Report Broad, Ambiguous, or Delegated: Constitutional Infirmities of the Bank Secrecy Act

Introduction: The Bank Secrecy Act

The Bank Secrecy Act (BSA) establishes a financial surveillance regime aimed at preventing money laundering through the U.S. financial system. It authorizes the Secretary of the Treasury to issue regulations requiring certain designated parties—called “financial institutions”—to collect and keep identifying information about their customers, to keep records of their customer’s transactions, to make that information available to law enforcement, to report certain information about transactions automatically to the Department of the Treasury without any legal process or warrant, and to develop an internal protocol to deny financial services to certain customers and to prevent customers from engaging in certain transactions.¹ Failure to register as a financial institution can be prosecuted as a strict liability felony.²

This Report argues that the Bank Secrecy Act is either (A) so broad as to criminalize everyday life,³ (B) so ambiguous as to make uncertain its application to millions of Americans, or (C) spared from being so broad or so ambiguous by the exercise of legislative authority delegated by Congress to the Treasury Department. Each of these alternative interpretations of the Bank Secrecy Act raises substantial constitutional concerns.

The Bank Secrecy Act is riddled with sweeping powers and expansive terms.⁴ For this report, however, we will mostly confine our analysis to the definition of financial institution. As discussed throughout, that single definition is fundamentally consequential. For example:

• Whether a person meets the definition of financial institution is the key determinant of whether they are subject to the lion’s share of the Bank Secrecy Act’s obligations.
• Failure to register as a financial institution can be prosecuted under racketeering laws as a strict liability felony with up to five years in jail as well as severe monetary penalties and moral condemnation. Ignorance of one’s status as a financial institution is not a defense.
• Whether the definition of financial institution includes only “third parties” (i.e. middlemen service providers like banks or telecommunications companies) is a key factor in determining whether the Act directs Treasury to engage in unconstitutional searches and seizures or, instead, merely directs Treasury to collect personal data from third parties in a way that does not constitute a search.
• Whether the definition of financial institution is ambiguous determines the amount of discretion afforded Treasury in interpreting the breadth of the statute’s coverage and the level of deference that such interpretations should be afforded by courts.
• Whether the definition of financial institution grants Treasury legislative authority to freely redefine the statute’s application will determine the constitutional propriety of Congress’s delegation of power to the Treasury Department.⁵

In the Bank Secrecy Act, “financial institution” has a 425-word definition divided into 26 distinct and very specific sub-categories of persons or businesses.⁶ Among these are insured banks; commercial banks; trust companies; agencies or branches of foreign banks in the United States; brokers or dealers in securities or commodities; investment bankers; currency exchanges; issuers, redeemers, or cashiers of travelers’ checks, checks, money orders, or similar instruments; operators of a credit card system; insurance companies; dealers in precious metals, stones, or jewels; pawnbrokers; travel agencies; telegraph companies; businesses engaged in vehicle sales; persons involved in real estate closings and settlements; the United States Postal Service; and casinos. The definition of financial institution also includes a particularly broad sub-category commonly referred to as “money transmitters” that is defined as

a licensed sender of money or any other person who engages as a business in the transmission of funds, including any person who engages as a business in an informal money transfer system or any network of people who engage as a business in facilitating the transfer of money domestically or internationally outside of the conventional financial institutions system[.]⁷

The definition of financial institution also includes two “catch-all” provisions. The Secretary of the Treasury is given the power to determine that either of the following are also financial institutions:

[A]ny business or agency which engages in any activity which the Secretary of the Treasury determines, by regulation, to be an activity which is similar to, related to, or a substitute for any activity in which any business described in this paragraph is authorized to engage[, and]

[A]ny other business designated by the Secretary whose cash transactions have a high degree of usefulness in criminal, tax, or regulatory matters.⁸

Outside of the definitions section of the Act there is an exemption power that allows the Secretary of the Treasury to craft “appropriate” exemptions from compliance with the requirements of the Act.⁹ While this exemption power does not enable the Secretary to narrow the definition of financial institution, its frequent use accomplishes the same result in practice.

As will be discussed throughout, the Secretary has used these powers extensively. The implementing regulations of the Bank Secrecy Act, at 31 C.F.R. Chapter X, include a distinct set of sub-definitions of financial institution that includes alternative definitions and terms for several of the statutory categories described above, as well as several new sub-categories of financial institution not described in the statute. The regulations also include several “facts and circumstances limitations” that are, in effect, flexible tests used by the Treasury Department to engage in an ad hoc narrowing of its own regulatory definition of financial institution. These facts and circumstances limitations appear within the regulation’s definitional section, but nowhere is it clear how such facts and circumstances determinations are to be made by the agency nor how precedential such determinations will be in future interpretations of the definition. In practice, persons are excluded from the definition (or they are not) according to these facts and circumstances in a variety of administrative procedures, such as through rulemaking, administrative ruling, informal guidance, or mere prosecutorial discretion.

This report will proceed in three parts: Breadth, Ambiguity, and Delegation.

In Breadth, we begin with a plain English reading of the statute’s definition of financial institution, focused on the sub-category of money transmitters. Second, we will re-read this definition in the context of the whole statute and use semantic canons of statutory construction to arrive at a “best reading” of the statute’s application. We find that a “best reading” leaves us with an extremely broad scheme that criminalizes everyday life. Though it is self-evident that a criminal law scheme this broad would be an anathema to the rule of law, we nevertheless describe why and in which context courts have historically found such broad statutes troublesome. We will contrast the concepts of textual ambiguity, vagueness, and breadth within the scholarly literature on statutory construction and due process. We find that the BSA fits neatly in the category of broad statutes struck down by courts for due process deficiencies.

In Part II, Ambiguity, we set aside our “best reading” of the statute and the attendant conclusion that it is absurdly broad, and, in the alternative, we proceed to a discussion of ambiguity. First, we show that courts and Treasury have repeatedly reached similar conclusions about the problematic breadth of the definition of financial institution but have found ways to narrow the statute’s application by treating this breadth as if it were merely textual ambiguity. We discuss why this invented ambiguity is a convenient article of faith, or ipse dixit, that enables courts and regulators to save the statute from probable unconstitutionality. We go on to discuss, following scholarly research on ambiguity, why this particular contrivance (moving from breadth to ambiguity) allows for the questionable use of several substantive canons of statutory construction: lenity, constitutional avoidance, and Chevron deference. Furthermore, we find that any attempt to narrow the statute’s application by use of these substantive canons would fail to adequately inform persons of their obligations under the law.

Finally, in Part III, Delegation, we look at two catch-all provisions within the statutory definition of financial institution that grant Treasury the power to expand the statute’s coverage, as well as a subsequent section of the Act that allows Treasury to craft “appropriate” exemptions. We find that whenever Treasury issues new rules or guidance that expands or contracts the definition of financial institution it cites these passages but does not make clear whether it is relying on the delegations in these passages to refine the definition of financial institution, or whether it is simply engaging in a process of agency statutory construction given the invented ambiguities discussed in the previous part. We discuss recent case law on non-delegation and find that these passages of the Bank Secrecy Act are likely constitutionally deficient because they are a highly consequential delegation of legislative power from Congress to the executive that lacks an intelligible principle.

Finally, we conclude with a brief discussion of how and whether these deficient parts can be severed from the statute as a whole, and what the implications may be from a policy perspective.

I. Breadth

A “Best Reading” of the Statute’s Application

As a starting point we analyze the text of the definitions section of the Bank Secrecy Act in order to uncover a working theory of the scope of the law’s application. Following an approach set out by Justice Kavanaugh, we build a “best reading” of the statute, first by reading the text as a non-technical English speaker would, and then by re-reading the text with the context of the whole statute in mind and utilizing long standing semantic canons of statutory construction to address any ambiguities.¹⁰

As described earlier, the portion of the Bank Secrecy Act that raises particularly consequential breadth and ambiguity concerns is the definition of “financial institution,” in particular subpart (R) of that definition, which begins:

[A] licensed sender of money or any other person who engages as a business in the transmission of funds, …¹¹

Set aside the first clause, “licensed sender of money” as it should be uncontroversial.¹² The statute is broad because of the clause that follows, “or any other person who engages as a business in the transmission of funds[.]” How would a hypothetical non-technical, ordinary English speaker interpret this part of the definition? Plainly, it appears that a person accepting payment for her labor could be included within this definition. When a barber receives cash in payment for a haircut or gives cash in change, is she or is she not engaged as a business—barber—in the transmission of funds—her receipt of payment or offering of change for cutting hair? While her business is not “money transmission” per se, it is, nonetheless, a business, and part of that business involves engaging in the transmission of funds. Should this definition be read to include only persons whose business is money transmission per se, or should it be read to additionally include persons involved in other businesses who engage in incidental transmissions of money? As absurd as the broader interpretation may seem, it is a genuine option chosen by regulators tasked with enforcing the law in some circumstances. As we will see, this interpretive flexibility is either a fundamental ambiguity in the statute or it is the mark of a statute with deliberately broad potential application.

A plain reading of the Bank Secrecy Act’s definition of financial institution Subpart (R) nevertheless suggests that all persons engaged in business and being paid or paying others may be, so defined, financial institutions. Therefore, a plain reading of the Bank Secrecy Act may mandate that every American who pays or is paid must register with the Treasury Department and regularly report the details of her monetary transactions as if she were a bank or other financial institution, and as if the people she paid or took payment from were her bank customers. As we will discuss in the subsequent subsection detailing the trouble with broad statutes, this would have a dangerously Orwellian logical outcome: effectively all Americans would have to register with a federal agency, surveil and collect information useful to law enforcement from their commercial counterparties, and report that information to the government sua sponte and without a warrant or subpoena. Your fellow Americans are your brothers and sisters, but taken together they are big brother.

Semantic Canons

Canons of statutory construction are longstanding tools used by lawyers and judges to arrive at reasonable and consistent interpretations of legislative text.¹³ Semantic canons provide rules for interpreting the English language¹⁴ as compared with contextual canons that provide rules for interpreting words within the context of the whole statute, or substantive canons that provide rules for interpreting statutes within our constitutional and rule of law traditions.¹⁵ Thanks to the increasing prominence of textualism in the judiciary, semantic canons have played an important role in several recent high profile cases at the Supreme Court.¹⁶

Our plain reading is broad verging on absurdity. While an ordinary English speaker would likely concede that the text could treat all paid or paying persons as financial institutions, she would also struggle, no doubt, with the unreasonableness of such a broad conclusion: every paid or paying American is a financial institution?! The alarmingly broad interpretation suggested by our plain reading is, however, bolstered rather than impugned when one applies semantic canons of statutory construction to resolve ambiguities.

For starters, general terms are to be given their general meaning:¹⁷ the use of the word “any” rather than “a” within “any person who engages as a business in the transmission of funds” suggests that this is intended by Congress to be a comprehensive category including any person who does these things.¹⁸ This broad interpretation is bolstered by virtue of the expansive “laundry list”¹⁹ that follows the clause:

any other person who engages as a business in the transmission of funds, including any person who engages as a business in an informal money transfer system or any network of people who engage as a business in facilitating the transfer of money domestically or internationally outside of the conventional financial institutions system[.]²⁰

As per semantic canon, “include” is presumed to be nonexclusive.²¹ “Any” is repeated twice more in the inclusive list, again suggesting that an expansive interpretation is intended within the law. The list includes “network of people” as an additional entity-type included in the list. If you, on your own, are not necessarily engaged in a transmission, but you are part of a network of people who are together so-engaged, you are also included.

While an ordinary English speaker might think it silly that a list of financial institutions would include individual persons who are not ordinarily thought of as financial institutions (e.g. banks or securities brokers), the inclusive list again contradicts that intuition, plainly stating that even transmission activities occurring “outside of the conventional financial institutions system” are expressly included within the definition.

It is tempting to construe the proviso “as a business” as creating a condition that the money transmitter definition should only encompass legal persons professionally transmitting money for others (e.g. Western Union or PayPal). However, the proviso canon suggests that “as a business” should be read merely to qualify the matter immediately preceding.²² Therefore, “as a business” merely qualifies how one “engages” not whether one’s business is “transmission of funds.” Again, the full clause is “any other person who engages as a business in the transmission of funds”; it is not “any other person who transmits funds as a business” or, even more narrowly, “any person who is in the business of transmitting funds for others.” Employing only semantic canons and an ordinary usage of the English language, the definition section of the Bank Secrecy Act apparently extends the category of regulated financial institutions to include anyone who transmits funds whether for her own personal purposes or on behalf of others, essentially every productive member of society. As we will see in later sections, sometimes this absurdly broad meaning is contradicted by regulators and courts, but at other times it is insisted upon in order to afford law enforcement purportedly necessary discretion to charge atypical money transmitters with the strict liability felony of failure to register.²³ This flexible and inconsistent interpretation leads to substantial confusion surrounding the actual rule of law with regard to money transmission regulation.

Contextual Canons

Is every paid or paying citizen truly a financial institution under the law? Perhaps not. As the Supreme Court in statutory construction cases repeatedly stresses, a passage must be read within the context of the whole statute. After all, this is the Bank Secrecy Act, not the Paid or Paying Citizens Secrecy Act. Even in light of contextual canons of construction, however, the resulting interpretation remains exceedingly broad.

Adjacent subcategories of the definition of financial institution (subcategories aside from money transmitter) provide context that Congress intended to go beyond the ordinary expectation of what the term “financial institution” might include. For example, a pawnbroker is a financial institution, as are casinos and the U.S. Postal Service.

Despite the explicitly non-standard types of financial institutions listed adjacent to the definition of money transmitter, it may, nonetheless, be tempting to narrowly interpret the specific definition of money transmitter in light of the general meaning of its super-category, financial institution, which is commonly understood as a regulated customer-facing business that provides financial services. The general/specific contextual canon, however, argues that this is inappropriate.²⁴ If there is conflict between the specific definition of “money transmitter” and the general term “financial institution”—and it isn’t even clear that there is in this case—this canon says the specific provision, an explicitly broad definition of money transmitter, prevails.

And what about the title of the Act that we alluded to earlier? Surely in the Bank Secrecy Act, rather than Paid or Paying Citizens Secrecy Act, we should read the definition of money transmitter as including only people whose business is bank-like (i.e. performing money transmission for others), not merely people who are engaged as a non-financial business in incidental transmissions of money. The “Bank Secrecy Act,” however, is merely the short title “popularly known”²⁵ to describe the law at Subchapter II of Chapter 53 within Title 31 of the U.S. Code. Short titles and popular titles do not generally get the benefit of the title-and-headings canon.²⁶ The Subchapter’s actual heading is simply “Records and Reports on Monetary Instruments Transactions,” the Chapter’s heading is “Monetary Transactions,” and Title 31 itself is simply “Money and Finance.” None of this suggests any strict qualification regarding whose monetary transactions are intended to be recorded or reported or by whom.

Neither could the prefatory-materials contextual canon²⁷ be used to narrow the definition; the subchapter’s statement of purpose is not tailored to any particular type of institution:

It is the purpose of this subchapter … to require certain reports or records where they have a high degree of usefulness in criminal, tax, or regulatory investigations or proceedings, or in the conduct of intelligence or counterintelligence activities, including analysis, to protect against international terrorism.²⁸

Records and reports of every paid or paying citizen’s financial transactions would certainly have a high degree of usefulness in investigations and no mention is made in the statement of purpose concerning who should and should not be obligated to keep those records or make those reports.

