dissenting:
Goodyear and its lawyers were accused in this case of perpetrating a fraud on the Haegers and the court. If sustained, those charges could of course severely damage the professional reputations of the lawyers involved. The district court accordingly approached the task of determining whether the charges were true with great thoroughness and care. After conducting a lengthy evidentiary hearing and reviewing multiple rounds of briefing, the court concluded that Goodyear and its lawyers acted in bad faith when they failed to produce test results that were responsive to the Haegers’ document requests. I agree with the majority that the district court’s misconduct findings are supported by the record, but I nevertheless conclude that the $2.7 million sanctions award must be vacated.
The district court’s finding of bad faith authorized it to levy sanctions under its inherent power. Chambers v. NASCO, Inc., 501 U.S. 32, 50, 111 S.Ct. 2128, 115 L.Ed.2d 27 (1991). Those sanctions could have taken one of two forms: punitive sanctions, which are criminal in nature and intended to vindicate the authority of the court; or compensatory sanctions, which are civil in nature and designed to compensate the injured party for losses sustained as a result of the misconduct. Miller v. City of Los Angeles, 661 F.3d 1024, 1029-30 (9th Cir.2011); F.J. Hanshaw Enterprises, Inc. v. Emerald River. Development, Inc., 244 F.3d 1128, 1136-42 (9th Cir.2001).1
The district court chose not to impose punitive sanctions. Doing so would have required the court to follow procedures applicable in criminal cases, such as appointing an independent prosecutor, affording the accused the right to a jury trial, and demanding proof of misconduct beyond a reasonable doubt. Miller, 661 F.3d at 1030. Compensatory sanctions, by contrast, may be imposed by the court acting alone after providing adequate notice and an opportunity to be heard. Losar v. Ford Motor Co., 399 F.3d 1101, 1110 (9th Cir.2005). That is the route the district court chose to follow here. The question for us is whether the court correctly labeled the sanctions compensatory. If it did not — if the sanctions are instead punitive — they cannot stand. See Miller, 661 F.3d at 1029-30; F.J. Hanshaw, 244 F.3d at 1141-42.
*1144In my view, the $2.7 million sanctions award cannot be deemed compensatory. The award could be compensatory only if the record reveals a causal connection between the misconduct the court found and the amount it awarded. See Miller, 661 F.3d at 1029-30. The $2.7 million award represents all of the attorney’s fees incurred by the Haegers after Goodyear breached its discovery obligations, including fees for the years of litigation that ensued before the parties settled on the first day of trial. The court purported to find the necessary causal link between the misconduct and the fees awarded on the theory that, if Goodyear had produced the test results when it was supposed to, “the case more likely than not would have settled much earlier.” I do not think that finding is supported by the record.
Our decision in Miller v. City of Los Angeles, 661 F.3d 1024 (9th Cir.2011), illustrates the deficiency. In that case, the district court found that defense counsel violated an in limine order by suggesting during closing arguments that the decedent was armed when the defendant police officer shot him. Id. at 1026. The trial ended in a hung jury. The district court awarded the plaintiffs all fees incurred during the trial as a compensatory sanction, presumably on the theory that defense counsel’s improper closing argument caused the jury to hang, thus necessitating a retrial and rendering all of the fees incurred during the first trial a waste. We concluded that the award could not be deemed compensatory. Id. at 1030. The record did not establish a causal connection between the lawyer’s misconduct and the jury’s inability to reach a verdict. It was simply impossible to know, on the record compiled in that case, why the jury could not reach a verdict, and the limited evidence available suggested that it was not because of defense counsel’s improper remarks. Id.
The record in this case is similarly devoid of evidence establishing a causal link between Goodyear’s misconduct and the fees awarded. It’s anyone’s guess how the litigation would have proceeded if Goodyear had disclosed all responsive test results from the start. The case might have settled right away, as the district court assumed, but that seems unlikely. The test results did not provide conclusive proof that the Haegers’ tire failed due to its defective design. To be sure, the test results were favorable to the Haegers: The results supported the Haegers’ theory that Goodyear sold tires that were prone to failure when used on motor homes at highway speeds, especially in hot driving conditions like those prevailing at the time of the Haegers’ accident in Arizona. But even if those test results had been put before the jury, Goodyear still planned to argue that the Haegers’ own tire, which had endured more than 40,000 miles of wear and tear, failed because it struck road debris, not because the tire was defective. And Goodyear has consistently maintained (whether rightly or wrongly) that the test results it concealed do not accurately predict tire behavior in real-world driving conditions.
