concurring in part and concurring in result:
I join the per curiam opinion except for part II.A, concerning the sufficiency of the evidence under the Hobbs Act, 18 U.S.C. § 1951. I also agree with the conclusion that the Hobbs Act reaches Lynch’s actions, but I believe the law of our circuit requires us to reach that result in a different fashion than the court does.
I
The majority’s analysis should be correct. In particular, United States v. Lopez, 514 U.S. 549, 115 S.Ct. 1624, 131 L.Ed.2d 626 (1995), handed down several months after United States v. Collins, 40 F.3d 95 (5th Cir.1994), substantially clarified the approach we are to take in evaluating federal criminal statutes enacted un*1163der Congress’s commerce power. Lopez drew a sharp line between activities Congress may regulate as per se interference with commerce and those it could regulate only through a showing of substantial interstate effect. 514 U.S. at 558-559, 115 S.Ct. 1624. Similarly, United States v. Morrison, 529 U.S. 598, 615-17, 120 S.Ct. 1740, 146 L.Ed.2d 658 (2000), struck down a Commerce Clause enactment as pertaining to activities with an insufficiently direct tie to interstate economic activity, but did not indicate that where the ties to interstate economic activity are direct, there is any Commerce Clause reason for limiting federal authority.
There is therefore no basis in the Supreme Court’s recent Commerce Clause jurisprudence for finding this case beyond the reach of the Hobbs Act, which all agree reaches to the boundaries of Congress’s constitutional authority. See Stirone v. United States, 361 U.S. 212, 215, 80 S.Ct. 270, 4 L.Ed.2d 252 (1960); United States v. Lynch, 282 F.3d 1049, 1051-52 (9th Cir.2002) (Lynch I). Where, as here, the defendant lured the victim over state lines, used interstate telecommunications systems to facilitate the crime, drove the stolen truck through several states, and used the stolen debit card to facilitate the transfer of funds from a Nevada bank to the defendant while he was in other states, the impact of the robbery on interstate commerce was direct and immediate. The circumstances here are thus entirely different from situations in which the prosecution relies on the origin of the products stolen or the likely aggregate impact of similar robberies on the market for certain products in other states. Cf., e.g., United States v. Perrotta, 313 F.3d 33, 36 (2d Cir.2002) (reversing Hobbs Act conviction because “the government must show something more than the victim’s employment at a company engaged in interstate commerce to support Hobbs Act jurisdiction”); United States v. Peterson, 236 F.3d 848, 853 (7th Cir.2001) (rejecting government theory that the Hobbs Act reaches a theft of currency because it was printed out of state); United States v. Wang, 222 F.3d 234, 240 (6th Cir.2000) (rejecting an aggregation theory that would otherwise create “a general federal police power with respect to the crimes of robbery and extortion”).
Our caselaw has gone off the rails, however. As much as I would prefer to conclude otherwise, there is in my view simply no escaping the fact that the earlier opinion in this case precludes the majority’s application of a “direct effects” theory.
Lynch I, relying on Collins, held unequivocally that robbery of individuals can only be prosecuted under the Hobbs Act if one of three criteria are met:
In United States v. Collins, the Fifth Circuit formulated a test for determining when the robbery or extortion of an individual would have the requisite de minimis effect. 40 F.3d at 100. Under this test, crimes directed toward an individual violate the Hobbs Act only if: (1) the acts deplete the assets of an individual who is directly and customarily engaged in interstate commerce; (2) if the acts cause or create the likelihood that the individual will deplete the assets of an entity engaged in interstate commerce; or (3) if the number of individuals victimized or the sum at stake is so large that there will be some cumulative effect on interstate commerce.
Id. at 1053 (emphasis added). I do not know how to read the italicized language as stating anything other than an unalterable requirement applicable in any robbery of an individual prosecuted under the Hobbs Act.
The majority reads Collins as intending the standard announced to apply only where the reliance is on indirect effects on commerce. That reading is possible. Col*1164lins did note that “[b]oth direct and indirect effects on interstate commerce may violate section 1951(a),” and the government’s theory in that case relied upon either the inference that the robbery affected the victim’s ability to conduct interstate business or the fact that the vehicle stolen had previously traveled in interstate commerce. 40 F.3d at 99.
The facts in Lynch I, however, were, of course, the very facts now before us, involving a crime that began by enticing the victim to travel interstate and ended with the interstate use of the stolen vehicle and credit card. Yet, the remand order in Lynch I, 282 F.3d at 1055, was absolutely clear that if the district court could not find Hobbs Act jurisdiction under one of those three categories, it had to dismiss the indictment.1 There is no mention in Lynch I of the possibility of direct effects on commerce, no citation to the reference to direct effects in Collins, and no indication whatever that the standard announced for this circuit was anything other than what the opinion said it was — the sine qua non requirement for Hobbs Act jurisdiction over robbery of individuals.
