The Trial Tax: Worse Than You Thought

When University of Pennsylvania lawprof David Abrams announced his research, it made news.  After all, it purported to be an empirical study that contradicted not only the accepted wisdom, but the experience of every criminal defense lawyer ever.  The trial tax was a myth.

Comes now (or relatively recently at least) David Abrams from Penn with an article that slays the sacred cow of the trial penalty by providing, you know, data. And the data is the best kind of data because inasmuch as it’s true, it is SURPRISING data. Specifically, Abrams argues that based on the study he performed. the data supports the view that in fact there’s a trial discount not a trial penalty.

Surprising isn’t the word for it. Shocking? Stunning? Astounding?  How about wrong?

Last year David Abrams, Penn, stunned the criminal law world with a study concluding that defendants actually receive shorter sentences at trial than they do for pleading guilty. Rather than “penalizing” those who exercise their right to trial, we actually punish those who plead guilty more harshly, turning decades of plea bargaining debates on their head.

Abrams was wrong, and he’s not the only one.

The most significant aspect of the adoration of empiricism is that data displaces deeply-held beliefs.  It doesn’t matter if everybody says so. What matters is what the data says, and at its core is empirical evidence showing that everybody has been wrong all along.  Data. It’s a thing.

Except that the experience in the trenches, case by case, defendant by defendant, was not just the opposite, but overwhelming so.  And indeed, this was a primary driver of federal pleas, the fact that a defendant who wanted to put the government to its burden of proof would, if he lost, pay dearly for doing so.  Have criminal defense lawyers been advising their clients falsely all these years? Have we assumed the “common wisdom” to our clients’ detriment?

No. Of course not.

My latest article reveals significant conceptual and statistical errors in the canonical methodology that cause most other studies, including those by the United States Sentencing Commission (USSC), to greatly underestimate the trial penalty. Underestimating the Trial Penalty: An Empirical Analysis of the Federal Trial Penalty and Critique of the Abrams Study. 84 Miss. L. J. 1195 (2015) (Selected through peer review). Where leading researchers report that the federal trial penalty is only around 3-15%, I find that the average federal trial defendant receives sentences around 64% longer than if they had pled guilty instead. In other words, federal defendants cannot exercise their constitutional right to trial unless they are willing to risk a 64% longer sentence, a heavy “penalty” indeed. Where Abrams reports that Chicago defendants pay a similar penalty for pleading guilty, I reveal that his data actually suggests that plea defendants receive shorter sentences than those who go to trial.

Assuming you just skipped over the quote, here’s the TL:dr:  Not only is there a trial tax, but the trial tax is a 64% longer sentence.  This is likely far more than anyone would have realized. Yeah, data is a thing.

Applying these findings, I demonstrate that the federal trial penalty is so large that only a tiny fraction of defendants could ever rationally choose to go to trial. In such a system, the constitutional right to trial by jury becomes less of a “right” and more of a trap for fools.

This is a critical point, one that impacts not only the advice of criminal defense lawyers, and the choices made by defendants, but the perspective of federal judges, legislators and those cool guys in the Department of Justice.  The price of challenging the government’s might is a 64% longer sentence.  Innocent? 64%. Overcharged? 64%. Guilty, but with explanation? 64%.  Constitutional rights violated along the way? Still 64%.

Not that all defendants (or their lawyers) are invariably rational beings, but as Andrew Kim makes abundantly clear, rolling the dice with the likelihood of a penalty of a 64% longer sentence is just plain nuts.

Kim goes on to explain his empirical methodology, together with the failings of Abrams’ statistical analysis.  If that’s your thing, it’s fascinating. It’s not mine.

Kim also notes the limitations of his own analysis, that it’s impossible to take into account all the underlying bargaining, such as charge bargaining (which is where the prosecutor threatens to add additional or higher level charges to the indictment if a defendant doesn’t cop a plea, thus making the situation worse than it would appear at the time negotiations occur).  But this gap would add to the 64% trial tax, not reduce it.