Then there is the ejusdem generis²⁹ contextual canon: read the definition of money transmitter narrowly because it is one example of a category of similar things described by the list of financial institutions. The argument would be that every other item in the list is a sort of intermediary that specializes in providing to its customers monetary transactions of one sort or another; therefore, “money transmitter” should be interpreted as limited to those who specialize in performing transmission as a business for other persons, rather than including persons who are simply transmitting their own funds.

As both Justice Kavanaugh and Justice Kagan³⁰ have stressed, however, the usage of ejusdem generis relies on the selection of an arbitrary common denominator amongst all the items in the list. The actual text of the definition of financial institution does not begin with a common qualifier (e.g. all list items are types of intermediaries specializing in monetary transactions). A court that presumes that the common denominators are “intermediaries” and “businesses dealing in money” would be inventing those common denominators out of whole cloth; that invention would be, in essence, policy-making. Narrowing the otherwise plain text of the definition in light of an arbitrary judgment concerning a term’s commonality with neighboring terms is merely judicial rewriting of the statute in an attempt to cabin a troublesomely broad definition.

The arbitrariness of our common denominator aside, it might not even be appropriate to label all of the list items as types of monetary-transaction-specialist intermediaries. Financial institution subpart (Q) is “a travel agency,”³¹ an intermediary to be sure but one that specializes in booking accommodations and arranging tours rather than engaging in monetary transactions. Subpart (N) is a “dealer in precious metals, stones, or jewels,”³² and while these valuables might well have numismatic qualities it seems inappropriate to call these dealers specialists in monetary transactions; they may be engaged exclusively in the retail sale of jewelry or other decorative arts rather than coins. Even more plainly, a “dealer in precious metals, stones, or jewels” is not inherently an intermediary of other persons’ transactions; she may conceivably be simply selling items from her own collection. She may be one of two people engaged in an arms-length transaction rather than a go-between or third party intermediary for several transactions. Finally, financial institution Subpart (Z) clearly contradicts the interpretation that this list includes only intermediaries specializing in monetary transactions per ejusdem generis. It allows the inclusion in the category of financial institution, “any other business designated by the Secretary whose cash transactions have a high degree of usefulness in criminal, tax, or regulatory matters.”³³ No matter the particular line of business—be it a baker, a banker, or a candlestick-maker—any substantial cash transactions made by a business could clearly have a high degree of usefulness in tax matters, especially if the business is under audit.

Plainly, the category of financial institution is unambiguously broad by definition and cannot arbitrarily be limited to a certain subset of businesses who specialize in intermediating monetary transactions. To claim that “intermediation” and “monetary transactions” are the common denominators of the list of financial institutions may be expedient in order to narrow the alarmingly broad definition of money transmission and the Orweallian collateral consequences of that breadth, but it would be a policy-making decision to do so, not merely a sensible interpretation of an otherwise ambiguous law. Accordingly, narrowing the definition by use of ejusdem generis or any of the other semantic or contextual canons discussed thus far would be an interpretive move that a majority of the Court, at least, should be hesitant to make.

In Part II we will discuss why several substantive canons of construction (lenity, constitutional avoidance, and Chevron deference) also cannot be used to narrow the definition because the predicate for their usage is a finding of ambiguity. Again, the text is plainly broad but it is not ambiguous. As we will now discuss, breadth is not equivalent to ambiguity, and breadth on its own is a substantial constitutional problem.

Why Breadth is a Problem

Laws that do not clearly announce their requirements or that cannot be understood by the ordinary people to whom they apply are antithetical to the rule of law. Constitutionally, due process is denied when ordinary people cannot or do not understand what conduct is prohibited under the law, when subjects must guess at the law’s meaning, and when those tasked with enforcing the law can freely shift their own interpretations to maximize their discretionary power. This is not controversial. Less clear, however, is whether a law that is extremely broad rather than ambiguous also undermines the rule of law and denies due process to those affected.

Legal Scholar Kiel Brennan-Marquez, in an expansive treatment of the subject titled “Extremely Broad Laws” argues that statutory breadth can be equally as troublesome as statutory ambiguity and recounts how courts have, though often sub-silentio, developed a void-for-breadth doctrine. Brennan-Marquez conjectures that courts are faced with few tools addressing problematic statutory breadth head-on, and therefore “have adopted a second-best solution: recharacterizing breadth as a species of linguistic uncertainty—vagueness or ambiguity—and repurposing tools designed for those problems, the void-for-vagueness doctrine and the rule of lenity, to constrain the reach of broad laws.”³⁴ Brennan-Marquez offers examples of why several laws cast as vague are, in fact, plainly clear but problematically broad. That analysis stands on its own and mirrors substantially the preceding analysis of the definition of financial institution in the Bank Secrecy Act. While it is tempting to argue that the definition only appears to treat every paid or paying individual as a financial institution because of textual ambiguity, the fact of the matter is that the text is not ambiguous: “any person who engages as a business in the transmission of funds” is a financial institution. The definition is broad, not ambiguous.

Brennan-Marquez cleverly personifies these two species of troublesome law with appropriate literary embodiments. A vague law, he writes, is Kafkaesque: legal pitfalls hide ’round every corner and the benefit of the law—certainty—is denied through obfuscation and doubt, a perpetual fear of its uncertain application. A broad law is Orwellian: the demands of the law are clear but those demands are so extensive as to entirely curtail human flourishing. It is the terror and immobility of living under a police state, Orwell’s Oceania, rather than the absurdity and self-imposed immobility of wasting one’s whole life waiting before a fearsomely guarded gate that was, in Kafka’s “Before the Law,” deceptively unlocked and free for passage all along.

Three cases decided by the Supreme Court exemplify the problem of broad laws and illustrate the Court’s uncertain approach to addressing those problems: Marinello, Yates, and Morales.

In Marinello, the government argued that a clause of the Internal Revenue Code subjected anyone who impedes the due administration of taxes to criminal penalties.³⁵ “Interpreted broadly,” wrote Justice Breyer for the majority, “the provision could apply to a person who pays a babysitter $41 per week in cash without withholding taxes, leaves a large cash tip in a restaurant, fails to keep donation receipts from every charity to which he or she contributes, or fails to provide every record to an accountant.”³⁶ The majority held against the government but reasoned that the law was ambiguous because “impede” could mean either the broad interpretation favored by the government (i.e. impede means to perform any act that makes it difficult to collect taxes, or “impede” could refer only to specific acts that create difficulty in some specific tax proceeding aside from the general collection of taxes, a proceeding that was pending or foreseeable by the defendant when they created the impediment). In other words, keeping bad records may not be criminal under the statute, but actively deleting emails in advance of an audit would be. Brennan-Marquez agrees with the majority’s concern over broad application but argues that

the problem, ultimately, is not one of uncertainty. All noncompliance with tax law is, ipso facto, an “impediment” of the “administration of [taxes].” The problem is that, taken seriously, the Omnibus Clause would convert even the most minute instances of shoddy record-keeping into obstruction of justice offenses carrying up to three years of incarceration. In other words, the problem—as Justice Kagan suggested during oral argument—is the law’s “ungodly br[eadth].”³⁷

In our Bank Secrecy Act context, the problem is not that the definition of “money transmitter” is uncertain. The problem is that, taken seriously, the definition would convert every paid or paying American into a “financial institution” and obligate each of us to register with the federal Financial Crimes Enforcement Network (FinCEN) as such, keep extensive records of all transactions, robustly identify all transactional counterparties, and regularly report any suspicious dealings we have had with other Americans or foreign persons. And we must do all of that on penalty of felony conviction, severe fines, and jail time. This is an Orwellian state of affairs rather than a Kafkaesque absurdity.

In Yates, the government argued that a clause of Sarbanes-Oxley, an anti-accounting-fraud law passed in response to the Enron scandal, enabled it to prosecute for obstruction of justice anyone who destroys any “tangible object” with the intent to impede an investigation.³⁸ In that case, a Florida boat captain threw several undersized snapper overboard in advance of a probable inspection by wildlife authorities. Justice Ginsbsburg wrote for the majority that it was ambiguous whether “tangible object” in the statute included fish or whether it referred only to tangible objects “used to record or preserve information.”³⁹ Given that ambiguity, the Court applied the rule of lenity and held in favor of Yates.

Contrary to our analysis of the Bank Secrecy Act in the previous section, the court used ejusdem generis to support its conclusion that “tangible object” should be narrowly construed to mean tangible objects used for recordkeeping. The Court reasoned that the terms adjacent to “tangible object” within the list of covered items were all accounting related, for example “record” and “document.”

“Money Transmitter,” like “tangible object,” is a very broad category in the Bank Secrecy Act, and widespread classification as such could lead to mass criminal noncompliance with our banking laws, just as the criminalization of any destruction of “tangible objects” could lead to mass criminal noncompliance with our accounting laws. If a court were to follow Yates, perhaps the definition of money transmitter should be interpreted narrowly, as if it were ambiguous, because the adjacent categories of financial institution all describe intermediating monetary transactional services. For reasons discussed earlier, however, the definition of money transmitter is not ambiguous (thereby barring the court from engaging in a fishing expedition to uncover its meaning through context) and neither do the adjacent types of financial institutions all share intermediation or monetary transactions as common denominators.

As Justice Kagan highlighted in her dissent in Yates, the selection of that common denominator to interpret all terms ejusdem generis is arbitrary:

The canon says you need a common denominator. But what is that common denominator? Is the common denominator things that preserve information? Or in the context of an evidence tampering statute, is the common denominator things that provide information to an investigator, things that tell an investigator, say something to an investigator about what the crime is?⁴⁰

In a law review article titled “Fixing Statutory Interpretation,” Justice Kavanaugh, then a judge of the United States Court of Appeals for the D.C. Circuit, lauded Justice Kagan for her dissent and, in light of the common denominator problem, advocated an abandonment of ejusdem generis altogether:

Judges should not be in the position of trying to devise the connective tissue or common denominator. I would consider tossing the ejusdem generis canon into the pile of fancy-sounding canons that warrant little weight in modern statutory interpretation.⁴¹

Moreover, as Justice Kagan’s dissent in Yates makes clear, the problem with the relevant law, Sarbanes Oxley, was not its ambiguity but its breadth:

the [Court] points to the breadth of [§ 1519] as [if] breadth were equivalent to ambiguity, . . . [i]t is not. Section 1519 is very broad. It is also very clear. Every traditional tool of statutory interpretation points in the same direction, toward ‘object’ meaning object [including fish]. Lenity offers no proper refuge from that straightforward (even though capacious) construction.⁴²

In the Bank Secrecy Act context, as in Yates, every traditional tool of statutory interpretation points in the same direction: toward “any … person who engages as a business in the transmission of funds” meaning any person transmitting funds as part of her business. In other words, any person who pays or is paid for her labor is a money transmitter. Just as Sarbanes-Oxley does not say “tangible object used to record information” the Bank Secrecy Act does not say “any person who engages in the transmission of funds because that transmission is her principal business” or “any person who is in the business of money transmission, i.e. regularly provides money transmission services to others.” A judge may wish it said those things, but the text is clearly much broader.

Finally, in Morales, the Court overturned a Chicago ordinance that criminalized loitering by any group of people that included at least one “criminal street gang member.”⁴³ Justice Stevens proffered that such a broad law could lead to the arrest of a father and son out to watch a baseball game if the son happens to be a member of a gang. Stevens, writing for the majority, found that the ordinance was unconstitutional because its vagueness deprived people of due process,

[s]ince the city cannot conceivably have meant to criminalize each instance a citizen stands in public with a gang member, the vagueness that dooms this ordinance is not the product of uncertainty about the normal meaning of ‘loitering,’ but rather about what loitering is covered by the ordinance and what is not.⁴⁴

As Brennan-Marquez writes, “The issue, in other words, was not lack of clarity. It was that the ordinance clearly swept in ‘a substantial amount of innocent conduct,’ leaving ordinary people to guess at ‘what was [actually] forbidden’ in practice.”

As with Morales, the issue with the Bank Secrecy Act is not a lack of clarity over what the normal meaning of “any … person who engages as a business in the transmission of funds” might be. The issue is that the definition clearly includes every paid or paying American, leaving ordinary people to guess at whether they are obligated, in actual practice, to register with a federal agency as a financial institution and report the details of their financial dealings accordingly. Of course, an average paid or paying American is not alone in making those guesses, she might rely on statements from the relevant regulators to guess what the statute actually covers in practice. As we will see in the final section on delegation, however, this suggests that the regulators have legislative power to rewrite the statute more narrowly, and this potentially unconstitutional delegation of legislative authority is the only mechanism by which the law is saved from being, as Justice Kagan might say, ungodly broad.

Speaking generally, Brennan-Marquez summarizes that broad laws are problematic to courts within three contexts: (1) When the severity of punishment does not match the moral culpability of the offense, as in both Yates or Morales where throwing fish overboard or paying the babysitter in cash can trigger felony criminal liability. (2) When the law takes a “shotgun approach” to social regulation and thereby “criminalizes everyday life,” as in Morales where an attempt to police gang violence allows police to arrest a father and son out at the ballgame. And (3) when the breadth of the law shades into other protected constitutional rights.

The breadth of the Bank Secrecy Act’s application checks all three of these boxes, indicating that it would likely face similar intense scrutiny from the Court:

1. Moral Culpability vs. Penalties. The BSA’s penalties do not always match the moral culpability of the offense. Even minor violations of the Act, such as failure to register as a money transmitter with no associated money laundering or wrongdoing, can be felonies. Additionally, liability can pierce the corporate veil, and the law does not require scienter to reach a conviction.⁴⁵ In several cases, courts have found defendants guilty for failure to register even when those defendants had no knowledge that the act in which they were engaged was money transmission, broadly defined, and therefore required registration.
2. Shotgun Application Criminalizing Everyday Life. The BSA has the laudable goal of preventing use and abuse of the American financial system by criminals and terrorists. As drafted however, it takes a shotgun approach to stopping money laundering. Taken literally it says that every paid or paying American must record and report the details of her transactions, innocent or not, so that the Treasury Department can make sure that none of them involve the proceeds of crime. In practice few register, but that breadth affords law enforcement a very useful tool: a person merely cashing a check for a friend or selling their cousin some bitcoin can be left alone, but if they happen to be cashing that check for or selling that bitcoin to a criminal (irrespective of whether they know they are a criminal) they can be charged with failure to register as a financial institution, a strict liability felony. If every paid or paying American might be a financial institution, then every American should probably register with the Treasury Department. Then these Americans would only need to worry about being charged with knowingly laundering money rather than strict liability failure to register. Needless to say, almost no one registers but we all still cash checks and accept or make cash payments, and as such the law criminalizes everyday life.
3. Breadth Shades Into Other Constitutional Rights. A requirement that every paid or paying American surveil and sua sponte report on her transactional counterparts to the government is likely to chill substantial amounts of free expression and reveal copious amounts of potentially self-incriminating personal data to the government without a particularized warrant. Breadth aside, the BSA raises substantial First, Fourth, and Fifth Amendment concerns.⁴⁶

The rule of law infirmities inherent in broad statutes are similar to those in ambiguous laws: neither an ambiguous statute nor an extremely broad statute is capable of preemptively warning ordinary citizens that there was some alternative path that could have spared them the condemnation of the state. As Hobbes wrote, the law should guide people into voluntary compliance, it should respect the dignity and agency of human beings by affording them a choice (even if that choice ultimately may be to break the law) rather than forcing them into certain behaviors through previously unannounced power:

For the use of laws (which are but rules authorized) is not to bind the people from all voluntary actions, but to direct and keep them in such a motion as not to hurt themselves by their own impetuous desires, rashness, or indiscretion; as hedges are set, not to stop travellers, but to keep them in the way.⁴⁷

A broad law is a hedge that entirely surrounds a traveler, entrapping them and forbidding any and all voluntary motion. An ambiguous law is a hidden hedge that the traveler can take no steps to avoid and over which she will inevitably stumble. Both types of laws, as Hobbes would likely have said, “are not good laws, but traps for money.” Their common infirmity is that no one ever expects them or the serious life-destroying consequences that their violation could entail should they be enforced. As a great English comedy troupe once said, “NOBODY expects the Spanish Inquisition!”⁴⁸  Or, as Brennan-Marqez puts it,

Past a certain threshold, it becomes impermissible to pursue regulatory goals—even legitimate ones—by criminalizing large swaths of normal activity. Doing so turns virtually everyone into a criminal, at least on paper; and it leaves ordinary people without a meaningful understanding, ex ante, of the legal risks our actions invite.⁴⁹

Additionally, there are further rule of law concerns inherent in a court choosing to narrow a broad law through an arbitrary interpretation rather than strike that law down. The infirmity of the broad law was its potential for arbitrary application and a choice made by the judiciary to, in one isolated case, narrow the law’s application is simply another species of arbitrary application. As the Court remarked in Kolender v. Lawson, a case dealing with a similarly broad anti-loitering statute as Morales:

Although this court, like all other institutions of the United States, is supportive of the law enforcement goals of the government and society, we cannot engage in unprincipled interpretation of the law, lest we foment lawlessness instead of compliance.⁵⁰

The ad-hockery necessary to cabin the application of a broad law in each case that comes before a court is generative of unpredictability and retroactivity, what Bentham would call dog law: “When your dog does anything you want to break him of, you wait till he does it, and then beat him for it. This is the way you make laws for your dog: and this is the way judges made law for you and me.”⁵¹ Rather than reform a broad statute by forcing Congress to revisit its definition and develop something narrower and comprehensible, an ad hoc narrowing by the courts either punishes the accused directly for the crime or else punishes the larger citizenry by adding one further wrinkle to the incomprehensible bricolage of so many ways the law does not actually mean what it appears to mean. Complicating a broad law with case-by-case carve outs ultimately transmutes breadth into ambiguity, thus perpetuating due process deficiencies in future cases. As the court held in U.S. v. Bucey, “Due process requires that penal statutes define criminal offenses with sufficient clarity that an ordinary person can understand what conduct is prohibited.”⁵²

As Justice Kagan argued in her dissent in Yates, the real problem with many purportedly ambiguous laws is overcriminalization.⁵³ As for the obstruction of justice charges in Yates itself, however, she suggested that the problem was not as grave as the majority suggested:

That brings to the surface the real issue: overcriminalization and excessive punishment in the U.S. Code.

Now as to this statute, I think the plurality somewhat—though only somewhat—exaggerates the matter. The plurality omits from its description of § 1519 the requirement that a person act “knowingly” and with “the intent to impede, obstruct, or influence” federal law enforcement. And in highlighting § 1519’s maximum penalty, the plurality glosses over the absence of any prescribed minimum. (Let’s not forget that Yates’s sentence was not 20 years, but 30 days.)⁵⁴

Unlike the law in Yates, however, failure to register as a financial institution is a strict liability crime with no requirement to act knowingly,⁵⁵ and the penalties for noncompliance are typically more severe.⁵⁶

Kagan’s primary complaint with the majority opinion is not, however, the equity of the result (throwing out Yate’s conviction) but rather that achieving that result is, itself, a legislative act forbidden to the court.

But whatever the wisdom or folly of § 1519, this Court does not get to rewrite the law. ‘Resolution of the pros and cons of whether a statute should sweep broadly or narrowly is for Congress.’ If judges disagree with Congress’s choice, we are perfectly entitled to say so—in lectures, in law review articles, and even in dicta. But we are not entitled to replace the statute Congress enacted with an alternative of our own design.⁵⁷

Again, the definition of financial institution in the Bank Secrecy Act is broad but it is not ambiguous. If a court chooses to recast that breadth as ambiguity in order to reinterpret the law as something less Orwellian than Congress’s original text, then the court is merely curing the patient’s illness by killing her with other wounds. The next section focuses on such a misguided attempt to save the BSA from its own breadth by inviting ambiguity where none exists in the text.

II. Ambiguity

In this section we will set aside our “best reading” of the statute and the attendant conclusion that it is absurdly broad, and, in the alternative, we will proceed to a discussion of ambiguity. First, we show how courts and the Treasury Department have repeatedly reached similar conclusions about the problematic breadth of the definition of financial institution but have found ways to narrow the statute’s application by treating this breadth as if it were merely textual ambiguity. We discuss why this invented ambiguity is a convenient article of faith, or ipse dixit, that enables courts and regulators to save the statute from probable unconstitutionality. We go on to discuss, following scholarly research on ambiguity, why this particular contrivance (moving from breadth to ambiguity) allows for the questionable use of several substantive canons of statutory construction: constitutional avoidance, legislative history, and Chevron deference. Furthermore, we find that any attempt to narrow the statute’s application by use of these substantive canons would fail to adequately inform persons of their obligations under the law.

Solving for Breadth by Inventing Ambiguity

In actual practice no one reads the definition of financial institution as encompassing every paid or paying American. Instead, ambiguity in the definition is invented as a remedy to the statute’s breadth. Again, in reality § 5312(a)(2)(R) reads, “any other person who engages as a business in the transmission of funds.” A plain reading would include a person accepting payment for her labor.

Nonetheless, the text of § 5312(a)(2)(R) is sometimes imagined to be something different, as if Congress had actually written, “engages in the business of the transmission of funds.” In this fantasy text, the modifying word “business” is not a general proviso that asks, “Are her funds transmissions business-related or personal?” But rather it is a specific limitation on the types of businesses that qualify a person as a money transmitter. This invented alternative proviso instead asks, “What is her primary business? Barber or money transmitter?” This interpretation suggests that the specific business we are looking for is one where there is intermediation of other people’s transmissions of funds, rather than mere participation in one’s own transmission of funds.

To be clear, Congress could have easily drafted the statute’s text to read, “any person who is in the business of intermediating other people’s monetary transmissions,” or more simply “is in the business of money transmission,” but Congress instead drafted the text as “any person who engages as a business in money transmission.” Nonetheless, agencies and courts have, when it is convenient, invented the narrow interpretation, “in the business of,” despite it not being the best reading of the actual statute’s text.

The Financial Crimes Enforcement Network (FinCEN), the bureau of the Treasury Department that administers Bank Secrecy Act policy through administrative rulings, guidance, and rulemakings at times emphasizes the breadth of the statute, enabling it to reach all manner of persons should their transactions be useful in investigations. In the regulatory definition of “money transmitter,” for example, both the narrow and broad interpretations are present. A money transmitter is “a person that provides money transmission services.”⁵⁸ This is narrow because that sounds like an intermediary; they are “in the business.” But a money transmitter is also, “any other person engaged in the transfer of funds.”⁵⁹ That sounds like everyone who has ever sent or received money.

FinCEN, nonetheless, usually narrows its interpretation of the statute by implying that there is ambiguity in the phrase “as a business,” reading it as “in the business of,” and then proceeding to differentiate between types of businesses that are intermediating other people’s transactions and those that are merely participating in their own transactions.

This arbitrary narrowing is most evident within the “facts and circumstances” limitations that were added to the regulatory definition of money transmitter through rulemaking.⁶⁰ When a business asks FinCEN whether they are or are not a money transmitter (a frequent occurrence given the surprising breadth of both the statutory and regulatory definitions), the agency will issue an administrative ruling that the business either is or is not engaged in money transmission. That ruling will generally point to the regulatory definition’s facts and circumstances limitations in order to engage in a rationalization of why they are or are not within the scope of the definition. The facts and circumstances limitation that most closely approximates our invented ambiguity (not “engages as a business” but “in the business of”) is as follows:

(ii) Facts and circumstances; Limitations. Whether a person is a money transmitter as described in this section is a matter of facts and circumstances. The term “money transmitter” shall not include a person that only … (F) Accepts and transmits funds only integral to the sale of goods or the provision of services, other than money transmission services, by the person who is accepting and transmitting the funds.⁶¹

Herein, our hypothetical barber could find some comfort. While the statutory definition seems to call her a money transmitter for being “engaged as a business in the transmission of funds,” and while the regulatory definition also includes her within “any person engaged in the transfer of funds,” this facts and circumstances limitation seems to carve her out. The barber is accepting or transmitting funds “only integral” to her haircutting business, a service that is something “other than money transmission.”

In practice, however, this is a difficult standard to adjudicate. If the service you provide is, for example, private bill paying services and you do money transmission integral to that service, is private bill paying a service “other than money transmission services?” FinCEN has said no in some cases and yes in others.⁶² Perhaps this lack of consistency is reasonable given the vagaries of the carve-out. After all, in order to define the negative “other than money transmission services” we need to define the positive “money transmission services,” and now we’re back at the original problem: everywhere else in the regulations and statute “money transmission” is broadly defined to cover just about anything. Thus this “only integral to” carve-out from the broad definition seems to rely on people not fitting into the original broad definition to begin with.

Another example: If you are regularly trading and therefore transmitting currency-like instruments for your own financial benefit (e.g. a foreign currency day-trader), are your transmissions “only integral to the sale of goods or the provision of services”? Seemingly no, they are integral to your own profits as a trader acting in your own best interest. Your trades are not providing a service or good to another person, they are a good (or bad) that you personally enjoy. Nonetheless, this personal currency trading activity has been found not to be money transmission within administrative rulings and, despite its ill-fit, the reason offered for the negative ruling is the “only integral to” limitation.

Practice aside, however, it seems contradictory to the plain meaning of the statute that any such facts and circumstances should be relevant. According to Congress a money transmitter includes “any other person who engages as a business in the transmission of funds.” If that’s the statutory law, why should these regulatory facts and circumstances yield narrower results? The answer seems to be that if the agency did not engage in such quasi-legislative narrowing, the surveillance regime would be too broad, ranging into the absurd and unconstitutional. As Treasury has often stated within its own rulemakings,

The definition of “financial institution” in sections 5312(a)(2) and (c)(1) is extremely broad.⁶³

And, with regard to its own regulatory definition:

FinCEN agrees that the breadth of the definition of money transmitter proposed in [this rulemaking] requires limitation to avoid both unnecessary burden and the extension of the Bank Secrecy Act to businesses whose money transmission activities either do not involve significant intermediation or are ancillary to the completion of other transactions.⁶⁴

But, in the very next sentence, note how this breadth—troublesome though it may be—affords Treasury badly needed flexibility to regularly adjust who it chooses to regulate in response to emergent conditions:

[T]he varieties of methods by which funds are transmitted and remitted by persons performing the function of financial intermediary for that purpose, as well as the pace of financial change, make any rigid definition both impossible and inadvisable.⁶⁵

Thus it seems our barber or, indeed, any paid or paying American will need to suffer a certain amount of uncertainty. The broad definition that potentially includes them is needed in order to investigate and prosecute, when occasionally necessary, all manner of illicit activity. But paid or paying Americans need not worry too much; in most cases they will not be included in the regulatory scheme according to individual determinations made with a flexible list of facts and circumstances. Yes, failing to register as a money transmitter is still a felony⁶⁶ and the definition seems to include you. But don’t worry too much about not registering, you can always write a letter to the Treasury and if it’s reasonable they will tell you why you are not captured by the definition. Thus, a statute that is Orwellian for its breadth (everyone must register and surveil their counterparts for the state) is transmuted into one that is Kafkaesque for its (invented) ambiguity (you may or may not be violating a serious law at any moment, but you can always ask the regulator to explain why you are or are not violating the law in this specific case and they will give you the answer, which applies specifically to you and cannot be relied upon by others, and then maybe they will prosecute you or maybe not; maybe you just shouldn’t have asked…).

Indeed, recognizing that a number of people might be captured by the definition, and that the Treasury Department cannot very well conduct a rulemaking for every necessary carve-out, past rulemakings plainly point worried businesses to this administrative ruling procedure:

[the comment from a worried business], like a number of other comments [in this rulemaking process], concerns the application of these rules in specific situations, for example, armored car companies. FinCEN does not believe it is appropriate to resolve those fact specific situations in the context of a general rulemaking, but is willing to consider them in the context of specific, fact based inquiries.⁶⁷

In Part III we will further discuss how and under what authority FinCEN engages in quasi-legislative narrowing (and occasional broadening) of the statutory definition. It is possible that the Treasury is not inventing ambiguity in order to engage in a flexible interpretation of the otherwise broad statutory definition. It is possible, as we shall discuss, that FinCEN is simply redefining the term as it wishes according to power delegated to it by Congress. This distinction between “we are interpreting the BSA as being broad/narrow” and “we are drafting our own definitions under the BSA as broad/narrow” is not made clear in the rulemakings we have reviewed. One plausible interpretation of the agency’s behavior, however, is that the agency is, contrary to our analysis in Part I, reading the BSA as ambiguous rather than broad, and engaging in a process of statutory interpretation to refine that ambiguity into something more definite (that also happens to be more narrow).

Courts have also, at turns, taken this approach to the BSA’s troublesome text. Dealing specifically with the breadth of the Bank Secrecy Act’s potential application, the First Circuit in Anzalone, treated the text as ambiguous rather than broad in order to apply the rule of lenity and find for the defendant, quoting United States v. Bass for support,

[The principle of lenity] is founded on two policies that have long been part of our tradition. First, “a fair warning should be given to the world in language that the common world will understand, of what the law intends to do if a certain line is passed. To make the warning fair, so far as possible the line should be clear.” Second, because of the seriousness of criminal penalties, and because criminal punishment usually represents the moral condemnation of the community, legislatures and not courts should define criminal activity. This policy embodies “the distinctive distaste against men languishing in prison unless the lawmaker has clearly said they should.” Thus, where there is ambiguity in a criminal statute, doubts are resolved in favor of the defendant.⁶⁸

It is no doubt true that the BSA, owing to its extremely broad application, fails to offer the world fair warning of its actual scope of application in practice, and it is no doubt true that the statute carries significant criminal penalties and moral shame for those found to be in violation. However, as we have explained, there is no ambiguity in the Bank Secrecy Act itself. The statute is clear and it is broad, meaning that there will be ambiguity in how the law is actually applied and enforced by the government in practice rather than in what the law actually says. As the court in Anzalone remarked, “The present ambiguity regarding coverage of the Reporting Act and its regulations has been created by the government itself.”⁶⁹ Therefore, contrary to the holding in Anzalone, if the statute is clear and ambiguity is derived from inconsistent executive interpretation, then application of the rule of lenity is inappropriate.⁷⁰ Lenity, in statutory construction, can only be used to resolve ambiguities, not to invent them where they otherwise do not exist.⁷¹

Though one may agree with the outcome in Anzalone—ordinary persons should not be found to be money transmitters—by choosing not to enforce the law in the specific case, the First Circuit is not clarifying the meaning of an ambiguous statute, it is simply adding noise to the randomness of a broad statute’s already uncertain application. The court has, in effect, simply rewritten the law, adding Anzolone’s particular facts and circumstances to FinCEN’s existing list of facts and circumstances limitations.