If anything, it seems more plausible to assume that the case would have proceeded to trial had the test results been timely disclosed. The Haegers’ grievance is that they accepted a low-ball settlement from Goodyear on the eve of trial under false pretenses. The concealed test results, they contend, would' have significantly strengthened their hand. That suggests the Haegers would have been willing to take their case to the jury if Goodyear had refused to increase its offer, but it does not suggest that Goodyear would have thrown in the towel and met the Haegers’ demands. In fact, the only relevant data point in the record supports the opposite conclusion. In the Schalmo case, one of *1145the other motor home accident suits involving the same allegedly defective tire, Goodyear produced the test results at issue, but the plaintiffs and Goodyear elected to take the case to trial (with the jury returning a sizeable verdict for the plaintiffs). Goodyear did not settle that case immediately upon disclosure of the test results, as the district court assumed would have happened here.
In short, we simply don’t know — and have no way of reliably figuring out — what would have happened if timely disclosure of the test results had occurred. Thus, I think the district court clearly erred in finding that “the case more likely than not would have settled much earlier” had Goodyear disclosed the test results when it should have.
The majority does not contend that a causal connection between Goodyear’s misconduct and the fees awarded has been shown here, as required for the sanctions to be deemed compensatory. The majority instead contends that any language from Miller purporting to impose such a requirement is dicta. Maj. op. at 1136-87. I don’t think that portion of Miller can be dismissed as dicta, but even if it could, that wouldn’t matter. Miller’s, discussion of causation did not break new ground; it simply reflects the well-established principle that a sanction can be deemed compensatory only if it compensates the injured party for losses sustained as a result of the sanctionable misconduct. See, e.g., Lasar, 399 F.3d at 1111; F.J. Hanshaw, 244 F.3d at 1142. What we said about causation in Miller — whether dicta or not — merely illustrates why the fees awarded in this case were not sustained as a result of Goodyear’s misconduct.
The majority reads Chambers v. NAS-CO, Inc., 501 U.S. 32, 111 S.Ct. 2123, 115 L.Ed.2d 27 (1991), as establishing a competing principle: that a fee award may be deemed compensatory even if the fees were not incurred as a result of the sanc-tionable misconduct, so long as the misconduct involves “frequent and severe abuses of the judicial system.” Maj. op. at 1138. The majority assumes that principle must be valid because, in its view, not all of the fees awarded to NASCO were incurred as a direct result of Chambers’ misconduct.
I see two problems with the majority’s reading of Chambers. First, it is by no means clear as a factual matter that the majority’s reading is correct. The district court in Chambers expressly held that Chambers’ misconduct began even before NASCO formally filed suit. After Chambers informed NASCO that he would not honor the agreement to sell his local television station, NASCO gave Chambers notice on a Friday that it intended to file suit the following Monday seeking specific performance. That advance notice was required by court rules because NASCO also intended to seek a temporary restraining order .preventing Chambers from disposing of the station pending resolution of the suit. 501 U.S. at 36, 111 S.Ct. 2123. Rather than acknowledge that he had no valid defense to the suit, and that he therefore had no business putting NASCO to the expense of filing it, Chambers embarked on what turned out to be a years-long campaign of bad-faith litigation misconduct, beginning with his efforts over the weekend to fraudulently transfer ownership of the station in order to deprive the district court of jurisdiction. Id. at 36-37, 111 S.Ct. 2123. Because the district court found that Chambers never had a good-faith basis for resisting the relief NASCO sought, and that all of the actions he took in “defending” the suit were aimed solely at obstructing and delaying the inevitable sale of the television station, it seems fair to say that all of NASCO’s attorney’s fees were incurred as a direct result of Chambers’ misconduct. See id. at 50-51, 111 S.Ct. 2123.
