United States v. Jennifer Riccardi

989 F.3d 476

• Court: Court of Appeals for the Sixth Circuit
• Decided: March 3, 2021
• Docket: 19-4232
• Status: Published

Opinion

RECOMMENDED FOR PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 21a0054p.06

UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT

┐ UNITED STATES OF AMERICA, │ Plaintiff-Appellee, │ v. > No. 19-4232 │ │ JENNIFER RICCARDI, │ Defendant-Appellant. │ ┘

Appeal from the United States District Court for the Northern District of Ohio at Cleveland. No. 1:18-cr-00740-1—John R. Adams, District Judge.

Argued: December 1, 2020

Decided and Filed: March 3, 2021

Before: DAUGHTREY, NALBANDIAN, and MURPHY, Circuit Judges. _________________

COUNSEL

ARGUED: Jeffrey C. Rager, RAGER LAW FIRM, PLLC, Lexington, Kentucky, for Appellant. Laura McMullen Ford, UNITED STATES ATTORNEY’S OFFICE, Cleveland, Ohio, for Appellee. ON BRIEF: Jeffrey C. Rager, RAGER LAW FIRM, PLLC, Lexington, Kentucky, for Appellant. Laura McMullen Ford, UNITED STATES ATTORNEY’S OFFICE, Cleveland, Ohio, for Appellee.

MURPHY, J., delivered the opinion of the court in which DAUGHTREY, J., joined. NALBANDIAN, J. (pp. 18–23), delivered a separate opinion concurring in part and in the judgment. No. 19-4232 United States v. Riccardi Page 2

_________________

OPINION _________________

MURPHY, Circuit Judge. Jennifer Riccardi, a postal employee, pleaded guilty to stealing 1,505 gift cards from the mail. Most of these gift cards had an average value of about $35 for a total value of about $47,000. The Sentencing Guidelines directed the district court to increase Riccardi’s guidelines range based on the amount of the “loss.” U.S.S.G. § 2B1.1(b)(1). Yet § 2B1.1 does not define the word “loss.” A search for its ordinary meaning might produce definitions such as “[t]he amount of something lost” or “[t]he harm or suffering caused by losing or being lost.” American Heritage Dictionary of the English Language 1063 (3d ed. 1992). Perhaps, then, the word is ambiguous on the margins. Does it, for example, cover only financial harms or emotional ones too? But one definition of “loss” that you will not find in any dictionary is the rule that the district court used for Riccardi’s stolen gift cards: a $500 minimum loss amount for each gift card no matter its actual value or the victim’s actual harm (which, for Riccardi, amounted to a total loss amount of $752,500).

Riccardi challenges the use of this $500 minimum loss amount, which comes from the Sentencing Commission’s commentary to § 2B1.1. The commentary instructs that the loss “shall be not less than $500” for each “unauthorized access device,” a phrase that Riccardi concedes covers stolen gift cards. U.S.S.G. § 2B1.1 cmt. n.3(F)(i). But guidelines commentary may only interpret, not add to, the guidelines themselves. United States v. Havis, 927 F.3d 382, 386 (6th Cir. 2019) (en banc) (per curiam). And even if there is some ambiguity in § 2B1.1’s use of the word “loss,” the commentary’s bright-line rule requiring a $500 loss amount for every gift card does not fall “within the zone of ambiguity” that exists. Kisor v. Wilkie, 139 S. Ct. 2400, 2416 (2019). So this bright-line rule cannot be considered a reasonable interpretation of—as opposed to an improper expansion beyond—§ 2B1.1’s text. We thus reverse Riccardi’s sentence and remand for resentencing without the use of the commentary’s automatic $500 minimum loss amount for every gift card. No. 19-4232 United States v. Riccardi Page 3

I

In September 2017, an Ohioan mailed a $25 Starbucks gift card from the City of Mentor in northeast Ohio. The gift card never reached its destination at an address in nearby Parma. The sender complained to the U.S. Postal Service, which opened an investigation. Investigators learned that supervisors at a Cleveland distribution center had been finding lots of opened mail in the processing area. The investigation led to an employee at the center: Jennifer Riccardi. When confronted, Riccardi admitted that she had been stealing mail that might contain cash or gift cards. A search of Riccardi’s home revealed that she had been doing so for quite some time. It uncovered over 100 pieces of mail that she had taken just that day, $42,102 in cash, and 1,505 gift cards. The gift cards were laid out on the floor of Riccardi’s home organized by the 230 or so merchants at which they could be redeemed.

Riccardi pleaded guilty to three counts: possessing stolen mail in violation of 18 U.S.C. § 1708; stealing mail as a postal employee in violation of 18 U.S.C. § 1709; and possessing 15 or more “unauthorized access devices” in violation of 18 U.S.C. § 1029(a)(3). In her plea agreement, Riccardi confessed that she had been stealing mail for a year. When sorting mail, she would set aside items that might contain cash or gift cards (such as colorful greeting cards) and sneak these items out of the distribution center during her break or at the end of her shift. Riccardi would use the stolen gift cards herself or sell them to others. Of the 1,505 gift cards found at her home, 1,322 had face values totaling about $47,000 for an average of about $35 each. The government did not identify the values of the remaining 183 gift cards.

Riccardi’s presentence report determined her guidelines range using U.S.S.G. § 2B1.1. When measuring the loss for the 1,505 gift cards, the report applied the $500 minimum loss amount from § 2B1.1’s commentary, id. § 2B1.1 cmt. n.3(F)(i), resulting in a total loss of $752,500. The presentence report thus bumped Riccardi up to a loss category of between $550,000 and $1,500,000 and increased her offense level by 14. This enhancement resulted in a guidelines range of 46 to 57 months’ imprisonment.

At sentencing, Riccardi objected to the use of this $500 minimum loss amount because most of the stolen gift cards were worth a fraction of that amount. The district court overruled No. 19-4232 United States v. Riccardi Page 4

her objection. After considering the sentencing factors, it imposed a sentence near the top of the guidelines range: 56 months’ imprisonment. The court also ordered Riccardi to pay $89,102 in restitution, an amount that included the $42,102 in cash found at her home and the value of the gift cards ($47,000). The court ordered this restitution even though Riccardi had already forfeited the cash and gift cards to the government. It reasoned that forfeiture and restitution were distinct obligations.

