A Role for Prosecutors in Criminal Justice Reform?

At Marginal Revolution, Tyler Cowen links to an article by Scott Bland titled, “George Soros’ quiet overhaul of the U.S. justice system.” Basically, the article discusses the possibility of achieving significant criminal justice reform by electing prosecutors sympathetic to progressive reform goals. Specifically, Bland notes Soros’ support for prosecutors that run on platforms of “reducing racial disparities in sentencing and directing some drug offenders to diversion programs instead of to trial.”

Could this work?

I’m skeptical. I always say that prosecutors are sentence maximizers, but their sphere of competition is limited to their district. If the average sentence associated with some crime falls across the district (because that charge becomes eligible form pre-trial diversion, for example) prosecutors will be indifferent. This rule change results in a level effect – the annual sentence level for each prosecutor falls, but the distribution of sentences is preserved. In other words, if prosecutor A’s average sentence length is 10% greater than prosecutor B’s average sentence length, this rule change does nothing to effect that. Of course, if you believe – like I do – that crime and punishment is overrated, expanding pre-trial diversion programs makes the world a marginally better place.

But I don’t believe the fundamental problem with the criminal justice system is sentence-maximizing prosecutors. I do believe the fundamental problem is low-quality police. Adding 87 people a year to a pre-trial diversion program is nothing compared to restructuring police incentives to align with the common social welfare goal of successfully prosecuting criminals. In northern Virginia, it is extremely common to find police officers with wildly different nolle prosequi rates across black and white defendants for the same crime type. You can probably guess which group is on the bad end of the disparity. It’s also common to find police who arrest people basically arbitrarily and consistently have virtually all of their cases dropped. Unfortunately, many defendants (especially when detained before their trial) face tremendous pressure to accept plea bargains just to get on with their lives.


Fix big problems first. I would much rather George Soros pour money into legislative efforts to tie police compensation (for example) to successful prosecution rates. It would probably cost less than what he’s doing now and it would certainly better serve progressive criminal justice reform goals.

One Weird Trick to Promote Persistent Information Asymmetry

When I first began researching the legal services market, I quickly became aware of the serious information asymmetry problem across the marketplace – especially for criminal defense. As I’ve probably mentioned before, defendants have no real way to estimate the quality of an attorney when considering hiring one for a criminal case (except obviously by consulting Blackstone Trial Analytics, LLC – the trusted name in attorney referrals and quantitative LSM analysis). This is a classic adverse selection problem. Defendants know (or can approximately make themselves aware of) the average outcome of a criminal charge; but they don’t know which attorneys contribute better than average outcomes and which contribute worse than average outcomes.

Ultimately, attorneys seem to set roughly comparable prices for their services and share the market. Of course many defendants would like to pay more for high quality attorneys and probably all defendants would like to avoid low quality attorneys (at least at the prevailing prices). I wondered why (high quality) attorneys didn’t try to solve this problem. They could, for example, publicize their records. But this wouldn’t work if many other attorneys simply didn’t publicize their own records. Probably the public would have a hard time interpreting the record in the context of suppliers of criminal defense services generally. This is especially true if people systematically overrate their probability of success at trial.

More plausibly, I thought, attorneys could make their fees contingent on case outcomes. For example, they could charge some variable amount (by quality) for plea bargains and more for trials – much more where the defendant wins. Obviously any specific deal is possible, including zero or even negative fees. These arrangements, I thought, would probably produce few poor incentives, remove some bad existing incentives and communicate important facts about quality to defendants.

Generally, when there appears to be an obvious and easy solution to fix an apparent market failure, a non-market failure lurks just behind it. This is one of those cases. Here’s the relevant ABA rule:

(d) A lawyer shall not enter into an arrangement for, charge, or collect:

(1) any fee in a domestic relations matter, the payment or amount of which is contingent upon the securing of a divorce or upon the amount of alimony or support, or property settlement in lieu thereof; or

(2) a contingent fee for representing a defendant in a criminal case


Claims About Monetary Policy

Sometimes it seems that virtually all claims about monetary policy are either incoherent or wrong. Take for example the following comments from a recent Washington Post article about the Fed:

  1. “LOL
    The FED never got us out of any recession, it’s Still here. The FED caused the last recession, why on earth would it fix it. GDP is a sham as the numbers are very cooked. Unemployment numbers are cooked and well stirred. Inflation is at 8% and this is brought to you by FED / Global government in conjunction with Corrupt government overspending and a complicit PRESS who lies and misleads on a daily basis.”
  2. “Want to improve economic growth? Start with reeling in Big Pharma. A few Epipens cost Thousands of dollars that could be used to purchase goods and services throughout the whole economy instead of being used to enrich a handful of greedy execs.”

