Saturday, March 24, 2012

Plea Bargains Generally and in Tax Cases; Moneyball (3/24/12)

In two cases, the Supreme Court has recognized the central role of plea bargaining in the criminal system and held the the right to effective assistance of counsel extends to advising the client in plea negotiations.  Missouri v. Frye, ___ U.S. ___, 2012 U.S. LEXIS 2321 (2012), here (failure to advise the defendant of a time-limited plea offer by the prosecutor); and Lafler v. Cooper, ___ U.S. ___, 2012 U.S. LEXIS 2322 (2012), here (providing the defendant highly questionable advice to reject a proffered plea and go to trial instead).  In an earlier case, the Court had anticipated this broader holding the context of a plea agreement where the attorney failed to advise the defendant of the collateral immigration consequences of the plea.  Padilla v. Kentucky, 559 U. S. ___, ___ (2010), here.

The crux of the holdings is (Missouri v. Frye):
The State's contentions are neither illogical nor without some persuasive force, yet they do not suffice to overcome a simple reality. Ninety-seven percent of federal convictions and ninety-four percent of state convictions are the result of guilty pleas. See Dept. of Justice, Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics Online, Table 5.22.2009, sourcebook/pdf/t5222009.pdf (all Internet materials as visited Mar. 1, 2012, and available in Clerk of Court's case file); Dept. of Justice, Bureau of Justice Statistics, S. Rosenmerkel, M. Durose, & D. Farole, Felony Sentences in State Courts, 2006-Statistical Tables, p. 1 (NCJ226846, rev. Nov. 2010),; Padilla, supra, at ___ (slip op., at 15) (recognizing pleas account for nearly 95% of all criminal convictions). The reality is that plea bargains have become so central to the administration of the criminal justice system that defense counsel have responsibilities in the plea bargain process, responsibilities that must be met to render the adequate assistance of counsel that the Sixth Amendment requires in the criminal process at critical stages. Because ours "is for the most part a system of pleas, not a system of trials," Lafler, post, at 11, it is insufficient simply to point to the guarantee of a fair trial as a backstop that inoculates any errors in the pretrial process. "To a large extent . . . horse trading [between prosecutor and defense counsel] determines who goes to jail and for how long. That is what plea bargaining is. It is not some adjunct to the criminal justice system; it is the criminal justice system."  Scott & Stuntz, Plea Bargaining as Contract, 101 Yale L. J. 1909, 1912 (1992). See also Barkow, Separation of Powers and the Criminal Law, 58 Stan. L. Rev. 989, 1034 (2006) ("[Defendants] who do take their case to trial and lose receive longer sentences than even Congress or the prosecutor might think appropriate, because the longer sentences exist on the books largely for bargaining purposes. This often results in individuals who accept a plea bargain receiving shorter sentences than other individuals who are less morally culpable but take a chance and go to trial" (footnote omitted)). In today's criminal justice system, therefore, the negotiation of a plea bargain, rather than the unfolding of a trial, is almost always the critical point for a defendant. 
To note the prevalence of plea bargaining is not to criticize it. The potential to conserve valuable prosecutorial resources and for defendants to admit their crimes and receive more favorable terms at sentencing means that a plea agreement can benefit both parties. In order that these benefits can be realized, however, criminal defendants require effective counsel during plea negotiations. "Anything less . . . might deny a defendant effective representation by counsel at the only stage when legal aid and advice would help him.'" Massiah, 377 U. S., at 204 (quoting Spano v. New York, 360 U. S. 315, 326 (1959) (Douglas, J., concurring)).
I think most criminal defense attorneys recognize the power of the reasoning.  Confidence in the fairness of the criminal justice process requires competence of counsel for plea agreements.

I do want to address statistics.  As quoted above, the Court observed the "simple reality" that "Ninety-seven percent of federal convictions and ninety-four percent of state convictions are the result of guilty pleas."  Since this is a federal tax crimes blog, I and, I think, readers of the blog, might like to know what the plea rate in federal tax crimes is.  We know the plea rate is high, but I have found it difficult to ascertain what the plea rate actually is.

