I. Introduction: Prosecution, Power, and Problems

Prosecutors wield an awesome power. They make the first (and sometimes the last) critical decisions on whether to deploy the ultimate power of the state—the power to punish—against particular targets. The degree to which and the ways in which prosecutorial power is checked largely define a society’s conformance to the rule of law.

Over the past three decades, the world has embraced the concept of the rule of law to an unprecedented extent. The phrase has become the touchstone of good government for people all over the world, from Harare to Hanoi, Kabul to Kinshasa, Lhasa to Lima, Tegucigulpa to Tashkent.

At the same time, however, the rule of law has been undermined in America in ways that have not been fully appreciated. Changes in the locus and dispersion of prosecutorial authority, increasingly numerous, complex, and malleable legal rules, and failing procedural checks on prosecutorial decisions have allowed prosecutors (along with other officials exercising prosecutorial authority) to impose drastic punishments on selected targets without constraints traditionally associated with the rule of law.

Recognizing prosecutors’ ability to bring the power of the state to bear against individuals in ways that especially threaten freedom, the legal system in the United States is designed to restrict prosecutorial power in numerous ways, including through constitutional constraints on the ways in which criminal law can be made and deployed. Beyond legal checks, practical considerations also influence (and to some extent constrain) prosecutors’ judgments. Budget strictures, public relations considerations, personnel regulations and hierarchies, and ultimately judicial controls all limit prosecutorial discretion.

But older notions of how prosecutorial power might be misused—and older controls put in place to prevent that misuse—have been outstripped by more recent developments. And the practical restraints that do exist still leave room for very significant—and very troubling—amounts of discretionary prosecutorial power. That power can be exercised to pursue the innocent, to impose punishment without trial or conviction, and to pressure targets to compromise or capitulate rather than bear the risks and costs of asserting their rights or their innocence.

The power of the government as prosecutor is not only abused in ordinary criminal cases where the poor and the powerless are subjected to the weight of the criminal law system. Prosecutorial power is also abused in high-profile cases against the powerful, sometimes to serve personal or political ends. It is abused as well in selecting and pursuing targets in the world of business, where prosecution substitutes for ordinary regulatory processes or overrides salutary competitive forces.

Finally, the divisions between federal and state authorities often are not observed in prosecutorial decisions. Federal officials intrude on areas of state competence, and state authorities bring charges that effectively turn an individual state prosecutor into the national decision-maker on issues of regulatory importance. This results in duplication of effort on the part of government agents (prosecutorial and regulatory) and extra burden on those targeted for prosecution (or for analogous civil penalties). More concerning, it also brings opportunities for gamesmanship that advances prosecutorial interests at the expense of clear, cogent legal rules. All of these aspects of the misuse of prosecutorial power threaten the proper functioning of our constitutional system and undermine the rule of law.

An earlier article evaluated problems associated with overcriminalization, especially problems flowing from the expansion of regulatory crimes.1 This paper explores other problems in the operation of prosecutions in America, in part through specific examples of misuse of prosecutorial power. It particularly focuses on the risks associated with prosecutorial discretion, including risks to government structure, personal liberty, and ordinary market competition.

II. Bad Cases, Bad Actors: The Joe Salvati Story2

On March 12, 1965, Edward “Teddy” Deegan was shot and killed in Chelsea, Massachusetts (just outside Boston), a victim of fights among New England organized crime families. The murder was committed by five men connected to the Patriarca crime organization: Vincent “Jimmy” Flemmi, Joe “the Animal” Barboza, Ronnie Cassesso, Roy French, and Romeo Martin. The five were seen together in a bar the evening of the Deegan murder, left shortly before the murder, and returned not long after it; a car belonging to Martin was seen by a police officer near the scene. Within a day, a Boston-based agent of the Federal Bureau of Investigation, Special Agent H. Paul Rico, had information identifying all five killers. He relayed that information to his superiors.

In the world we all imagine, the rest of the story should have been the swift arrest, trial, and conviction of the five killers. That is not what happened, for reasons rooted in a long-running project by the FBI and Department of Justice to develop evidence against leading members of organized criminal enterprises. That project, begun several years before the Deegan murder, included cultivating informants who could provide information and testimony that could be used to convict high-level organized crime targets. Jimmy Flemmi was recruited as a cooperating informant—and assigned to Special Agent Rico and his partner, Dennis Condon—on the same day Teddy Deegan was killed. Jimmy Flemmi’s brother, Stephen “the Rifleman” Flemmi, became a highly prized cooperating figure in this program about six months later. Barboza, with serious new criminal charges hanging over him, became an informant two years later.

Barboza provided the key evidence in the Deegan murder case, which had languished because, despite what was known to the FBI, local law enforcement officials had not been able to put together a solid case. In addition to the testimony of informants at the time of the killing, the FBI had ample evidence from secret wiretaps on Raymond Patriarca that corroborated its information about the murder. With the evidence that was available to the Chelsea police, the Boston police, state troopers, and Suffolk County law enforcement officials, there would have been more than enough to prosecute the killers. Barboza, however, did not want to face capital charges and did not want to put Jimmy Flemmi, his best friend, in jeopardy. He constructed a story, which changed repeatedly and significantly over the next year, that correctly identified French, Cassesso, and Martin as participants in the killing, and also added four others to the event who had no connection to it. These included two high-ranking members of the Patriarca family: Peter Limone, who had warned Deegan that Barboza and Flemmi intended to kill him, and Henry Tameleo, who had no evident connection to the matter at all. The other two men named by Barboza were Louis Greco, who had intervened in a confrontation between Barboza and another man, and Joe Salvati, who owed Barboza money and refused to pay back the full amount (as calculated by Barboza).