This practice in the courts is, indeed, prone to becoming just as piecemeal and chaotic as the administrative rulings at the agency. While the First Circuit,⁷² as well as the Seventh Circuit a in similar case,⁷³ chose to narrowly interpret the BSA’s breadth through invented ambiguity, the Second,⁷⁴ Ninth,⁷⁵ and Eleventh⁷⁶ Circuits found no ambiguity and held for the government. What is an ordinary citizen to do? Should the potential application of a federal law with severe criminal penalties and moral condemnation depend upon the answer to the question: in which Circuit does the defendant live? Of course not.

Speaking broadly, should courts be engaged in policymaking under the guise of interpreting ambiguity at all? To offer a glimpse of how fraught this can be, consider U.S. v. Goldberg.⁷⁷ In that case a court decided that it would be appropriate for the judiciary to determine what dollar amount of money transmitted should constitute money transmission as a “business”:

[W]e disagree with the district judge’s view that the indictment failed to substantiate that the defendants engaged in the “business” of dealing in currency because it charged that but a single transaction was involved. Rather, the government charged that the interrupted $200,000 transaction was anticipated by the defendants to be but the first of a series of laundering transactions to take place over the course of several months; that that first $200,000 was less than 10% of the total amount defendants agreed to launder; and that defendants agreed to reduce their 30% laundering fee by more than half for the first $200,000 in anticipation of the fees they would receive for the remainder of the planned $3 million. Thus, while we agree with the district court that the terms “engage as a business” or “deal in” imply that ordinarily proof of more than one transaction would be required, cf. United States v. Tarr, 589 F.2d 55, 59 (1st Cir. 1978) (“deal[ing] in” or “engag[ing] in the business of” firearms, within the meaning of 18 U.S.C. § 922(a)(1) (1982), normally implies more than one isolated transaction, although court could “conceive of a single transaction sufficiently large” to be within statute), we conclude that the indictment alleged that a series of transactions had been agreed on and that it is a permissible inference that persons who agree to launder a total of $3 million over a three-month period for fees totaling some $900,000 are doing business.⁷⁸

That seems reasonable as a policy matter, but is policymaking the appropriate role of a court? Should a court decide a policy matter that could, by consequence, apply or neglect to apply a statute with serious criminal liability to entire swaths of the population? By citing the facts in this case, “$3 Million over a three-month period,”⁷⁹ and holding that the defendant’s actions qualified as money transmission, the court is, in effect, creating a de novo minimum threshold for money transmission in the Second Circuit. Is that the threshold in other circuits now? And while we may have little sympathy for a defendant who is knowingly promising to launder $3 million, that defendant should probably just be charged with money laundering or attempted money laundering, not with failure to register as a financial institution.

Also, note that when the court analyzes “engages as a business”⁸⁰ in the text of the BSA, it cites for authority another statute (regulating gun sellers) that reads “engaging in the business of.”⁸¹ This is, perhaps, the most blatant invention of ambiguity to resolve breadth. The BSA does not say that one has to be “in the business of” money transmission to be regulated as such, it merely says that people “engaged as a business” in money transmission will be regulated. The statute does not ask if the money transmission activity (whether measured in terms of dollar amount or number of transactions) rises to the level of “doing it as a business” it merely asks if some business, perhaps a barber, has engaged in a transmission. Why would the court willingly delve into a policy discussion over “how much is too much?” when a plain reading of the statute clearly includes the defendant in the definition of money transmitter? Probably because that plain reading would also include a kid selling lemonade in that definition and invented ambiguity and a policymaking exercise about minimum dollar thresholds is preferable to such absurd breadth, even though it does violence to our constitutional separation of powers and the predictability of law.

Just as in Morales (loitering with a gang member is a felony),⁸² Marinello (paying the babysitter in cash is obstruction of justice),⁸³ and Yates (throwing a fish overboard is accounting fraud),⁸⁴ the court would rather pretend that Congress did not speak clearly than suggest (as is somewhat obvious) that Congress said something clear and also absurdly broad.

Other courts, faced with this problem do not engage in policymaking but are content to apply the policy decisions made by FinCEN:

Moreover, “money transmitter” is, by definition, limited to certain “facts and circumstances.” 31 C.F.R. § 103.11(uu)(5)(ii). The Financial Crimes Enforcement Network (“FinCEN”) explained in its Federal Register notice that many commenters sought clarification of the definition of money transmitter and objected to any interpretation that would cause businesses that simply transmit funds as part of their other business activities to be categorized as money transmitters. See Amendment to the Bank Secrecy Act Regulations — Definitions Relating to, and Registration of, Money Services Businesses, 64 Fed.Reg. 45, 438, 45, 442 (Aug. 20, 1999) (to be codified at 31 C.F.R. pt. 103). FinCEN agreed that “the breadth of the definition of money transmitter proposed in the Notice requires limitation to avoid both unnecessary burden and the extension of the Bank Secrecy Act to businesses whose money transmission activities either do not involve significant intermediation or are ancillary to the completion of other transactions.” Id. at 45, 442-43. The government has not presented any evidence of significant intermediation here. Thus, we must overturn the 18 U.S.C. § 1957(a) conviction as well.⁸⁵

This also seems like an excellent policy. Money transmission registration and the attendant criminal felony liability for failure to register should only apply to persons who are engaged in “significant intermediation” of other people’s transactions. But setting aside the possibility (discussed in Part III) that this policy is being set under delegated authority to FinCEN from Congress, if this is merely FinCEN’s interpretation of the BSA, why should it carry any weight with the court? Why should FinCEN’s invented ambiguity be acceptable when the court’s is not? Is this an example of interpretive deference to agency interpretation? Would that even be an acceptable usage of deference canons? This is the subject of the next section.

Ipse Dixit

Public skepticism of an impartial judiciary is a perennial concern. In “Fixing Statutory Interpretation,” then-judge Kavanaugh argued that at least some of that skepticism is warranted given that statutory interpretation often hinges on a finding that this or that legislative text is “ambiguous” and can therefore be interpreted in a way that suits the biases of the judge. After explaining why several recent high-profile cases hinged on an arbitrary determination that the relevant text was ambiguous, Kavanaugh writes,

All of these cases came down to what turns out to be an entirely personal question, one subject to a certain sort of ipse dixit: is the language clear, or is it ambiguous? No wonder people suspect that judges’ personal views are infecting these kinds of cases. We have set up a system where that suspicion is almost inevitable because the reality is almost inevitable.

Of course, in characterizing some of these decisions as examples of the problem, I am not in any way suggesting that the judges who authored them acted in an improper or political manner. To the contrary: most judges apply the doctrine as faithfully as possible. But too much of current statutory interpretation revolves around personally instinctive assessments of clarity versus ambiguity, as these cases amply show. It is difficult to make these assessments in a neutral, evenhanded way, or for different judges to reach the same assessments consistently. And even if judges could make threshold findings of ambiguity in a neutral way, they still would have trouble convincing the public that they were acting impartially. It is all but impossible to communicate clarity versus ambiguity determinations in a reasoned and accountable way — especially when those determinations lead directly to the results in controversial cases. Perhaps unsurprisingly, then, over time a number of Supreme Court Justices have expressed frustration with the difficulty — and arbitrariness — of the threshold inquiry.⁸⁶

In the remainder of “Fixing Statutory Interpretation,” Kavanaugh advocates abandoning modes of statutory interpretation that rely on this threshold inquiry, is the text ambiguous? As Kavanaugh puts it,

A number of canons of statutory interpretation depend on an initial evaluation of whether the statutory text is clear or ambiguous. But because it is so difficult to make those clarity versus ambiguity determinations in a coherent, evenhanded way, courts should reduce the number of canons of construction that depend on an initial finding of ambiguity.⁸⁷

Kavanaugh lists three “ambiguity-dependent canons” ripe for removal as part of his project to fix statutory interpretation: constitutional avoidance, legislative history, and Chevron deference.⁸⁸ As we have seen already in the context of the Bank Secrecy Act, as well as in Marinello and Yates, lenity is another ambiguity-dependent canon that we can add to this list. We will now briefly discuss how these canons could be used to “save” the Bank Secrecy Act from its otherwise absurdly broad application and then return to the subject of whether such interpretive acts, laden as they must be with arbitrary and personal determinations from judges, actually improve due process and the rule of law.

Forbidden Canons: Lenity, Constitutional Avoidance, Chevron

Lenity

The “rule of lenity,” also known as “strict construction” is an ancient common law doctrine directing courts to construe statutory ambiguities in favor of criminal defendants.⁸⁹ If the definition of financial institution was, in fact, ambiguous, and if a defendant was being charged with failure to register as a financial institution, we would expect courts to resolve ambiguities within that definition in favor of those defendants. If the ambiguity was whether “as a business” refers to whether someone is engaged in the business of money transmission or else simply transmitting money as a business, then the rule of lenity would direct the court to offer a charitable reading of the statute for the defendant. If she was merely a barber being paid for a haircut, then she is transmitting money as a business but is not in the business of money transmission itself and therefore does not meet the definition of financial institution as interpreted using the lenity canon. As a result, she should not be charged with failure to register as a financial institution.

In the previous section we discussed how some courts have chosen to narrow the Bank Secrecy Act’s application using the rule of lenity while others have not. As we discussed, the First Circuit in Anzalone applied the rule of lenity and held for the defendant despite a transparent admission in the opinion that ambiguity was not present in the statute, but rather derived from the inconsistency of its administration by the government.⁹⁰ Meanwhile, the Second Circuit in Goldberg, a case with a similar fact pattern as Anzalone, refused to apply the rule of lenity and held for the government because it found no ambiguity in the broad but clear language of the BSA: “Since the statutory and regulatory provisions unambiguously cover the defendants’ alleged conduct here, the rule of lenity does not come into play.”⁹¹

As per Kavanaugh’s concerns, these diverging opinions in neighboring circuits could certainly be perceived as bias creeping into the criminal justice system: the judge in Anzalone could be painted as a civil libertarian who is more concerned with overcriminalization than with stopping money laundering, while the judge in Goldberg could be painted as a law-and-order conservative more concerned with stopping crime than the collateral consequences of a large prison population and the abuse of discretion by law enforcement. However, these are unsupported accusations of political bias; the more likely explanation for the divergence in holdings is subtle differences in the facts of each case combined with different appraisals of the Bank Secrecy Act’s level of ambiguity as drafted and as applied. The decision over whether a statute is ambiguous, as Kavanaugh writes, is a matter of “personally instinctive assessments of clarity versus ambiguity.”⁹² It is this inherently dogmatic assertion about ambiguity versus clarity that drives inconsistent holdings, not an improper desire by judges to change policies to better match their political biases. To improve statutory interpretation, says Kavanaugh, we should avoid relying on canons of construction, such as lenity, that rely on an initial finding of ambiguity.⁹³ This would lead to more consistent outcomes in criminal cases involving the Bank Secrecy Act and it would also reveal the true breadth of the Act, something that may necessitate outright constitutional challenges to the law rather than challenges based on ambiguity over the law’s application to a particular defendant.

Constitutional Avoidance

In its original formulation, the constitutional avoidance canon directs courts as follows: “as between two possible interpretations of a statute, by one of which it would be unconstitutional and by the other valid, [a court’s] plain duty is to adopt that which will save the Act.”⁹⁴ A more modern formulation of the constitutional avoidance canon suggests that if there is ambiguity between an interpretation that is constitutional and one that “raises doubts” over the act’s constitutionality, then the court should also proceed by choosing the interpretation of the statute that does not raise such doubts.⁹⁵ Justice Brandeis offers the canonical formulation of this modern approach, “When the validity of an act of the Congress is drawn in question, and even if a serious doubt of constitutionality is raised, it is a cardinal principle that this Court will first ascertain whether a construction of the statute is fairly possible by which the question may be avoided.”⁹⁶ Thus the modern canon seeks to avoid even the question of constitutionality if a plausible interpretation of an ambiguous statute would allow.

As repeated ad nauseum, the Bank Secrecy Act’s definition of financial institution is broad but not ambiguous and therefore offers no opportunity to choose between “two possible interpretations for the statute.” As we have discussed, some administrative as well as judicial rulings have, nonetheless, invented an ambiguity surrounding the meaning of “as a business” within the “money transmitter” sub-definition of financial institution. This artificial ambiguity offers two possible interpretations: (1) financial institutions include persons in the business of money transmission (i.e. third party intermediaries transmitting money on behalf of one or more customers) or (2) financial institutions include persons who are transmitting money as a business (i.e. persons engaged in any business who are transmitting money either as recipient, sender, or a third party intermediary). If one of these interpretations would make the Bank Secrecy Act unconstitutional or (in the modern formula) raise questions about its constitutionality, then the alternative interpretation should be taken by the courts. As we will unpack below, our second interpretation raises clear questions about constitutionality and, indeed, would likely be simply unconstitutional.

A full analysis of the constitutionality of the Bank Secrecy Act is beyond the scope of this paper and we direct the reader to our previous work.⁹⁷ In brief, the Bank Secrecy Act creates a system of warrantless search and surveillance for records of financial transactions. Warrantless searches are generally prohibited under the Fourth Amendment. Nonetheless, the BSA has been found to be constitutional because it only mandates the search and seizure of transactional records that have already been voluntarily entrusted to a third party. Under the third-party doctrine of Fourth Amendment jurisprudence, the subjects of these transactional records (bank customers) have forgone their reasonable expectation of privacy over those records by virtue of handing them over to a third party (their bank) and therefore a warrant is no longer required for the government to obtain those records.⁹⁸

If we take the second and broader interpretation of “as a business” described above, then ordinary persons engaged in non-intermediated money transmission would be required to report the details of their transactions to the government sua sponte and without a warrant. Because these persons are not third parties to transactions—they are one of the two parties transacting directly—it can no longer be maintained that the records being searched have been voluntarily handed over to any third party. There is no third party and, therefore, the warrant requirement of the Fourth Amendment must be observed for governments to obtain those records. The BSA explicitly calls for warrantless reports and recordkeeping and therefore would be unconstitutional as applied to persons not acting as intermediaries.

Therefore, the constitutional avoidance canon directs courts to find that “as a business” within the definition of money transmission must limit the definition of money transmission only to persons providing money transmission services to others (e.g. third party intermediaries) and not to persons transmitting or receiving their own money on their own behalf. To our knowledge no court has made this interpretation although the argument seems strong.⁹⁹

Fear that the Act is unconstitutional, however, may be a driving force behind agency interpretation of the BSA. As briefly discussed earlier, FinCEN has described in rulemaking its own reservations over the breadth of the BSA, and those concerns focus specifically on whether reporting requirements could extend to persons or businesses not engaged in transaction intermediation. FinCEN has found “that the breadth of the definition of money transmitter proposed in [this rulemaking] requires limitation to avoid both unnecessary burden and the extension of the Bank Secrecy Act to businesses whose money transmission activities … do not involve significant intermediation[.]”¹⁰⁰ At no point does FinCEN or any other executive agency suggest that extension of the Bank Secrecy Act beyond intermediaries would be unconstitutional or even an inappropriate interpretation of the statute. However, we should not expect the agency to so clearly point out the limitations of its own power if ambiguity surrounding those limitations is advantageous to maintaining maximum flexibility and discretion in enforcement. It is the Court’s duty to point out and enforce those limitations and it is Congress’s duty to set clear limitations from the start.

