Full opinion text
PRADO, Circuit Judge: A jury convicted former Enron Corporation CEO Jeffrey K. Skilling (“Skilling”) for conspiracy, securities fraud, making false representations to auditors, and insider trading. Skilling argues that the government prosecuted him using an invalid legal theory, that the district court used erroneous jury instructions, that the jury was biased, that prosecutors engaged in unconstitutional misconduct, and that his sentence is improper. We affirm the con-fictions, vacate the sentence, and remand for resentencing. I. Factual Background Skilling’s rise at Enron began when he founded Enron’s Wholesale business in 1990. In 1997, he became Enron’s President and Chief Operating Officer and joined the Board of Directors. In February 2001, he became Enron’s CEO, and on August 14, 2001, Skilling resigned from Enron. About four months after Skilling’s departure, Enron crashed into sudden bankruptcy. An initial investigation uncovered an elaborate conspiracy to deceive investors about the state of Enron’s fiscal health. That conspiracy allegedly included overstating the company’s financial situation for more than two years in an attempt to ensure that Enron’s short-run stock price remained artificially high. With Congress looking on, the President appointed a team of investigators, the Enron Task Force. The investigation led to criminal charges against Skilling and many others. According to the government, the conspiracy, led by Skilling and Ken Lay (“Lay”), Enron’s CEO until Skilling took over (and again after his abrupt exit), worked to manipulate Enron’s earnings to satisfy Wall Street’s expectations. Other top Enron officials were key players in the unlawful scheme, including Richard Cau-sey (“Causey”), the Chief Accounting Officer (“CAO”); Andrew Fastow (“Fastow”), the Chief Financial Officer (“CFO”); and Ben Glisan (“Glisan”), the Treasurer. A. Conspiracy and Securities Fraud Several of Skilling’s convictions stem from allegations of conspiracy and securities fraud. The government presented evidence that Skilling engaged in fraud in several of Enron’s business endeavors. As an international, multi-billion dollar enterprise, Enron had elaborate financial dealings. At the time of its bankruptcy, the company was comprised of four major businesses: Wholesale, which bought and sold energy; Transportation and Distribution, which owned energy networks; Retail, or Enron Energy Services (“EES”), which sold energy to end-users; and Broadband, or Enron Broadband Services (“EBS”), which bought and sold bandwidth capacity. The government alleged that Skilling took specific fraudulent actions with respect to Wholesale, EES, and EBS. Wholesale, the most profitable division, accounted for nearly 90% of Enron’s revenue. The government presented evidence to show that the conspirators lied about the nature of Wholesale, calling it a “logistics company,” even though it was a much more economically volatile “trading company.” Construing Wholesale as a “logistics company” had important ramifications for how investors valued the division. In fact, Skilling reportedly told Ken Rice (“Rice”), EBS’s CEO, that if investors perceived Enron as a trading company, its stock would “get whacked.” The alleged artifice also included masking the losses of Enron’s other struggling subdivisions by shifting the losses to Wholesale. That made the struggling divisions appear financially sound and thus encouraged additional investment. EES was a retail undertaking that Enron created to sell natural gas to customers in deregulated markets. Although Enron had high expectations for EES’s profitability after its initial start-up period, EES did not meet these expectations. As of the fall of 2000, various utilities in California owed Enron substantial fees, which Enron had already booked as profits under its “mark-to-market accounting.” The utilities, however, were suffering heavy financial losses and stopped paying these fees. Under general accounting rules, Enron should have recorded a loss of hundreds of millions of dollars based on the failure of the utilities to pay the fees, but Skilling and his co-conspirators tried to hide the harm by transferring the losses to Wholesale so that EES would continue to show promise, at least on paper. The government claims that Skilling hid EES’s other problems as well. For example, in early 2001, EES employees allegedly realized that Enron was not properly valuing EES’s contracts and that, again because of Enron’s use of mark-to-market accounting, Enron would need to record a loss of many millions of dollars. Skilling allegedly told David Delainey, who was in charge of EES, to “bleed the contract issues over time” instead of recognizing the loss all at once. Skilling again concealed EES’s losses within Wholesale in late March 2001, after the California Public Utilities Commission decided to add a surcharge to electricity. Enron lost hundreds of millions of dollars as a result of this surcharge because, under its contracts, it could not pass the extra fees on to its customers. After Skill-ing was consulted and signed off, Enron shifted the EES losses to Wholesale by transfering EES’s risk-management books to Wholesale. Business at EES did not improve, and by August 2001, when Skill-ing left Enron, EES had lost over $700 million in that year alone. Enron failed to account for these losses properly, making EES appear to be in better financial shape than it really was. EBS was Enron’s attempt to enter the telecommunications industry. Enron invested more than $1 billion in EBS, but lost money every quarter as EBS struggled to meet earnings targets. The government claims that in 2000, EBS managed to reach its earnings targets, but only by means of transactions afield from its core business (such as selling and monetizing corporate assets). Skilling allegedly hid from investors EBS’s failure to meet earnings targets through core business activities. The government claims that Skilling knew EBS was struggling, at least based on its record of performance, but that he wanted to announce to the investing public that EBS was doing well and would do even better in 2001. Although EBS’s executives said it was impossible, Skilling set EBS’s earnings targets for 2001 to be a loss of only $65 million. EBS’s personnel initially thought a loss estimate of nearly $500 million was more realistic, and, even with the best circumstances, they projected losses of at least $110 million. Rice, EBS’s CEO, warned Skilling that the earnings targets for EBS were wrong, but Skilling apparently would not change them. Skilling told Rice that certain international assets were not producing sufficiently and that “we really need to hang in there for a year or two until EES and EBS could pick up the slack,” because Enron “didn’t need any more bad news.” In 2001, EBS was projected to lose $35 million in the first quarter, but Rice quickly realized that losses would actually be around $150 million. Skilling allegedly found out but would not budge on earnings targets, instead authorizing EBS to fire employees and engage in more non-core business to boost revenues. It worked for the first quarter, although Rice likened the monetizations to “one more hit of crack cocaine on these earnings.” EBS faired a little better in the second quarter of 2001, reporting losses of approximately $100 million. Given that EBS continued to lose money, however, Enron decided to merge EBS into