The FBI officials who had followed the case knew that Barboza’s testimony was false. (Rico would turn out to be more closely aligned with criminal associates than with law enforcement colleagues, and was later indicted for murder in an unrelated case; he apparently helped suppress evidence or craft false testimony in a number of cases.) In this case, Barboza’s testimony conflicted with all nine contemporaneous reports prepared by federal and state officials based on evidence they had collected prior to Barboza becoming a state’s witness. It changed in ways that were hardly credible but were necessary to fit the police reports. For instance, when Barboza found out that a police report had placed someone who looked just like Jimmy Flemmi in a car near the murder, Barboza claimed that the person was Joe Salvati—even though the police described someone with a pronounced bald spot and Salvati had a full head of hair. According to Barboza, Salvati (who was not an associate of the crowd that committed the murder) wore a bald wig, even though none of the other suspects was wearing a disguise. The state officials may not have known, as federal officials did, that Barboza was lying, but they did little to assure that his testimony fit all the evidence they had. In the end, a deal was struck for a reduced set of charges against Barboza. More serious charges were filed against the other men, with prosecutors seeking the death penalty for each; four were sentenced to death and Salvati to life imprisonment.

None of the men convicted in the case was executed—Massachusetts abolished the death penalty while appeals were pending—but two of them died in jail. Joe Salvati served 29 years before his sentence was commuted; Peter Limone served more than 30 years.

The ultimate release of Salvati and Limone and the public recognition that they were falsely convicted is not so much an affirmation that the American legal system works as tribute to Good Samaritan serendipity. After Joe Salvati’s appeals were exhausted, his wife, Marie Salvati, asked Medford attorney Victor Garo to see if anything could be done for her husband, a man she protested was convicted of a crime he did not commit. A skeptical Garo took a retainer from Mrs. Salvati, looked into the case, returned the retainer, and worked for nearly 35 years without pay to secure Joe’s release, to have his conviction reversed, and to secure compensation for a life turned inside out. Garo’s efforts met resistance from state and federal authorities at every turn. Decade after decade, at every level, government officials who knew what had happened did everything they could to hide the facts, while those who were not complicit in framing Salvati and his co-defendants did not want to look into Garo’s claims.

After more than 15 years of losing every legal challenge, having every door slammed, and failing to make any progress through judicial and administrative channels, Garo took his case to the public with the help of respected Boston newsman Dan Rea. A series of special investigative reports on local and then national television raised the profile of Salvati’s case. Although the initial response from federal and state prosecutors was hostile, even to the point of bringing pressure on Rea’s employer to curtail these reports, the heightened public attention led to federal investigations into corruption and abuse in the Boston FBI office and congressional inquiries. Eventually, these, along with court proceedings based on newly acquired evidence corroborating Salvati’s and Limone’s contentions, secured vacation of their convictions and state decisions not to retry.

Victor Garo’s commitment to justice and to his client is extraordinary. Without that, Joe Salvati and Peter Limone might have died in jail, as Henry Tameleo and Louis Greco did. But Garo’s fight on his client’s behalf did not only require righting the initial wrongs. Prosecutors and investigators, their superiors and their successors—including many who were in no way complicit in framing the four men—continued to the very end to resist efforts aimed at discovering what had happened and righting the injustice. Focusing on the FBI’s role, U.S. District Judge Nancy Gertner described both the initial wrong and the continuing wrong in this case:

The plaintiffs were convicted of Deegan’s murder based on the perjured testimony of Joseph . . . Barboza. . . . The FBI agents “handling” Barboza . . . and their superiors—all the way up to the FBI Director—knew that Barboza would perjure himself. They knew this because Barboza, a killer many times over, had told them so—directly and indirectly. Barboza’s testimony about the plaintiffs contradicted every shred of evidence in the FBI’s possession at the time—and the FBI had extraordinary information. . . .

Nor did the FBI’s misconduct stop after the plaintiffs were convicted. The plaintiffs appealed, filed motions for a new trial, . . . sought commutations, appeared before parole boards, seeking clemency from the governor . . . On each occasion, when asked about the plaintiffs, on each occasion when the FBI could have disclosed the truth—the perfidy of Barboza and their complicity in it—they did not. This was so even as more and more evidence surfaced casting more and more doubt on these convictions. In the 1970s, for example, Barboza tried to recant his testimony, not in all cases in which he had participated, but only as to the plaintiffs in this case—the very men the FBI knew to be innocent. In the 1980s, Agent Rico was found by a court to have suborned the perjury of another witness under similar circumstances. Yet, there was still no FBI investigation, no searching inquiry to see if an injustice had been done in this case.3

This case is testament to the power wielded by officials who oversee and support prosecution, and to the risk that those who see themselves as society’s bulwark against wrongdoing will come to view the law as an impediment to just outcomes. When people with power believe that they and their judgments of right and wrong are above the law, there is always a danger they will use their power recklessly or malevolently. The long-delayed course of justice reveals the power of natural human instincts allied with institutional design: inertia, reluctance to admit mistakes, and unwillingness to expose one’s own corner of the bureaucracy to ridicule or liability.

The Salvati story is shocking and extraordinary because it exposes the extreme lengths to which the particular government officials who “handled” Barboza and the Flemmi brothers went: sanctioning perjury, blatantly corrupting the legal process, even overlooking murder, with utter disregard for the effects their actions had on the lives of others. This story is shocking because we know that the people who suffered punishment were actually innocent—and plenty of people in positions of authority knew or should have known that.