Riccardi raises two challenges on appeal. She argues that the district court should not have applied the $500 minimum loss amount to each of the gift cards and that it should have offset her restitution obligation with the amounts that she forfeited. We address each argument in turn.

II. “Loss” Amount

The guideline for theft offenses—U.S.S.G. § 2B1.1—starts with a base offense level of 6. Id. § 2B1.1(a)(2). It then lists a variety of offense characteristics that can affect this offense level, ranging from the number of victims involved, id. § 2B1.1(b)(2)(A)(i), to the possession of a firearm, id. § 2B1.1(b)(16)(B). As relevant here, courts must “increase the offense level” in incremental amounts based on the “loss” from the offense. Id. § 2B1.1(b)(1). If the loss is “[m]ore than $6,500,” § 2B1.1 instructs courts to add 2 to the offense level. Id. § 2B1.1(b)(1)(B). If the loss is “[m]ore than $15,000,” it instructs them to add 4. Id. § 2B1.1(b)(1)(C). The guideline continues in this fashion up to a loss amount of “[m]ore than $550,000,000,” for which it directs courts to increase the offense level by 30. Id. § 2B1.1(b)(1)(P).

The government bears the burden to prove the amount of the loss by a preponderance of the evidence. See, e.g., United States v. Jones, 641 F.3d 706, 712 (6th Cir. 2011); United States v. Rothwell, 387 F.3d 579, 582 (6th Cir. 2004). We treat the district court’s “determination of the amount of loss” as a factual finding and thus review it under a deferential clear-error standard. United States v. Warshak, 631 F.3d 266, 328 (6th Cir. 2010). But we review de novo the district court’s “methodology for calculating” the loss and its interpretation of the guidelines. Id.; United States v. Thomas, 933 F.3d 605, 608 (6th Cir. 2019). A misinterpretation of a No. 19-4232 United States v. Riccardi Page 5

guideline can result in a procedurally unreasonable sentence. See, e.g., United States v. Stubblefield, 682 F.3d 502, 510 (6th Cir. 2012); cf. Rosales-Mireles v. United States, 138 S. Ct. 1897, 1907–08 (2018).

Here, the government did not attempt to meet its burden to prove the loss from Riccardi’s theft by relying on factual evidence about the total amount that Riccardi stole or the total harm that her victims suffered. Instead, the government sought to meet its burden by relying on a legal rule that treats the “loss” for each of the 1,505 gift cards as $500 even though most of the gift cards had values averaging about $35. Riccardi raises two challenges to the use of this $500 mandatory minimum. She first argues that the district court misread § 2B1.1’s commentary by applying the $500 minimum to the gift cards that had face values below that amount. Even if the $500 minimum applied, Riccardi next argues that the commentary’s mandatory minimum conflicts with § 2B1.1’s text. We review both legal arguments de novo. See Warshak, 631 F.3d at 328.

A

Riccardi initially claims that the district court wrongly applied the $500 minimum loss amount under the plain language of § 2B1.1’s commentary. She misreads the commentary.

Although § 2B1.1 directs district courts to increase the offense level based on the amount of the “loss,” the guideline itself leaves this critical word undefined. U.S.S.G. § 2B1.1(b)(1). The Sentencing Commission instead added guidance over how to determine the “loss” in commentary accompanying § 2B1.1. Application Note 3 provides a detailed code for “the determination of loss under subsection (b)(1).” Id. § 2B1.1 cmt. n.3. This application note sets a general rule that “loss” means “the greater of actual loss or intended loss.” Id. § 2B1.1 cmt. n.3(A). It defines “actual loss” to mean “the reasonably foreseeable pecuniary harm that resulted from the offense” and “intended loss” to mean “the pecuniary harm that the defendant purposely sought to inflict[.]” Id. § 2B1.1 cmt. n.3(A)(i)–(ii). Yet Application Note 3 later orders courts to use “special rules” “to assist in determining loss” in specific types of cases, including those involving “unauthorized access devices.” Id. § 2B1.1 cmt. n.3(F). It states: “In a case involving any . . . unauthorized access device, loss includes any unauthorized charges made with No. 19-4232 United States v. Riccardi Page 6

the . . . unauthorized access device and shall be not less than $500 per access device.” Id. § 2B1.1 cmt. n.3(F)(i).

Application Note 3(F)(i)’s special rule applies here. This rule gives “unauthorized access device” the definition from 18 U.S.C. § 1029(e)(3). Id. § 2B1.1 cmt. nn.3(F)(i), 10(A). Section 1029(e) defines “unauthorized access device” as a “stolen” access device, and it defines “access device” as “any card” “or other means of account access that can be used” “to obtain money, goods, services, or any other thing of value[.]” 18 U.S.C. § 1029(e)(1), (3). Critically, Riccardi admitted in her plea agreement that the gift cards she stole were “unauthorized access devices” under § 1029(e). Cf. United States v. Truong, 587 F.3d 1049, 1051–52 (9th Cir. 2009) (per curiam). So we need not consider that issue. And although most of the stolen gift cards had face values below $500, Application Note 3(F)(i) instructs courts that the loss “shall be not less than $500 per” unauthorized access device. U.S.S.G. § 2B1.1 cmt. n.3(F)(i). The imperative “shall” signals a duty to impose this $500 minimum loss amount, no matter the actual loss amount. See Babb v. Wilkie, 140 S. Ct. 1168, 1173 (2020); SAS Inst., Inc. v. Iancu, 138 S. Ct. 1348, 1354 (2018). Because Riccardi admitted to stealing 1,505 gift cards, her loss amount could not be less than $752,500 (1,505 times $500) under Application Note 3.