Moreover, our court has since relied on Lynch as adopting the rule governing Hobbs Act jurisdiction for robbery of individuals, not a rule limited to cases of indirect impact on commerce:
Specifically, [Lynch I\ held that when the government brings charges for robbery under the Hobbs Act, and the target of the robbery was an individual, the government must show that the defendant (1) stole from a person directly and customarily engaged in interstate commerce; (2) created a likelihood that the assets of an entity engaged in interstate commerce would be depleted; or (3) victimized a large number of individuals or took a sum so large that there was some cumulative effect on interstate commerce.
United States v. Rodriguez, 360 F.3d 949, 955 (9th Cir.2004) (emphasis added, internal quotation marks omitted) (citing Lynch 1, 282 F.3d at 1055).
While there is no law-of-the-case problem here, as the per curiam opinion notes, ante at 1157-58, we cannot avoid the fact that the rule concerning Hobbs Act jurisdiction announced in Lynch I is the law of the circuit.2 As a three-judge panel, we are powerless to rewrite Lynch I to apply, as it should, to (at most) indirect-effects cases. See Coalition of Clergy, Lawyers, & Professors v. Bush, 310 F.3d 1153, 1162 n. 3 (9th Cir.2002). Rather, our only recourse would be to call as a panel for en *1165banc review of Lynch I, explaining that the standard there adopted cannot be reconciled with Commerce Clause principles as announced in recent Supreme Court cases. See 9th Cir. Gen. Order 5.2(b).
There are good grounds, in addition to the palpable conflict with Supreme Court Commerce Clause jurisprudence, for hearing Lynch I en banc. For one thing, as the majority opinion suggests, Lynch I is impossible to reconcile with an earlier (but post-Lopez) case from this circuit, United States v. Atcheson, 94 F.3d 1237 (9th Cir.1996). In Atcheson, we affirmed a Hobbs Act conviction for a scam in which the defendants made appointments with contractors, waited for them to arrive for a “business meeting,” and then held them hostage and robbed them of cash, jewelry, and credit and ATM cards. We readily found a “direct impact on interstate commerce” and held that, “[where the crime itself directly affects interstate commerce, as in the Hobbs Act, no requirement of a substantial effect is necessary to empower Congress to regulate the activity under the Commerce Clause].” Id. at 1242-43 (emphasis added).
Lynch I distinguished Atcheson on the ground that the victims in Atcheson were targeted qua “business men and women.” 282 F.3d at 1054 (quoting Atcheson, 94 F.3d at 1240). But Atcheson never mentioned that the victims were engaged in interstate business, and did not rely on their status as business people as the basis for finding that robbing them of personal property established an effect on commerce.3 Instead, the basis for Atcheson’s holding was that the theft, not the victims, affected interstate commerce, because the credit cards stolen were from out-of-state financial institutions, the defendants made out-of-state telephone calls to facilitate the use of those cards, one victim’s property was fenced across state lines, and another victim’s funds were withdrawn from a bank headquartered out of state — factors considerably less substantial than the direct interstate effects in this case. In short, the distinction of Atcheson in Lynch I simply does not hold water; there was no basis for deciding the two cases under different standards despite their similar facts.
Moreover, as the majority notes, ante at 1156-57, the Eleventh Circuit, in United States v. Carcione, 272 F.3d 1297 (11th Cir.2001), had, before Lynch I was decided, addressed facts not distinguishable for Commerce Clause purposes from those here and held that the facts were sufficient to establish a direct effect on commerce, and that the Collins standard, previously adopted in the Eleventh Circuit, did not apply. Lynch I therefore created an in-tercircuit as well as an intracircuit conflict.
I therefore believe that there are compelling grounds for en banc consideration of the rule announced in Lynch I. The panel should have requested that the court hear this case en banc so as to reject the “only if’ standard. The alternative the majority adopts, of rewriting Lynch I, may be rough justice, as Lynch I, in my view, rewrote Atcheson. But appellate courts should not be in the business of pursuing correct results by ignoring established stare decisis principles, even to right a perceived prior breach of those same principles.
In short, while I agree with the majority’s Commerce Clause analysis, I am simply unwilling, as much as I might like to do so, to join the majority in recasting Lynch *1166I unilaterally rather than calling upon the court sitting en banc to do so.
II
Applying Lynch I rather than revising it, I would reach the same result as does the majority. Rodriguez supplies the basis for doing so.