But if the overall picture is extreme, basic threads in the fabric of this tale are not unusual. Law enforcement officials often see their role as partisans in a fight against bad people, and that can lead them to stretch legal limits to secure and preserve convictions. This good guys against bad guys view of the process can be exacerbated when law enforcement officers are rewarded for turning informants, generating arrests and prosecutions, and winning cases. That reward structure can push less well-grounded officials to take things a bit too far, to cut corners, to worry too little about the real world impact of their decisions. And officials who are not complicit in the problem too often lack incentives to ask hard questions about their colleagues, as thirty years of denials, resistance, and disregard by otherwise innocent officials in the Salvati case make clear. Frequent unsubstantiated complaints against law enforcers may inure them to complaints from those truly wronged, which may not be so easily distinguished from baseless complaints. Law enforcement officials do need a degree of independence, but there must be mechanisms in place that provide safety valves to guard against abuses and better ways to identify instances in which the system is failing.

Those concerned with the rule of law should see in this story not simply the immoral behavior of a small coterie of government officers. Instead, we should see the natural risks that attend large concentrations of discretionary power. While our government structures largely constrain grants of discretionary power, they leave huge amounts of it in the hands of prosecutors. Some degree of discretion is doubtless necessary to protect society against dangerous individuals, but vigilance is needed to make sure that power is not abused. Without checks, protectors too easily can become tyrants. James Madison made that understanding a centerpiece of his vision for our government;4 it should remain central to our government today.

III. Hide and Seek: The Stevens Prosecution5

The Salvati case lies at one extreme on the spectrum of prosecution misbehavior, with the government’s entire case resting on testimony known to be false at the time. A more common type of problem involves prosecutions that rest not on wholly fabricated evidence but on distorted evidence—evidence that is undercut by information not produced in court. Imagine that a child is accused of cheating on an exam; her exam answers are identical to those of another student. But the teacher who is supposed to decide the fate of the accused is not told that the other student not only has significantly lower grades overall, but has also been suspended twice in the past two years for copying from classmates’ exam papers. That missing information does not prove that the accused student was not cheating, but it certainly casts the matter in a different light.

In criminal cases, prosecutors have a special obligation to disclose information that could help persuade jurors that the accused is innocent. A line of court decisions tracing back to the 1963 case of Brady v. Maryland,6 spells out the sorts of information that might reasonably be thought to be exculpatory, either directly or indirectly. In Brady itself, the defendant admitted his involvement in the crime, but claimed that his companion had actually done the killing. The prosecution did not reveal that the companion in fact had confessed to exactly that set of facts. In addition to requiring that such direct evidence be turned over to the defense, the Brady line of cases mandates that prosecutors inform defense counsel, on request, of other information in the government’s hands that would make witness testimony less credible, such as prior inconsistent statements (earlier statements by a witness that are inconsistent with her trial testimony). Prosecutors must also reveal agreements with witnesses that might cast doubt on the truth of their testimony, such as an agreement to give a witness a reduced sentence if his testimony leads to conviction of other defendants.

The Brady rules are built around the idea that criminal cases should not be seen as games, but as honest efforts to establish the truth. That should be the goal for all government officers.

That is a fitting aspirational statement, but it hardly describes the reality of criminal prosecutions. Just as defense lawyers see their job as providing the best defense for their clients and letting the judge or jury sort out the findings to assess guilt or innocence, prosecutors often see their task as making the case for conviction and punishment. Each side measures success by wins and losses. Prosecutors are motivated by winning, not by making sure that defendants have the fairest chance to escape punishment. Good lawyers all want results to be just, but they see their individual roles as making the system work by doing their own partisan job with intelligence, skill, and zeal. With able counsel on both sides, the system tends to work, but not because prosecutors are indifferent about outcomes. Brady rules do not change the basic nature of the game; they simply change some of its parameters. And the players still recognize that, for them, it is a game.

One relatively high-profile example is the federal government’s prosecution of Alaska Senator Ted Stevens. Stevens was accused of violating ethics laws by accepting and failing to report gifts, principally in the form of thousands of dollars of improvements to his Alaska home. The home improvements were arranged and paid for by a company owned by Stevens’ longtime friend Bill Allen, whose company had benefited from federal largesse brought back to Alaska by Stevens, whose position on the Appropriations Committee and forty years of Senate service made him a formidable source of Alaskan pork. Allen’s company was by no means unique in its receipt of federal money.

A critical aspect of the case was Stevens’ contention that he had asked Allen to bill him for the costs associated with the work on his house. The defense relied not only on Stevens’ own testimony, but on a letter he had sent to Allen. After thanking him for, among other things, the work on his home, Stevens said:

You owe me a bill—remember Toricelli, my friend. Friendship is one thing—compliance with the ethics rules entirely different. I asked Bob [Person] to talk to you about this, so don’t get PO’d at him—it’s [sic] just has to be done right.7

The reference to New Jersey Senator Bob Torricelli is a caution to Allen. Shortly before Stevens wrote the note to Allen in October 2002, Torricelli gave up his own reelection effort, largely because of the fallout from his being admonished by the Senate Ethics Committee for failing to report gifts from businessman David Chang. (The reference should have been even more of a caution to Allen than to Stevens, as Chang served an 18-month prison term for illegal campaign contributions while the criminal investigation of Torricelli was dropped.)

At Stevens’ trial, Allen testified that the letter was a sham, intended to provide an appearance of conforming to the rules but understood by both men to be insincere. According to Allen, no bill was expected and had it been forthcoming, Stevens would not have paid. Not only did this contradict the Senator’s testimony; it also contradicted statements that Allen had given previously, including two statements he had made to the FBI. The prosecution, despite repeated demands by the defense lawyers for all Brady evidence (and orders from the trial judge to comply), did not reveal Allen’s inconsistent statements to the defense and even removed the contradictory statements from an FBI report that was eventually turned over to defense lawyers. When pressed, the government lawyers sent a letter to the defense lawyers explaining that in prior statements Allen had said he did not think that Stevens would have paid a bill for the amount of the renovations; though the letter contained some slightly modifying language as well, it conveyed the impression that the prior statements had been consistent with what Allen would testify to in court. Some of the material making clear that Allen was making conflicting statements finally were turned over to the defense team at 11 p.m. the night before it was to begin its cross-examination; more of the critical material was only provided five months later, after the verdict in the case.