Generally speaking, Kavanaugh summarizes the legal academy’s discontent with the constitutional avoidance canon in “Fixing Statutory Interpretation”:

Judge Easterbrook has described “the canon of construing statutes to avoid constitutional doubt” as “wholly illegitimate.” Noting that constitutional “doubt is pervasive,” he explains that the constitutional avoidance canon “acts as a roving commission to rewrite statutes to taste.” As a result, the canon “is simultaneously unfaithful to the statutory text and an affront to both of the political branches.” Likewise, Judge Posner criticizes the canon for “creat[ing] a judge-made constitutional ‘penumbra’ that has much the same prohibitory effect as the judge-made (or at least judge amplified) Constitution itself.”¹⁰¹

Here, as in our discussion of lenity, and the problem of invented ambiguity to address breadth identified by Brennan Marquez, we see that the core concern is judicial overreach and the de facto power of legislation that such interpretive moves afford courts. Especially in the context of a law with far reaching and severe criminal sanctions and moral condemnation such as the Bank Secrecy Act, courts and executive officials should not hold the power to, at will, expand or contract the statute’s applicability according to their personal biases. It is for the democratically elected legislature and the legislature alone to make such difficult policy choices.

Justice Kavanaugh goes on to explain how one might proceed in the absence of a constitutional avoidance canon:

If the constitutional avoidance canon were jettisoned, judges could instead determine the best reading of the statute based on the words of the statute, the context, and the agreed-upon canons of interpretation. If that reading turned out to be unconstitutional, then judges could say as much and determine the appropriate remedy by applying proper severability principles.¹⁰²

In the case of the Bank Secrecy Act’s definition of financial institution, it would follow that a best reading of the statute makes every paid or paying American into a financial institution, requires them to engage in warrantless data collection on behalf of the government, and is, therefore, unconstitutional. Proper severability principles could, at that point, require that the court strike any sub-categories of the financial institution definition that attempt to regulate non-intermediaries. This would mean that, for a time, even intermediaries who once fit within any of those stricken definitions would no longer be required to comply with the Bank Secrecy Act. For a time “money transmitter” would no longer be a category of BSA-regulated financial institutions, but businesses like “insured banks,” of course, would remain subject to the BSA as those categories are well-drafted and inclusive only of true intermediaries.

In turn, given the seriousness of the crimes involved and the real value of financial surveillance techniques to discourage and prosecute those crimes, we can reasonably expect Congress to quickly address any gaps created by a finding of unconstitutionality. Moreover, with the valuable signal sent by a court refusing to improve a bad law through creative interpretation, Congress would know that it needs to work harder to draft a better, narrower definition of financial institution that, likely, would focus on the question of intermediation rather than mere participation in money transmission.

Chevron and Major Questions

Chevron deference is a doctrine of judicial deference compelling federal courts to defer to an agency’s interpretation of an ambiguous statute.¹⁰³ Classically, it is a two step review process. In step one, the court must determine whether the statutory text under interpretation is in fact ambiguous or whether the text is clear and conveys Congress’s expressed intent. If the text is clear, then the agency is not allowed to deviate from that text; it must carry out the expressed intent found in the legislation. If, however, the text is ambiguous, then the court moves to step two.

In step two, the court must determine whether Congress implicitly or explicitly created the aforementioned textual ambiguity. As Justice Scalia observed,

An ambiguity in a statute committed to agency implementation can be attributed to either of two congressional desires: (1) Congress intended a particular result, but was not clear about it [implicit ambiguity]; or (2) Congress had no particular intent on the subject, but meant to leave its resolution to the agency [explicit ambiguity].¹⁰⁴

If the ambiguity is found to be explicit then the court must defer to the agency’s interpretation unless it is manifestly contrary to the statute or is arbitrary and capricious. As Justice Stevens wrote for a unanimous Court in Chevron, “If Congress has explicitly left a gap for the agency to fill, there is an express delegation of authority to the agency to elucidate a specific provision of the statute by regulation. Such legislative regulations are given controlling weight unless they are arbitrary, capricious, or manifestly contrary to the statute.”¹⁰⁵

If the ambiguity is found to be implicit, then, as the Court held in Chevron, “a court may not substitute its own construction of a statutory provision for a reasonable interpretation made by the administrator of an agency.”¹⁰⁶ In practice, step two will always leave room for deference to agency interpretation, merely directing the court to review that interpretation under either a reasonableness standard or an arbitrary and capricious standard.

There is longstanding dissatisfaction with Chevron’s step two inquiry from former and current members of the Court and the legal academy. Justice Scalia argued that while deference to agency interpretation of explicit ambiguities makes sense,¹⁰⁷ deference to Agency interpretation of implicit ambiguities does not: “what we have is genuinely a question of law, properly to be resolved by the courts.”¹⁰⁸

Justice Gorsuch has similarly argued that the step two inquiry transfers all responsibility for statutory interpretation from the courts to the agency:

Of course, some role remains for judges even under Chevron. At Chevron step one, judges decide whether the statute is “ambiguous,” and at step two they decide whether the agency’s view is “reasonable.” But where in all this does a court interpret the law and say what it is? When does a court independently decide what the statute means and whether it has or has not vested a legal right in a person? Where Chevron applies that job seems to have gone extinct.¹⁰⁹

Columbia Law School professor Philip Hamburger agrees that this abnegation of interpretive authority is constitutionally problematic: “When judges defer to agency interpretations, they depart from their judicial office or duty, under Article III of the Constitution, to exercise their own independent judgement.”¹¹⁰

Justice Gorsuch has also noted that Chevron creates instability in statutory interpretation as successive executive administrations are free to reinterpret the law: “The founders were wary of [the] costs [of agency interpretation], knowing that, when unchecked by independent courts exercising the job of declaring the law’s meaning, executives throughout history had sought to exploit ambiguous laws as license for their own prerogative.”¹¹¹

Additionally, in a significant recent decision, King v. Burwell, the Court held that Chevron deference does not apply in cases involving “question[s] of deep economic and political significance.”¹¹² King is an application of the so-called major questions doctrine, which we will discuss below.

Turning to the ostensible ambiguity in the definition of financial institution within the BSA, we can rehearse a Chevron analysis and find practical examples of these critiques of Chevron. Financial institution is defined such that it may have both explicit and implicit ambiguity. Subpart (a)(2)(Y) indicates that some explicit ambiguity was intended by Congress; it allows the Secretary of the Treasury to determine whether persons are performing activities “similar to, related to, or substitute for” the listed activities in the definition, and to include those persons in the definition of financial institution via rulemaking. As we have discussed throughout, however, the particular statutory definition of “money transmitter” at (a)(2)(R) carries potential implicit ambiguity.

There would seemingly be little debate amongst the critical judges and academics noted above that the explicit ambiguity of the term financial institution, which enables Treasury to identify activities that are “similar” “related” or “substitutes for” defined financial institutions, warrants Chevron deference. As Justice Scalia observed with respect to explicit ambiguity generally, “what we have is the conferral of discretion upon the agency, and the only question of law presented to the courts is whether the agency has acted within the scope of its discretion.”¹¹³ A court could review particular agency rulemakings to determine whether the agency has faithfully identified an activity as a substitute for, say, banking, or whether that identification is contrary to the statute, or arbitrary and capricious.

However, with the major questions doctrine in play, even this Chevron outcome is not assured. The list of businesses that perform services “similar to, related to, or substitute for” any of the 26 explicitly defined statutory sub-categories of financial institutions seems to contain multitudes. Chef Julian Serrano’s Picaso restaurant is undeniably “related” to the Bellagio Casino (a financial institution explicitly defined) to which it is attached. FedEx and UPS are, no doubt, substitutes for the United States Postal Service (another explicitly identified financial institution). And the service collectively provided by thousands of individual persons comprising the peer-to-peer Bitcoin network is, no doubt, similar to money transmission. But can the regulator use its explicitly delegated authority to add restaurants, parcel delivery providers, and people running Bitcoin software on their home computers to the list of financial institutions? A mattress’s capacity for cash hoarding is a well understood substitute for a bank, so what of mattress discounters?

In these hypotheticals, the major questions doctrine could intervene in step one of Chevron as it did in Brown and Williamson. In that case the Court found that Chevron would typically apply to an FDA determination that tobacco is a drug and can be regulated accordingly.¹¹⁴ The Court noted that the dispute involved “an administrative agency’s construction of a statute that it administers” and the FDA’s definition of “drug” was explicitly ambiguous to allow for agency gap-filling: “articles (other than food) intended to affect the structure or function of the body.”¹¹⁵ The Court, however, found that classifying tobacco as a drug would have far reaching economic and social consequences and argued “that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion.”¹¹⁶

Similarly, classifying restaurants, parcel delivery providers, and people running Bitcoin software on their home computers as financial institutions may appear to be directly within explicit authority granted by Congress to the agency to regulate activities “similar to, related to, or substitute for” enumerated financial institutions. However, the far reaching economic and social consequences of such a decree might suggest that Congress must have directly intended for some limitation against an absurdly broad category of financial institution, further suggesting that there was never any ambiguity over whether Congress intended super broad application; it did not. Therefore the analysis ends at Chevron step one, the agency interpretation is negated by the direct intent of Congress.

The major questions doctrine could also be used as it was UARG v. EPA to intervene in step two of Chevron. There the Court found that the EPA could not include stationary sources of greenhouse gasses, including apartment buildings, in their framework for emission standards regulation.¹¹⁷ The Court found that the scope of to-be-included emissions sources under agency interpretation “would be inconsistent with—in fact, would overthrow—the Act’s structure and design” owing to the fact that it would lead to a massive increase in permit applications, billions of dollars in administrative costs, and “decade-long delays” that would cause “construction projects to grind to a halt nationwide.”¹¹⁸ The Court argued that these substantial costs and consequences proved that the agency’s interpretation, while answering a real ambiguity per step one, was nonetheless not “reasonable” per step two. The same would be true of requiring restaurants nationwide to register as financial institutions, asking parcel services to report the beneficial ownership information of the packages they deliver, or asking thousands of ordinary persons running Bitcoin software to file suspicious activity reports whenever their computer seems to be under a heavier load than usual.

Just as including all greenhouse gas sources as stationary pollution emitters “would bring about an enormous and transformative expansion in EPA’s regulatory authority without clear congressional authorization,”¹¹⁹ so too would widening the field of financial institutions bring about a transformative expansion of Treasury’s surveillance authority without clear congressional authorization. Just as the burden of regulating every stationary greenhouse gas source would “overthrow” the structure and design of our federal environmental laws, so too would regulating a sizable number of U.S. businesses and individuals overthrow the structure and design of our federal anti-money-laundering laws.

Recently, in West Virginia v. EPA, the Court chose to forgo any Chevron analysis of agency interpretation because the major questions doctrine so clearly applied to a case where the agency “assert[ed] highly consequential power beyond what Congress could reasonably be understood to have granted.”¹²⁰ In that case the EPA was claiming authority under the Clean Air Act to order coal-fired power plants to cease producing electricity. The Court held that the plain language of the statute allowed the agency to mandate the use of technological methods to reduce pollution at plants but did not allow the agency to simply shut down plants altogether. Citing a range of past decisions including Brown and Williamson and UARG, the Court in West Virginia v. EPA articulated a clear and generalizable set of factors for when the major questions doctrine applies: cases where an agency has (1) “claim[ed] to discover in a long-extant statute an unheralded power”, (2) claimed power that “represents a transformative expansion in [its] regulatory authority”, (3) “located that newfound power in the vague language of an ancillary provision of the Act … that was designed to function as a gap filler and had rarely been used in the preceding decades”, and (4) where the Agency’s discovery “allowed it to adopt a regulatory program that Congress had conspicuously and repeatedly declined to enact itself.”¹²¹

These factors would all be met in a case where the Treasury sought to use the Bank Secrecy Act to regulate ordinary persons, cryptocurrency software developers, or other persons not historically understood to be financial institutions as financial institutions under their regulations. We can go through each factor in turn:

1. Agency “claim[ed] to discover in a long-extant statute an unheralded power.” As discussed throughout this report, the Bank Secrecy Act became law in 1976 and has historically been used merely to regulate actual financial institutions, i.e. banks and securities brokers.
2. Agency claimed power that “represents a transformative expansion in [its] regulatory authority.” As we have discussed, the application of bank-like surveillance and recordkeeping requirements to the activities of ordinary persons and non-bank businesses would represent a massive transformation of Treasury’s powers, from an agency that merely supervises a defined category of government-chartered financial intermediaries to an agency that can police and surveils any citizen for any financial dealings and transactions.
3. Agency “located that newfound power in the vague language of an ancillary provision of the Act … that was designed to function as a gap filler and had rarely been used in the preceding decades.” Any hypothetical expansion of Treasury’s surveillance powers would be rooted in the provisions of the definition of financial institution that allow the identification of new institutions as needed to fill gaps. While some reasonable usage of these gap-filling provisions have been made over the decades since Congress passed the BSA, never has the Treasury gone so far as to identify ordinary commercial activities of private citizens as within their jurisdiction.
4. Agency’s discovery “allowed it to adopt a regulatory program that Congress had conspicuously and repeatedly declined to enact itself.” In the case of cryptocurrency activities that are not already covered within the definition of financial institution, individual members of Congress have at various times contemplated both the explicit inclusion of certain activities¹²² as well as safe harbors to explicitly exclude these activities from Treasury’s jurisdiction.¹²³  Much as with the regulation of coal power plants in West Virginia v. EPA however, Congress itself has conspicuously and repeatedly declined to enact any laws that would put non-intermediated cryptocurrency transactions within the ambit of the Bank Secrecy Act.

Even if we set aside the major questions doctrine, we can return to Justice Kavanaugh’s original critique of ambiguity-dependent canons of interpretation generally. If findings of ambiguity are inherently arbitrary and risk politizing or randomizing the use of highly consequential tools for statutory interpretation, then courts should minimize the extent to which they find ambiguity in statutes. Chevron’s first step asks whether the statute is ambiguous and directs courts to ignore a contrary agency interpretation if the statute is non-ambiguous. If we were to apply Chevron to an agency interpretation of the definition of money transmitter, a court could refuse to grant deference by ending the analysis at Chevron step one.

As we have repeated throughout this paper, the definition of money transmitter is not ambiguous at all. It clearly states that both “licensed sender(s) of money” as well as “any other person who engages as a business in the transmission of funds” are money transmitters. How could Congress be more clear? Accordingly, any agency interpretation that narrows or broadens this plain language interpretation would be entitled to no deference. This is problematic from a policy perspective because that plain language interpretation is, in practice, exceedingly broad and would effectively require every paid or paying American to register with FinCEN and surveil and report on the details of financial transactions made with their fellow Americans. As with the constitutional avoidance canon above, however, Kavanaugh’s critique suggests that we should not allow an Agency to save the statute from what Justice Kagan has called “ungodly breadth”¹²⁴ by developing an arbitrary interpretation of legislative text that was never ambiguous at the start. Instead, the Court should interpret the law honestly and, if it is problematically broad, it should overturn the statute on constitutional grounds as described in the previous section.