Riccardi counters that another part of Application Note 3 tells courts to calculate the loss based on all available information, including the “fair market value of the property” stolen. U.S.S.G. § 2B1.1 cmt. n.3(C)(i). Because Riccardi and the government agree on the value of most of the gift cards, she argues that this portion of Application Note 3 should control over the $500 minimum. This contention has things backwards. The specific governs the general in the interpretation of a legal text. See RadLAX Gateway Hotel, LLC v. Amalgamated Bank, 566 U.S. 639, 645 (2012). Because Application Note 3(F)(i) adopts a “special rule” (including the $500 minimum) for access devices, it trumps the general framework for other types of property. Our precedent has repeatedly made this point in response to arguments seeking to sidestep this $500 minimum. See United States v. Moon, 808 F.3d 1085, 1091 (6th Cir. 2015); United States v. Lyles, 506 F. App’x 440, 445 (6th Cir. 2012); United States v. Gilmore, 431 F. App’x 428, 430–31 (6th Cir. 2011). No. 19-4232 United States v. Riccardi Page 7

Riccardi next argues that the purpose behind this $500 minimum does not fit gift cards. Commentary to the original § 2B1.1 imposed a minimum loss amount of $100 per card only for stolen credit cards. U.S.S.G. § 2B1.1 cmt. n.4 (1987). In 1998, Congress told the Commission to review and amend the guidelines “to provide an appropriate penalty for offenses involving the cloning of wireless telephones[.]” Wireless Telephone Protection Act, Pub. L. No. 105-172, § 2(e)(1), 112 Stat. 53, 55 (1998). A report from this review suggested that the average loss per stolen credit card was over $1,000 and that the average loss per cloned phone was around $800. Econ. Crimes Pol’y Team, U.S. Sent’g Comm’n, Cellular Phone Cloning Final Report, at 27 & n.49 (Jan. 25, 2000). This information led the Commission to increase to $500 the commentary’s minimum loss amount for all access devices, not just credit cards. 65 Fed. Reg. 26,880, 26,895 (May 9, 2000); U.S.S.G. §§ 2B1.1 cmt. n.2, 2F1.1 cmt. n.17 (2000). Riccardi argues that the Commission had cloned phones in mind when making these changes and that the average loss for a stolen gift card is likely well below $500.

Whether or not Riccardi correctly describes this commentary’s background, the alleged purpose does not change things. Riccardi has conceded that gift cards are “access devices,” and the commentary requires a minimum loss of “$500 per access device.” U.S.S.G. § 2B1.1 cmt. n.3(F)(i). So “even the most formidable argument concerning the [commentary’s] purposes [cannot] overcome the clarity we find in [its] text.” Kloeckner v. Solis, 568 U.S. 41, 55 n.4 (2012); see RadLAX, 566 U.S. at 649.

Riccardi also argues that it would be “unreasonable” to apply this $500 minimum in cases in which the court knows that the actual value of a stolen gift card falls below $500. Yet this “reasonableness” argument does not implicate the proper reading of Application Note 3(F)(i) and so does not belong in Riccardi’s procedural-reasonableness challenge. See Stubblefield, 682 F.3d at 510. (No interpretive rule allows us to depart from the plain text when we find it “unreasonable.”) Rather, Riccardi’s argument is a disguised substantive-reasonableness challenge to her sentence under the sentencing factors in 18 U.S.C. § 3553(a). See, e.g., Moon, 808 F.3d at 1092–93. That is, Riccardi’s argument “goes to the [district] court’s authority to vary up or down from the guidelines range, not to the meaning of the guidelines.” United States v. Kozerski, 969 F.3d 310, 314 (6th Cir. 2020); cf. United States v. Murphy, 815 F. App’x 918, No. 19-4232 United States v. Riccardi Page 8

922 (6th Cir. 2020). But Riccardi waived any substantive-reasonableness challenge in her plea agreement.

B

That does not end matters. For whatever reason, the Commission opted to place its $500 minimum in § 2B1.1’s commentary, not in § 2B1.1. So Riccardi alternatively asserts that the $500 minimum conflicts with § 2B1.1. We agree. Commentary may only interpret the guideline. And a $500 mandatory minimum cannot be described as an interpretation of the word “loss.” Rather, it is a substantive legislative rule that belongs in the guideline itself to have force.

We start with the basic differences between the guidelines and the commentary. The Sentencing Reform Act of 1984, Pub. L. 98-473, Title II, ch. II, 98 Stat. 1987, tasked the Commission with creating “guidelines” that contain sentencing ranges for various categories of offenses. 28 U.S.C. § 994(a)(1), (b)(1); Stinson v. United States, 508 U.S. 36, 40–41 (1993). These administratively adopted guidelines significantly affected individual liberty because Congress required district courts to follow them when choosing the length of a defendant’s prison term. 18 U.S.C. § 3553(b)(1); United States v. Havis, 927 F.3d 382, 385–86 (6th Cir. 2019) (en banc) (per curiam). Congress thus included several procedural safeguards to act as a check on the sentencing rules that the Commission put in the guidelines. Congress required the Commission to submit the original guidelines for its review and to give it six months to review all amendments. See Sentencing Reform Act, § 235(a)(1), 98 Stat. at 2031–32; 28 U.S.C. § 994(p). It also required the amendments to go through notice-and-comment rulemaking. 28 U.S.C. § 994(x). And while the guidelines have been only advisory since United States v. Booker, 543 U.S. 220 (2005), they still significantly affect individual liberty because a court must use them as the initial benchmark for a proper sentence. Havis, 927 F.3d at 385.

Since the beginning, the Commission has also included “application notes” in “commentary” that accompanies the guidelines. See, e.g., U.S.S.G. § 2B1.1 cmt. nn.1–8 (1987). An original guideline explained that this “commentary” “may serve a number of purposes.” Id. § 1B1.7. Among other things, “it may interpret the guideline or explain how it is to be applied.” No. 19-4232 United States v. Riccardi Page 9

Id. Yet the Sentencing Reform Act did not mention the “commentary,” and later amendments have made only passing reference to it. Sentencing Reform Act, 98 Stat. at 1987–2040; Stinson, 508 U.S. at 41 (citing 18 U.S.C. § 3553(b)). To amend the commentary, then, the Commission need not follow the same procedures that govern changes to the substantive rules in the guidelines themselves (congressional review and notice-and-comment rulemaking). Havis, 927 F.3d at 386. That fact led some circuit courts to hold originally that they were not bound by the commentary’s interpretation of the guidelines. See Stinson, 508 U.S. at 39–40 & 40 n.2.

The Supreme Court rejected this view in Stinson. Analogizing to administrative law, the Court viewed the guidelines as the “equivalent of legislative rules adopted by federal agencies.” Id. at 45. And it viewed the commentary as “akin to an agency’s interpretation of its own legislative rules.” Id. It thus found that the commentary deserved the deference given to an agency’s interpretation of its regulations—what was then known as Seminole Rock deference but now goes by Auer deference. Id.; see Auer v. Robbins, 519 U.S. 452, 461 (1997); Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 414 (1945). Applying Auer’s test, Stinson held that the commentary’s interpretation of a guideline “must be given ‘controlling weight unless it is plainly erroneous or inconsistent with the” guideline. 508 U.S. at 38 (quoting Seminole Rock, 325 U.S. at 414). Stinson added that the Commission could effectively amend a guideline by amending the commentary so long as “the guideline which the commentary interprets will bear the [amended] construction.” Id. at 46.