The government also failed to disclose a mountain of other information potentially helpful to the defense.8 A witness referred to early on by the government, but whose testimony (as the government lawyers found out during their preparation sessions with him) turned out not to be helpful to the government was sent back to Alaska without informing the defense or the judge. Another witness, Dave Anderson, was called to the stand at the last minute to establish the extent of the work done on Stevens’ home (after fabricated records on that issue, introduced by the government, were excluded from consideration). Prosecutors did not reveal the immunity deal they had struck with Anderson, which Anderson denied on the witness stand. They also did not reveal that one of the FBI agents working with the prosecution team on the case was involved in an intimate personal relationship with Bill Allen during the preparation for and trial of the case.

The Stevens defense team was led by Brendan Sullivan, one of the nation’s premier defense lawyers who gained broad national fame as counsel for Lieutenant Colonel Oliver North during the congressional Iran-Contra hearings in the 1980s. Sullivan, both meticulous and intelligent, left no stone unturned in his own preparation. The government, which can often point to lapses in defense demands for information, had no such excuse in the Stevens case. In fact, Sullivan pressed relentlessly for exculpatory information, believing Stevens was innocent and aware that Stevens’ political career hung in the balance. The government had filed its charges against Stevens with a senatorial reelection contest just over the horizon. That led to a defense decision to embark on the equivalent of a forced march, with a goal of acquittal before voters went to the polls.

As it happened, the jury convicted Stevens a week before the election. Although he had led in polls prior to the verdict, Stevens narrowly lost the election; voters not wanting to send a convicted senator to Washington probably made the difference between victory and defeat. The defense asked the Justice Department to open an investigation into the prosecution’s conduct and filed a motion for a new trial. The extent of the government misconduct, while suspected by Sullivan, only came to light after the trial when an FBI agent who was disturbed by the way the case unfolded filed a whistleblower complaint within the Justice Department. After the change in administration, the new Attorney General, Eric Holder, stepped in to halt the proceedings, ordering the prosecution to dismiss the matter—a Democrat ordering his department to stop prosecuting a Republican on charges brought during a Republican presidency. The judge who had overseen the case was not mollified, ordering an independent inquiry of the prosecution’s behavior.9

The legacy of the Stevens case in part was a change in the composition of the Senate. Without the misconduct, Stevens almost certainly would not have been convicted; without the conviction, he would have been reelected; with a Republican in that seat, the Democrats would have had just 59 senators in their caucus for the first session of the 111th Congress, one shy of the number needed to move legislation forward. With many sweeping legislative initiatives moving haltingly through the Congress when they moved at all and Republicans solidly opposed to most of them emanating from the other end of Pennsylvania Avenue, this chain of events—starting with prosecution misbehavior—may have altered history in ways no one could have foreseen.

Commentary following the Stevens case has concentrated on what special factors might have led to egregious misbehavior from the Public Integrity Section of Justice, which had charge of the case.10 Some have pointed to problems with the “culture” of the Justice Department encouraging its lawyers, especially those associated with keeping other public officials in line, to see themselves as too pure to be questioned. Whatever truth there may be to that accusation, the Stevens case is doubtless the tip of a much larger iceberg. What it uncovers extends far beyond the confines of the Public Integrity Section or the Justice Department as a whole.

Prosecutors become invested in the cases they bring. They are not neutral toward the evidence that comes before them. Information that outside observers would label exculpatory in an instant might be seen by prosecutors as equivocal. That is not necessarily the result of a conscious effort to subvert Brady—though in cases like the Stevens prosecution it comes awfully close. The difference is one of perspective. After living with a case, putting their time and energy and effort into it, even identifying their own career success with victory in a high-stakes case, prosecutors tend to become so convinced of the justice of their cause that they lose the ability to make unbiased assessments of issues related to their case. That is precisely why due process requires an impartial decision-maker. The same happens on the other side of the courtroom, but the defense rarely has the evidence and tools at its disposal to put the prosecution at a disadvantage.

The result of living with and becoming personally committed to a case is much like what happens in a late-night game of Scrabble. If you look at the letters long enough, and want a really good word to put down badly enough, you begin to see words that aren’t there, that aren’t real words, that in any other context you would know in an instant aren’t real; but when you’re ready to put down the triple word score worth 66 points, you would swear they are legitimate words. That is why there are dictionaries and third parties to referee disagreements in Scrabble, and judges to settle disputes in court.

IV. Regulatory Crime: No Accounting for Prosecutorial Judgment11

Perhaps the most common and costly problems of prosecutorial judgment involve exercises of prosecutorial discretion in the context of regulatory crimes. The field of regulatory crimes has exploded over the past few decades, with estimates of the number of criminal provisions and criminally-enforceable regulations reaching into the hundreds of thousands. That is a far cry from the Ten Commandments or the small set of common law crimes that were presumed to be known by all citizens. Virtually all commentators, including those who support the current rules on prosecution of regulatory crimes, recognize that the range of regulations is so vast and the regulations themselves are so difficult to know that the prosecution of these crimes (either in the criminal courts or through pursuit of civil fines) is inevitably a highly selective matter.

In this context, the most critical judgments often will be determinations of which potential targets to prosecute, what charges to bring, and what level of investment to make in the particular case. These decisions—which are treated as exercises of prosecutorial discretion, outside the purview of judicial review or other effective control—hold the prospect of being final decisions on matters of individual businesses’ life or death.