The first case is obviously incoherent. Ayer would call this “literal nonsense.” How must the commenter view the world for his view to be correct? GDP is a “sham” and the numbers are “very cooked”; of course if this is true we would expect the government (given broad latitude over GDP figures) to release more robust figures. Ditto for unemployment numbers. The entire last sentence is unfalsifiable/silly.

The second case is an example of just being plain wrong. The commenter is probably relying on some idea that consumption is an “important” component of aggregate demand and investment is a trivial component. The commenter’s theory is at least testable.

Obviously these comments are from laypeople unschooled in macroeconomic theory. I suppose we can forgive them, although I find that intellectually unsatisfying. At any rate, it’s more surprising when you see this from real economists.

John P. Hussman argues for the following view of monetary policy:


Remembering the Quantity Theory of Money (QTM):


Velocity is defined as Py (nominal GDP) over M (money supply).

The only way V could fall in direct proportion to the rise in M is if increasing M had no effect on nominal GDP (NGDP). In other words, we drop the classical advice that nominals only effect nominals and replace this with nominals only effect reals.  Hussman’s story is further complicated by the fact that markets actually react to increases in the monetary base as if they are expansionary (they are). If the announcement of an increase in the monetary base causes asset prices to rise, NGDP has increased.

Honestly, I’m not sure how Hussman’s graph was constructed. A graph of the size of the (indexed) monetary base versus M1 velocity is reproduced below:

fredgraph (3)

I think the “dogmatic monetarist” story looks more plausible.

Juries: 12 Increasingly Angry Men

One of the most enjoyable things about economic analysis is that it often yields surprising facts about cherished institutions. For example, the jury selection process in the United States probably burdens defendants disproportionately in criminal trials – most people believe jury selection serves to mitigate the problem of biased juries.

Why? Imagine the distribution of jurors by how sympathetic they are to the defendant. I would guess the distribution isn’t normal with a mean of “indifferent”; instead, the average juror probably starts from a relatively unsympathetic position. Assume for a particular defendant, 60% of the population has some bias against the defendant and 40% has some bias in favor. If the jury is randomly selected, we would expect 6/10 jurors to be relatively anti-defendant. If the prosecutor and defense attorney can identify the jurors biases and are allowed one challenge each, the final jury should be 6.2/10 anti-defendant. Every additional challenge increases the frequency of anti-defendant jurors in the panel.

Apparently Scots law embraces (wisely, I think) random jury selection. The quotation below is from Peter Duff’s “The Scottish Criminal Jury: A Very Peculiar Institution”:

There is no equivalent to the voir dire procedure in Scotland, a fact which might surprise some American readers. The strong opposition of the Scottish criminal justice system to any procedure of this type is well illustrated by the observations of the Appeal Court in McCadden v. H. M. Advocate:

There may never be a process which eliminates the possibility of personal prejudices existing among jurors, the nearest practical one (and it is not foolproof) being possibly the “vetting” of jurors, a system against which the law of Scotland has steadfastly closed the doors. Evidence of how it is used and abused in countries in which it is operated only tends to confirm the wisdom of that decision.

The court went on to observe that it should not be “lightly assumed” that jurors will pursue their prejudices in defiance of their oath and the directions of the judge. On a more practical note, the court pointed out that the broad base from which jurors are drawn means that any prejudices and biases tend to cancel each other out, and further, that the majority verdict, whereby a bare eight to-seven vote either way suffices, ensures that it is unlikely that one prejudiced juror can affect the outcome of the case.

Miscellaneous Articles

Below are a few interesting articles for those interested in LSM theory or empirical work (they are in no particular order):

Akerlof, The Market for “Lemons”: Quality, Uncertainty and the Market Mechanism

The seminal work on adverse selection.

Jennifer Bennett Shinall, Slipping Away from Justice: The Effect of Attorney Skill on Trial Outcomes

Where the author gets close to living in living in my ratex prosecutor world. I’ll write about this paper (and Abrams and Yoon)

Abrams and Yoon, The Luck of the Draw: Using Random Case Assignment to Investigate Attorney Ability

Investigating attorney ability.