Wholesale. Ultimately, Enron lost the entire $1 billion that it had initially invested in EBS. B. False Representations About Enron’s Finances Many of the allegations of fraud also stem from Skilling’s representations to investors about the financial standing of Wholesale, EES, and EBS. Skilling, as a high ranking corporate officer, held conference calls with investors to update them on the company’s progress. The government claims that Skilling misled investors during these calls. For example, on January 22, 2001, Enron released its earnings report for the previous quarter, and Skilling told investors that “the situation in California [regarding the utilities] had little impact on fourth quarter results. Let me repeat that. For Enron, the situation in California had little impact on fourth quarter results.” Skilling also stated that “nothing can happen in California that would jeopardize” earnings targets. However, when he made these statements, Skilling allegedly knew that the California utilities likely could not pay the fees that Enron was expecting and that Enron might have to write off a loss of hundreds of millions of dollars. He also listened silently as Mark Koenig (“Koe-nig”), Enron’s Director of Investor Relations, assured investors that non-core business revenues were a “fairly small” amount of EBS’s earnings, which the government alleges was not actually the case. Three days later, Skilling spoke at Enron’s annual analysts conference, claiming that EES and EBS, like Enron’s other major businesses, had “sustainable high earnings power.” Skilling argues that this statement was merely harmless puffery. At the conference, he also reasserted that Wholesale was “not a trading business. We are a logistics company.” On March 23, 2001, Enron held a special conference call with analysts. Enron’s stock price had been declining, and investors began surmising that EBS was having financial difficulties. Skilling comforted investors, saying that EBS was “having a great quarter” and that Enron was “highly confident” that EES would meet its earning target. According to the government, however, Skilling knew both divisions were in extreme financial turmoil. On April 17, 2001, Skilling hosted another conference call in which he explained the transfer of EES’s risk-management books to Wholesale by saying that there was “such capacity in our wholesale business that were — we just weren’t taking advantage of that in managing our portfolio at the retail side. And this retail portfolio has gotten so big so fast that we needed to get the best — the best hands working on risk management there.” In fact, the government claims that Skilling used the transfer to hide losses. Skilling also said that the “first quarter results were great” at EES, even though they were down substantially, and he again praised EBS, explaining that there was a “very strong development of the marketplace in the commoditization of bandwidth” and “we’re feeling very good about the development of this business.” Skilling was silent again while Rice and Koenig understated EBS’s non-core revenues. On July 17, 2001, Skilling told investors that EES “had an outstanding second quarter” and was “firmly on track to achieve” its earnings targets. That quarter alone, EES lost hundreds of millions of dollars. Skilling reiterated that the EES reorganization was based on a concern for management efficiency, while the government contends that the only purpose of the EES reorganization was to hide EES’s losses. C. Manipulating Enron’s Reserves Skilling also allegedly committed fraud when he manipulated Enron’s reserves to hit specific earnings targets in the fourth quarter of 1999, the second quarter of 2000, and the fourth quarter of 2000. Stock analysts made various projections regarding the earnings that Enron would announce each quarter, and the average of these estimates was known as the “consensus estimate.” The government claims that Skilling was particularly committed to hitting or beating the consensus estimate. In January 2000, the consensus estimate for the fourth quarter of 1999 was earnings of 30<t per share, which Enron could meet based on its earnings for that quarter. The day before the company was to announce its earnings, however, Koenig brought Skilling unwelcome news: the consensus estimate had jumped a penny per share. Skilling purportedly decided to announce earnings high enough to reach the estimate, even though the increase was not merited by any change in Enron’s underlying financial portrait. Skilling allegedly took a similar unwarranted action at the end of the second quarter of 2000. At that time, the consensus estimate was 32<t per share. A draft earnings report showed that Enron was going to announce earnings that met the estimate. Skilling, however, wanted to beat the consensus estimate by reporting 34<c per share. To do that, he allegedly told Wholesale to increase its earnings by $7 million, and then by an additional $7 million. Wholesale acquiesced both times, reopening its books and adding $14 million from a reserve account that it had set aside to cover potential liabilities. The government claims that Enron did not have a business reason for using its reserves to increase Wholesale’s earnings, instead doing so solely to exceed analysts’ expectations. The government contends that Skilling improperly used the reserve accounts again later that year. Wholesale’s business was very profitable during the second half of 2000, and by late December it had placed over $850 million in reserves. The decision to put that money away was not based on feared future liabilities; instead, Wholesale set that money aside specifically to use in the event of an unfavorable consensus estimate. At the end of the fourth quarter, Skilling ordered that Enron recall some of that money to guarantee that Enron could announce a specific level of earnings. D. Third-Party Entities LJM and LJM2 Another avenue of Skilling’s alleged fraud came from his use of pseudo third-party entity LJM (and later LJM2) to improperly hedge its investments, doing so through four “secret” oral side deals. Fastow, Enron’s CFO, proposed to Skilling and Causey that they create an entity to help Enron more easily meet market expectations. The impetus for creating LJM was the $200 million that Enron had received from an investment in a company called Rhythms Net. Enron wanted to book that money, but it was possible that the value of the investment would drop, meaning that the asset’s expected value would have to be reduced. By transferring the asset to a pseudo third-party, however, Enron could hedge its investment without needing to pay market rates for this hedging service. The government claims that LJM became that pseudo third-party. Enron contributed $234 million of its own shares in seed money to form LJM. Fastow, who was LJM’s general partner, contributed $1 million, and outside investors contributed $15 million. Enron encountered a potential problem, however, with using LJM to hedge its investments. Fastow faced a conflict of interest, because he was both Enron’s CFO and LJM’s general partner. Before Enron could sign a deal with LJM, Enron’s code of conduct required the Office of the Chairman, which consisted of Skilling and Lay, to waive the conflict rules, and Enron would have to disclose this waiver in its Securities and Exchange Commission (“SEC”) filings. The Office of the Chairman granted the waiver, but allegedly not without first