Rodriguez involved a government sting operation in which the defendant agreed to take part in a theft of a drug “stash house” from which he and co-defendants expected to steal twenty-five kilograms of cocaine. Upholding Rodriguez’s Hobbs Act conviction, we held that “an intended robbery of cocaine from narcotics traffickers is the robbery of a business, and [we] do not require the government to make the heightened showing under Lynch [I ]” for robberies of businesses — only of individuals. 360 F.3d at 955-56. Noting that “the trafficking of narcotics is a federally-regulated activity implicating interstate commerce,” Rodriguez held that “federal jurisdiction exists to apply the Hobbs Act to conspiracies involving the theft of cocaine from narcotics traffickers.” Id. at 956.
Although the Lynch I panel failed to note it, the evidence at Lynch’s federal trial showed that the robbery and murder of Brian Carreiro stemmed from the unraveling of Carreiro, Lynch, and Larry Pizzichiello’s drug trafficking ring. Pizzi-chiello testified that Lynch owed Carreiro money from past deals. Lynch and Pizzi-chiello’s motive in killing Carreiro was, at least in part, to cancel that debt. While I cannot discern from the record whether Lynch and Pizziehiello succeeded in taking drugs from Carreiro along with his money, truck, and life, the activities charged under the Hobbs Act were plainly crimes against a drug trafficker in, at least in part, his “business” capacity. We therefore need not consider the Collins factors — not because those factors pertain only to indirect effects, as the majority holds, but because the robbery of Carreiro was not, under Rodriguez, the robbery of an individual.
Establishing federal jurisdiction by distinguishing among the activities of robbery victims, rather than looking to the actual or expected interstate effects of the robbery itself, is a rule that has no basis in the language of the Hobbs Act or in broader principles of federalism. Cf. Lynch I, 282 F.3d at 1051-52 (discussing federalism backdrop to Commerce Clause analysis). Atcheson, Lynch I, and Rodriguez, however, require us to focus on Carreiro’s involvement in drug trafficking, not on the interstate aspects of the crime itself. I would therefore affirm the Hobbs Act conviction on the basis of Rodriguez’s singling out of drug trafficking-related crimes as outside the purview of the rule regarding robbery of individuals announced in Lynch I.
I recognize that this result is not consistent with the remand language of Lynch I. Law of the case, however, gives way when there are intervening legal developments affecting the propriety of an earlier order in a case. Jeffries v. Wood, 114 F.3d 1484, 1489 (9th Cir.1997) (en banc). Rodriguez is, in my view, such a legal development. That Rodriguez interprets the very opinion, Lynch I, that would otherwise govern this appeal as law of the case does not seem to me to matter; Rodriguez is a plausible reading of Lynch I, and should be applied to all later appeals, including this one. Applying an after-decided case that is consistent with Lynch I is quite different from applying a rule of law different from the one announced in Lynch I.
CONCLUSION
Keeping the court’s precedent coherent when we sit in three-judge panels is not always an easy task. When one panel has strayed from our rules designed to achieve that end, as I believe the panel in Lynch I *1167did, it is tempting to restore order by bending the rules a second time. In the long run, however, the court and the litigants are better served if the second panel is scrupulous in applying our principles regarding the respective roles of three-judge and en banc panels, for principles serially ignored will eventually atrophy entirely. I therefore cannot concur in the majority’s treatment of Lynch I, although I entirely agree with its Commerce Clause analysis.
. The Lynch I panel explained the scope of the remand thusly:
We therefore vacate the district court’s denial of Lynch’s Rule 29 motion and remand for a determination whether the evidence presented at trial supports the conclusion that Lynch (1) stole from a person "directly and customarily engaged in interstate commerce;” (2) created a likelihood that the assets of an entity engaged in interstate commerce would be depleted; or (3) victimized a large number of individuals or took a sum so large that there was "some cumulative effect on interstate commerce.” [Collins,] 40 F.3d at 100. If the district court concludes there is federal jurisdiction under the new test, it should again deny Lynch's Rule 29 motion; if not, it should grant the motion and dismiss his indictment with prejudice.
282 F.3d at 1055.
. It is especially regrettable that the Collins "test” should have been adopted in this form, since Collins itself provided no principled argument for why those categories and no others should be sufficient for Hobbs Act jurisdiction in robberies of individuals. The Collins factors were simply those categories of conduct the Fifth Circuit had previously found susceptible to Hobbs Act jurisdiction; why it strung the factors together as an "only if” test is elusive. See Collins, 40 F.3d at 100.
. The defendants in Atcheson had traveled from Utah to Idaho to initiate their scheme, but our opinion in Atcheson does not report that any of the victims were found to have moved interstate to attend the "business meeting.” See Atcheson, 94 F.3d at 1240.