Invented Ambiguity: Summarizing Agency Power to Broaden or Narrow the BSA

Treasury’s more narrow interpretation of “money transmitter” may not, therefore, be worthy of Chevron deference at all. However, it may be defensible on entirely different grounds. If we assume that (a) explicit ambiguities in the definition of financial institution only allow the agency to broaden the coverage of definition but not to narrow it, and (b) that the explicit definition of money transmitter is, by plain reading, already so broad as to include every paid or paying American, then there is only one remaining provision of the Bank Secrecy Act that empowers Treasury to narrow the application of the law’s onerous and constitutionally-suspect surveillance requirements: § 5318. That section reads:

The Secretary of the Treasury may … prescribe an appropriate exemption from a requirement under this subchapter and regulations prescribed under this subchapter. The Secretary may revoke an exemption under this paragraph or paragraph (5) by actually or constructively notifying the parties affected. A revocation is effective during judicial review.¹²⁵

This language is not, it bears emphasizing, a part of the definition of financial institution found at § 5312;¹²⁶ it is instead found in the section describing discretionary powers granted to the Secretary. Therefore, we have, in total, three ways that the Treasury can adjust the scope of the statute’s application:

1. 1. through agency interpretation of explicit ambiguity at (a)(2)(Y)¹²⁷ and (a)(2)(Z)¹²⁸—the passages allowing the Secretary to determine that some activities are “similar to, related to, or substitute for”¹²⁹ enumerated financial institutions and to regulate businesses whose “cash transactions have a high degree of usefulness”¹³⁰ in investigations,
   2. through agency interpretation of implicit ambiguity at (a)(2)(R)¹³¹—the very broad and allegedly ambiguous definition of money transmitter, or
   3. by prescribing “appropriate” exemptions from the requirements of the statutory scheme as a whole under authority found in § 5318.¹³²

To be clear, option 1 only allows FinCEN to broaden the statute. It capitalizes on explicit ambiguity in the definition of financial institution wherein Congress effectively said, “the definition may be broader than we have stated if the agency finds that certain activities are similar to, related to, or substitute for the activities we’ve already defined, or if the agency finds that certain transactions not already covered in our definition have a ‘high degree of usefulness’ in investigations.”

Option 2 could be used to broaden or narrow the statute because it argues that Congress was not clear about the definition of money transmitter and that the agency can interpret that ambiguity by either broadening or narrowing its interpretation of precisely to whom “any other person who engages as a business in the transmission of funds” refers.

Option 3 allows the Treasury to grant exemptions and can therefore only be used to narrow the application of the law (but the power to revoke exemptions could return the statute’s application to its original breadth). Unlike options 1 and 2, option 3 does not alter the definition of financial institution. It simply says that some financial institutions will remain classified as such but can be granted “appropriate”¹³³ exemptions from compliance with the Bank Secrecy Act.

We have already explained why options 1 and 2 may or may not be entitled to Chevron deference. Broadening the law via option 1 is subject to a Chevron review and Treasury’s interpretations may be upheld so long as they are not arbitrary and capricious or manifestly contrary to the intent of the statute. However, the major questions doctrine may preclude deference if the interpretation offered by Treasury “would be inconsistent with—in fact, would overthrow—the Act’s structure and design”¹³⁴ or if the court found “that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion.” Broadening or narrowing the law via option 2 is also subject to Chevron review but because the ambiguity of (a)(2)(R) is implicit rather than explicit, agency interpretations must meet a reasonableness standard. Here, however, there is a weak case for finding ambiguity in Step One of Chevron as a plain language reading of (a)(2)(R) suggests that the text is broad but not ambiguous. Without ambiguity, the agency is not entitled to any deference. This denial of Chevron deference at Step One would be consistent with Justice Kavanaugh’s preference for avoiding the use of ambiguity dependent canons.

Narrowing the law’s broad application via option 3 (crafting exemptions), is not—unlike options 1 and 2—subject to Chevron review because it is not agency interpretation; it is an agency decision to craft an “appropriate”¹³⁵ exemption from application of the statute to certain already defined financial institutions utilizing powers explicitly delegated to the agency by § 5318.

Having three separate avenues for broadening or narrowing the coverage of a burdensome law that carries felony liability is complicated enough, but the situation is even worse in practice. Historically, whenever the Treasury has engaged in rulemakings, guidance, or administrative rulings that broaden or narrow the scope of the Bank Secrecy Act’s application, it has neglected to specify precisely which of these three very different statutorily granted or implied powers it is exercising.¹³⁶ Typically, it will cite several sections at once to support the general proposition that the Treasury has the power to broaden or narrow the category of financial institutions in order to “more carefully tailor the application of the Bank Secrecy Act to a major, if little understood, part of the financial sector in the United States.”¹³⁷ To defend this flexibility, the agency will often cite some general legislative intent from Congress. For example, “The Congress has long-recognized the need generally to address problems of abuse by money launderers of ‘non-bank’ financial institutions.”¹³⁸ Typically it will not, however, explain upon which of the three sections any particular ruling is reliant. As we have now discussed, and as is summarized in the chart below, each of these passages is distinct and while some may confer authority to Treasury for agency interpretations worthy of Chevron deference some may not.

5312(a)(2)(Y) and (Z)	Can broaden application by finding that certain activities are “substitute, similar, or related” or that certain transactions are useful.	Depends on finding of Explicit Ambiguity	Chevron Review Applies Step One: obvious case of explicit ambiguity in definition that Congress intended agency to resolve
5312 (a)(2)(R)	Can broaden or narrow application by interpreting to whom “any other person who engages as a business in the transmission of funds” refers	Depends on finding of implicit ambiguity	Chevron Review Applies Step One: poor case for ambiguity. Analysis would likely end here with denial of deference
5318	Can narrow application (or return to original breadth) by prescribing or revoking “appropriate” exemptions.	Not ambiguity dependent.	Chevron does not apply because this is a delegation of power to make exemptions, not interpret the law.

 

Again, § 5318 isn’t even a legislative grant of interpretive authority explicit or implicit, so Chevron is not relevant. Instead, it is a clear delegation of authority to make exemptions from compliance with the law. So far, we have looked comprehensively at the breadth and ambiguity infirmities inherent in the Bank Secrecy Act. Now with § 5318 in mind we turn to a discussion of delegation.

III. Delegation

[T]here is no denying the impressive sweep of the authority conferred upon the Secretary by the Bank Secrecy Act of 1970. While an Act conferring such broad authority over transactions such as these might well surprise or even shock those who lived in an earlier era, the latter did not live to see the time when bank accounts would join chocolate, cheese, and watches as a symbol of the Swiss economy. Nor did they live to see the heavy utilization of our domestic banking system by the minions of organized crime as well as by millions of legitimate businessmen.¹³⁹

In the previous section, we argued that the Bank Secrecy Act’s troublesome breadth of application should not be narrowed by a court through the invention of ambiguity and the use of associated substantive canons or through deference to agency interpretation of ambiguity via Chevron. In the section before that, we discussed how the BSA’s breadth raises grave due process concerns that could render the law unconstitutional. If this analysis holds, then there is only one remaining avenue by which the broad statute could be cabined in an effort to remedy due process deficiencies: delegation of authority to the Treasury Department to rewrite its application. As we shall see in this final section, however, this delegation raises its own constitutional infirmities. If all that saves the BSA’s definition of financial institution from unconstitutional breadth is the executive branch’s delegated power to rewrite the statute, then that delegation itself is likely unconstitutional.

What Powers are Delegated in the BSA?

What powers, precisely, are actually delegated to the Secretary of the Treasury in the BSA? As we will discuss throughout this final section, the BSA grants the Secretary the power to:

1. Broaden the Category of Covered Entities: The Secretary may expand the definition of financial institution using powers delegated by § 5312(a)(2)(Y) by determining, “through regulation,” that to-be-included persons are engaged in activities “similar to, related to, or substitute for” activities performed by defined financial institutions, or using § 5312(a)(2)(Z) by “designating” (presumably by regulation or by mere order) “any other business … whose cash transactions have a high degree of usefulness in criminal, tax, or regulatory matters.”
2. Exempt Entities: The Secretary can craft exemptions for otherwise covered financial institutions using powers delegated by § 5318. She “may … prescribe an appropriate exemption” from the requirements of the Bank Secrecy Act and any regulations thereunder promulgated.
3. Expand Collected Data: The Secretary can expand the range of customer data collected by financial institutions using powers delegated by § 5326 if “reasonable grounds exist for concluding that additional recordkeeping and reporting requirements are necessary to carry out the purposes of [the Bank Secrecy Act] or to prevent evasions thereof.”

Each of these powers can be used to collect or cease collection of millions of records of financial transactions from millions of Americans. As a warrantless surveillance regime the BSA almost certainly dwarfs national security letters and FISA court orders.¹⁴⁰ Each of these powers can be wielded by the Secretary whenever she deems an expansion or contraction of the surveillance regime “reasonable,” in the case of §5326, or “appropriate,” in the case of §5318. None of these powers are subject to judicial oversight, a subpoena, or a warrant. The BSA is silent as to whether any of these determinations can be appealed. Before we look at the constitutionality of these delegated powers, we will quickly review some important details of the statutory scheme.

Expansion vs. Exemption of Regulated Financial Institutions

In the previous section we discussed uncertainty over whether FinCEN’s facts and circumstances limitations to the definition of money transmitter (and the associated administrative rulings) were either (a) agency interpretation of an ambiguous definition within the statute, or (b) a new, narrower, definition that the agency has the delegated authority to promulgate. The somewhat strange fact of the matter is that while the BSA delegates tremendous power to broaden the definition of financial institutions, it does not delegate any power to narrow it. It is true that the Treasury can create an “appropriate exemption” from “a requirement” imposed by the BSA, but an exemption from an obligation is not the same as an exclusion from a definition that triggers obligations. Plainly, the text of §5318 says that of those included in the definition, the Treasury can exempt from compliance anyone it deems “appropriate” to exempt.¹⁴¹ It does not say that the Treasury can use this power to exempt in order to change the definitions to include fewer persons.

As such, existing rulemaking and guidance that has narrowed the statutory definition of financial institutions is likely not legal unless we find that the BSA’s definitions are ambiguous (and we do not), or if the narrowing rulemaking or ruling is framed as an exemption from otherwise applicable obligations to comply. The fact of the matter is that this is not how these rulemakings and administrative rulings have been framed.¹⁴² They have been framed as an answer to the questions, “Am I a money transmitter?” or “Am I a financial institution?” rather than “I know I am a money transmitter, but can I be exempted under §5318?” Therefore, we will not even discuss the BSA’s delegation of power to narrow the definition of financial institution because it simply does not exist.

Past narrowings through rulemaking and administrative ruling could be challenged as entirely lacking in statutory authority (because, as discussed earlier, the statute is not ambiguous and Chevron does not apply) rather than as an unconstitutionally delegated legislative power. Of course, these rulings are not likely to be challenged by persons who are happy to have been excluded from the definition of financial institution (because they have found a way out of an onerous surveillance scheme) and not able to be challenged by persons who are denied an exclusion because the statute does clearly permit broadening the category of entities even if it doesn’t permit narrowing.

That said, the powers to (1) expand the definition of financial institutions and (2) offer financial institutions exemptions from obligations are, unquestionably, delegated powers that may be unconstitutional. So too is a third statutory section that we will now briefly characterize.

Expansion of Data Collection

Up to this point we have focused primarily on the definitional and exemption sections of the Bank Secrecy Act in order to uncover the statute’s breadth and purported ambiguity. To analyze the Bank Secrecy Act with respect to the non-delegation doctrine, we will deal with these sections as well, but we will additionally look at a third section: § 5326. Records of certain domestic transactions.¹⁴³ In §5326, the statute gives the Secretary of the Treasury discretion to order any regulated financial institution or group of financial institutions within a “geographic area” to collect records of “any” customer transaction “equal to or greater than an amount which the Secretary may prescribe.”¹⁴⁴ Further, the financial institution can be ordered to maintain those records for a period of time specified by the Secretary, and to report to the Secretary the details of those records “in the manner and to the extent specified in the order.”¹⁴⁵ The only statutory condition on the Secretary’s power to issue these orders is that “reasonable grounds [must] exist for concluding that additional recordkeeping and reporting requirements are necessary to carry out the purposes of this subtitle or to prevent evasions thereof.”¹⁴⁶

While the officiousness of this language goes some distance to make this surveillance-ordering power appear reasonable, formalized, and limited, the section effectively empowers the Secretary to order any financial institution to collect, retain, and report any details about any customer transaction to the Secretary for any specified amount of time. There is no judicial oversight of these orders, there is no ability for financial institutions (let alone their customers) to challenge these orders, and while they are time-limited to 180 days they can be renewed indefinitely. If the order specifies customer information not already collected by the financial institution because it relates to the customer of a bank customer (e.g. a customer of a money transmitting business which, in turn, is a customer of a bank subject to a §5326 order), the statute empowers the Secretary to order the financial institution to demand additional customer and transaction information from their customers and to report to Treasury any failure to supply the additional information.

Finally, these orders can come with a gag order: the statute forbids the officers, directors, employees or agents of subject financial institutions from disclosing “the existence of, or terms of, the order to any person except as prescribed by the Secretary.”¹⁴⁷ The statute does not specify any means by which persons subject to the gag order can challenge the gag order.

Altogether, the BSA grants the Secretary tremendous power to choose who will be deputized to perform warrantless financial transaction surveillance on behalf of the government. The Secretary can hypothetically deputize every American to surveil herself and her transactional counterparties. The Secretary can also arbitrarily (subject only to it being “appropriate”) exempt any of these deputized persons from needing to surveil and report any financial transactions at all. And the Secretary can order these deputized persons to include in their surveillance ambit “any” transaction of any amount “equal to or greater than an amount which the Secretary may prescribe.” The Secretary can do this all while the deputized person is bound by law to keep the order secret. The whole of the Bank Secrecy Act applies (or it doesn’t) subject to the Secretary’s choices. It is a delegated power to legislate on an ad hoc basis who must become an arm of the state’s warrantless surveillance apparatus, who is exempt, and how much data they must collect. Is this delegation constitutional?

Gundy: Poultry Revived

The delegation of legislative power from Congress to the Secretary of the Treasury described above is unprecedented in its scope and degree of discretion. Nonetheless one would be justified in wondering whether that matters. Other potentially relevant constitutional deficiencies are equally apparent: the lack of any particularized warrant for this data collection (as would be required under the Fourth Amendment), the lack of a third party if individuals are deputized to surveil their immediate counterparties (as would otherwise absolve the search from the warrant requirement under the third-party doctrine), the First Amendment implications of an unchallengeable gag order, and the Fifth Amendment implications of a statute that could be wielded to demand that suspects reveal their own transactional details to investigators.

The non-delegation doctrine, at least as compared with the Fourth and First Amendments, is a branch of constitutional law that has seemingly atrophied rather than strengthened over the course of twentieth century jurisprudence. Any first year law student knows that “all” legislative power is “vested”¹⁴⁸ in Congress and cannot be delegated from Congress to the executive branch, but any second year law student knows that the last time the Court struck down a delegation was in 1935 in the Schechter Poultry case.¹⁴⁹ Non-delegation has been a non-doctrine for almost exactly as long as beer has come in cans and Social Security checks in the mail. Until June of 2019 we would have agreed that a non-delegation challenge to the Bank Secrecy Act would be a non-starter, but then Gundy happened.

Before we look at the Gundy case specifically and what it means for delegations such as those in the BSA, let’s briefly review the state of the non-delegation doctrine before Gundy.

At its heart, the non-delegation doctrine is simply a part of our constitutional separation of powers. Article I of the Constitution vests “all legislative powers” in Congress.¹⁵⁰ The negative implication of this vesting is that legislative power cannot, therefore, reside in the executive or judicial branches. Further, because the Constitution is the highest law of the land, Congress cannot simply divest itself of that exclusive power by passing a law that hands legislative power over to another branch in a delegation.