One of the most noted examples of this phenomenon is the federal government’s 2005 prosecution of Arthur Andersen LLP (Andersen), one of the nation’s leading accounting firms founded almost a century before, and one of the surviving “Big Five” firms at the time of the prosecution.9 The events that led to Andersen’s prosecution started with the downfall of the energy and services conglomerate Enron. Enron was formed from the merger of natural gas pipeline firms in the 1980s, quickly changing its name and location. It aggressively expanded, branched out into other ventures, and grew to be a firm that reportedly was generating revenues in excess of $100 billion, making it one of the top ten firms among the Fortune 500. As it turned out, the firm’s profitability was due in part to overstating asset values and moving liabilities off-balance-sheet to a series of limited liability “special purpose entities,” owned in part by Enron employees. Andersen was Enron’s auditing firm.

When rumors of problems at Enron began to edge into public speculation on its financial issues, an account manager at Andersen (and her supervising partner) reminded others at the firm who had worked on Enron matters that it would be good to comply with Andersen’slong-established (but rarely implemented) document retention policy. That policy called for retention only of a single, centrally filed copy of information relevant to client service, not necessarily including drafts and notes, but also cautioned that documents should not be destroyed if the firm is “advised of litigation or subpoenas regarding a particular engagement.”12 While the invitation to implement the firm’s policy resulted in the destruction of thousands of documents (doubtless in the expectation that the documents otherwise would become subject to discovery in litigation or regulatory investigations relating to Enron), the shredding was stopped—on the direction of the same individuals at Andersen who had reminded employees of the firm’s document retention and destruction policy—the day after Andersen was notified formally of an investigation by the SEC and served with a subpoena for records.

There is no doubt that the reminder about the company’s policy respecting documents was intended to prevent anyone outside the firm—including regulatory and investigatory authorities as well as potential private litigants, reporters, and anyone outside the accounting firm interested in its relationship with Enron—from discovering what Andersen employees were thinking about, worrying about, and talking about in the privacy of the firm. That, of course, is the entire purpose of having policies about document retention: to preserve what is important and necessary for future uses while reducing the risk that disclosures of background discussions among members of a team working on a problem will inhibit free discussion of all aspects of the issues being addressed. As the Supreme Court said, that is the same reason given for protecting communications between lawyers and clients and similar legal privileges.

The applicable law in the Andersen prosecution—a statute dealing with witness tampering—did not make any document destruction a crime; in relevant part, it punished only the knowing use of physical force, threats, or corrupt persuasion of another with an intent to cause the person to withhold or alter documents in order to impede an official proceeding. The Supreme Court unanimously read the instruction as applying to a very limited class of cases, in which the intention to undermine a proceeding for corrupt purposes was plain, something that fit with the other parts of the provision punishing threats and physical force directed at intimidating potential witnesses and undermining official proceedings. It earlier had held that advising a client to withhold information that a lawyer thought was protected from disclosure to the Internal Revenue Service did not violate the law.13 The Court had no trouble seeing that encouraging fellow firm members to dispose of documents that were not public records and were not (at the time) the objects of subpoenas served to the firm, especially within the contours of a document handling policy, is far closer to the case of advice to a client on what is protected from disclosure than to threatening witnesses with harm if they testify honestly or produce documents they are under a legal duty to provide.

The prosecutors had pushed for a very different reading of the statute, one that would have criminalized any action that made any possible investigation or prosecution of anyone more difficult. Even if it is understandable that prosecutors would want to have the ability to deploy a flexible and powerful tool to make it easier to enforce rules that they believe are beneficial, the decision to bring criminal charges against Andersen and the mindset respecting the law that supported this prosecution are hard to defend from a broader perspective. A criminal law as flexible as the prosecution sought would threaten a huge range of decisions that are consistent with ordinary, prudent business behavior. It would make the prosecutor the ultimate authority over business regulation.14 Any individual federal prosecutor could credibly threaten almost any firm with criminal liability, which in turn would bring enormous pressure to capitulate to whatever penalties and restrictions the prosecutor might demand. Even then, the prosecutor would be free to accept or reject the corporate surrender, deciding in the fashion of spectators in the Roman Coliseum who got thumbs up and who did not.

Andersen’s treatment is a cautionary tale. Like any accounting firm, it had value as an enterprise that could certify whether the businesses it audited complied with established rules; it was not an investigative enterprise, just one that was supposed to make sure that the books it audited met accounting criteria and produced the results announced so far as could be determined from the information provided. The firm can be faulted for failing to ask more probing questions of Enron and of other companies it audited. Prosecutors evidently believed that this failure reflected business concerns that more probing inquiries might jeopardize the contribution Enron and other clients made to Andersen’s bottom line. But treating the firm as a partner in a criminal enterprise—especially under a law designed to address serious efforts to intimidate and corrupt judicial proceedings—does something quite different from seeking to channel business decisions in a more publicly beneficial direction. It stigmatizes questionable business judgments as indefensible assaults on public good.

In the end, the government’s decision to prosecute Andersen deprived the company of the public-trust-signaling value on which its accounting business depended. Clients fled even before any neutral authority had a chance to consider the charges. In short order, it lost almost its entire revenue base along with more than 96 percent of its employees (down from 85,000 employees in 2001 to less than 3,000 by the end of 2002) and was effectively finished as a going enterprise. Further, the prosecution came after Andersen’s management made clear it was prepared to accede to pretty much any terms the prosecutors wanted, including removal of the partner in charge of the Enron account and termination of the two individuals who had recommended implementation of the document retention policy.