On its face, Stinson’s plain-error test seemed to require courts to give great deference to the commentary. By way of analogy, the plain-error test that applies to unpreserved arguments on appeal requires a legal error to “be clear or obvious, rather than subject to reasonable dispute.” Puckett v. United States, 556 U.S. 129, 135 (2009). Unsurprisingly, then, we have previously been quick to give “controlling weight” to the commentary without asking whether a guideline could bear the construction that the commentary gave it. See, e.g., United States v. Ednie, 707 F. App’x 366, 371–72 (6th Cir. 2017); United States v. Jarman, 144 F.3d 912, 914 (6th Cir. 1998). Perhaps for this reason, defendants have not previously “challenge[d] the general validity” of the $500 minimum loss amount at issue here. Gilmore, 431 F. App’x at 430; see Moon, 808 F.3d at 1091. No. 19-4232 United States v. Riccardi Page 10

Recently, however, the Supreme Court clarified Auer’s narrow scope in the related context of an agency’s interpretation of its regulations. See Kisor v. Wilkie, 139 S. Ct. 2400, 2414–18 (2019). Kisor acknowledged that the Court’s “classic” plain-error phrasing of Auer’s test “suggest[ed] a caricature of the doctrine, in which deference is ‘reflexive.’” Id. at 2415 (citation omitted). Yet Kisor cautioned that a court should not reflexively defer to an agency’s interpretation. Before doing so, a court must find that the regulation is “genuinely ambiguous, even after [the] court has resorted to all the standard tools of interpretation” to eliminate that ambiguity. Id. at 2414. The agency’s interpretation also “must come within the zone of ambiguity the court has identified after employing all its interpretive tools.” Id. at 2416.

Should Kisor affect our approach to the commentary? We think so for both a simple reason and a more complicated one. As a simple matter, Stinson analogized to agency interpretations of regulations when adopting Seminole Rock’s plain-error test for the commentary. 508 U.S. at 45. Stinson thus told courts to follow basic administrative-law concepts despite Congress’s decision to locate the relevant agency (the Commission) in the judicial branch rather than the executive branch. See id.; cf. Mistretta v. United States, 488 U.S. 361, 384–85 (1989). So Kisor’s clarification of the plain-error test applies just as much to Stinson (and the Commission’s guidelines) as it does to Auer (and an agency’s regulations). Indeed, Kisor itself cited Stinson as a decision applying Seminole Rock deference before Auer. Kisor, 139 S. Ct. at 2411 n.3.

The more complex reason follows from Kisor’s response to a notice-and-comment concern raised by the challenger in that case. When asking the Court to overrule Auer, the challenger argued that Auer allowed an agency to freely change a legislative rule (a change that otherwise requires notice-and-comment rulemaking) simply by changing its interpretation of the rule without using that type of rulemaking. Id. at 2420. Kisor rejected the challenger’s premise—that an agency could willy-nilly change a legislative rule simply by changing its interpretation. Why? Precisely because of the limits that Kisor imposed: Before deferring to the changed reading of the rule, a court must “first decide whether the rule is clear; if it is not, whether the agency’s reading falls within its zone of ambiguity; and even if the reading does so, whether it should receive deference.” Id. In other words, Kisor’s limitations on Auer deference No. 19-4232 United States v. Riccardi Page 11

restrict an agency’s power to adopt a new legislative rule under the guise of reinterpreting an old one.

The same concern applies here, so Kisor’s response should too. See Havis, 927 F.3d at 386. Only the guidelines (not the commentary) must go through notice-and-comment rulemaking. 28 U.S.C. § 994(x). So if the Commission could freely amend the guidelines by amending the commentary, it could avoid these notice-and-comment obligations. The healthy judicial review that Kisor contemplates thus will restrict the Commission’s ability to do so.

We are not alone in this conclusion. The en banc Third Circuit recently adopted the same view. See United States v. Nasir, 982 F.3d 144, 158 (3d Cir. 2020) (en banc). It recognized that its pre-Kisor cases had upheld commentary expanding the guidelines. Id. Yet these cases could not stand after Kisor, the court found, because it “cut back on what had been understood to be uncritical and broad deference to agency interpretations of regulations[.]” Id. As a concurrence put it, Kisor must awake us “from our slumber of reflexive deference” to the commentary. Id. at 177 (Bibas, J., concurring in part).

We thus do not immediately defer to Application Note 3(F)(i). Rather, we first ask whether § 2B1.1 is “genuinely ambiguous.” Kisor, 139 S. Ct. at 2415. Section 2B1.1’s language tells courts to “increase the offense level” in incremental amounts based on the amount of the “loss” (measured in dollars). U.S.S.G. § 2B1.1(b)(1). Where, as here, a legal text does not define a term, we generally “give the term its ordinary meaning.” United States v. Zabawa, 719 F.3d 555, 559 (6th Cir. 2013); United States v. Sands, 948 F.3d 709, 713–14 (6th Cir. 2020). And “dictionaries are a good place to start” to identify the range of meanings that a reasonable person would understand a word like “loss” to have. Zabawa, 719 F.3d at 559. One dictionary defines the word to mean, among other things, the “amount of something lost” or the “harm or suffering caused by losing or being lost.” American Heritage Dictionary of the English Language 1063 (3d ed. 1992). Another says it can mean “the damage, trouble, disadvantage, [or] deprivation . . . caused by losing something” or “the person, thing, or amount lost.” Webster’s New World College Dictionary 799 (3d ed. 1996). A third defines it as “the being No. 19-4232 United States v. Riccardi Page 12

deprived of, or the failure to keep (a possession, appurtenance, right, quality, faculty, or the like),” the “[d]imunition of one’s possessions or advantages,” or the “detriment or disadvantage involved in being deprived of something[.]” 9 Oxford English Dictionary 37 (2d ed. 1989).