While judges complain that nobody goes to trial anymore, and that prosecutors control the sentence to be imposed through charging decisions, this outrageously high penalty for exercising the constitutional right to go to trial raises some serious questions about the role judges play in this fiasco. Not every crime has a mandatory minimum, and post-Booker, judges have far more sentencing discretion than ever over the decades when the Supreme Court let us be confused about the Guidelines.

So why are you pounding defendants with a penalty like this for taking a case to trial, judges?  Why do you hate constitutional rights so much?

And still, there’s a deeply insidious problem lurking behind the numbers.  If these plea bargains, without the 64% trial tax, are sufficient as far as the government is concerned to satisfy the legitimate purposes of sentencing, and they must be unless the government is in the business of setting criminals free to prey upon society, then how could there be any justification to add an additional 64% punishment on top of sentences that already fulfill the purposes of § 3553(a)?

In other words, we’ve all been scammed by sentences that are not merely Draconian, but unlawful according to the purposes Congress and the Supreme Court say must apply.  If 60 months will do, then 121 fails. You can’t have it both ways, no matter what the government says in its sentencing memo.

It’s time to shake off the life plus cancer sentences, and start recognizing that the government might be the one selling the lie, but the courts have been running the cash register.

21 thoughts on “The Trial Tax: Worse Than You Thought

  1. pml

    So who is correct here, LawProf A or LawProf B? My guess would be a persons view would depend on which side of the isle he sits on.

      1. pml

        Like you always say, your Blawg, your rules, your opinion, but like always there are those that disagree with your biased opinion.

        1. SHG Post author

          And I let them comment and express that disagreement because I’m a swell guy. Do you think I should ignore the methodology flaws and instead embrace your biased opinion?

    1. paul

      The one who treated the data with more scientific rigor. The methodology is important. That’s probably why it was linked.

    2. Glen

      It’s really surprising that the assessment of objectively empirical data might depend upon “which side of the isle” one sits.

      But then again, maybe not, given that mathematical illiteracy is so widespread and is encouraged by our educational system. How may lawyers became lawyers because math was just too hard?

  2. John Barleycorn

    Heck, and here I always knew this quandary had something to do with an experienced CDL’s innate fear of all the stuff that can go wrong with hydraulics if the load isn’t balanced just right and /or just a little bit too heavy and your average prosecutor’s inability to even understand hydraulics let alone know what to do with even a basic ratchet action tractor jack other than threaten to add another pulley to the mechanism he/she plans to use while hoisting said tractor jack over someone’s head, not to leverage the load to serve justice mind you, but to simply drop the fifty pounds of iron on the defendant’s skull because they enjoy the sound of iron cracking skulls and when they miss, once every few years or so, it’s not as though they are gonna break the tractor jack.

    You know one of these days when you get your readership up to snuff, (you might consider more vulgarity and if that doesn’t work perhaps more generic spiritual references that can thread the needle of all those different gOd feelz certain robes consult when they have the sniffles) and capture nine out of ten circuit court judges, it’s possible that 7 out of ten judges, regardless of what bench they happen to be perched upon, might start insisting that every time they hear the words nolo contendere uttered, or take a more disgraceful plea, they insist the prosecutor give a play by play, in chambers, of the first few rounds that got us to the “justice climax” and then insist that the when they return to open court that the prosecutor whip out his muleta and prove up on what’s holding up the justice above the cape.

    If the defendant’s CDL gores the poor bastard, well then the case is dismissed before the gavel comes down.

    In every corner of your guild, it seems, some fundamental principles are slipping a little too often. You all better do something about that before people start to notice.

    P.S. Bullfighters speak Latin right?
    P.S.S. Post title are very important, especially on Fridays. No more slacking.

    1. Mort

      (you might consider more vulgarity and if that doesn’t work perhaps more generic spiritual references that can thread the needle of all those different gOd feelz certain robes consult when they have the sniffles)

      Or more pics of hot chicks. That usually helps.

  3. Richard G. Kopf


    As you might imagine, I have problems with the study you cite. Let me give you one example

    The author doesn’t like the “acceptance of responsibility” reduction set forth in the Guidelines. The author uses that reduction to drive up his percentages. By including “acceptance of responsibility” as a “trial penalty” this bumps his numbers from 28.3% to 64.0%. (See table 3.)