creating controversy within Enron’s senior leadership. After forming LJM to hedge the Rhythms Net investment, Enron’s first deal with LJM involved an interest in a Brazilian power plant, known as Cuiaba, that Enron sold to LJM in 1999. Enron was concerned that its South American unit would not meet its earnings targets, so it decided to sell its interest in Cuiaba to obtain additional revenue. Initially, Enron tried to find a third party to buy the interest in Cuiaba, but was unsuccessful. Skilling, the government claims, then called Fastow and tried to sell Enron’s interest in Cuiaba to LJM. Fastow at first was not interested; not only was the asset a bad investment, there was no time for due diligence. Skilling allegedly replied, “Don’t worry. I’ll make sure that you’re all right on the project. You won’t lose any money.” That oral understanding did not appear in the deal write-up, and the accountants treated it as a real sale, even though, based on the alleged oral understanding between Skilling and Fas-tow, it was not a legitimate sale. Enron eventually bought back the Cuia-ba interest, paying full value (notwithstanding depreciation) plus 13%. Allegedly to camouflage the deal by avoiding round numbers, LJM received an extra $42,000 above the 13%. To allay further scrutiny, before Enron bought back the interest, Fastow apparently “sold” his interest in LJM to Michael Kopper (“Kop-per”), a former Enron executive. That way, Enron did not need to describe the Cuiaba buyback as a transaction with a related party. For this to work, however, Skilling needed to honor his secret oral promise to Fastow that LJM would not lose money on the deal, despite Kopper becoming LJM’s chief; the government claims that Skilling orally confirmed that he would do so. The second allegedly fraudulent secret side deal between Enron and LJM involved the sale of Nigerian barges. The government claims that in the waning days of 1999, LJM warehoused assets for Enron, allowing Enron to claim earnings during 1999 while it arranged for permanent buyers after the end of the year. With respect to the Nigerian barges deal, in late 1999, Enron sought to sell its interest in a group of barges anchored off the coast of Nigeria to meet an earnings target at the end of the quarter. Most investors were nervous about putting money into Nigeria, so Enron could not find a buyer. Skilling allegedly called Fastow into his office and asked for LJM to buy the barges, again saying he would “make sure” that LJM would not lose money. Fastow initially was reluctant, because he was trying to raise money for LJM and did not want to scare off investors. However, he told Skilling that LJM would purchase the Nigerian barges if Enron could not find another investor within six months. Fastow, on behalf of Enron, then arranged for Merrill Lynch to buy the Nigerian barges from Enron. Although Merrill Lynch did not really want to purchase the barges, Fastow purportedly made an oral “guarantee” — although he likely did not use that word — that Merrill Lynch would have to hold the barges for only six months in return for a risk-free profit. Because Merrill Lynch’s investment was not at risk, the government alleges that Enron improperly treated the transaction as a sale and should not have recorded any earnings from the deal. At the end of the six months, Enron still could not find another buyer, so LJM purchased the interest from Merrill Lynch, apparently without even negotiating over price. Causey allegedly assured LJM that it would earn a guaranteed return on the deal, meaning that there was no transfer of risk to LJM. In sum, the government asserts that the sale of the Nigerian barges was not a true sale and that Enron improperly recognized earnings from the deal. The third secret oral side deal involved the “Raptors,” which were special purpose entities (“SPE”) that would hedge assets for Enron, meaning that if the assets decreased in value, a Raptor would cover the difference — at least on paper. The government asserts that for Enron to validly hedge an asset, a third party must own at least three percent of the SPE’s equity, and that equity must be at risk. The government alleges that LJM acted as that “third party” even though it was not really a separate entity, agreeing to provide the equity that would be at risk. The government contends that to fund the Raptors, LJM contributed $30 million and Enron spent $400 million of its own stock. Apparently, Fastow was at first against mixing LJM with the Raptors, but the government argues that Skilling convinced him through a “secret” side deal, whereby after funding the Raptors, LJM would recoup its $30 million and would receive an additional $11 million. To do this, Enron agreed to pay $41 million to the Raptors to purchase a “put” on the stock Enron had contributed to the Raptors. The Raptors then transferred that $41 million to LJM. In return, the Raptors had to pay Enron if the price of the stock Enron used to capitalize the Raptors fell below a certain level. LJM thus had nothing at risk; if the hedged assets dropped in value, the hedging Raptor, not LJM, was liable for the loss, and LJM recouped the capital it initially invested in the Raptors, along with an additional $11 million. To make this transaction work, however, Enron’s accountants had to sign off. Arthur Andersen, Enron’s external auditor, approved the $41 million payment to LJM once it was described as a return “on” capital to LJM and not a return “of’ capital. The auditors allegedly were not told that LJM and Enron did not haggle over the value of the hedged assets or that LJM was not actually a true third party. The government terms this entire deal— both Enron’s “put” and LJM’s return promise to allow Enron to hedge an asset at any value — the “quid pro quo.” The Raptors were also involved in other transactions. For example, in 2000, Enron hedged EBS’s investment in Avici, an internet company, into one of the Raptors. That asset, at its highest value, was worth around $160 million, but the value was decreasing. Enron allegedly hedged the Avici interest with a Raptor at its highest value. The asset plummeted in value over the next few months, but because of the allegedly fraudulent hedge, Enron did not record the loss. Under the government’s theory, Skilling knew all about the deceit and even told Fastow to “[k]eep it up.” The alleged fraud continued throughout 2000. In fact, by the end of 2000, another one of the Raptors lost more than $100 million because of similar hedging. In total, Enron’s use of the Raptors allegedly kept nearly $500 million in losses off of Enron’s books in 2000. The fourth “secret” side deal the government presented was “Global Galactic.” Global Galactic was not actually a single deal but instead was the name that Fastow gave to the three-page handwritten list of his undocumented side deals with Enron. Fastow claimed that he created the list to keep track of the various deals and to ensure that he was on the same page as Enron’s management on the substance of these deals. Skilling asserts that he had no connection to the deals on this list. E. False Representations to Auditors and Insider Trading In addition to conspiracy and securities fraud, the jury also convicted Skilling for making false representations to auditors and insider trading. Skilling was required to sign Enron’s SEC reports. The government claims that Skilling knew of the many false statements within Enron’s