Roach, Indigent Defense Counsel, Attorney Quality, and Defendant Outcomes

Kahan, Hoffman, Evans, Devins, Lucci and Cheng, “Ideology” or “Situation Sense”? An Experimental Investigation of Motivated Reasoning and Professional Judgment

Landes, An Economic Analysis of the Courts

Baker and Mezzetti, Prosecutorial Resources, Plea Bargaining, and the Decision to Go to Trial

Weimer, Plea Bargaining and the Decision to go to Trial: The Application of a Rational Choice Model

Yang, Resource Constraints and the Criminal Justice System: Evidence from Judicial Vacancies


Why Are Public Defenders So Good?

In my last post, I provided a graph suggesting public defenders have above average win-rates. Most people find this surprising. Actually, this fits neatly into a model of the LSM where defense attorneys are profit maximizers and public defenders are sentence minimizers. Profit maximization does not imply sentence minimization. Instead, defense attorneys focus on “Win-Stay” and “Lose-Stay” outcomes. To see what I mean, consider Bayes’ Rule


All of this means that the probability of A conditional on B equals the probability of B conditional on A multiplied by the probability you assign to A, over the probability of B conditional on A multiplied by the probability you assign to A plus the probability of B conditional on not-A multiplied by the probability you assign to not-A. If the above isn’t clear, check out Bryan Caplan’s excellent lecture notes or this post at Econlog.

Here’s an example relevant to the defense attorney profit maximization problem:

P(A|B) = P(Attorney is Optimal|Bad Case Outcome)
P(B|A)=  P(Bad Case Outcome|Attorney is Optimal)
P(A) = Probability Attorney is Optimal
P(~A)= Probability Attorney is not Optimal

The profit maximizing attorney wants to persuade clients with bad outcomes that their attorney was still the correct choice. This way, the attorney still has access to that client’s network (and of course for future cases with the same client). In order to do this, attorneys should focus on increasing their clients’ subjectively held belief that they are high quality and increasing the clients’ belief that bad outcomes with high quality attorneys are common. For simplicity, let’s assume that the attorney’s clients will stay with them or recommend them to others if the attorney wins their case.

These incentives create a potential agent-principal problem in the attorney-defendant relationship. If an hour of signaling “I have a great win-rate” does more to increase the probability of Lose-Stay outcomes than an hour of work increasing the probability of winning, the attorney will invest too little (from the defendant’s perspective) in actually winning.

Public defenders, as sentence minimizers, don’t have this problem. Basically – and this can be seen in the data – the average public defender is a better agent than the average private defense attorney. Of course public defenders have obvious weaknesses – essentially zero budget for non-procedural trial inputs, for example. But with respect to procedural inputs, they should behave as if they have been given infinitely large budgets.


The Randazza Critique: An Objection

From commenter marcorandazza on a review of a firm in Massachusetts that apparently profiles attorney win-loss records:

This is utterly moronic.

When I hear of a lawyer who “never lost a case,” I hear “this lawyer sucks.”

Any idiot can pick winners. The real skill comes when you take a hard case. Take a dozen hard cases in a row, and lose them all. I’ve certainly taken my share of cases where I had to tell the client “you have a really low chance of winning, but your case is important” or just “you’re in deep shit, but I’ll do my best.”

I’m proud of many of my “losses.” I brag about them. In fact, one of the best things that ever happened to my career was losing a case 0-7 in the Florida Supreme Court. Ever since that case, I’ve handled lots of cases that hang on its logic. And, when my adversary inevitably makes the argument I made in that case, I say “your honor, I love my adversary’s argument. In fact, I wish that it were the law. However, I made the exact same argument to the Florida Supreme Court and the court ruled against me 0-7”


These two dipshits should just go backpack around Europe. Their idea is trash.

I frequently encounter two chief criticisms to attorney win-rate analysis:

  • Trial inputs and output are highly heterogeneous and classifying outcomes as wins or losses is extremely complicated.
  • Many attorneys select their case portfolios for easy cases to give themselves an artificially high win-rate.

I think (1) is a better criticism and I’ve addressed it briefly before (I’ll write about it later as well). Basically, I think the argument that heterogeneous trial inputs and outputs is a serious impediment to win-rate analysis is overrated. It’s technically complicated but theoretically straightforward to consider different types of trial outcomes (e.g. accounting for plea bargains) and analyze sentence quality controlling for relevant variables.

Criticism (2) seems to me much weaker but oddly more commonly voiced. The most important theoretical objection:

Win-rates are not commonly analyzed so I’m not sure why attorneys would be aggressively pursuing high win-rates at the expense of income from potential clients.

