David Abrams, recently published in the Journal of Empirical Legal Studies, casts doubt upon one of the classic assumptions of the criminal process: That plea bargains pay off for defendants.
More than 90% of all criminal cases, in CA and elsewhere, end in plea bargains rather than in a jury trial. Rather than this being an aberration, it is, as some commentators believe, a necessary mechanism to account for the cost and hassles of an impractical and unsustainable jury system.
The common understanding of the plea bargain system is that each party to the agreement gains and loses something by the bargain. The prosecution is prepared to offer a sentence that is less than what the defendant might receive from the judge in return for an expedited and less expensive resolution of the matter, leaving prosecutors with more time to devote to cases on trial. The defendant, however, gives up his/her right to trial for the certainty that s/he will not incur a "trial penalty", that is, be sentenced more harshly by the judge if he or she is convicted.
But it turns out this may not be true.
In Is Pleading Really a Bargain?, Abrams runs regressions on a dataset from Cook County in an effort to predict which trial strategy (trial or plea bargain) yields a more lenient sentence. The results, as described in the abstract below, are surprising.
A criminal defendant's decision of whether to accept a plea bargain is one with serious consequences both for his or her immediate and long‐term future. Conventional wisdom suggests that defendants are better served by entering into a plea bargain, to avoid what is known as the “trial penalty.” In this article I present evidence that this notion is likely mistaken. In OLS regressions using data from Cook County state courts, I find that a risk‐neutral defendant seeking to minimize his or her expected sentence would do substantially better by rejecting a plea bargain. I also employ an IV approach to the question and, while the instrument is weak, the results are consistent with the OLS: defendants are better off going to trial.
Admittedly, there are some methodological problems with Abrams' piece. Since he's using court data, he cannot appropriately control for self selection of cases; it may well be that defendants who chose to go to trial did so because they, or their defense attorneys, thought they had a better chance with the judge. Nonetheless, his analysis is impressive.
Abrams offers two possible explanations for his data. The first is the availability heuristic. Defendants perceive trials as being more lengthy and more harsh, because they are exposed to sensationalized trials via the media. The second is the difference in interest between defendant and defense attorney, which I expect grows when public defense offices are weighed down with caseload and slashed budgets.
I have a third possible explanation, which I believe is at least as plausible. In a world of mass incarceration and normalized, mechanical sentences with little discretion, bargaining is more like buying groceries at a supermarket than at a Middle Eastern bazaar (this analogy is Malcolm Feeley's). In this sort of situation, the bargain price comes to manifest exactly what the prosecution expects from the court given the vast amount of evidence predicting it. The cases that go to trial are cases in which the defense believes there are enough unique features to take them out of the "normal crimes" category and make them seem special enough to the judge to warrant a downward departure from the acceptable range. And so, since so few cases go to trial, the ones that do appear special and benefit from the special attention. Some research by the late Yael Hassin, which compared actual parole committees to computers in terms of predictions of dangerousness in early releases, suggests that providing agencies with more discretion (in parole, sentencing, and the like) yields more merciful and lenient results. If so, it is not surprising that judicial attention, in a universe of otherwise mechanized sentencing, yields more lenient sentences.