The decision to prosecute was especially striking given that, as Richard Epstein observed, “There was no evidence that any of the actions taken [by the Andersen employees] were done to shield Andersen or its partners from criminal liability . . . [as opposed to] protecting Andersen’s reputation as an auditor.”15 Perhaps the prosecutors had a sense that, in the wake of a series of high-profile stock-market shocks traced to corporate accounting fraud, the public would be well-served—and faith in the stock market would be restored—by prosecuting a well-known accounting firm. Whatever signal was received by the public, another message was communicated to the business world. Rather than a step in the direction of assuring greater fidelity to accounting standards, Andersen’s prosecution instructed the business community that federal law enforcement officials could and would issue a corporate death sentence if they chose.

The Arthur Andersen case is a cautionary tale in another sense. Not only does it show the power that individual prosecutors can wield in exercising discretion to bring charges, to publicize them, and to pursue proceedings even after securing every possible concession; it also shows how prosecutors can wield power that seems much more logically reposed in the SEC, a regulatory agency with a mandate that encompasses the sort of oversight associated with assuring appropriate standards for assessment of public companies, including the Enrons of the world. In fact, the SEC did pursue cases against both Enron and Arthur Andersen, and eventually revoked Andersen’s license to provide services essential under the securities laws.

Regulatory proceedings and administratively-initiated court cases that can result in civil penalties are in many ways quite similar to prosecutions.16 They operate under different rules of procedure and are subject to distinctive practical and political constraints and inducements. Competitors can entreat agencies to deploy these alternative forms of corporate sanction for reasons that are not aligned with broader public interests; the threat of severe sanctions can have similar in terrorem effect on enforcement targets; and the discretion to pick and choose among potential targets can produce similar issues respecting the scope of official discretion. The use of administrative processes, which have fewer protections for enforcement targets than criminal proceedings, can raise special problems, especially when related criminal charges can still be pursued. Civil liability can generate high costs to firms that limit their willingness to engage in some beneficial behaviors, and it may not deter the corporate misbehavior that imposes widespread costs on others. In short, whether one is concerned with overdeterrence or underdeterrence of corporate misconduct, administrative regulation and civil liability both have serious flaws.

The use of civil sanctions, however, does differ in at least one important respect from the use of criminal penalties: civil litigation does not generally carry the same signal of serious misconduct as criminal prosecution, a signal that threatens severe reputational harm regardless of the ultimate outcome of the case. That is the essential lesson of the Andersen prosecution.

V. States as Nations: Tales of Eliot Spitzer17

While many of the most talked-about prosecutions of the past several decades have been the work of federal prosecutors, state prosecutors handle the vast majority of criminal cases and account for a number of visible abuses of power as well.

Criticism of and concern with state prosecutions should not be taken as a total indictment of the criminal justice system, which is both relatively predictable and generally fair. An enormous number of state court cases—more than 50 million each year—are routine traffic cases, and another 30 million or so are relatively routine criminal cases.18 More than 90 percent of these are disposed of by plea agreements, the bulk of these reached between overwhelmed and underpaid prosecutors and public defenders, court-appointed counsel, and lawyers trying to make a living in the sort of practices that seldom make news.

Yet, there are some state prosecutors—particularly a few famous or infamous state attorneys general—who have made careers of pursuing targets that are both powerful and unpopular. Sometimes these targets are officials or well-established citizens whose behavior quite clearly constituted what anyone would understand to be crimes: theft, bribery, murder, extortion. The “crusading prosecutor” is a tried and true story-line for novels, TV shows, and movies, drawn in some measure from real life; it is a type that depends on bravery, commitment, and dedication to legal ideals. But other times, the prosecutor targets individuals and enterprises that are engaged in what looks very much like ordinary business behavior, often in highly regulated industries such as financial services, health care, and insurance.

The fact that these targets are already subject to extensive regulation, very often at the federal level, makes them at once questionable subjects for state prosecutors’ attention and inviting targets. These are apt to be politically popular prosecutions because the regulations signal public concern about the industries, public skepticism that they can be trusted to operate free from government supervision, and public understanding of their importance. Moreover, where primary regulatory authority is in federal hands, a state official can play the role of outsider shining a light on the awful job other government officials have done. It is no surprise that crusading attorneys general frequently have high public approval ratings—and frequently move on to the governor’s mansion after highly visible crusades.

The “poster child” for the crusading state attorney general is Eliot Spitzer, former New York Attorney General and then Governor of New York before his dramatic political fall from grace. As Attorney General, Mr. Spitzer’s office brought cases against leading firms in investment banking, mutual funds, insurance, and other industries. The prosecutions were almost always predicated, at least in part, on a combination of charges under New York’s Martin Act.19 This 1920s-vintage law gives prosecutors who choose to use it extraordinary power, as it punishes a broad and largely unspecified range of activities in connection with (among other things) advice, advertisement, purchase, or sale of securities, without any requirement of intent to defraud. This includes anything that is deemed “contrary to the plain rules of common honesty” and anything “tending to deceive or mislead the public.”20

With this broad charter to police almost any activity in the financial arena that he deemed questionable, Spitzer was able to select among an extraordinarily broad array of possible targets. No matter how earnestly a company’s management tried to stay within the bounds of legal strictures and generally-accepted business practices, firms were unable to defend against charges under the Martin Act. While particular companies no doubt act questionably or even commit outright fraud, businesses generally counted on prosecutors’ ability to distinguish the few truly bad actors from the general run of folks trying to succeed in making profits and providing useful services—those who occupy the vast expanse between Mother Teresa and Carlo Ponzi.

Spitzer’s prosecutions, however, pushed the line of what is acceptable beyond ordinary business practice to something more like an aspirational goal—almost to the point of requiring firms to recommend and design only investments that would turn out to be good. Businesses facing Spitzer’s charges and their lawyers—along with many neutral observers—were almost always skeptical of the Attorney General’s concept of legally permissible acts—especially because his office purported to distinguish criminal conduct from ordinary business behavior.