SEC filings. For example, Enron characterized the money it obtained from selling the Cuiaba interest and from the Nigerian barges deal as legitimate income, when in fact these deals did not represent true sales. Likewise, although Enron had finalized the terms of its buyback of the Cuiaba interest much earlier, Enron allegedly delayed the transaction so Fastow could sell his interest in LJM to Kopper, with the goal of allowing Enron to avoid disclosing the buyback as a related-party transaction in its SEC reports. The government also claims that Enron kept Arthur Andersen in the dark about the false statements in its SEC filings. Skilling and other members of Enron’s management were required to give Arthur Andersen certain management-representation letters, which actually contained known falsehoods, thus negating the validity of the audits. The assertedly false statements included claims that the auditors had access to all financial records and that Enron had disclosed all related-party transactions. When Skilling resigned from Enron in August 2001, Enron’s internal financial numbers were ghastly. On September 6, 2001, Skilling allegedly called his broker and tried to sell 200,000 shares of Enron stock. The sale did not go through, however, because the SEC still listed Skilling as an Enron “affiliate.” This designation meant that the broker had to disclose the sale to the SEC and the public, which Skilling wished to avoid. Skilling allegedly told his broker to wait to complete the transaction until he obtained a letter from Enron saying that he was no longer a part of Enron’s management. Skilling sent that letter to his broker on September 10. Because of the terrorist attacks of September 11, Skilling was unable to sell his shares until September 17, the first day the markets reopened. He allegedly called his broker on that day, reiterated his order to sell, and told him he did not want the people at Enron to know about it. When Skilling testified before the SEC in December 2001, he stated that he sold the shares because he became “scared” after the September 11 attack and that “[t]here was no other reason other than September 11th that I sold the stock.” The government contends that Skilling’s testimony was a lie and that the sale amounted to insider trading. II. Trial and Sentence According to the government, Skilling’s conduct constitutes multiple instances of criminal activity. In July 2004, a grand jury returned a superseding indictment charging Skilling, Lay, and Causey with various counts of conspiracy, securities fraud, wire fraud, and insider trading. The indictment charged Skilling with one count of conspiracy to commit securities and wire fraud, fourteen counts of securities fraud, four counts of wire fraud, six counts of making false representations to auditors, and ten counts of insider trading. Several weeks before trial began, Causey pleaded guilty to one count of securities fraud, and the government dropped four counts against Skilling that involved Cau-sey. Skilling and Lay went to trial, and, at the close of its case, the government eliminated four additional counts. At trial, Skilling argued that he did not break any laws, that he was loyal to Enron, and that he consistently relied on competent legal and accounting advice; he characterized any falsehoods in his statements to analysts as immaterial in content and context. He also challenged the veracity of the government’s witnesses, such as Fastow and Glisan. For example, Skill-ing claimed that Fastow’s testimony regarding Skilling and Fastow’s alleged shared understanding about Cuiaba and the Nigerian barges — “bear hugs,” in Fas-tow’s words — did not reflect what Skilling said but merely consisted of Fastow’s misinterpretations. Skilling also questioned the validity of the so-called Global Galactic list. In May 2006, the jury found Skilling guilty of nineteen counts: one count of conspiracy, twelve counts of securities fraud, five counts of making false statements, and one count of insider trading; the jury acquitted Skilling of nine counts of insider trading. The jury convicted Lay of every count against him. The district court sentenced Skilling to 292 months’ imprisonment, three years’ supervised release, and $45 million in restitution. III. Honest-Services Fraud Allegation On appeal, Skilling argues that we must reverse all of his convictions because the government used an invalid theory of “honest-services fraud” to convict him. The jury convicted Skilling of one count of conspiracy. The indictment and the government’s theory allowed for three objects of the conspiracy: to commit (1) securities fraud, (2) wire fraud to deprive Enron and its shareholders of money and property, and (3) wire fraud to deprive Enron and its shareholders of the honest services owed by its employees. Because the jury returned a general verdict, we cannot know on which of the three objects it relied. In Yates v. United States, 354 U.S. 298, 312, 77 S.Ct. 1064, 1 L.Ed.2d 1356 (1957), the Supreme Court held that where a jury returns a general verdict of guilt that might rest on multiple legal theories, at least one insufficient in law and the others sufficient, the verdict must be set aside. In such a situation, we cannot trust the jury to have chosen the legally sufficient theory and to have ignored the insufficient one, because “[j]urors are not generally equipped to determine whether a particular theory of conviction submitted to them is contrary to law.” Griffin v. United States, 502 U.S. 46, 59, 112 S.Ct. 466, 116 L.Ed.2d 371 (1991). In Yates, the defendant was charged with a single count of conspiracy with two objects: (1) to advocate the overthrow of the federal government and (2) to organize the Communist Party of the United States. Yates, 354 U.S. at 300, 77 S.Ct. 1064. The “organize” object, however, was time-barred and thus legally insufficient. Id. at 304-11, 77 S.Ct. 1064. The Court overturned the conviction, because it was not apparent whether the jury’s verdict rested on the legally sufficient “advocate” object or the legally insufficient “organize” object. Id. at 312, 77 S.Ct. 1064. Likewise, if any of the three objects of Skilling’s conspiracy offers a legally insufficient theory, we must set aside his conviction to avoid the possibility that the verdict rests on the insufficient theory. Skilling avers that the honest-services fraud object of the conspiracy count is legally insufficient, mandating the reversal of the conspiracy conviction. He claims that this would also taint the convictions that rely upon the conspiracy count. We review de novo whether a theory of conviction is legally insufficient. United States v. Phillips, 219 F.3d 404, 409 (5th Cir.2000) (“We review questions of law and application of statutes de novo.”). The honest-services statute provides that “the term ‘scheme or artifice to defraud’ includes a scheme or artifice to deprive another of the intangible right of honest services.” 