The most important empirical objection:

Public defenders are the least likely of all criminal defense attorneys to select their cases and they tend to have win-rates above the median. If attorneys were biasing their win-rates by heavily selecting their client portfolios, we would definitely not expect to see something like this:



Judge William Blackstone on the proper orientation toward new ventures:

The general expectation of so numerous and respectable an audience, the novelty, and (I may add) the importance of the duty required from this chair, must unavoidably be productive of great diffidence and apprehensions in him who has the honour to be placed in it. He must be sensible how much will depend upon his conduct in the infancy of a study, which is now first adopted by public academical authority; which has generally been reputed (however unjustly) of a dry and unfruitful nature; and of which the theoretical elementary parts, have hitherto received a very moderate share of cultivation. He cannot but reflect that, if either his plan of instruction be crude and injudicious, or the execution of it lame and superficial, it will cast a damp upon the farther progress of this most useful and most rational branch of learning; and may defeat for a time the public-spirited design of our wise and munificent benefactor. And this he must more especially dread, when he feels by experience how unequal his abilities are (unassisted by preceding examples) to complete, in the manner he could wish, so extensive and arduous a task; since he freely confesses, that his former more private attempts have fallen very short of his own ideas of perfection. And yet the candour he has already experienced, and this last transcendent mark of regard, his present nomination by the free and unanimous suffrage of a great and learned university, (an honour to be ever remembered with the deepest and most affectionate gratitude,) these testimonies of your public judgment must entirely supersede his own, and forbid him to believe himself totally insufficient for the labour at least of this employment. One thing he will venture to hope for, and it certainly shall be his constant aim, by diligence and attention to atone for his other defects: esteeming, that the best return which he can possibly make for your favourable opinion of his capacity, will be his unwearied endeavours in some little degree to deserve it.

On a related note, Blackstone Trial Analytics is now open for utility maximization business.


Should Public Defender Caseloads Matter for Indigent Defense Outcomes?

Unsurprisingly, I believe the answer is yes; surprisingly, I expect heavy caseloads should improve defense outcomes. My reasoning:

  • Public defenders (PDs) behave as if they are not budget constrained (with respect to trial inputs they supply themselves).
  • Heavy caseloads should increase the probability of seeing any given case continued on a given day.
  • The average indigent defendant is more likely than the average wealthy defendant to have a prior criminal record, and the marginal disutility of additional criminal charges is probably strongly diminishing. In other words, prosecutors quickly lose (or often lack) the ability to tempt indigent defendants with plea bargains that offer features like amended charges with better labor market signaling. Indigent defendants have more taste for trials, which are costly to prosecutors.

(1) strongly suggests that PD clients are relatively more expensive to prosecute, especially at general district court levels where things like (costly) expert testimony are less common. (3) also suggests a level effect; when indigent defendants and relatively wealthy defendants have identical case details, the indigent defendants will probably do somewhat better in terms of trial outcomes as a group. (2) implies that as PD caseloads increase, the probability of continuances across the PD portfolio increase. Sentence maximizing prosecutors will know this, and will be induced to offer more favorable pleas or drop charges as needed.

Thoughts on Utah v. Strieff

From the Syllabus:

Narcotics detective Douglas Fackrell conducted surveillance on a South Salt Lake City residence based on an anonymous tip about drug activity. The number of people he observed making brief visits to the house over the course of a week made him suspicious that the occupants were dealing drugs. After observing respondent Edward Strieff leave the residence, Officer Fackrell detained Strieff at a nearby parking lot, identifying himself and asking Strieff what he was doing at the house. He then requested Strieff’s identification and relayed the information to a police dispatcher, who informed him that Strieff had an outstanding arrest warrant for a traffic violation. Officer Fackrell arrested Strieff, searched him, and found methamphetamine and drug paraphernalia. Strieff moved to suppress the evidence, arguing that it was derived from an unlawful investigatory stop. The trial court denied the motion, and the Utah Court of Appeals affirmed. The Utah Supreme Court reversed, however, and ordered the evidence suppressed.

Held: The evidence Officer Fackrell seized incident to Strieff’s arrest is admissible based on an application of the attenuation factors from Brown v. Illinois, 422 U. S. 590.

In previous posts, I’ve argued that judges are expert adjudicators. Just as we should trust the average view of plumbers over our own views on plumbing, or the average view of economists on economics, we should trust the average views of judges on adjudicating. So is the above decision optimal? Actually, I doubt that it is. The Justices on the Supreme Court voted 5-3 against suppression, but if you factor in the opinions of each judge who considered the case in lower courts, the vote appears to be 9-8 in favor of suppression.

Additionally, the majority flexes its consequentialism muscle, arguing (Thomas writing) that the exclusionary rule is socially costly. Paging Dr. Landsburg.