Objections to Spitzer’s view of the law were seldom tested, because rather than simply arguing his positions to judges, trying to persuade them that he had the law’s interpretation and application right in the cases he brought, Spitzer deployed heavier artillery to overcome his targets’ objections and effectively to coerce settlement without trial. As Daniel Gross explained:

Spitzer viewed his targets not as criminals who needed to be jailed but as professionals (and firms) with assets, careers, and reputations to protect. So he didn’t simply indict. He issued press releases. . . . When Spitzer [targeting Merrill Lynch] published a press release detailing “a shocking betrayal of trust by one of Wall Street’s most trusted names,” Merrill Lynch lost $5 billion in market value in a few days and quickly settled. Getting the rest of the investment banking world to go along was then a relatively easy matter.21

The result ultimately was a “global settlement” of charges against ten leading investment banking firms, with the firms agreeing to pay substantial fines (more than $1 billion), make structural changes to their research and investment operations, and change some practices respecting initial public offerings of securities.

The problems with the use of high-pressure tactics to secure, in Gross’s words, “punishment, remedy, and structural change” all in one tidy package are two-fold. First, that use of prosecutorial pressure bypasses neutral review from judges who are responsible for reading the law and assuring that it applies, that it is used in a predictable manner, and that it is not twisted to fit a given official’s own interests. The essence of the rule of law is that sort of principled predictability, and the structures of government that support and reinforce predictable, neutral, general application of law are prized precisely because they promote the rule of law.22 Prosecutors who are sure they are doing the right thing are no substitute for the limited powers and well-defined processes that prevent tyranny.

The second problem with this kind of prosecution is that it upends federalism. Spitzer took it upon himself to regulate industries that have important components located in New York but that operate nationally and affect national finance and commerce. Spitzer and his associates used the threat of criminal and civil liability to force changes in business practices and structures without the sort of hearings, input, consideration, and expertise that are typically required for construction of far-reaching regulation of business practices.23 But more than that, construction of such regulation by a single set of state officials—even if they adopted better suited processes to construct the regulations—risks letting parochial visions control national enterprises.

The Constitution assigns responsibility for matters that primarily affect a single state to that state’s officials, while giving federal officers power over matters that have broad spillover effects across state lines. Spitzer’s targets overwhelmingly fell into the category of entities and behaviors better regulated at the national than the state level. Yet his office’s solutions to perceived problems routinely imposed conditions on targets that had national impact. In fact, making global changes was the essence of all of the office’s major initiatives.

There are certainly some matters of national scope that state prosecutors can play a role in, especially on a collaborative basis with one another or with federal authorities where each prosecutor is protecting the interests of his or her own state’s citizens. And there are matters where one state’s citizens might have a specific interest that could be protected in litigation by its state attorney general, as with environmental harms that are concentrated in specific locations. But those types of litigation are radically different from the one-prosecutor structural reform cases brought by Spitzer, using the location of firms to leverage changes wide nationwide effects.

VI. From Traditions to Solutions

Traditionally, prosecutors have enjoyed broad, unreviewable discretion to decide whom to bring into court, to select the charges, to determine what resources to invest in each case, and to choose when and how to settle. This level of discretion been defended as necessary to protect enforcement officials against constant engagement over the choice of enforcement priorities and resources. It has also been defended as benign: in theory, “type 1” errors (decisions to prosecute someone who is not guilty) will be corrected by not guilty verdicts, and “type 2” errors (failure to prosecute those who are guilty) are simply an inevitable result of limited law enforcement resources. Moreover, not all crimes deserve prosecution, and prosecutors should be able to consider excuses or justifications in deciding whether to bring charges. Changes in the set of government officials who exercise prosecutorial authority, expansion in the number of criminal offenses, declines in the clarity and predictability of criminal offenses, however, have undermined justifications for prosecutorial discretion.

These changes have greatly amplified prosecutors’ power. Their range of targets, of possible charges, and of prosecution tactics is vastly larger while the proportion of potential targets that have meaningful criminal intent or are engaged in obviously criminal conduct has become smaller. Beyond that, in a world of instant and constant communication, where news travels immediately around the globe and markets react instantaneously to the information available, prosecutorial misbehavior can have dramatic consequences for which ex post correctives are inadequate. Those who think that prosecutors’ mistakes or misuse of their office is benign should consider what happened to Joe Salvati, Ted Stevens, Arthur Andersen, and Eliot Spitzer’s targets.

A number of steps could be taken to make matters better from a rule of law standpoint. The most important—but also the least likely—corrective would be a dramatic reduction in the number of crimes on the books, a step that would lessen target-and-charge selection options and reduce prosecutorial leverage. This would also realign the balance of legislative versus prosecutorial control over law. A second possible corrective would be to restore requirements for intentionality or a degree of knowledge of criminality or at least a reason to expect someone ignorant of the law to have had knowledge of its requirements. This too is unlikely to happen.

A third possible change would be to create or increase impediments to misuse of prosecutorial powers, whether through greater penalties for the sort of misbehavior observed in the Ted Stevens prosecution or through checking the prosecution via press release that characterized much of Eliot Spitzer’s work as New York Attorney General.

Finally, judges might be more skeptical of rules that broadly vouchsafe prosecutors’ discretion, whether in facilitating inquiries before irreparable damage is done or in assuring more searching scrutiny of the legal assertions on which prosecutions—frequently in connection with regulatory crimes—are based.

Over the past few decades, the contours of our legal system have diverged sharply from conditions that made broad prosecutorial discretion defensible. Personal liberty and economic rationality, touchstones of our heritage and of our future, depend on preventing the sort of prosecution abuses that easily occur but cannot easily be corrected. Prosecutors can be generally respected, even applauded; crusaders can be admired, even lionized. But freedom to live, work, and play under predictable, neutral rules deserves a higher place in our pantheon of values. It deserves not just our respect but our protection. That would preserve what our founding generation gave us, what our “greatest generation” fought for, and what future generations should inherit.