These definitions show that “loss” can mean different things in different contexts. The word might include emotional harms, as in the statement that the children “bore up bravely under the [loss] of both parents[.]” Webster’s Third New International Dictionary 1338 (1986). Or it might include just economic harms, as in the statement that my friend was “forced to sell all the stock at a [loss].” Id. (Another part of § 2B1.1’s commentary does, in fact, read § 2B1.1 as limited to economic harms. See U.S.S.G. § 2B1.1 cmt. n.3(A)(iii); Kozerski, 969 F.3d at 313.) Even in the economic realm, the word might cover only the precise value of, say, a gift card that is stolen (the “amount of something lost”). American Heritage Dictionary, supra, at 1063. Or it might include the costs associated with obtaining a replacement gift card, including the time and expense from a second trip to the store (“the damage, trouble, disadvantage, [or] deprivation . . . caused by losing something”). Webster’s New World College Dictionary, supra, at 799.

In this case, however, we need not decide whether one clear meaning of the word “loss” emerges from the potential options after applying “the ‘traditional tools’ of construction” to § 2B1.1. Kisor, 139 S. Ct. at 2415 (citation omitted). No matter the word’s meaning, the commentary’s $500 minimum loss amount for gift cards does not fall “within the zone of [any] ambiguity” in this guideline. Id. at 2416; cf. MCI Telecomms. Corp. v. Am. Tel. & Tel. Co., 512 U.S. 218, 225–29 (1994). No reasonable person would define the “loss” from a stolen gift card as an automatic $500. Rather, the “amount” of the loss or “damage” to the victim from a gift-card theft in any case will turn on such fact-dependent things as the value of the gift card or the costs of replacing it. American Heritage Dictionary, supra, at 1063; Webster’s New World College Dictionary, supra, at 799. This case proves the point. It is undisputed that 1,322 of Riccardi’s stolen gift cards had total face values of $47,000 for an average value of about $35. And the government identifies no evidence suggesting that the total “damage” from this theft approached the $752,500 required by the commentary’s mandatory $500 loss amount. No. 19-4232 United States v. Riccardi Page 13

Our conclusion is reinforced by caselaw distinguishing “legislative rules” (which must proceed through notice-and-comment rulemaking) from “interpretive rules” (which need not proceed through that rulemaking) under the Administrative Procedure Act. See generally Perez v. Mortg. Bankers Ass’n, 575 U.S. 92, 95–97 (2015); 5 U.S.C. § 553(b). Precedent in that context recognizes that a specific numeric amount like the $500 in this case generally will not qualify as a mere “interpretation” of general nonnumeric language. See Catholic Health Initiatives v. Sebelius, 617 F.3d 490, 495 (D.C. Cir. 2010). An agency, for instance, did not simply “interpret” a rule requiring parties to use “structurally sound” facilities to house dangerous animals when it concluded that this rule mandated an eight-foot fence. See Hoctor v. U.S. Dep’t of Agric., 82 F.3d 165, 169–71 (7th Cir. 1996). Rather, “when an agency wants to state a principle ‘in numerical terms,’ terms that cannot be derived from a particular record, the agency is legislating and should act through rulemaking.” Catholic Health Initiatives, 617 F.3d at 495 (quoting Henry J. Friendly, Watchman, What of the Night?, in Benchmarks 144–45 (1967)).

The same logic applies here. The commentary’s bright-line $500 loss amount cannot “be derived from [§ 2B1.1] by a process reasonably described as interpretation.” Hoctor, 82 F.3d at 170. The Commission’s decision to adopt this minimum loss amount was instead a substantive policy choice, one presumably based on empirical factors like the difficulty of determining actual losses in cases involving “access devices” or the “average” loss in those types of cases. Yet if the Commission seeks to keep individuals behind bars for longer periods of time based on this type of “fictional” loss amount, this substantive policy decision belongs in the guidelines, not in the commentary. Lyles, 506 F. App’x at 445; see Havis, 927 F.3d at 385–86.

The government’s responses do not change things. It does not argue meaningfully that a $500 minimum amount for gift cards qualifies as an “interpretation” of the word loss. It instead suggests that calculating the loss amount will often prove challenging and that district courts must make estimates. It thus relies on the standard of review for a district court’s finding about the amount of the loss: A defendant “must carry the heavy burden of persuading this Court that the [district court’s] evaluation of the loss was not only inaccurate, but was outside the realm of permissible computations.” United States v. Jackson, 25 F.3d 327, 330 (6th Cir. 1994); see, e.g., No. 19-4232 United States v. Riccardi Page 14

United States v. Gray, 521 F.3d 514, 543 (6th Cir. 2008). The government places undue reliance on this standard of review. Yes, it may sometimes be difficult to estimate the actual amount of loss. And yes, when a district court makes a record-based factual finding about the amount of the loss, we review its finding under a deferential clear-error standard. See Jackson, 25 F.3d at 330. But the district court’s $500-loss-per-gift-card finding was not tied to its view of the evidence or the amount of the actual loss; it was tied to the legal requirement in Application Note 3(F)(i). And we review the purely legal question whether this requirement comports with § 2B1.1 without the slightest deference to the district court. See, e.g., Havis, 927 F.3d at 384–87.

The government next turns to precedent interpreting the guidelines. It notes that our unpublished Murphy decision upheld a separate part of Application Note 3 indicating that “loss” includes not just actual financial harm but also financial harm intended by the defendant. See 815 F. App’x at 924. Yet Murphy did not address Kisor’s recent clarification about the limited nature of Auer. And, regardless, whether “loss” in § 2B1.1 can be read to include intended loss says nothing about whether it also can be read to mean an automatic $500 loss amount for all stolen gift cards no matter the facts. Nothing in Murphy implies that § 2B1.l can be read in that broader way.

The government also attempts to distinguish our en banc Havis decision. That case involved a guideline phrase (controlled substance offense) that had a definition in the guideline itself, not in the commentary. 927 F.3d at 384. The guideline did not include “attempt” crimes in the definition of “controlled substance offense,” but the Commission’s commentary enlarged the definition to cover those crimes. Id. We held that the commentary qualified as an improper attempt to expand, not interpret, the guidelines. Id. at 386–87. Here, by contrast, the government argues that § 2B1.1 does not itself define the word “loss” and so “leaves the job of defining its parameters to the corresponding Application notes.” Appellee’s Br. 22. Not so. Just because a word in a guideline does not come with its own guideline definition does not leave us at sea about its meaning or give the Commission license to define the term however it likes in the commentary. Courts presume that an undefined word comes with its ordinary meaning, not an unusual one. See, e.g., Asgrow Seed Co. v. Winterboer, 513 U.S. 179, 187 (1995); Zabawa, 719 F.3d at 559. That bedrock interpretive insight applies to words in the guidelines too. Sands, 948 No. 19-4232 United States v. Riccardi Page 15

F.3d at 713–14; see also, e.g., United States v. Ward, 972 F.3d 364, 369 (4th Cir. 2020); United States v. Crittenden, 372 F.3d 706, 708 (5th Cir. 2004). And the commentary’s unusual “definition” of loss conflicts with the ordinary definition that we must follow.