I am aware that, at one time, DOJ Tax claimed a 97% conviction rate.  That is not the same as a 97% plea rate, but obviously the great bulk of the 97% conviction rate (if that is an accurate rate) would be pleas.  I have previously expressed skepticism about that conviction rate because I could not make it square with other publicly available data.  (Of course, as Benjamin Disreali mused, “there are lies, damned lies, and statistics,” so worrying about these statistics is perhaps a fool's errand; but perhaps not since the Supreme Court finds such statistics important.)

DOJ Tax’s claim as to a 97% conviction rate was  presented in a report titled: Eileen J. O’Connor, Assistant Attorney General, Statement to the Senate Finance Committee, titled Filing Your Taxes: an Ounce of Prevention Is Worth a Pound of Cure 3 (4/12/07).  The report is here and, as indicated the statement is at page 3, here: I could not reconcile the claim statistics with IRS statistics, however.  So, I made a written request to DOJ Tax for an explanation and the request was declined.

I expect that DOJ Tax has good statistics on (i) the conviction rate and (ii) the plea rate.  I think that DOJ Tax should publish those statistics.

Perhaps the issue is not that important -- does it really make a difference whether the conviction rate is 90 or 97% or the plea rate is 90 or 97%?  Perhaps it does.  It could suggest that, despite the lore that tax cases have a high rate of conviction, they really don't relative to general federal tax system.  It could mean that DOJ Tax is not that selective in picking cases to indict.  It could mean that DOJ Tax criminal prosecutors (including their AUSA colleagues in trying these cases) are not as good as nontax federal criminal prosecutors.  It could mean that convictions in tax cases are just harder to achieve than for the other federal crimes.  At any rate, it would at least be interesting to get the real statistics and see if we can play Moneyball with them.

For example, and just playing with the numbers we have (albeit suspect), assuming arguendo the claimed  conviction rate of 97% and the year for that statistic is representative, we might extend those statistics working from the known to the unknown and make some interesting speculations.  Let’s say 95% of federal tax crimes plead.  This means that only 5% of the tax indictments are tried.  Then, assuming a 97% conviction rate, it would appear the only 2 of 5 that go to trial end in conviction.  In other words, for those cases going to trial the defendant has an implied success rate of 60%.  And what that might further imply is that for at least some of the 95% pleading (say the 10% portion that are the weakest cases for the Government), the defendants might have been better off had they gone to trial.  But this is just speculation from a statistical data set that, I think, overstates the conviction rate (or, if correct for the year claimed, was not a representative year).

Other resources on the Supreme Court's Decisions in Missouri v. Frye and Lafler v. Cooper:

  • Erica Goode, Stronger Hand for Judges in the ‘Bazaar’ of Plea Deals (NYT 3/22/12), here.


  1. A layman might take your skepticism of Assistant Attorney General O'Connor's claim which she made to the Senate Finance Committee that DOJ Tax has a 97% conviction rate to be an assertion that she lied to that august group of Senators, especially given your pointed statement about your assumptions, i.e. "DOJ has good statistics... [and they] should publish them".

    It seems she was just telling the Senate to look at the IRS if they wanted to know what was going on because the DOJ just rubber stamps IRS recommendations. At least that's how I view rates beyond 90% approval on anything.

    GAO did a study Additional Time Needed to Complete Offshore TAx Evasion Examinations around the same time (March 2007). They had some interesting things to say about how it is inside IRS given the time pressure caused by the statute of limitations and their surprisingly limited budget.

    As a layman I find your "speculations" somewhat - er, speculative. If your assumption that "95% of federal tax crimes plead" is out by only 1 part in 20, the actual plea rate could be 90% or 100% depending which way the error goes which has a fantastic effect on your conclusion. But I assume you have solid ground under your feet.

    Another NYT take on the situation is Adam Liptak's Justices' Ruling Expands Rights of Accused in Plea Bargains Everyone including the court seems to be viewing this as if the plea bargains are almost always better deals than what anyone could get a judge to actually hand down.

    Given my recent study of my own choices after I "committed" my crimes of not filling out FBARs I had no idea existed I look at this Supreme Court decision the way you seem to be leaning. I probably go much farther than you - to me the OVDI plea bargain seems a preposterously bad deal if it is assumed the person really did not know. And given Geithner's demonstration of how much evidence can be piled up against a guy in a tax matter yet he still walks out confirmed as Treasury Secretary because the US Senate officially bought his line that he did not know, things look bad for the IRS politically and legally here.

