were made at margins even higher than the partial AFA rates selected does not constitute substantial evidence in the light of the contradictory evidence suggesting a much lower average rate. See id. at 1363-64 (rejecting same argument). The AFA rates thus remain untied to Fairmont's commercial reality.

2. Deterrence

Commerce's attempts to justify the partial AFA rates by invoking the need for deterrence is unavailing. Commerce expresses a concern that selecting a lower rate based upon a greater portion of the available data would entice respondents to game the system by declining to report transactions with high dumping margins with the expectation that the ultimate AFA rate will be lower than the rate they otherwise would have received. Third Remand Results at 15-17. That concern might in fact justify a relatively greater increase in the rate in order to ensure compliance in other cases, but that justification does not appear to apply in this case because there is no evidence that Fairmont engaged in such strategic behavior. Dongguan II, 904 F.Supp.2d at 1364. Rather, Fairmont's failure to train its employee in how to identify in-scope sales resulted in a "perfunctory evaluation of its own records" that led to some in-scope sales going unreported. Dongguan I, 865 F.Supp.2d at 1229-30. Commerce, in exercising its discretion in calculating the AFA margins here, should focus on its legitimate concern with deterring this type of carelessness. Furthermore, Commerce's logic would dictate using the highest margin available as the AFA rate, which the court already has rejected. See Dongguan II, 904 F.Supp.2d at 1364 (rejecting Commerce's use of the highest reported CONNUM-specific margin below 216.01% in each of the four product groups containing unreported sales).

The court likewise rejects Commerce's reliance on Gallant Ocean to support its determination that the rates do not go beyond what is necessary to ensure compliance. Commerce takes the statement in Gallant Ocean "that a rate over five times the highest rate imposed on similar products is far beyond an amount sufficient to deter ... future non-compliance," 602 F.3d at 1324, and reasons that the AFA rates here are acceptable because they are less than five times the overall rate of Fairmont's reported sales of the products at issue. Third Remand Results at 19. But the Court of Appeals for the Federal Circuit did not set any bright-line rule. It stated only "that a rate over five times the highest rate imposed on similar products is far beyond an amount sufficient to deter... future non-compliance." Gallant Ocean, 602 F.3d at 1324 (emphasis added). The court also notes that the AFA rate of 57.64% at issue in Gallant Ocean was markedly lower than the AFA rates at issue in this case. See id. Although the AFA rates here are less than five times the overall margin for the reported sales, they are still several times higher, and the AFA rates are all well over 100%. The court already has stated that the fact that "the selected AFA margins are many times higher than the weighted-average margin for the reported sales ... indicates that the deterrence factor applied is far beyond the amount necessary to deter future non-compliance." Dongguan III, 931 F.Supp.2d at 1355 (citing Gallant Ocean, 602 F.3d at 1324). Commerce has not meaningfully lowered the AFA rates, and the court remains of the view that Commerce's methodology is flawed in that the selected AFA rates are far beyond the amount necessary to deter future non-compliance.

CONCLUSION

For the reasons stated above, Commerce's four partial AFA rates are not supported by substantial evidence. To assist Commerce in finally bringing this case to an end, the court will explain further what is required for the rates to be supported by substantial evidence. The evidence here shows that the product categories Commerce chose to correspond to the unreported sales encompass a wide variety of specific products, with dramatically differing dumping margins. In order to account for these variations, Commerce must use a much greater portion of the reported sales in order to achieve an AFA rate consistent with Fairmont's commercial reality. Alternatively, Commerce may use some other reasonable methodology for tying the unreported sales here to the reported sales used to calculate the AFA rates (such as through a comparison of the U.S. prices), avoiding the problem created by these fluctuations altogether. Commerce shall file its remand determination with the court before or on September 17, 2014. The parties shall have until October 17, 2014, to file objections, and the government will have until October 31, 2014, to file its response.