The government thus falls back on a policy argument. When financial fraudsters get caught in access-device thefts, the government often encounters difficulty uncovering the full extent of their crimes. It thus believes that a $500 minimum loss amount is a reasonable compromise to account for these difficulties. We need not take issue with the government’s policy points to reject their application here. Perhaps some crimes do have difficult-to-quantify losses. Cf. United States v. Carver, 916 F.3d 398, 404 (4th Cir. 2019). And it is not unusual for a statutory term to include an unusual statutory definition that departs from its ordinary meaning. See, e.g., Tanzin v. Tanvir, 141 S. Ct. 486, 490 (2020); Digit. Realty Tr., Inc. v. Somers, 138 S. Ct. 767, 776 (2018). Thus, nothing we say here would prevent the Commission from adopting its $500 minimum amount for access devices by placing this legislative rule in the guideline itself. We hold only that the Commission may not make this kind of substantive policy choice in the commentary and claim that its choice represents nothing more than an “interpretation” of the guideline.

We end by flagging one issue that the government did not raise. It appears that the Commission sent the amendment adopting this $500 minimum amount to Congress for its review and added it to the commentary using notice-and-comment rulemaking. See 65 Fed. Reg. 26,880, 26,895 (May 9, 2000); 65 Fed. Reg. 2663, 2668 (Jan. 18, 2000). Should we overlook that this $500 minimum sits in the commentary given that the Commission may have met the procedural checks required for it to amend the guidelines themselves? We think not. By placing this loss amount in the commentary, the Commission has retained the power to adjust it tomorrow without satisfying the same procedural safeguards. See Stinson, 508 U.S. at 39–46. So the normal administrative principles should apply. Under those principles, this $500 minimum loss amount for gift cards does not “fall ‘within the bounds of reasonable interpretation’” of § 2B1.1’s text. Kisor, 139 S. Ct. at 2416 (citation omitted). The district court thus should not have used it. No. 19-4232 United States v. Riccardi Page 16

III. Restitution

Apart from her 56-month sentence, Riccardi also challenges the district court’s order requiring her to pay $89,102 in restitution. She argues that the court should have offset this amount with the cash and gift cards that she forfeited to the government. But we cannot consider Riccardi’s argument because she waived the right to raise it in her plea agreement.

A criminal defendant “may waive any right, even a constitutional right, by means of a plea agreement.” United States v. Winans, 748 F.3d 268, 270 (6th Cir. 2014) (citation omitted). That includes the right to appeal a sentence. Id. When deciding whether a defendant has waived this right, we interpret the plea agreement using traditional principles of contract interpretation. Id.

These rules foreclose Riccardi’s appeal of the restitution order. Her plea agreement unambiguously stated that she “expressly and voluntarily waive[d]” the right “to appeal the conviction or sentence[.]” And “restitution is a part of a defendant’s sentence.” United States v. Rafidi, 730 F. App’x 338, 340 (6th Cir. 2018). So we have repeatedly read a waiver of the right to appeal a “sentence” as including a “restitution” order. Id. at 342; United States v. Grundy, 844 F.3d 613, 616 (6th Cir. 2016); United States v. Black, 652 F. App’x 376, 379 (6th Cir. 2016); Winans, 748 F.3d at 271; United States v. Patel, 577 F. App’x 568, 572 (6th Cir. 2014) (per curiam); United States v. Curry, 547 F. App’x 768, 770–71 (6th Cir. 2013); United States v. Reese, 509 F. App’x 494, 499 (6th Cir. 2012); United States v. Gibney, 519 F.3d 301, 306 (6th Cir. 2008); United States v. Sharp, 442 F.3d 946, 952 (6th Cir. 2006).

To be sure, we did not read an appeal waiver in this fashion in United States v. Smith, 344 F.3d 479 (6th Cir. 2003). Riccardi’s waiver nevertheless falls within our usual rule, not within Smith’s exception. There, we held that a defendant’s plea agreement did not waive a right to appeal the loss calculations supporting the restitution order when the agreement indicated that the defendant waived the right “to appeal any sentence which is within the parameters of this agreement[.]” Id. at 483. Because the agreement’s “parameters” did not include a method for calculating the loss, we read this ambiguous phrase against the government and held that the defendant’s appeal of the loss calculation fell outside the waiver. Id. Here, by contrast, No. 19-4232 United States v. Riccardi Page 17

Riccardi’s waiver unambiguously covers the right to appeal any sentence without limitation. It does not cover only a sentence within the plea agreement’s “parameters.” So Smith does Riccardi no good.

True enough, Riccardi’s plea agreement also included several exceptions to this appeal waiver for certain claims (hence why she could appeal the district court’s use of the $500 minimum loss amount). But Riccardi gives us no reason to conclude that any exception covers her challenge to the restitution order. Some exceptions—e.g., those permitting Riccardi to bring ineffective-assistance-of-counsel or prosecutorial-misconduct claims or to appeal certain guidelines calculations—have no relevance to her restitution challenge.

That leaves Riccardi with the right to appeal (1) “any sentence to the extent it exceeds the maximum of the sentencing imprisonment range determined under the advisory Sentencing Guidelines” or (2) “any punishment in excess of the statutory maximum[.]” Yet the first exception—for a sentence exceeding the guidelines “imprisonment range”—does not apply because restitution is not “imprisonment,” and the guidelines do not provide restitution “ranges.” See Sharp, 442 F.3d at 952. Riccardi also offers no explanation why the second exception—for punishments exceeding the “statutory maximum”—applies. Indeed, our cases have noted that the “restitution statutes do not contain a maximum penalty[.]” Id.; see United States v. Bradley, 969 F.3d 585, 591–92 (6th Cir. 2020).