18 U.S.C. § 1346. That is, the statute defines the “scheme or artifice to defraud” language found in the substantive mail and wire fraud statutes, 18 U.S.C. §§ 1341 and 1343, respectively, to include the substantive crime of depriving another of one’s honest services. Thus, wherever mail or wire fraud is an object of a conspiracy, there are two possible objects that can be charged: use of the mails or wires to deprive another of (1) property or money or (2) honest services. See 18 U.S.C. §§ 1341,1343. Although § 1346 defines “scheme or artifice to defraud,” it offers no definition of “honest services.” Before the enactment of § 1346 in 1988, courts read the notion of honest services into the wire and mail fraud statutes. Both § 1341 and § 1343 read, “Whoever, having devised or intending to devise any scheme or artifice to defraud, or for obtaining money or property .... ” Courts read the disjunctive between “to defraud” and “for obtaining money or property” as indicating two separate objects of the scheme or artifice. McNally v. United States, 483 U.S. 350, 358, 107 S.Ct. 2875, 97 L.Ed.2d 292 (1987). Money or property was one object, and courts construed “to defraud” to include schemes whose object was the deprivation of intangible rights such as honest services. Id. With McNally, the Supreme Court ended prosecution for honest-services fraud as a part of mail fraud. Id. at 358-60, 107 S.Ct. 2875. The disjunctive phrase in the mail-fraud statute did not indicate that there were two objects of the fraud; rather, the Court “read § 1341 as limited in scope to the protection of property rights.” Id. at 360, 107 S.Ct. 2875. In response, Congress enacted § 1346, which this court and others have held was intended to overturn McNally with respect to the inclusion of the right to honest services under §§ 1341 and 1343 and to restore “the honest-services doctrine developed in the years leading up to McNally.” United States v. Brumley, 116 F.3d 728, 733 (5th Cir.1997) (en banc); see also United States v. Rybicki, 354 F.3d 124, 136-37 (2d Cir.2003) (en banc). Thus, to determine what constituted “honest-services” fraud, we looked to our pre-McNally precedent and then set forth the rule for this circuit. Brumley, 116 F.3d at 733-34 (‘We decide today that services must be owed under state law and that the government must prove in a federal prosecution that they were in fact not delivered.”). In United States v. Gray, 96 F.3d 769 (5th Cir.1996), we considered the honest-services statute in the context of a private employer and employees. A jury convicted three assistant basketball coaches at Baylor University (“Baylor”) of conspiracy to commit mail and wire fraud by depriving Baylor of its honest services through a scheme to obtain credits and scholarships for players in violation of National Collegiate Athletic Association (“NCAA”) rules. Id. at 772. The coaches argued that honest-services fraud “improperly criminalize[d] mere deceit,” because the coaches had broken no law, only a private association’s rules, and “they lacked the requisite intent to either harm the victims or to obtain personal benefit .... Essentially, [the coaches] argue[d] that their scheme was not intended to harm Baylor but rather to help Baylor by ensuring a successful basketball team.” Id. at 774. We rejected the argument and noted that a breach of fiduciary duty of honesty or loyalty involving a violation of the duty to disclose could only result in criminal mail fraud where the information withheld from the employer was material and that, where the employer was in the private sector, information should be deemed material if the employee had reason to believe the information would lead a reasonable employer to change its business conduct. Id. at 774-75 (quoting United States v. Ballard, 680 F.2d 352, 353 (5th Cir.1982) (per curiam) (on petition for rehearing)). We concluded that the information the coaches withheld, namely that they were cheating, was material; had Baylor been aware of their actions, it undoubtedly would have changed its business conduct by recruiting players who satisfied NCAA requirements. Id. at 775. In light of our conclusions, the coaches’ conspiracy to violate NCAA rules was a federal crime; that their intent was to help rather than harm the university was of no consequence. Id. at 774-75. In United States v. Brown, 459 F.3d 509 (5th Cir.2006), we again addressed honest-services fraud and refined our jurisprudence. The defendants arranged for the Nigerian barges deal we described above, whereby Enron “sold” three energy-producing barges to Merrill Lynch. Id. at 513. The “purchase” allowed Enron to book, as earnings, the money it received from Merrill Lynch, thereby helping Enron meet its earnings targets. Id. at 513-16. In return, Enron assured Merrill Lynch a fixed rate of return on the investment — a flat fee — and promised that Enron or a third party would repurchase the barges within six months. Id. Such an agreement, however, would have rendered the deal a loan from Merrill Lynch to Enron, because Merrill Lynch had no equity at stake. Id. As a loan, the transaction could have no positive impact on Enron’s earnings, meaning that it was fraudulent for Enron to book any earnings from the deal. Id. After surveying our prior honest-services jurisprudence, we reversed the defendants’ convictions, concluding that their conduct did not fall within the bounds of the honest-services statute. Id. at 523. In particular, we held that where an employer intentionally aligns the interests of the employee with a specified corporate goal, where the employee perceives his pursuit of that goal as mutually benefiting him and his employer, and where the employee’s conduct is consistent with that perception of the mutual interest, such conduct is beyond the reach of the honest-services theory of fraud as it has hitherto been applied. Id. at 522. Importantly, we expounded upon our understanding of honest-services fraud by providing a crucial distinction from the facts in Gray: Gray is distinguishable both factually and legally. Gray is dissimilar to this case in part because the opinion recognizes nothing akin to Enron’s corporate incentive policy coupled with senior executive support for the deal (the deal was sanctioned by Fastow, Enron’s Chief Financial Officer), which together created an understanding that Enron had a corporate interest in, and was a willing beneficiary of, the scheme. The opinion in Gray presents only the coaches’ own belief that their scheme benefited the university; no one or any authority outside the cadre of coaches encouraged, approved, or even knew of the wrongdoing. Id. at 522 n. 13. Given the gloss this passage places on Gray, we can distill the holding in Brown to be the following: when an employer (1) creates a particular goal, (2) aligns the employees’ interests with the employer’s interest in achieving that goal, and (3) has higher-level management sanction improper conduct to reach the goal, then lower-level employees following their boss’s direction are not liable for honest-services fraud. Thus, we reversed the convictions of the employees in Brown because they were acting both in the corporate interest and at the direction of their employer. Id. at 522. In essence, Brown created an exception for honest-services fraud where an employer not only aligns its interests with the interests of its employees but also sanctions the fraudulent conduct, i.e., where the corporate deci-sionmakers, who supervised the employees being prosecuted, specifically authorized the activity. Skilling does not contest that he owed Enron a fiduciary duty. Instead, he contends that his conduct did not breach that duty, because his fraud was in the corporate interest and therefore was not self-dealing. In particular, Skilling asserts that he did not engage in his conduct in secret. Skilling further latches onto our discussion in Brown regarding corporate interest and contends that