1  Ronald A. Cass, Overcriminalization: Administrative Regulation, Prosecutorial Discretion, and the Rule of Law, 15 Engage (No. 2) 14 (July 2014).


2  The facts set forth in this section come primarily from Judge Nancy Gertner’s account in Limone v. United States, 497 F.Supp.2d 143 (D. Mass. 2007), aff’d, 579 F.3d 79 (1st Cir. 2009), supplemented by information from contemporaneous news reports.


3  See Limone, 497 F.Supp.2d at 152-53.


4  See, e.g., Federalist Nos. 47 (Madison), 51 (Madison).


5  The facts set forth in this section are primarily based on U.S. Department of Justice, Office of Professional Responsibility, Report On Investigation of Allegations of Prosecutorial Misconduct in United States v. Theodore F. Stevens, Crim. No. 08-231 (D.D.C. 2009) (EGS) (Aug. 15, 2011), available at http://www.leahy.senate.gov/imo/media/doc/052412-081511Report.pdf (Stevens Report), and are well-documented in contemporaneous news reports.


6  373 U.S. 83 (1963).


7  See Anna Stolley Persky, A Cautionary Tale: The Ted Stevens Prosecution, 24 The Washington Lawyer 18, 22 (No. 2, Oct. 2009).


8  See Stevens Report, supra note 5.


9  The result was the almost 700-page Stevens Report, supra note 5. The report summarized misconduct that supported suspension of two prosecutors. A subsequent decision of the Merit Systems Protection Board, without contradicting the findings of the report, concluded that the Department had violated required procedures in issuing the suspensions.


10  See, e.g., Persky, supra; Jeffrey Toobin, Casualties of Justice, The New Yorker, Jan. 3, 2011, available at http://www.newyorker.com/magazine/2011/01/03/casualties-of-justice (“the Stevens case . . . was a profoundly unjust use of government power against an individual—a case flawed in both conception and execution”). See also Rob Cary, Recalling the Injustice Done to Sen. Ted Stevens, Roll Call, Oct. 28, 2014, available at http://www.rollcall.com/news/recalling_the_injustice_done_to_sen_ted_stevens_commentary-237407-1.html (Mr. Cary, one of the lawyers representing Sen. Stevens, is not an impartial commentator, but his observations pointedly capture the sense of dismay with the prosecutors’ performance also expressed by others).


11  The salient facts of the case are recounted, among other places, in the Supreme Court decision reversing the firm’s conviction, Arthur Andersen, LLP v. United States, 544 U.S. 696 (2005).


12  Andersen Policy Statement No. 780—Notification of Litigation, excerpted in Arthur Andersen, LLP v. United States, supra, 544 U.S. at n. 4.


13  See Upjohn Co. v. United States, 449 U.S. 383 (1981).


14  For a similar prosecution based on an excessively broad construction of a law (in this case, a law that actually was addressed to financial regulation, though applied in a very different and wholly unpredictable context), and a similar rejection from the Supreme Court, see, e.g., Yates v. United States, No. 13-7451, --- U.S. --- (Feb. 25, 2015).


15  Richard A. Epstein, Deferred Prosecution Agreements on Trial: Lessons from the Law of Unconstitutional Conditions, in Prosecutors in the Boardroom: Using Criminal Law to Regulate Corporate Conduct 38, 47 (New York University Press, Anthony S. Barkow & Rachel E. Barkow eds. 2011) (Barkow & Barkow).


16  For discussion of the similarities and differences between corporate criminal liability and regulatory or civil proceedings, see, e.g., Samuel Buell, Potentially Perverse Effects of Corporate Criminal Liability, in Barkow & Barkow, supra, at 87-107.


17  Most of the facts in this section, are readily available in public news reports, and can also be found in Peter Elkind, Rough Justice: The Rise and Fall of Eliot Spitzer (Penguin Books 2010).


18  This figure is roughly 300 times the 90,000-100,000 criminal cases each year in federal courts. See, e.g., Administrative Office of the United States Courts, Federal Judicial Caseload Statistics: 2013, available at http://www.uscourts.gov/statistics-reports/federal-judicial-caseload-statistics-2013; U.S. Department of Justice, Bureau of Justice Statistics, State Court Caseload Statistics, available at http://www.bjs.gov/index.cfm?ty=tp&tid=30 (State Court Statistics).


19  N.Y. Gen. Bus. Law. Art. 23-A, §352 et seq.


20  See, e.g., People v. Sala, 258 A.D.2d 182, 193 (3d Dept. 1999), aff’d 95 N.Y.2d 254 (N.Y. Ct. App. 2000); People v. Federated Radio Corp., 244 N.Y. 33, 38-39 (N.Y. Ct. App. 1926).


21  See Daniel Gross, Eliot Spitzer: How New York’s Attorney General Became the Most Powerful Man on Wall Street, Slate (Oct. 21, 2004), available at http://www.slate.com/articles/news_and_politics/assessment/2004/10/eliot_spitzer.single.html.


22  See, e.g., Federalist Nos. 47 (Madison) (noting that, while the accumulation of powers “in the same hands . . . may justly be pronounced the very essence of tyranny,” the structures of the Constitution is consciously designed to prevent that result), 51 (Madison) (elaborating on the reasons that the Constitution provides incentives and opportunities for different officials and institutions to check tendencies of others to aggrandizement); Ronald A. Cass, The Rule of Law in America 4-19, 28-29 (Johns Hopkins University Press 2001).


23  The argument against Lone Ranger lawmaking without process is not an endorsement of administrative regulation in its present form. The litany of problems associated with the regulatory state is long, well-analyzed, and raises serious concerns. Still, defects in one process do not make an alternative preferable.