    It seems to me the vast bulk of cases of the type of small fry OVDI is netting would without doubt be better off had they prepared themselves to face trials. For one thing DOJ couldn't possibly justify proceeding in all the many thousands of cases that would be generated. And for another the fact that FBAR prosecutions put the burden of proof on IRS coupled with a presumed IRS reluctance to create unfavorable precedents makes me wonder if there have been any cases at all where the guy in court even remotely resembles Moby Minnow.

    People who are in OVDI now looking seriously at opting out who think they might have difficulty convincing the IRS should take heart at this Supreme Court activity. If someone is in there who actually did not know FBAR existed they got bad advice. Even if they are as "guilty" as Geithner things look better than people here seem to believe. The Supremes now say that because the "justice" system isn't what they were pretending it was, i.e. for all practical purposes there no longer are trials - its just people being herded into OVDI type plea bargains under threat of harsh consequences should they dare the system to actually grant them a fair trial everywhere, and, given the Supreme Court pretence everyone is entitled to competent counsel when the outcome could be serious, what choice is left the IRS but to let anyone who wants out, out?

    1. I want to make clear that I am not asserting that the AAG Tax lied to Congress. I don't know.

      I am asserting that her statement to Congress does not appear to be consistent with other statistics I saw from other sources. These other sources were not head to head -- the old apples, oranges thing -- comparisons, but there did seem to be some inconsistency even taking that into consideration. There may be an explanation for the seeming inconsistency. I asked the AAG Tax in writing to explain further. She refused. I draw no inference from her refusal to explain her own numbers. I am, however, skeptical and would still like to see the data and assumptions underlying the statistic.

      That is all I intended to say.

      Jack Townsend

    2. I think her writing in the report is eloquent enough.

      "FY 2006
      Prosecution authorized (defendants)
      Prosecution declined
      Percentage of targets declined
      Grand Jury Investigation authorized*

      As the chart reflects, the percentage of targets that we decline to prosecute is relatively constant and low, ranging from three to six percent over this period. Moreover, our conviction rate is high—on the order of 97 percent for the 2006 fiscal year. The low percentage of declined cases and the high conviction rate suggests that the IRS is applying the appropriate standards in its case selection and that, as a general proposition, it is satisfactorily investigating the cases that it refers. Your staff asked us to comment on whether we are overly selective in evaluating IRS referrals. In a word, “no.” The Tax Division’s declination rate is not, by any definition, high. The Tax Division’s judgment in authorizing tax prosecutions is invaluable. It would be inappropriate to bring charges that cannot be proven in a court of law and tax cases can be difficult to prove. In addition, because of the significance of deterrence in tax law enforcement, it would be counterproductive to prosecute tax cases that might not result in a conviction."

      What is remarkable is that the IRS recommended only 1,180 taxpayers for prosecution. That is 1180/134,372,678 taxpayers = 1/100,000 approximately

      And yes, I do believe that 97% includes plea bargains.

      How many went to trial is a good question. Per her commentary, though, the point of this is deterrent, nothing more. So to divulge too much perhaps weakens that deterrent.

  2. After studying the Senate Geithner confirmation hearing partial transcript and other documents released by the Senate on the matter, I saw how adamant Geithner was that although he worked for the same employer for four years and filed taxes under the same assumptions and was nailed by an IRS audit for two of those years i.e. 2003 and 2004, it did not enter his mind that he was not nailed for the two other years, i.e. 2001 and 2002, because IRS could not touch him because of the statute of limitations. He told the Senators he had no idea. Whatever his accountant told him was the word of God. The accountant just told him he had no obligation to pay 2001 and 2002 back taxes and that's all he was interested in knowing it seems. Suddenly, when Obama tells him he's going to be the Secretary of the Treasury, he for the first time realizes that it was the statute of limitations that stopped the IRS from nailing him for 2001 and 2002. Independently of any desire to make himself look better at his upcoming Senate confirmation hearing, he told the US Senate, he decides he isn't going to hide behind the statute of limitations, no, he's going to pay up, just as he would have had he found out at any other time.

    Geithner was not telling lies. The President of the United States was standing by his side assuring the Senators his choice for Secretary of the Treasury had merely made the "mistakes of an honest man".