This simplistic summary of the non-delegation doctrine ignores the metaphysical difficulty of determining when a power is, in fact, legislative rather than executive or judicial.¹⁵¹ It also ignores the fact that some minor legislative delegations may be imperative for the executive to perform its Constitutional duties.¹⁵² The Court has, nonetheless, always recognized some form of the non-delegation doctrine, where things get complicated is in these more nuanced questions. To abbreviate the multi-century history of non-delegation, we will simply point out two tests that emerged in important cases, Wayman and its progeny, and Hampton and its progeny.

Wayman and Progeny: Detail-filling, Fact-finding, and Non-legislative Delegations

In Wayman,¹⁵³ Chief Justice Marshall set out the original articulation of a non-delegation doctrine, explaining that there are “important subjects [of law and policy], which must be entirely regulated by the legislature itself,” but also “those of less interest, in which a general provision may be made [by Congress], and power given to those who are to act . . . to fill up the details.”¹⁵⁴ This “fill up the details” standard was refined in subsequent cases to include the validation of statutes that called on the executive to fill in rather minute details of a statute, for example the particular design of tax stamps for margarine packaging,¹⁵⁵ as well as more consequential details, such as the details of what specific “use and occupancy” rules are needed to protect public forests from “destruction” and “depredations.”¹⁵⁶

As Justice Gorsuch highlights in his dissent in Gundy,¹⁵⁷ that line of detail-filling non-delegation analysis is aptly summarized in Yakus, where the court asks whether the language of the delegation is “‘sufficiently definite and precise to enable Congress, the courts, and the public to ascertain’ whether Congress’s guidance has been followed.”¹⁵⁸ Thus a non-delegation analysis following Wayman and Yakus would require that Congress offer clear instructions that explain both (a) to what end power should be used (e.g. it should be used to protect forests from destruction and degradation), as well as (b) by what means power should be used (e.g. it should be used to create use and occupancy rules). Further, as per Yakus, the results of any exercise of power should be comparable to those clear instructions such that both the legal profession but also the lay public can ascertain whether Congress’s guidance has been followed. For example, a restriction on atmospheric polluters on neighboring private land would not be a use and occupancy rule for the public forest being regulated. Similarly, an occupancy rule meant to increase housing supply in a neighboring town would not be a rule made to prevent forest destruction and depredations. In both these hypotheticals we can judge the choices of the executive because Congress set out reasonably precise standards for means and ends when it delegated power. With means and ends specified all that is left to the executive is detail-filling.

Additionally, early cases allowed the legislature to make a law’s application conditional on executive “fact-finding.” For example, Congress passed a law enabling the construction of the Brooklyn Bridge if and only if the executive found that such a bridge would not interfere with navigation of the East River.¹⁵⁹ When this fact-finding condition was challenged as an impermissible delegation of legislative power, the Court disagreed, holding that the law “simply declared that, upon a certain fact being established, the bridge should be deemed a lawful structure, and employed the secretary of war as an agent to ascertain that fact.”¹⁶⁰ This is not a detail-filling example because there are no details to fill in with regard to policy choices. The only thing that is authorized is a bridge; there’s no range of policy choices left to the executive. The statute does not say, for example, create a link between Manhattan and Queens how you see fit, be it by ferry system, subterranean rail tunnel, aerial tram, or similar conveyance. It is merely a fact-finding contingency on Congress’s explicit choice: build a bridge if you find that it would not impede navigation. Nonetheless, the Yakus standard could be applicable here as well. We have a delegation that is specific enough that a member of the public could read it and determine the executive’s obedience: if, for example, the executive decides not to build a bridge for other reasons after concluding that navigation would be not be impaired, then it has acted legislatively outside the clearly delegated authority.

Finally, early cases allowed Congress to delegate powers to the executive and judicial branches when those delegated powers were already constitutionally granted powers to the executive or judicial branches respectively. For example, Congress can delegate the quasi-legislative power to regulate courtroom practice to Article III Courts, which already have plenary authority over the judicial process. Similarly, Congress can delegate quasi-legislative power to negotiate treaties and tariffs to the executive who already has plenary Constitutional power over the conduct of foreign affairs.

Justice Gorsuch neatly summarizes these three acceptable delegations from original non-delegation case law as, detail-filling, fact-finding, and non-legislative delegations, and argues that up until the mid twentieth century, only delegations of these sorts would be permitted by the Court.

Hampton and Progeny: An “Intelligible Principle”

In Hampton, Chief Justice Taft largely engaged in the aforementioned Wayman-derived analysis. However, he summarized his holding by remarking that in “lay[ing] down by legislative act an intelligible principle to which the [executive official] is directed to conform” Congress avoids making an unconstitutional delegation.¹⁶¹ As with many pithy turns of phrase, this “intelligible principle” language eventually came to subsume the prior full analysis; it became a useful (if perhaps impoverishing) analytical shorthand.

Courts through the twentieth century began truncating their full separation of powers analysis into a simpler question: does the statute lay down an “intelligible principle?” This simplified and relaxed standard for delegations matched well with a growing cultural and political presumption that citizen legislatures could not be composed of technical experts, but that the essential nature of governing in the modern world nonetheless required technical expertise.¹⁶² Accordingly, Congress could not be expected to learn the science of air pollution or nutrition and should, therefore, delegate power to executive branch agencies in order to form better and necessarily technocratic policies.

Justice Gorsuch does not believe that the “intelligible principle” doctrine supplanted the earlier Wayman-derived standards of detail-filling, fact-finding, and non-legislative delegation.¹⁶³ Rather he argues that the “intelligible principle” shorthand has simply impoverished the analysis and, in a few cases, led to the upholding of some delegations that should not have been acceptable under traditional analysis. Gorsuch, in his Gundy dissent, is very transparently signaling that he would prefer a return to more rigorous non-delegation analysis, such as is found in the standard in Yakus, where the court asks whether the language of the delegation is “‘sufficiently definite and precise to enable Congress, the courts, and the public to ascertain’ whether Congress’s guidance has been followed.”¹⁶⁴

Gorsuch was joined in his dissent by Justice Thomas and Chief Justice Roberts. Justice Alito wrote a concurrence to the majority where he suggested a willingness to engage in more strict analysis if a majority of the Court went along. Justice Kavanaugh joined the Court seven days after oral arguments were heard and therefore did not take part in the decision. In a statement published with the Court’s November 25, 2019 orders, Justice Kavanaugh specified that Justice Gorsuch’s analysis of the non-delegation doctrine in Gundy “may warrant further consideration in future cases.”¹⁶⁵ If, as appears likely in some future delegation case, Justices Kavanaugh and Alito were to join the dissenters in Gundy, then the balance of the court would shift in favor of a more rigorous non-delegation review. We will therefore proceed with a look at Gorsuch’s dissent in Gundy in order to examine how the three delegations in the BSA (to expand the definition of financial institutions, to offer exemptions, and to expand the range of data collected) might fair if they are challenged and adjudicated by the current Court.

Gundy

In Gundy, section 20913 of the Sex Offender Registration and Notification Act (SORNA) was challenged as an unconstitutional delegation of legislative authority to the Attorney General.¹⁶⁶ SORNA, in general, requires convicted sex offenders to register for a national database.¹⁶⁷ Section 20913(d), specifically, empowers the Attorney General to determine how and whether “pre-Act offenders” (persons convicted of a sex crime before SORNA’s passage) should be made to register.¹⁶⁸

In rough summary, the argument that this is not an impermissible delegation, favored by the plurality, is that the Attorney General is not permitted by law to choose whether or not pre-Act offenders should be made to register because the statute is meant to be “comprehensive.”¹⁶⁹ Therefore, despite the open-ended language at 20913(d), Congress could not have intended to leave such a substantial number of offenders outside of its scope, and could not be presumed to leave such a substantial and legislative choice up to the unfettered discretion of the Attorney General. The suspect passage in 20913(d), argues the plurality, should be interpreted as ordering the Attorney General “to apply SORNA to pre-Act offenders as soon as he thought it feasible to do so.”¹⁷⁰ Thus the plurality claims that there is an intelligible principle by which the Attorney General is bound: he must apply registration requirements to pre-Act offenders as soon as feasible. It bears noting that the actual language of 20913 has no such “as soon as feasible” condition,¹⁷¹ but the plurality argues that support for this interpretation can be found in the statute’s declaration of purpose:

So begin at the beginning, with the “[d]eclaration of purpose” that is SORNA’s first sentence. §20901. There, Congress announced … that “to protect the public,” it was “establish[ing] a comprehensive national system for the registration” of “sex offenders and offenders against children.” §20901. The term “comprehensive” has a clear meaning—something that is all-encompassing or sweeping. … That description could not fit the system SORNA created if the Attorney General could decline, for any reason or no reason at all, to apply SORNA to all pre-Act offenders.¹⁷²

The counter-argument, favored by the dissenting justices, is that section 20913(d) is an unconstitutional delegation of legislative power because a plain reading interpretation of that section does, in fact, empower the Attorney General to choose whether or not the statute should apply to pre-Act offenders and offers no such feasibility standard or other intelligible principle to be used by the Attorney General in making that choice. As Justice Gorsuch’s dissenting opinion describes:

For all half-million pre-Act offenders, the law says only this, in 34 U. S. C. §20913(d): “The Attorney General shall have the authority to specify the applicability of the requirements of this subchapter to sex offenders convicted before the enactment of this chapter . . . and to prescribe rules for the registration of any such sex offender.” Yes, that’s it. The breadth of the authority Congress granted to the Attorney General in these few words can only be described as vast.¹⁷³

In response to the plurality’s expansive use of statutory construction to limit the scope of the delegation by reference to SORNA’s declaration of purpose, Justice Gorsuch argues:

Unable to muster a feasibility standard from the only statutory provision addressing pre-Act offenders, the plurality invites us to hunt in other and more unlikely corners. It points first to SORNA’s “[d]eclaration of purpose,” … In fact, this provision doesn’t purport to guide the Attorney General’s discretion at all. Instead, it simply declares what Congress believed the rest of the statute’s enacted provisions had already “establishe[d],” without the need for any action by the Attorney General. And by now surely we must all agree that broad and sweeping statements like these about “a statute’s ‘basic purpose’ are . . . inadequate to overcome the words of its text regarding the specific issue under consideration.” While those adopting SORNA might have declared that they hoped and wished for a “comprehensive national system,” the fact remains that the law they actually adopted for pre-Act offenders leaves everything to the Attorney General. Hopes and dreams are not laws.¹⁷⁴

Going further, Gorsuch argues that even if said feasibility standard existed, it might not be sufficiently clear as a condition for legislative delegation to, as per Yakus, enable the public to determine the Attorney General’s fidelity to Congress’s instructions:

But even this new dream of a statute wouldn’t be free from doubt. A statute directing an agency to regulate private conduct to the extent “feasible” can have many possible meanings: It might refer to “technological” feasibility, “economic” feasibility, “administrative” feasibility, or even “political” feasibility. Such an “evasive standard” could threaten the separation of powers if it effectively allowed the agency to make the “important policy choices” that belong to Congress while frustrating “meaningful judicial review.” And that seems exactly the case here, where the Attorney General is left free to make all the important policy decisions and it is difficult to see what standard a court might later use to judge whether he exceeded the bounds of the authority given to him.¹⁷⁵

With this dissenting opinion in mind, let us now turn to the delegations within the Bank Secrecy Act: expansion of the definition of financial institution, exemption from compliance, and expansion of the scope of collected data.

Expansion of the Definition of Financial Institution

The Secretary may expand the definition of financial institution using powers delegated at § 5312(a)(2)(Y) by determining, “through regulation,” that to-be-included persons are engaged in activities “similar to, related to, or substitute for” activities performed by defined financial institutions, or using § 5312(a)(2)(Z) by “designating” (presumably by regulation or by mere order) “any other business … whose cash transactions have a high degree of usefulness in criminal, tax, or regulatory matters.”¹⁷⁶

Of our three delegations, this one has the most statutory text to work with. Can we, per Justice Gorsuch, frame this as a detail-filling, fact-finding, or non-legislative delegation?

As with SORNA, this passage grants sweeping power to the executive branch. As with the half-million Americans who may or may not be subject to sex-offender registration, there are millions upon millions of Americans who may be classified—under § 5312(a)(2)(Y)—as financial institutions, pending a determination by the Secretary that they are engaged in activities “similar to, related to, or substitute for” activities performed by any of the other twenty-some, already defined financial institutions. As thoroughly discussed in the first two sections of this report, the definition of money transmitter (one species of these defined financial institutions) is already broad enough to potentially encompass every paid or paying American. If you expand that already broad definition to anyone whose activities are “similar to, related to, or a substitute for … any … person who engages as a business in the transmission of funds” then the category of persons who may be subject to the substantial surveillance and reporting obligations of the BSA dwarfs the number of pre-Act offenders in limbo with regard to SORNA.¹⁷⁷

The situation is even worse with respect to § 5312(a)(2)(Z). It requires no creative reasoning or strained statutory interpretation to argue that any and every American business has cash transactions that would “have a high degree of usefulness in criminal, tax, or regulatory matters.”¹⁷⁸ Every business in the country could, therefore, be a mere “designation” away from being swept (or not swept) into the scope of the BSA’s surveillance strictures at the Secretary’s discretion.

Unlike SORNA, where the decision to include or not include persons within the statutory scheme is entirely unqualified (except by a feasibility standard invented by the plurality), the decision to identity a person or group of persons as a financial institution under § 5312(a)(2)(Y) must (a) be made through regulation (subject to notice and comment and the other strictures of the Administrative Procedures Act) and (b) can only extend to persons “similar to, related to, or a substitute for” other defined financial institutions. This begins to look like detail-filling rather than unfettered legislative prerogative. As with the details for how the executive can develop “occupancy and usage” rules to prevent the “destruction” and “depredation” of public forests, perhaps it is acceptable for the Secretary to determine the details of who, amongst a range of persons, is acting in a manner similar to a bank or money transmitter in order to prevent money laundering through banks and bank substitutes.

While this is a credible argument, it goes too far. First, this argument relies on our ability to countenance that the following is a mere detail in the statutory scheme that is the Bank Secrecy Act: whether every or any American should be deputized to surveil and report to the government the details of her financial transactions without a warrant or other due process. Returning to the original articulation of the detail-filling jurisprudence of non-delegation, Justice Marshall made a distinction between “important subjects, which must be entirely regulated by the legislature itself,” and “those of less interest, in which a general provision may be made, and power given to those who are to act . . . to fill up the details.”¹⁷⁹ What exactly is “of less interest” may surely be debated in several non-delegation contexts, but the decision to subject millions of Americans to a warrantless mass surveillance regime hardly seems anywhere near that category; it is, self-evidently, an “important subject.”

Second, detail-filling delegations are subject to the test articulated in Yakus, “that Congress must set forth standards ‘sufficiently definite and precise to enable Congress, the courts, and the public to ascertain’ whether Congress’s guidance has been followed.”¹⁸⁰ Does the language “similar to, related to, or a substitute for” have sufficient precision to enable the public to ascertain whether Congress’s guidance is being followed when new entities are classified as financial institutions? As Parts I and II of this report have cataloged, empirically the answer is almost certainly no. Of particular note is the history of statements made by FinCEN with regard to its own statutory authority as well as several bewildered and concerned citizen commentators in rulemakings that have sought to expand the category pursuant to that authority.