    OK, one can debate that acceptance-reduction on policy grounds. But that tells us nothing about legally unjustified trial penalties. For the same crime, and same criminal history category (or criminal history score in points), I would be more interested in knowing the precentage of sentences imposed at (or below) the low-end of the Guidelines after trial (and without acceptance) compared to the percentage of sentences given above the low-end of the Guidelines after trial (and without acceptance).

    I could go on but I am tired. It’s Friday, and I want a drink.

    All the best.


    PS Despite some of my doubts about it, I compliment Professor Kim for his thoughtful and well-conceived paper. It is evidence of what academics can do for those of us in practice or on the bench when they strive to give us good empirical research.

    1. SHG Post author

      You raise an excellent question about how many post-trial sentences come in below guidelines. There is still the charge bargaining/superceder problem, but we could remove all cases that weren’t tried on the original charges, and the 2-3 point acceptance of responsibility, but let’s chalk that up to a semantic difference, the plea benefit rather than the trial tax, not to mention why a defendant shouldn’t be entitled to acceptance post-trial because he challenges a constitutional violation, quantity, an enhanced element, but now we’re left with about 12 trials in our universe for analysis.

      Can there be an apples to apples comparison? Can there be one with statistical significance? Can the million other factors that differentiate cases be controlled? Beats me.

    2. Rick Horowitz

      Legally-justified or not, how does this alter the original point? The point, as I understood it, was that the data supports understanding the trial tax being 64%.

      Does it change the tax if the reason given is that 35.7% of the tax comes from explicitly refusing to take a plea, er, I mean for not accepting responsibility?

      Or am I misunderstanding something? (Since I handle state cases, I don’t deal with these complex federal sentencing guidelines, so maybe I’m missing something.)

      1. mb

        Scott can correct me if I’m wrong, but I think there are two points to the post, and that you are not adequately delineating between the two. One point is that it’s dumb to suggest that those who plead guilty are punished more harshly than those convicted at trial, as this would necessarily mean that lawyers on both sides are acting irrationally. Prosecutors would be risking trial and acquittal without making their best offer and defense attorneys would be advising clients to eschew trial, and the chance of acquittal, without gaining anything.

        The second point is Scott’s characterization of the disparity as a “tax” on the exercise of the Constitutional right to trial. I had never thought of it that way, but there’s a good argument that it is, at least some of the time, a fair description. How often it is fair to call it a tax vs how often it would be more fair to call it the benefit of a plea negotiation hinges on the reasons for the disparity in individual cases.

        1. SHG Post author

          If the disparity was only the 2-3 point reduction for acceptance, and the post-trial sentence was based on the same guidelines, less the 2-3 reduction, I would be happy to call it the benefit of the plea rather than the trial tax. But it never seems to happen that way.

    3. losingtrader

      RGK ,
      I promise the maths won’t get easier after that drink, since I suspect I know what it is.
      Maybe you’ll even toss away the idea that looking at guidelines is ethical as a starting point., or if not that, then at least it’s more fun to wing it.

  4. Nigel Declan

    Abrams’ fatal flaw (or, at least, one of many) was that he included acquittals in his data set of sentences. As a result, every defendant who was acquitted was credited as receiving a sentence of 0. His argument was that the proper measure of the “trial penalty” should properly include the possibility of success at trial, but this ignores the fact that plea offers can profoundly affect the decision to go to trial (as well as other factors he doesn’t account for).

    If a defendant who is looking at a 20 year sentence if convicted at trial is offered a plea of 18 years, he may prefer to go to trial much more than if he was offered a plea of 5 years. Furthermore, if the defendant in the first case is convicted, the “trial penalty” is 2 years, whereas if he is acquitted, the “trial penalty” is -18 years, meaning that even a small percentage of acquittals can significantly skew the average downwards.

    Abrams may or may not be skilled at crunching numbers, but his lack of understanding of the subject matter he is studying renders his conclusion meaningless.