In short, if Riccardi “wished to reserve h[er] right to appeal the restitution order, [s]he should have negotiated for that right in h[er] plea agreement.” Rafidi, 730 F. App’x at 342 (quoting Sharp, 442 F.3d at 952). She has given us no basis to conclude on appeal that she did so.

* * *

We reverse Riccardi’s 56-month sentence, dismiss her separate challenge to the restitution order, and remand for resentencing consistent with this opinion. No. 19-4232 United States v. Riccardi Page 18

___________________________________________________

CONCURRING IN PART AND IN THE JUDGMENT ___________________________________________________

NALBANDIAN, J., concurring in part and in the judgment. I agree that application note 3(F) is not a permissible interpretation of “loss” in U.S.S.G. § 2B1.1(b)(1). But I part ways because I believe that Stinson rather than Kisor dictates this conclusion. Stinson did not instruct courts to apply the Seminole Rock1 standard no matter how it evolved. It established a free- standing deference standard that it analogized to Seminole Rock deference, but it did not call for courts to strictly apply Seminole Rock deference in sentencing guidelines cases. Unless the Supreme Court directs otherwise, I would continue to apply Stinson deference to guideline commentary cases rather than Kisor.

I.

Stinson sets out the standard for deferring to sentencing guideline commentary. Stinson v. United States, 508 U.S. 36 (1993). Stinson builds up to its holding by analyzing “analogies” that courts had used to describe the sentencing guideline commentary’s legal status. Id. at 43–45. Some suggested the comments were like a drafter’s “contemporaneous statement of intent.” Id. at 43. Others thought that “analogy to” Chevron captured the idea. Id. at 44. But the Court rejected these comparisons. Id.

The Court favored an “analogy” that—although “not precise”—it thought best described the commentary: “an agency’s interpretation of its own legislative rule.” Id. As far as the “functional purpose of commentary,” the guidelines were “equivalent” because they helped others to interpret and apply rules that were “within the Commission’s particular area of concern and expertise and which the Commission itself has the first responsibility to formulate and announce.” Id. at 45. This feature made the commentary “akin to” the agency’s interpretation of its own rules and entitled to “controlling weight” if constitutional and consistent with federal statute. Id. (quoting Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 414 (1945)).

1 I use Seminole Rock to refer both to Bowles v. Seminole Rock & Sand Co., 325 U.S. 410 (1945) and its successor case, Auer v. Robbins, 519 U.S. 452 (1997). No. 19-4232 United States v. Riccardi Page 19

The Court then explained that the Sentencing Reform Act also supported affording deference to guideline commentary. Id. at 45–46. Congress had tasked the Commission with periodically reviewing, revising, and clarifying the guidelines. Id. Amending the commentary is one way to accomplish that job, and courts should defer when the Commission pursues its statutory mission that way. Id. at 46.

Two aspects of this opinion suggest that the Court did not intend that Stinson and Seminole Rock would march in lockstep. First, the Court’s language describes the relationship between agency interpretation and guideline commentary as one of analogy, not equivalency. That falls short of a warrant to cross-apply cases interpreting Seminole Rock into the Stinson context. Second, the Court based the deferential standard not only on that similarity, but also on the Sentencing Commission’s unique obligation to review and revise the Guidelines. So coupling Stinson and Seminole Rock leaves a swath of the Court’s reasoning behind.

Consistent with this view, our cases do not apply “Seminole Rock deference” or “Auer deference” in guidelines commentary cases. Instead, we have repeatedly applied Stinson deference as its own free-standing directive. Commentary is authoritative “as long as the interpretation ‘does not violate the Constitution or a federal statute’ and is not ‘plainly erroneous or inconsistent with’ the provision’s text.”2 The cases that cite Stinson together with Seminole Rock or Auer are the exception, not the rule.3 Though Stinson considered Seminole Rock in deciding to extend deference to guideline commentary, we have viewed Stinson deference as creating an independent standard since its inception.

2 United States v. Buchanan, 933 F.3d 501, 514 n.2 (6th Cir. 2019) (quoting Stinson, 508 U.S. at 45); see also, e.g., United States v. Douglas, 634 F.3d 852, 862 (6th Cir. 2011); United States v. Hawkins, 554 F.3d 615, 618 (6th Cir. 2009); United States v. Jiles, 259 F.3d 477, 480 (6th Cir. 2001); United States v. Alexander, 88 F.3d 427, 431 (6th Cir. 1996); United States v. West, 59 F.3d 32, 34 (6th Cir. 1995); United States v. Lavoie, 19 F.3d 1102, 1104 (6th Cir. 1994); United States v. Garza, 999 F.2d 1048, 1053 (6th Cir. 1993). 3 Only three majority opinions in the circuit have cited either Seminole Rock or Auer together with Stinson in the same opinion. One was in a “quoting” parenthetical accompanying a Stinson quotation. United States v. Hayter Oil Co. of Greeneville, Tenn., 51 F.3d 1265, 1274 (6th Cir. 1995). The other was a footnote in a later vacated panel opinion. United States v. Havis, 907 F.3d 439, 443 (6th Cir. 2018), reh’g en banc granted, opinion vacated, 921 F.3d 628 (6th Cir. 2019), and on reh’g en banc, 927 F.3d 382 (6th Cir. 2019). The third only cites Auer for deferring to agency reading of rules. United States v. Owen, 940 F.3d 308, 314 (6th Cir. 2019). No. 19-4232 United States v. Riccardi Page 20

In Kisor, the Court addressed “Auer deference” and took the “opportunity to restate, and somewhat expand on” the Auer principles. Kisor v. Wilkie, 139 S. Ct. 2400, 2414 (2019). It did not mention the sentencing guidelines or suggest that its ruling disturbed Stinson.4 For this reason, two Fifth Circuit cases have rejected the idea that Kisor’s holding about Auer changed Seminole Rock through legal contagion. United States v. Cruz-Flores, 799 F. App’x 245, 246 (5th Cir. 2020) (per curiam); United States v. Vivar-Lopez, 788 F. App’x 300, 301 (5th Cir. 2019) (per curiam).

For all the reasons above, I would apply Stinson without factoring in Kisor’s guidance. Perhaps, in the end, the Supreme Court will vindicate the thoughtful and well-reasoned majority opinion. But for now, I would leave it to the Supreme Court to expand its own precedent especially because, I believe, the result is the same in this case.