As Treasury has often stated within its own rulemakings:

The definition of ‘‘financial institution’’ in sections 5312(a)(2) and (c)(1) is extremely broad.¹⁸¹

And, with regard to its own regulatory definition:

FinCEN agrees that the breadth of the definition of money transmitter proposed in [this rulemaking] requires limitation to avoid both unnecessary burden and the extension of the Bank Secrecy Act to businesses whose money transmission activities either do not involve significant intermediation or are ancillary to the completion of other transactions.¹⁸²

And, in response to a concerned business offering comment in a rulemaking to expand the definition:

[the comment from a worried business], like a number of other comments [in this rulemaking process], concerns the application of these rules in specific situations, for example, armored car companies. FinCEN does not believe it is appropriate to resolve those fact specific situations in the context of a general rulemaking, but is willing to consider them in the context of specific, fact based inquiries.¹⁸³

And there is a library with dozens of said fact-based inquiries where FinCEN has had to determine, according to its own “facts and circumstances” limitations, whether some business is or is not a money transmitter and subject to the law.¹⁸⁴

It is clear that the statute has plenty enough powerful rulemaking or ruling authority to go around with respect to who can be classified as a financial institution, but there is little in the way of clear statutory limitations that FinCEN can utilize to justify its decisions and avoid over-application. Indeed FinCEN has, quite reasonably, made up several limitations out of whole cloth. For example, the statute says nothing about financial institutions being a category limited to persons performing activities involving “significant intermediation” as described above. The Fourth Amendment might provide such a limitation in the form of a warrant requirement for searches of information not held by a third party (i.e. intermediation), but the statute does not. FinCEN’s rules provide limitations in the form of “facts and circumstances” that exclude persons from the otherwise capacious definition of money transmitter, but the statute never mentions any of those facts or circumstances.

Limitations abound in FinCEN’s regulations, but none of them are in the statute. Indeed, if the definition of money transmitter is broad but not ambiguous, as we argue in Part I, then these limitations have been extralegally crafted in contravention of the statute. As we are discussing, there is a delegated power to broaden the definition of financial institution but there is no mention in the text of any delegated authority to narrow it. A plain reading of the text indicates that Congress neglected to set any explicit limit on Treasury’s delegated authority to broaden the definition.

How do we compare the “facts and circumstances” limitations, or the several administrative rulings clarifying the scope of “money transmission” to the specified delegation standard in the statute in order to determine if the executive is remaining faithful to the stated will of Congress and is merely detail-filling? That standard, again, is that only those who are “similar to, related to, or a substitute for” a financial institution shall be included in the definition. How do we make these comparisons with respect to rulemakings that neglect to specify whether they are interpreting ambiguous language in an existing definition or else utilizing that delegated power to expand the definition? We cannot make these comparisons in any meaningful way because the standard is not “sufficiently definite and precise to enable Congress, the courts, and the public to ascertain’ whether Congress’s guidance has been followed.”¹⁸⁵

The standard for expansion of the definition under § 5312(a)(2)(Z) is even worse. If it is true that everyone’s transactions would be useful to a tax investigation, then the statute grants the Secretary the power to apply or not apply the BSA’s requirement to anyone at all.

Exempt Entities

The Secretary can craft exemptions for otherwise covered financial institutions using powers delegated at § 5318. She “may … prescribe an appropriate exemption” from the requirements of the Bank Secrecy Act and any regulations thereunder promulgated.¹⁸⁶

As with SORNA, this exemption granting power is utterly unconstrained by the statute. Of the millions of persons to whom the BSA may apply, the Secretary is free to exempt any or all of them from compliance altogether. The only condition on that power to selectively free people from a burdensome surveillance regime is that it must only include “appropriate” exemptions. Needless to say, “appropriate” is not defined in the statute, and no further standards are specified. Merriam-Webster defines appropriate as “especially suitable or compatible,”¹⁸⁷ but Merriam-Webster cannot tell us what the statute also neglects to specify: suitable or compatible with whom? To what? We hesitate to continue this analysis because it is absurd.

How could any exemption granted under this section be judged by “Congress, the courts, and the public”¹⁸⁸ to determine whether Congress’s instructions are being followed? As with the plurality in Gundy, we would have to search elsewhere in the statute for a meaningful limitation on this power, for a gloss on what is meant by “appropriate.” Short of finding one, this is legislative power pure and simple: it is the power to arbitrarily and at will relieve private parties from surveillance obligations to which they would otherwise be subject under the law.

Could these exemptions be granted in order to do detail-filing or fact-finding? Perhaps they could be either. Perhaps the Secretary might uncover facts that show that some particular financial institution is somehow immune from money laundering thanks to a radical new technology, and that, as such, it would be appropriate to exempt it from BSA obligations. But this standard could also be used to craft an exemption that was “appropriate” because it saved the agency money and minimized the burden to taxpayers, or appropriate because it was made via some arbitrarily formalized process developed by FinCEN for granting exemptions: The Appropriate Process. Appropriateness, especially without a given frame of reference, is not a standard. And determining what is and is not “appropriate” in a vacuum is not detail-filling or fact-finding; it is legislating.

The record of past exemptions also does not improve our confidence that there is some standard set down by Congress by which we might judge appropriateness. In 2002, FinCEN granted a “temporary exemption” from establishing anti-money laundering programs to all of the following businesses: “dealers in precious metals, stones, or jewels; pawnbrokers; loan or finance companies; private bankers; insurance companies; travel agencies; telegraph companies; sellers of vehicles, including automobiles, airplanes, and boats; persons engaged in real estate closings and settlements; certain investment companies; commodity pool operators; and commodity trading advisors.”¹⁸⁹

The stated reason for the exemption was that it was “practical” because of the pervasive lack of precision in existing definitions and the possibility for surprise and confusion if obligations are applied to everyone who appears to fit within those broad definitions. Of those exempted FinCEN said,

[A]lthough included within the list of financial institutions in the BSA, these businesses have never been defined for purposes of the BSA. For example, does a ‘dealer in precious metals, stones, or jewels’ include a jewelry counter at a department store and a kiosk in a shopping mall that sells gold and silver earrings, bracelets, and necklaces, as well as a diamond merchant? Similarly, does ‘a business engaged in vehicle sales, including automobile, airplane, and boat sales’ include businesses selling motorcycles, motorbikes, or snowmobiles? Treasury and FinCEN do not believe it is sound regulatory policy to subject the broad categories of BSA “financial institutions” to the requirements of BSA section 5318(h)(1) without specifically defining the businesses that will be subject to those requirements.¹⁹⁰

Appropriate? Sure. FinCEN also said that the exemption was necessary (and perhaps therefore “appropriate”) because they needed more time to “identify the nature and scope of the money laundering or terrorist financing risks associated with these businesses.”¹⁹¹ FinCEN was rightly concerned about the impact of rushing to apply the BSA’s requirements to potentially vulnerable small businesses:

An inadequate understanding of the affected industries could result in poorly conceived regulations that impose unreasonable regulatory burdens with little or no corresponding antimoney laundering benefits. Finally, Treasury and FinCEN are aware that many of these financial institutions are sole proprietors or small businesses, and that any regulations affecting them must recognize this fact.¹⁹²

All of this sounds very appropriate, but is it what Congress wanted? Is it detail-filling or fact-finding? Or is it a legislative decision to “temporarily” avoid over-application of an extremely costly regulatory regime to a large and vulnerable section of the American economy?

To our knowledge this temporary exemption persists to this day, some 18 years later.¹⁹³ As Legal Counsel for Global Financial Integrity Heather A. Lowe testified before the House Financial Services Subcommittee on Financial Institutions and Consumer Credit on June 28, 2017:

Technically, persons involved in real estate closings are already classified as FIs per the definition established by the USA PATRIOT Act in 2001, but they were given a “temporary exemption” (which had no sunset clause) from AML compliance requirements in 2002. Despite Treasury conducting a comment period with respect to AML compliance in the real estate sector in 2003, they have not removed that temporary exemption. Congress should consider doing so.¹⁹⁴

Respectfully, perhaps Congress should also consider doing its job to reasonably specify the limits of a warrantless financial surveillance dragnet, instead of delegating the arbitrary power to make carve-outs from a disastrously broad definition to FinCEN.

Expand Collected Data

The Secretary can expand the range of customer data collected by financial institutions using powers delegated at § 5326 if “reasonable grounds exist for concluding that additional recordkeeping and reporting requirements are necessary to carry out the purposes of [the Bank Secrecy Act] or to prevent evasions thereof.”¹⁹⁵ Here again we have a delegation that lacks sufficient limitation or instruction to qualify as detail-filling. The purposes of the BSA are as follows: “to require certain reports or records where they have a high degree of usefulness in criminal, tax, or regulatory investigations or proceedings, or in the conduct of intelligence or counterintelligence activities, including analysis, to protect against international terrorism.”¹⁹⁶

As with § 5312(a)(2)(Z), if “reasonable grounds” means simply that the records have a “high degree of usefulness” in a tax investigation, then any and all records from any taxpayer at any time would seemingly be fair game. As with our previous delegations, this is not a power to determine some regulatory detail “of less interest” or to find facts upon which the law’s application might be contingent. This is the heart of the regulatory scheme: the power to determine, at will, if certain transaction details are or are not to be handed over to the government without a warrant or judicial review. Furthermore, in the case of § 5326, it is the power to demand that information in secret, subject to a gag order that has no specified avenue for challenge in the statute.

Justice Gorsuch offers a comprehensive review of the evolution of the non-delegation doctrine in his dissent in Gundy. While reviewing Touby v. United States, a more recent case in the “intelligible principle” lineage from Hampton, he offers a particularly clear articulation of the doctrine as a whole:

Touby may have at least begun to point us back in the direction of the right questions. To determine whether a statute provides an intelligible principle, we must ask: Does the statute assign to the executive only the responsibility to make factual findings? Does it set forth the facts that the executive must consider and the criteria against which to measure them? And most importantly, did Congress, and not the Executive Branch, make the policy judgments?¹⁹⁷

Throughout this report we have recounted all the many ways that the Bank Secrecy Act, owing to its incredible breadth, has been interpreted and reinterpreted, expanded and contracted by the courts and by FinCEN. Often these have been good and necessary decisions made to avoid over-application of a dangerously invasive warrantless surveillance regime, to literally prevent our otherwise inevitable drift towards a society where everyone is obligated to spy-on and rat-out everyone else. These are not, however, factual findings. These are not decisions made according to some limiting and sensible criteria set out in the statute. These are policy judgements of serious gravity and they’ve been made by the executive branch with powers unconstitutionally delegated.

The constitutionality of the Bank Secrecy Act was challenged on First, Fourth, and Fifth Amendment grounds not long after its passage in 1971. In California Bankers Association v. Shultz a divided court found that, as applied, the BSA did not violate the First, Fourth, or Fifth Amendment rights of the plaintiffs.¹⁹⁸ Justice Brennan joined Justice Douglas in a fierce critique of the plurality’s decision to effectively deny a citizen’s reasonable expectation of privacy over their bank records:

Since the banking transactions of an individual give a fairly accurate account of his religion, ideology, opinions, and interests, a regulation impounding them and making them automatically available to all federal investigative agencies is a sledge-hammer approach to a problem that only a delicate scalpel can manage. Where fundamental personal rights are involved — as is true when as here the Government gets large access to one’s beliefs, ideas, politics, religion, cultural concerns, and the like — the Act should be “narrowly drawn” to meet the precise evil. Bank accounts at times harbor criminal plans. But we only rush with the crowd when we vent on our banks and their customers the devastating and leveling requirements of the present Act. I am not yet ready to agree that America is so possessed with evil that we must level all constitutional barriers to give our civil authorities the tools to catch criminals.¹⁹⁹

But Justice Brennan also offered a separate dissenting opinion where he pointed toward an even greater constitutional infirmity than chilled speech from surveillance or warrantless data collection, the indefinite delegation of legislative power:

Although the Secretary has by regulation limited the meaning of “monetary instruments,” 31 C.F.R. § 103.11, and invoked the section only where the transaction involves more than $10,000, see 31 C.F.R. § 103.22, this in no way alters the fundamental vice of the statute.

That vice, see concurring opinion in United States v. Robel, supra, is the delegation of power to the Secretary in broad and indefinite terms under a statute that lays down criminal sanctions and potentially affects fundamental rights. See Bantam Books, Inc. v. Sullivan, 372 U.S. 58 (1963); Cantwell v. Connecticut, 310 U.S. 296, 304-307 (1940). My view in Robel applies here:

“Formulation of policy is a legislature’s primary responsibility, entrusted to it by the electorate, and to the extent Congress delegates authority under indefinite standards, this policy-making function is passed on to other agencies, often not answerable or responsive in the same degree to the people. ‘[S]tandards of permissible statutory vagueness are strict . . .’ in protected areas. NAACP v. Button, 371 U.S., at 432. ‘Without explicit action by lawmakers, decisions of great constitutional import and effect would be relegated by default to administrators who, under our system of government, are not endowed with authority to decide them.’ Greene v. McElroy, 360 U.S. 474, 507.” 389 U.S., at 276.

In the case of the Bank Secrecy Act, also potentially involving First, Fourth, and Fifth Amendment rights of the vast majority of our citizenry, it exceeds Congress’ constitutional power of delegation to empower the Secretary of the Treasury to require whatever reports and records he believes to be possessed of a “high degree of usefulness” where the purpose is to further “criminal, tax, or regulatory investigations or proceedings.”²⁰⁰

Justice Brennan’s prescient analysis came decades before credit cards and online payments made the intimate details of nearly every day-to-day transaction visible to banks. It was a full 20 years before the BSA was amended to include suspicious activity reporting for domestic transactions, and 35 years before those reports would top one million annually. Justice Brennan was correct in 1972 that the BSA gave unchecked authority to the Secretary of the Treasury to legislate who should and should not be surveilled, and that delegation was never challenged.²⁰¹ Today, after repeated and arbitrary expansions of the category of “financial institution,” after “temporary” exemptions that have lasted decades, and after an unknown number of secret data collection orders, it is time to test the law again.

Conclusion

Recent scholarship on ambiguity within statutory construction encourages us to avoid a finding of ambiguity if one is not suggested by the best reading of the statute. We find that the definition of financial institution is not ambiguous. It is, instead, very broad.

Recent scholarship on breadth suggests that broad statutes raise serious due process concerns and encourages us to dismantle an unconstitutionally broad statute rather than narrowly construe its text and avoid due process issues at the level of administration and enforcement. If the definition of financial institution is not ambiguous, then the Treasury Department cannot narrowly construe the definition relying merely on Chevron deference.

While the Bank Secrecy Act may not be ambiguous, it does delegate substantial power to the Secretary of the Treasury. Recent supreme court jurisprudence suggests that a naked delegation of unconstrained discretion to apply or not apply a law to large swaths of the population is unconstitutional.

There may be an impulse to ignore these textual infirmities or treat this critical analysis as academic or overly pedantic. Remember, however, that the Bank Secrecy Act authorizes one of the most sweeping warrantless data collections in our nation’s history. It implicates the personal private records of at least every American who has a bank account. It vacuums up these personal records automatically without any judicial finding of cause, particularized accusation, or reasonable suspicion. It deputizes our largest financial corporations as arms of federal law enforcement against their customers. And it does all of this, or it doesn’t, at the complete and unchecked discretion of unelected officials in the executive branch. Broad, ambiguous, or delegated, there’s much to find constitutionally troubling in our nation’s financial surveillance dragnet.

Acknowledgements

Thank you to Andrea O’Sull⁠i⁠van who provided invaluable research assistance, and to my Coin Center colleagues Neeraj Agrawal, Landon Zinda, Robin Weisman, and Jerry Brito for indispensable discussions and comments on early drafts of this paper.

Notes