*1146Second, even if some portion of NAS-CO’s attorney’s fees were not incurred as a direct result of Chambers’ misconduct, the majority incorrectly assumes that the Supreme Court upheld the award as purely compensatory. The sanction imposed there was not purely compensatory; it served the “dual purpose” of (1) vindicating the court’s own authority and (2) “mak[ing] the prevailing party whole for expenses caused by his opponent’s obstinacy.” Id. at 46, 111 S.Ct. 2123 (internal quotation marks omitted). The first of these purposes, we have subsequently held, is the domain of punitive sanctions, and the Court in Chambers left no doubt that punishment was indeed a key purpose of the sanctions imposed in that case. See id. at 55 n. 17, 111 S.Ct. 2123 (“the sanctions imposed on Chambers were aimed at punishing not only the harm done to NAS-CO, but also the harm done to the court itself’). Because it was partly punitive, the sanctions award did not need to be limited to fees directly caused by Chambers’ misconduct.
I concede that the district court imposed the sanctions in Chambers without applying the heightened procedural protections we have subsequently held are necessary before punitive sanctions may be imposed, and that the Supreme Court nonetheless affirmed. I don’t think we can read anything into that fact. The defendants in Chambers did not raise any due process arguments, and the Supreme Court therefore did not address whether the process afforded the defendants was adequate. Moreover, the law has changed since Chambers was decided. A few years later the Court issued International Union, United Mine Workers v. Bagwell, 512 U.S. 821, 114 S.Ct. 2552, 129 L.Ed.2d 642 (1994), the case from which we first derived the rule that imposition of punitive sanctions must be accompanied by the procedural protections applicable in criminal cases. See F.J. Hanshaw, 244 F.3d at 1137-38. If any doubts lingered about whether Chambers authorizes imposition of so-called “dual purpose” sanctions without following the procedures applicable in criminal cases, we put those to rest in Miller. The dissent in Miller made that very argument, 661 F.3d at 1039 (Ikuta, J., dissenting), but the panel majority implicitly rejected it. See id. at 1030.
None of this is to suggest that compensatory sanctions can’t be fashioned at all. There may well be other ways to calculate the losses sustained by the Haegers as a result of the misconduct. For example, the most direct loss the Haegers sustained is that they probably settled their case for less than it was really worth. It might be possible to use the Schalmo case, and others like it if they exist, to calculate the difference between what the Haegers actually received in settlement and what they likely would have received — whether through an enhanced settlement or a jury verdict — if the test results had been disclosed in a timely manner. But going down that path would obviously be fraught with proof problems of its own.
Alternatively, instead of attempting to calculate lost settlement value, the district court could again focus on attorney’s fees incurred by the Haegers, limiting the award to fees that can be linked in a non-speculative way to the misconduct. The fees that most readily spring to mind are those wasted on expert discovery that took place under the mistaken assumption that key test results supporting the Haegers’ liability theory did not exist. Those and other fees similarly traceable to the misconduct are no doubt comparatively small, but I don’t think the district court was right in suggesting that .calculating them would be an impossible task. Those fees can be calculated; it’s just that they may produce a sanction smaller than seems *1147warranted given the severity of the misconduct the district court found.
If the sanctions that can properly be deemed compensatory seem too paltry under the circumstances, the district court could still fashion an award of punitive sanctions, so long as it applies the corresponding heightened procedural protections. See Miller, 661 F.3d at 1030-31; F.J. Hanshaw, 244 F.3d at 1141-42. Because Goodyear and its lawyers were not afforded those protections before punitive sanctions were imposed, I dissent from the majority’s affirmance of the $2.7 million award.
. Sanctions may also be civil in nature if they are "designed to compel future compliance with a court order.” International Union, United Mine Workers v. Bagwell, 512 U.S. 821, 827, 114 S.Ct. 2552, 129 L.Ed.2d 642 (1994). The sanctions imposed here could not serve that function because the litigation between the Haegers and Goodyear had long since ended.