II.

Declining to apply Kisor does not rubber-stamp the commentary. Although Stinson does not require “genuine ambiguity” before deferring, Kisor, 139 S. Ct. at 2415, it limits its deference to interpretations that the guideline “will bear.” Stinson, 508 U.S. at 46. It also requires that the comment “not run afoul of the Constitution or a federal statute” and that it “is not ‘plainly erroneous or inconsistent’ with” the guideline’s text. Id. at 47.

These are not toothless commands. Recently, we struck down a comment that added attempt crimes to the definition of a “controlled substance offense” in the guidelines. United States v. Havis, 927 F.3d 382, 383–84 (6th Cir. 2019) (en banc) (per curiam). We held that if “no term in [the guideline’s text] would bear” the commentary’s interpretation, then the comment added to rather than interpreted the text. Id. at 386. In other words, the step before applying deference is asking, “[I]s this really an ‘interpretation’ at all?” Id. at 387.

At least nine circuits have struck down commentary under Stinson. Consider the Offense Statutory Maximum (OSM) cases. The OSM table lists ranges of incarceration terms, and each

4 To be fair, Kisor lists Stinson in a footnote as one case “applying” Seminole Rock deference. Kisor, 139 S. Ct. at 2412 n.3. I do not read that footnote as a command, however, to strictly apply such deference in these cases. No. 19-4232 United States v. Riccardi Page 21

range corresponds with an Offense Level. U.S.S.G. § 4B1.1. To use the table, courts determine the maximum statutory sentence a criminal could receive for his crimes, and they use the corresponding Offense Level as the starting point for the guidelines sentence calculation. Before commentary guidance to the contrary, courts interpreted the OSM as the maximum sentence allowed by statute after applying any career-offender enhancements. But then the Sentencing Commission amended § 4B1.1’s Application Note 2 to define the OSM as the maximum sentence allowed before applying any career-offender enhancements.

Several circuits rejected the amended commentary. Our own circuit wrote that it “contravene[d] federal law” by conflicting with the “plain language of [28 U.S.C.] § 994(h).” United States v. Branham, 97 F.3d 835, 845–46 (6th Cir. 1996). The Tenth Circuit agreed that the phrase “maximum term authorized” in the statute unambiguously meant the sentence with enhancements included. United States v. Novey, 78 F.3d 1483, 1487 (10th Cir. 1996). Three other circuits rejected the commentary as inconsistent with the statute as well.5 And the Supreme Court ultimately affirmed this view, noting that, “If the Commission’s revised commentary is at odds with § 994(h)’s plain language, it must give way.” United States v. LaBonte, 520 U.S. 751, 757 (1997) (citing Stinson, 508 U.S. at 38).

Circuit courts also rejected commentary in “crime of violence” cases. Under the guideline at the time, a crime of violence included any crimes with use of force as an element as well as “burglary of a dwelling, arson, or extortion, . . . use of explosives, or [crime that] otherwise involves conduct that presents a serious potential risk of physical injury to another.” U.S.S.G. § 4B1.2 (Nov. 2015). The commentary purported to add in “kidnapping, aggravated assault, forcible sex offenses, robbery . . . [and] extortionate extension of credit.” § 4B1.2 cmt. n. 1 (Nov. 2015). But the Fourth Circuit held that the commentary could not add in an offense if it did not fit with the guideline text. United States v. Shell, 789 F.3d 335, 339 (4th Cir. 2015). And a few months later, the First Circuit held that commentary adding possession of a machine gun to the definition of “crime of violence” could not survive after the guideline’s

5 See United States v. McQuilkin, 97 F.3d 723, 731 (3d Cir. 1996); United States v. Hernandez, 79 F.3d 584, 595 (7th Cir. 1996); United States v. Fountain, 83 F.3d 946, 951 (8th Cir. 1996). No. 19-4232 United States v. Riccardi Page 22

residual clause—the commentary’s textual hook—was no longer valid. United States v. Soto- Rivera, 811 F.3d 53, 59–61 (1st Cir. 2016).6

And the examples go on. Commentary must “yield the road” when it tells courts to overlook an enhancement that the guideline text says to apply. See United States v. Chuong Van Duong, 665 F.3d 364, 368 (1st Cir. 2012); United States v. Ordonez, 305 F. App’x 980, 985 (4th Cir. 2009) (per curiam). Courts must include all intended losses in their loss calculations even if the commentary opines that certain losses should not count. See United States v. Henderson, 19 F.3d 917, 928 (5th Cir. 1994). The statutory command to reference the guideline for the specific crime committed overrides a commentary’s instruction to reference a different guideline. See United States v. Smith, 184 F.3d 415, 418–19 (5th Cir. 1999); United States v. Fields, 242 F.3d 393, 398 (D.C. Cir. 2001).

The point: Stinson requires that commentary interpret the guidelines, not contradict or add to them. So the question for us is whether the comment here heeds those bounds. I conclude it does not for the same reasons that the majority opinion well articulates. Maj. Op. at 11–15. Ascribing a certain number to “loss” is not a definition. It is a policy call. Imagine that the commentary required court to calculate any credit card theft as a million-dollar loss. That would be unreasonable, but not because it is an inferior definition of “loss.” A loss may be millions just as well as it may be $500. Rather, the million-dollar provision would be an unreasonable policy choice.

Finally, I would note that this decision does not impact comments that properly amplify the different aspects of “loss.” Other commentary in §2B1.1(b)(1) addresses losses unique to certain crimes by specifying types of loss the court should consider. For a product substitution case, that includes any “reasonably foreseeable” costs that the crime causes, such as “making substitute transactions,” “handling or disposing of the product delivered,” and “actual or potential disruption to the victim’s business operations.” § 2B1.1, cmt. n. 3 at (A)(v)(I). And in procurement fraud cases, courts should include “reasonably foreseeable administrative costs . . . of repeating or correct[ing] the procurement action affected.” Id. at (A)(v)(II). This kind of

6 The Seventh Circuit held the same when it addressed this issue en banc. United States v. Rollins, 836 F.3d 737, 743 (7th Cir. 2016) (en banc). No. 19-4232 United States v. Riccardi Page 23

commentary enhances the definition of “loss” by emphasizing particular aspects rather than making a policy call disguised as an interpretation.

For all these reasons, I concur in the result but base my reasoning on Stinson rather than Kisor.