his actions were not fraudulent because he acted in pursuit of Enron’s goals of achieving a higher stock price. If this were the correct reading of Brown, that decision would be in irreconcilable conflict with Gray, which binds us, as the basketball coaches in Gray acted pursuant to their employer’s interest of having a winning basketball team. “When two panel opinions appear in conflict, it is the earlier which controls,” Harvey v. Blake, 913 F.2d 226, 228 n. 2 (5th Cir.1990), as “one panel of this court cannot overrule the decision of another panel,” United States v. Darrington, 351 F.3d 632, 634 (5th Cir.2003). Skilling misconstrues our holding in Brown, however, because he fails to recognize the manner in which the court in Brown explicitly distinguished Gray. As we noted above, Gray and Brown present different facts; in Gray, the basketball coaches acted on their own volition, without any direction from their supervisors, while in Brown, a lower-level Enron employee acted at the direction of Fastow, who as a decisionmaker had the authority to tell his employee that Enron sanctioned the particular fraud in question. See Gray, 96 F.3d at 775; see also Brown, 459 F.3d at 522 & n. 13. The difference is that in Brawn, the employee undertook the specific fraud in question at the direction of the employer, while this did not occur in Gray. In essence, because the Enron decisionmaker in Brown sanctioned the specific fraudulent conduct of its employee, the employee (and the other conspirators) did not deprive Enron of its honest services. Thus, for example, had the basketball coaches in Gray showed that the President of Baylor University or other decisionmakers specifically directed their fraudulent conduct, then they would not have been liable for honest-services fraud. Applying this rule, Skilling’s convictions must stand. First, Enron created a goal of meeting certain earnings projections. Second, Enron aligned its interests with Skilling’s personal interests, e.g., through his compensation structure, leading Skill-ing to undertake fraudulent means to achieve the goal. Third — and fatally to Skilling’s argument — no one at Enron sanctioned Skilling’s improper conduct. That is, Skilling does not allege that the Board of Directors or any other decision-maker specifically directed the improper means that he undertook to achieve his goals. Of course, a senior executive cannot wear his “executive” hat to sanction a fraudulent scheme and then wear his “employee” hat to perpetuate that fraud. Therefore, it is not a matter of Skilling setting the corporation’s policy himself. Instead, the question is whether anyone who supervised Skilling specifically directed his actions — such as how Fastow sanctioned the scheme in Brown. Skilling never alleged that he engaged in his conduct at the explicit direction of anyone, and therefore he cannot avail himself of the exception from Brown. That the Board of Directors approved several of the fraudulent transactions is of no moment. Tacitly approving a sale is not the same as having senior executives direct their lower-level employees to engage in fraudulent conduct. Skilling reads a requirement of secrecy into the holding in Brown, asserting that it is not honest-services fraud if the employer knows of the fraud in question. This argument is unavailing, for two reasons. First, a requirement of secrecy (or lack thereof) does not appear in Brown. Second, it makes no difference that the Board of Directors knew of Skilling’s conduct if Enron (through the Board or its senior executives) did not actually direct Skilling to undertake the fraudulent means to achieve his goals. The elements of honest-services wire fraud applicable here are (1) a material breach of a fiduciary duty imposed under state law, including duties defined by the employer-employee relationship, (2) that results in a detriment to the employer. Brown sheds light on the employer-employee relationship by creating an exception for when the employer specifically directs the fraudulent conduct. Further, it is a sufficient detriment for an employee, contrary to his duty of honesty, to withhold material information, i.e., information that he had reason to believe would lead a reasonable employer to change its conduct. Accordingly, the jury was entitled to convict Skilling of conspiracy to commit honest-services wire fraud on these elements. Thus, although we do not know on which alleged object of the conspiracy the jury based its verdict, there is no risk that Skilling was convicted of conspiracy based on a legally insufficient theory, and the jury was entitled to convict on any or all of the three objects. IV. Jury Instructions Skilling raises four alleged errors arising from either the instructions that the district court gave to the jury or Skilling’s proposed instructions that the court rejected. Specifically, Skilling asserts that the district court improperly (1) instructed the jury on “deliberate ignorance,” (2) denied the jury adequate guidance on the legal meaning of “materiality” in the context of the charges against him, (3) denied his proposed “side deal” instruction, and (4) denied his proposed “good faith” instruction. In assessing a jury instruction, we consider whether it is a “correct statement of the law,” United States v. Pompa, 434 F.3d 800, 805 (5th Cir.2005) (internal quotation marks omitted), whether it “clearly instructs jurors,” id., and whether it is “factually supportable,” United States v. Mendoza-Medina, 346 F.3d 121, 132 (5th Cir.2003) (“[T]he court may not instruct the jury on a charge that is not supported by evidence.”) (internal quotation marks omitted). We review for abuse of discretion and, in deciding whether the evidence reasonably supports the charge, we construe the evidence in the light most favorable to the government. United States v. Fuchs, 467 F.3d 889, 901 (5th Cir.2006) (citing cases), cert. denied, 549 U.S. 1272, 127 S.Ct. 1502, 167 L.Ed.2d 241 (2007). Additionally, we afford the district court “substantial latitude in formulating the charge.” United States v. Pettigrew, 77 F.3d 1500, 1510 (5th Cir.1996). A district court’s error in giving the jury instructions is subject to harmless error review. United States v. Edelkind, 525 F.3d 388, 397 (5th Cir.), cert. denied, — U.S.-, 129 S.Ct. 246, 172 L.Ed.2d 186 (2008); United States v. Ibarra-Zelaya, 465 F.3d 596, 607 (5th Cir.2006), cert. denied, 549 U.S. 1138, 127 S.Ct. 992, 166 L.Ed.2d 749, and cert. denied, — U.S.-, 127 S.Ct. 3056, 168 L.Ed.2d 767, and cert. denied, — U.S. -, 127 S.Ct. 3056, 168 L.Ed.2d 767 (2007). Moreover, a court’s rejection of a proposed jury instruction constitutes “reversible error only where the requested instruction is substantially correct; the actual charge given the jury did not substantially cover the content of the proposed instruction; and where the omission of the proposed instruction would ‘seriously impair the defendant’s ability to present a defense.’ ” United States v. Loe, 248 F.3d 449, 459 (5th Cir.2001) (quoting Pettigrew, 77 F.3d at 1510). A. Deliberate Ignorance Skilling claims that the district court erred in giving a “deliberate ignorance” instruction. The instruction, borrowed verbatim from Fifth Circuit Pattern Instruction 1.37, informed jurors that they could infer a defendant’s knowledge of facts from evidence that he was deliberately ignorant of such facts. Skilling objected to the instruction and argues that the evidence did not support it. “Deliberate ignorance” amounts to a half-step between the highest mens rea standard of “knowledge” and the lower standards of “recklessness” and “negligence.” It “ ‘denotes a conscious effort to avoid positive knowledge of a fact which is an element of an offense charged, the defendant choosing to remain ignorant so he can plead lack of positive knowledge in the event he should be caught.’ ” . United States v. Lara-Velasquez, 919 F.2d 946, 951 (5th Cir.1990) (quoting United States v. Restrepo-Granda, 575 F.2d 524, 528 (5th Cir.1978)). Thus, we have noted that “[t]he key aspect of deliberate ignorance is the conscious action of the defendant — -the defendant [must have] consciously attempted to escape confirmation of conditions or events he strongly suspected to exist.” Id. In simplest terms, deliberate ignorance is reflected in a criminal defendant’s actions that suggest, in effect, “Don’t tell me, I don’t want to know.” Id. (citing United States v. de Luna, 815 F.2d 301, 302 (5th Cir.1987)). The instruction’s purpose “is to inform the jury that it may consider evidence of the defendant’s charade of ignorance as circumstantial proof of guilty knowledge.” Id. Although we have frequently upheld the use of deliberate ignorance instructions, we have just as frequently warned of a risk inherent in them: Because the instruction permits a jury to convict a defendant without a finding that the defendant was actually aware of the existence of illegal conduct, the deliberate ignorance instruction poses the risk that a jury might convict the defendant on a lesser negligence standard— the defendant should have been aware of the illegal conduct. Id. The source of this risk is the potential for confusion about the degree of “deliberateness” required to convert ordinary, innocent ignorance into guilty knowledge. The concern is that once a jury learns that it can convict a defendant despite evidence of a lack of knowledge, it will be misled into thinking that it can convict based on negligent or reckless ignorance rather than intentional ignorance. In other words, the jury may erroneously apply a lesser mens rea requirement: a “should have known” standard of knowledge. In light of that concern, a district court should give the deliberate ignorance instruction only in the “rare” instance where there is significant evidence of deliberate ignorance. Id. “[T]he district court should not instruct the jury on deliberate ignorance when the evidence raises only the inferences that the defendant had actual knowledge or no knowledge at all of the facts in question.” Id. Rather, for the instruction to be warranted, “[t]he evidence at trial must raise two inferences: (1) the defendant was subjectively aware of a high probability of the existence of the illegal conduct; and (2) the defendant purposely contrived to avoid learning of the illegal conduct.” Lara-Velasquez, 919 F.2d at 951. Thus, where there is no such evidence, a district court should not give the instruction because there is no basis for finding deliberate ignorance. In such a case, it is usually harmful, because it is likely to lead the jury to find that the defendant had the requisite knowledge when he in fact did not. Cf. Lara-Velasquez, 919 F.2d at 951 (describing the risk inherent in the deliberate ignorance instruction). For example, in United States v. Ojebode, 957 F.2d 1218, 1229 (5th Cir.1992), we reversed a conviction for illegally importing heroin into the United States based on an improper deliberate ignorance instruction. The defendant carried drugs on a flight from Germany to Mexico that happened to have a layover in Houston, and he was arrested with the drugs during the layover. Id. at 1221-22. He argued that he did not knowingly bring the heroin into the United States because he did not know the flight stopped in Houston. Id. at 1223-24. The district court gave a deliberate ignorance instruction, and the defendant was convicted. Id. at 1229. We reversed, because there was no evidence that the defendant had “tried to avoid learning of the flight’s scheduled landing in Houston,” “refused to view the posted flight schedule!,] or absented himself from places where he would be likely to learn of his Lufthansa Flight’s likely stops.” Id. In other words, there was no purposeful contrivance to avoid learning of a relevant fact, so there was insufficient evidence of deliberate ignorance. Id. The instruction therefore posed too great a risk that the jury would convict for his negligent ignorance — i.e., that he should have known where the flight was headed. See id. Skilling argues that there was no evidence that he was deliberately ignorant of any illegal acts. Rather, he claims that he “never denied knowledge of the relevant underlying acts,” and that at trial he asserted only that those acts were not illegal. He claims that he “agreed” at trial with the government’s characterization of him as knowing everything that went on at Enron. He maintains that his “defense was not that he was unaware of fraud, but that there was no fraud.” Because his knowledge of allegedly illegal actions was never in dispute, he claims that “[njeither side argued [he] subjectively suspected criminal behavior but purposely contrived to avoid learning about it.” Even if Skilling is correct that there is little evidence to support a deliberate ignorance instruction, however, any error in the district court’s decision to give the instruction was harmless. This is because the peril to be avoided in cases reversing convictions based on the deliberate ignorance instruction is not present here. By his own admission, Skilling claims that he knew of the allegedly illegal acts, so there is no risk that a jury would rely on the deliberate ignorance instruction to find that he should have known of the acts. Consequently, even if the district court erred in giving the deliberate ignorance instruction in the sense that the instruction was “not supported by evidence,” Mendoza-Medina, 346 F.3d at 132 (internal quotation marks omitted), any such error was necessarily harmless, as we have repeatedly deemed deliberate ignorance instructions harmless where there is “substantial evidence of actual knowledge.” United States v. Threadgill, 172 F.3d 357, 369 (5th Cir.1999) (internal quotation marks omitted). Perhaps recognizing that difficulty, Skilling frames the harmfully misleading effect of the deliberate ignorance instruction as a concern that “the jury may well have decided [Skilling] should have known the transactions were fraudulent or merely should have known they were bad business decisions.” In so arguing, Skilling conflates the requisite mental state for the allegedly fraudulent acts (“knowingly”) with the requisite mental state for the fraud as a whole (“willfully and with the intent to defraud”). The traditional concern over the deliberate ignorance instruction is that the jury will misconceive the former mental state, which governs the act element — that is, the jury will erroneously assume that a defendant “knowingly” acquiesced in an act because he “should have known” about it. Skilling’s concern, however, is different: he suggests that the deliberate ignorance instruction caused the jury to misconceive the latter mental state, which governs the specific-intent element. He asserts that the jury could have erroneously assumed that he had the specific intent to defraud because he “should have known” of the acts about which he had, in