Judges Can't Tame Deferred Prosecution Agreements

Lately, some members of the federal judiciary have begun to voice misgivings about “Deferred Prosecution Agreements” in corporate criminal cases. That attention is long overdue. One has to wonder, though, how much can come of it.

Deferred Prosecution Agreements (“DPAs”) were originally created for low-level, often youthful individual offenders: instead of putting them into the slammer, put them in a rehab program. Stipulate at the front end that if they mend their ways and behave for the duration, you’ll drop the pending prosecution. Fifteen or so years ago, though, federal prosecutors began using DPAs to combat corporate criminal behavior, or allegations thereof. Prosecutors seize on some vaguely worded statutory criminal provision. Faced with a prospect of destruction, the corporate targets agree to a DPA. They promise to reform their “corporate culture,” agree to outside monitoring, and pay a ton of money to government agencies and various third parties. A billion dollars is no longer a large sum in this particular corner of the law.

How common is this? Hard to say: there is no central inventory. (Why should the government keep track of what it’s doing?) However, the law firm of Gibson Dunn & Crutcher keeps track of DPAs and publishes the results. The numbers, and the amounts, have risen sharply over the years. 2015 is shaping up as a record year.

Funny thing: apparently, no one much likes DPAs. Corporate counsel, as well as some academic critics, have characterized the practice as little more than extortion. (Jennifer Arlen of NYU Law School has a very insightful rule-of-law analysis here.) On the other side, Senator Elizabeth Warren and her squadrons have denounced DPAs as a cheap “Get-Out-Of-Jail” card: compared to the alleged misconduct and the penalties that would be due on conviction, the settlements are ridiculously cheap. Plus, it’s about time that the guilty corporate officials spend some time in Leavenworth. (The Department of Justice has recently announced its intent to pursue that objective, on top of mo’ money.)

For all I know, both sides may be right. I’m perfectly happy to believe that Wall Street—a preferred target for prosecutors—is a den of thieves. Then again, I’m also prepared to believe that government agencies act on monetary incentives; and many of the “indictments” look like unadulterated nonsense, or payback for conduct that the government itself incentivized. Like, mortgage “fraud.” The fact that you don’t know what to believe, it seems to me, is itself a non-trivial cost of the system.

What do judges have to do with this? The Speedy Trial Act requires commencement of trial within 70 days unless the parties agree to a deferment in writing, subject to judicial approval. Usually DPA approval is like notarization: in and out the door in five minutes. However, in a characteristically thoughtful opinion, Judge John Gleeson (Eastern District, New York—one of the very best district judges to grace the bench) has voiced grave misgivings about that rubber-stamp approach.  And this past February, in United States v. Fokker Services B.V., Judge Richard Leon in D.C. rejected a proposed DPA between the Department of Justice and a corporate defendant as woefully inadequate. That decision is pending on appeal; the D.C. Circuit heard arguments earlier this month.

When Article III judges are being recruited as recorders of deeds, you can be pretty sure that something has gone wrong. So by all means, have them take a closer look at these bargains.

Suppose they do, though: then, what? In the end, no federal judge will want to make a habit of scrutinizing competently lawyered agreements (as these are), let alone monitor them when neither party really wants them monitored (the usual scenario in these situations). Besides: if the judges look too hard, the parties will migrate into Non-Prosecution Agreements—without any judicial involvement (and yes, those, too, exist). But the real problem may be still deeper.

DPA’s have all the trappings of a legal settlement. The value of the case, standard Law & Econ models tell us, is a function of the money at stake and the odds at trial; and unless the parties differ greatly in their estimation, the case will settle within some range, somewhere along the way. In that scenario, it makes sense for courts to take a fairly casual glance: make sure that no one made a dumb mistake or acted under duress or agreed to something unlawful, and be done. But that model just doesn’t work for DPAs. The indictment is the penalty. For the target that’s a sunk cost (to reputation). What a subsequent agreement buys is something like expungement—“we neither admit nor deny culpability”—and, going forward, indemnity (assuming the firm pays up and goes through the rehab motions). That’s a very different calculus. Moreover, neither party has any idea of the odds at trial because none of this stuff has ever gone there. (The Arthur Andersen case did. But that was eons ago, and at least the markets no longer seem to care.) Run that through standard settlement models: anything can happen.

Judges may very well want to take a closer look because somehow, this just doesn’t look like an arms-length settlement. It may be extortion, or a sweetheart deal. But how on earth is a judge going to tell? The usual judicial impulse, exemplified by Judge Leon’s well-meant decision, is to examine whether the agreed-upon penalties fit the admitted facts or violations. But that’s misleading: the target will agree to the “facts” or “violations” only in exchange for a lower penalty, and the prosecutor will demand payment for each non-fact. You’d expect a bifurcated “settlement” distribution: billions for nothing, peanuts for admissions to stuff that looks serious but may or may not have happened. Closer judicial scrutiny might compress the range of permissible bargains. But what have we gained?

If not settlement incentives, what does determine DPA terms and values? Let me try a thought, parasitical on standard literature of some of my colleagues’ musings.

DPA are most common in industries that are joined at the hip with government: banks, pharmaceuticals, other contractors.  (Next up, carmakers.) In that context, DPAs may be best understood not as settlements but as “relational” contracts—not one-shot, arms-length transactions, but rather rough frameworks that shape, or partake of, a continuous relationship that rests mostly on implicit understandings among long-term partners. DPAs aren’t calculated to let the parties go their separate ways. They’re calculated to let them continue doing business again, going forward. You can’t really comprehend the terms of a given DPA unless you understand the broader context and the implicit understandings.

Every major DPA is accompanied by press releases. A truthful document of that sort might read as follows:

We (the government) have given you (corporation) a lot of valuable stuff over the years. We need some of it back. So we pretend to believe you did something wrong, and you pretend to acknowledge that that may be so (without admitting it). We pretend to reform your corporate culture; you affirmatively agree to hold meetings, produce PowerPoint slides, and hire our friends to “monitor” you. We both pretend that this has something to do with law. And as for the money: we both pretend that the “penalty” will deter you from doing things that we term “misconduct” for purposes of this bargain.

That’s the bargain, and it’s purely transactional. How is a judge supposed to review it, and for what?