    1. Mark Draughn

      Abrams has a background in Economics, and his paper was investigating the “shadow of the law” idea, which is an economics-style argument about choices and knowledge that implies that under ideal conditions trials and plea bargaining should obtain the same average results. Being economics-curious myself, I found it fascinating, but…that’s not what criminal defense lawyers mean by the “trial tax.”

      Abrams is talking about statistical expectations and decisions under uncertainty, but criminal lawyers are talking about the observation that if they force the state to prove its case by going to trial, which is supposedly a right the defendant has, the defendant will receive a harsher sentence that was offered as a plea. By that definition, a small trial tax shows up even in Abrams’ own data set, where non-zero sentences average 2.44 years for a plea and 2.91 years for a conviction at trial.

      (Abrams data also shows that the number of charges is about 10 percent higher for defendants who go to trial, which may be the result of charge bargaining.)

  5. Marc R

    Wow I can’t believe the study was attempted! The only data could be offenders pleading to X versus those sentenced for X after trial. But that’s not how pleas work. As you alluded, pleas usually involve the charge itself. Instead of agg battery client pleads to simple battery. So his appropriate comparator would be against the trial tax of one convicted by a jury of agg battery not of simple battery. A lot of pleas involve man mins or avoiding a felony conviction, so the study would have to look at original charges or informations versus those pleading to the same charge. And that’s the rub; looking at plea deals the author needed to look at original charges. But with that there’s an added problem of not knowing if the original charge was down filed by good lawyer negotiating or whether the intake and line prosecutor disagreed, or whether evidence was suppressed pre plea or a victim didn’t want to press charges so a plea was agreed before a deposition, or any number of factors charges are modified after filing that are lowered by a plea but have nothing to do with the plea deal. Some things defy statistical analysis or maybe this can be analyzed but the author didn’t introduce proper controls. I can think of some ways to make the study stronger like the above factors but because of too many unknowns as a defense lawyer I don’t know how a trial tax can be known other than a number of researches following a number of different CDLs in different jurisdictions using the same metrics and of the same quality with clients with similar criminal histories and departments with similar officer report and witnessing quality. Ergo, I’ll act as if there’s a trial tax until I believe there’s not a trial tax. I hope this isn’t a primer for defense lawyers to follow. A little knowledge is far more dangerous than no knowledge and just canceling the study or explaining it’s flaws should be in the appendix.

  6. losingtrader

    Screw the studies. From everyone’s comments something important will always be left out of the calculation.

    Me, I’ve decided to try to pre-negotiate a sentence for all the crimes I want to commit. Either that or move to South Africa:
    Cape Town – Thieves will soon be able to evade jail by simply striking a deal with their victims under the National Prosecuting Authority (NPA)’s proposed restorative justice programme.

  7. JonCB

    I kind of question whether the benchmark of plea bargains is “sufficient as far as the government is concerned to satisfy the legitimate purposes of sentencing”. It depends on what your theory behind jailing someone is (e.g. rehabilitation vs punishment/scale balancing vs deterrent) but i see two fundamental theories behind the plea bargain.

    Firstly, and the one i most commonly hear, is “I can almost certainly convict you but i will trade you a less than sufficient sentence in return for not forcing me to spend the money to actually convict you”.

    Secondly, there is the more bayesian rational (but potentially unlikely conscious) idea that “I believe i can convict but i’m only N Percent sure i can convict. So rather than risk an M percent chance of no conviction to get the full sentence you deserve, i will accept a 100 percent chance of this lesser sentence”.

    I don’t think in either of these cases the idea that the punishment is sufficient (regardless of the theory of sufficiency) is present. Of course i am not a Lawyer (or even American) so i could be very wrong. Otherwise a thought provoking article. Glad i came here from Popehat.

    1. SHG Post author

      Please bear in mind that this is a law blog, and the issue is one with which American criminal defense lawyers are deeply familiar. This issue has been addressed in great depth many times here, and it’s probably best not to leap in blindly to issues where you have nothing illuminating to offer.

Comments are closed.