fact, admitted knowing. Although plausible in the abstract, the likelihood of confusion in this atypical instance is not substantial, and it becomes even less so when we consider the deliberate ignorance instruction in context with the other jury instructions. Viewed as a whole, the jury instructions made it clear that the deliberate ignorance instruction related only to the act element and not to the specific-intent element. The district court provided the deliberate ignorance instruction as part of the definition of “knowingly,” a word that applied only to the act element of the fraud offense. In contrast, the definition of “willfully” governed solely the specific-intent element. The latter definition was emphatic that neither mere negligence nor bad business judgment suffices to establish the specific intent to defraud, and that instruction contained no deliberate ignorance definition. The “willfully” definition required the jury to find that the defendant acted “with bad purpose either to disobey or disregard the law.” Accordingly, there is no significant likelihood of confusion, and any error in issuing the deliberate ignorance instruction was harmless. B. Materiality Skilling next asserts that the district court erred in failing to provide the jury with adequate guidance on the legal meaning of “materiality” with regard to the charges against him. After explaining to the jury that false statements or omissions can support a fraud conviction only if they are material, the court instructed the jury on the definition of materiality. Skilling argues that the instruction was insufficient to convey to a “lay juror who has never invested in stock in his or her life” what a reasonable investor would consider important. In particular, Skilling contends that the court erred by (1) refusing to instruct the jury about statements that constitute “puffery” and are immaterial as a matter of law, and (2) denying his proposed supplemental instruction as unnecessary in light of the instruction the court ultimately gave. He argues that both instructions were necessary because “reasonable investors disregard all sorts of information that a lay juror might mistakenly consider material, especially when viewed in hindsight,” and that the district court’s refusal to give each constitutes reversible error. We address each contention in turn. 1. Puffery Skilling first challenges the district court’s materiality instruction on the ground that the court should have specifically instructed the jury on puffery. He sought to have the court tell the jury that even if a statement is false or misleading, it is mere “puffery” and therefore immaterial if it is “so lacking in specificity, or so clearly constituting the opinions of the speaker, that no reasonable investor could find the statement important to the total mix of information he or she would consider when making an investment decision.” Such statements, he notes, include “generalized, positive statements about [a] company’s competitive strengths ... and future prospects.” Rosenzweig v. Azurix Corp., 332 F.3d 854, 869 (5th Cir.2003). Although Skilling is correct that “an expression of opinion not made as a representation of fact,” can constitute puffery, Mfg. Research Corp. v. Greenlee Tool Co., 693 F.2d 1037, 1040 (11th Cir.1982) (citing Gulf Oil Corp. v. FTC, 150 F.2d 106, 109 (5th Cir.1945)), not all such statements of opinion are properly classified as puffery. Similarly, although Skilling correctly points out that “generalized, positive statements about [a] company’s competitive strengths ... and future prospects” can in some cases constitute immaterial puffery, such statements of opinion by corporate insiders are not per se immaterial. In Virginia Bankshares, Inc. v. Sandberg, 501 U.S. 1083, 1090, 1098, 111 S.Ct. 2749, 115 L.Ed.2d 929 (1991), the Supreme Court considered statements that a merger proposal would give shareholders “a high value for their shares,” and held that such statements could be deemed material. The Court explained, [i]t is no answer to argue, as petitioners do, that the quoted statement on which liability was predicated did not express a reason in dollars and cents, but focused instead on the “indefinite and unverifiable” term, “high” value, much like the similar claim that the merger’s terms were “fair” to shareholders. The objection ignores the fact that such concluso-ry terms in a commercial context are reasonably understood to rest on a factual basis that justifies them as accurate, the absence of which renders them misleading. Provable facts either furnish good reasons to make a conclusory commercial judgment, or they count against it, and expressions of such judgments can be uttered with knowledge of truth or falsity just like more definite statements, and defended or attacked through the orthodox evidentiary process that either substantiates their underlying justifications or tends to disprove their existence .... In this case, whether $42 was “high,” and the proposal “fair” to the minority shareholders, depended on whether provable facts about the Bank’s assets, and about actual and potential levels of operation, substantiated a value that was above, below, or more or less at the $42 figure, when assessed in accordance with recognized methods of valuation. Id. at 1093-94, 111 S.Ct. 2749 (footnote omitted). Virginia Bankshares thus instructs that conclusory statements of reasons, belief, or opinion — e.g., “high value” and “fair” — may be so contrary to the verifiable historical facts that they falsely “misstate the speaker’s [true] reasons” and “mislead about the stated subject matter.” Id. at 1095, 111 S.Ct. 2749. Indeed, the Supreme Court expressly explained that “there is no room to deny that a statement of belief by corporate directors about a recommended course of action, or an explanation of their reasons for recommending it,” can be material. Id. at 1090-91, 111 S.Ct. 2749. Skilling’s statements about the financial health of Enron were similar to those deemed potentially material in Virginia Bankshares. For example, at the 2001 analyst conference (count 23), Skilling claimed that all of Enron’s businesses, including EES and EBS, were “uniquely strong franchises with sustainable high earnings power.” He also characterized Wholesale as a “stable, high-growth business” and “not a trading business.” Similarly, on the March 23, 2001 analyst call (count 24), he claimed EBS was having “a great quarter on the intermediation side of the bandwidth business.” Summarizing EBS, he said that there was “essentially strong growth on the intermediation side, strong growth on the content services side, in terms of people, budgets, the whole thing.” The jury was entitled to find those and similar statements material. The government presented evidence of contrary, verifiable historical facts regarding the actual condition of EES, EBS, and Wholesale at the time Skilling made these statements: EES was facing a potentially enormous loss; EBS had an unsupportable cost structure, was losing money, was reducing the number of its employees, and had few customers or profitable deals; and Wholesale was heavily dependent on unstable, speculative trading. Moreover, in addressing the question of “whether statements of reasons, opinions, or beliefs are statements Vith respect to ... fact[s]’ so as to fall within the strictures of [the securities laws],” id. at 1091, 111 S.Ct. 2749, the Supreme Court has concluded that such st