    So in that spirit, it is obvious, O'Connor isn't even making a mistake. She is an honest woman. Any discrepancy between what she said and statistics you can actually lay your hands on must be the mistakes of dishonest bureaucrats or something.

    1. John Doe

      May I suggest that your long multi-paragraph rants against Geithner are frankly getting a little tiresome and repetitive. It is Jack's call whether he posts them, but most have nothing to do with the main themes on this blog. Even comments regarding what should happen in OVDI (because Geithner did the same etc. etc.) are not particularly helpful, since its what is happening, not what you think should happen that concerns many here.

    2. I agree with Anon. This is just rant, while this blog is typically fact based and thoughtful.

      Many of us have personal reservations about hypocrisy of politicians and political appointees, but I fail to see what this has to do with the present discussion, and, more importantly, how it advances understanding of the subject.

  3. Some comments express skepticism of my skepticism of the Government claim of 97% conviction rate for the year.

    The real problem is with statistics. As the quote quips, there are lies, damned lies and statistics.

    My first job as a lawyer was with DOJ Tax Appellate. Our responsibility was to brief and argue cases in the Federal Courts of Appeals. A fellow attorney claimed a perfect record -- all wins on all issues -- in his then 3+ year tenure. That was better than any of the other goods lawyers experienced over a similar period. He was proud of it. Then, in my research, I happened to come across one of his cases that he had lost. I asked him about it. He pronounced that he did not include that case in the data set he used to calculate his perfect record because the reviewer had so changed his brief in the review process that the loss was the reviewer's fault, not his. There were other exclusions from his perfect record data set.

    There were further potential layers of inquiry. At least at that time, new lawyers in the Appellate Section were given easy cases -- all taxpayer appeals from taxpayer adverse decisions at the trial level with easy issues. From my perspective, it those cases were easy wins. After getting a few cases under the belt and earning the confidence of the Appellate Front Office and the reviewers, the better attorneys would be given more difficult cases not so easily won -- some government appeals from losses (the appellant generally has an uphill battle) and some taxpayer appeals in difficult cases (say split opinions from the Tax Court). An attorney had to prove his or her mettle to get the more difficult cases; if he or she did not, then he or she would continue to get run of the mine cases that generally were easy wins. Difficult cases have higher risk of loss. So the persons handling more difficult cases would not have even close to a perfect records and still would be the best lawyers in the section.

    Now, focusing on the criminal statistics, there are several phenomena at play in what appears to be an outstanding posted success record. It is not particularly important whether the success rate is 97% or some lesser figure (say 88 or 90%); it still a high conviction rate. This reflects DOJ Tax’s selectivity for cases -- it prosecutes relatively few cases and picks cases that are highly likely for conviction. This accounts in major part for the high guilty plea rate. Another factor in guilty plea rates, of course, is the incentives in the Government's plea bargain policies and the Sentencing Guidelines which incentivizes pleas. So the conviction bias (at whatever level it is) is built into the system.

    Although probabilities in cases are difficult to quantify, they can be useful to illustrate concepts for analysis. If the Government routinely accepted cases that had a probability of conviction of 80%, at least within that dataset, there would only be an 80% conviction rate (give or take a range for error). At least as I understand the Government's imagination, an 80% or, heaven forbid, something less would be a disaster for the Government. (The Government does not want the taxpayers to think that there is anything less than a virtual certainty of conviction, if prosecuted; it is in military metaphor, an expert rather than just a a sharpshooter or marksman.) Accordingly, it chooses to prosecute cases with virtual certainty of conviction (the only quibble I have is what percentage is virtual certainty).

    Bottom line, I would like to know how the DOJ constructs it database to state 97% conviction rate. Does it exclude certain cases with a high nonconviction rate? If so, why? Was the particular year a fluke or representative of other years? How do the statistics compare to IRS statistics of conviction rates for the cases handled by CI? If they don't compare, why not?

    The devil may be in the details.

    Jack Townsend


Please make sure that your comment is relevant to the blog entry. For those regular commenters on the blog who otherwise do not want to identify by name, readers would find it helpful if you would choose a unique anonymous indentifier other than just Anonymous. This will help readers identify other comments from a trusted source, so to speak.