New York Federal Criminal Practice Blog
June 9, 2008

Second Circuit Holds Government Did Not Breach Plea Agreement by Supporting Higher than Estimated Guidelines Level at Sentencing

What a district court giveth in United States v. Allen, 2008 WL 1944549 (S.D.N.Y. April 30, 2008) discussed here, the Second Circuit taketh in United States v. Habbas, 05-6142-cr, 2008 WL 2220676 (2d Cir. May 30, 2008).  In a decision underlining the need for defense attorneys to secure clarifying language, either orally or in writing, as to the binding nature of Guidelines estimates in plea agreements, the Court held in Habbas that the government did not breach a plea agreement by supporting a higher Guidelines level than it had estimated in a plea agreement, even though the facts justifying the increased level were known to the government at the time of the original estimate. 


Habbas’ co-defendant had pled to obstruction of justice arising out of his involvement in an elaborate conspiracy to frame an individual for assault.  The government estimated his Guidelines level in the plea agreement at 16, yielding a sentencing range of 27 to 33 months.  The Probation Department, however, added a 4-level leadership role enhancement to this estimate, producing a Guidelines level of 20 and a sentencing range of 41 to 51 months (i.e. approximately 50% higher than the government’s estimated range).  At sentencing, the government supported the 4-level increase, saying that its failure to include it in the plea agreement was a “mistake.”  The court ignored both the government’s and Probation Department’s estimate, imposing a sentence of 96 months, in light of the cruelty and cynicism of the crime (the victim had spent 7 weeks in custody before the plot unraveled).


Rejecting the defendant’s argument that the government’s support of the higher Guidelines level was a breach of the plea agreement, the Court gave three reasons: 

(a)     No Binding Language in Plea Agreement

First, the Court held that the plea agreement at issue clearly put the defendant on notice that the estimate in the plea agreement was not binding, and the government was “likely to advocate for a higher sentence.”  For example, the plea agreement stated that its “estimate . . . is not binding on the [United States Attorney’s] Office” and that “the government reserves the right to argue for a sentence beyond that called for by the Guidelines based on the factors set forth in 18 U.S.C. § 3553(a).” 

It should be noted that plea agreements from the E.D.N.Y., where Habbas originated, routinely contain the phrase about the non-binding nature of the estimate, but this phrase is commonly understood as a reservation of the government’s right to argue a different estimate based on new facts, or some exotic guideline no-one had anticipated.  It has never been viewed as carte blanche to permit routine adjustments on meat and potatoes issues like the defendant’s role in the offense.  Moreover, reserving a right to argue for a higher sentence under § 3553(a) factors is not the same as reserving a right to alter the Guidelines calculation.

At least recognizing that their analysis of the plea agreement language here has wider implications, the Court notes that “in certain circumstances government deviation from its prior estimate could conceivably produce serious unfairness.”  But surely a sentencing range that is 50% higher than originally estimated is serious unfairness?  And what about the “serious unfairness” to the defendant who presumably pled guilty in reliance on the estimate in the plea agreement?  There is no mention in the decision as to whether the defendant was offered the right to withdraw his guilty plea.

In response to that argument, the Court claims that this is not a case of the government “revers[ing] its position regarding the applicability or effect of a particular provision, upsetting a reasonable reliance by the defendant on the government’s stated position.”  But surely back-tracking on the absence of a role adjustment is precisely a reversal of position.  The Guidelines, as the Court notes in Habbas, may be “complex,” but role adjustments are Federal Sentencing 101. 

(b)    No Bad Faith

As a fallback position, the Court points out that there was no suggestion that the government had acted in bad faith.  The government, under “the pressures to prepare a Pimentel estimate” to “accommodate” the defendant simply “failed to notice the possible applicability” of the aggravating role enhancement. 

Putting aside the Court’s confusing reference to United States v. Pimentel, 932 F.2d 1029, 1034 (2d Cir. 1991) (which as the Second Circuit Blog notes in commenting on Habbas generally refers to a non-binding informational letter from the government containing a guidelines estimate that is not a plea agreement at all), the Court’s statements here fail to recognize many realities of plea bargaining.  The estimates in plea agreements are not simply hastily included as “accommodations” to defendants.  They are the meat of the plea agreement, often hammered out over days and weeks of negotiations with defense counsel.  And they also go through at least one if not several layers of oversight at the U.S. Attorney’s office.  Far from being hasty “favors” to the defendant, they are the negotiated basis of his/her decision to plead guilty under a plea agreement. 

As the district court pointed out in Allen, where the government back-tracked from its Pimentel estimate, misleading a defendant of their potential liability under the Sentencing Guidelines, whether in bad faith or not, “damages the integrity of our justice system.”  Unlike the non-binding Pimentel letter at issue in Allen, the plea agreement at issue in Habbas is a contract, to which one applies principles of contract construction.  This is the first time I’ve heard of the theory that a contract may be repudiated by one party based on a unilateral, good faith “mistake,” that was not fraudulently induced by the other contracting party. 

(c)    No Harm to Defendant

Finally, the defendant was not harmed by the government’s changed position, because the district court essentially ignored the government’s and Probation Department’s calculation in imposing a sentence of eight years.  In other words, the dispute over the plea agreement in this case was academic.  So, why, oh why, did the Court need to weigh in with an unnecessary and confusing analysis of that issue?


The Court suggests in a footnote that in plea agreements going forward, the government reserve the right to change the Guidelines estimate with new facts and to make “good-faith changes” based on existing facts if “further study shows the applicability of guideline provisions not considered in making the estimate.”  This suggestion risks making plea agreements next to worthless.  What use is a today-only, cursory, hasty and completely reversible Guidelines estimate to a defendant making the profound decision to plead guilty and subject him/herself to a prison sentence? 

A more productive approach might have been to permit changes to the estimate based on new facts or an unanticipated application of some arcane, little-used provision of the Guidelines that the parties had not addressed.  In addition, should the estimate change on that basis, a defendant should be permitted to withdraw his/her plea.

At the very least, Habbas underlines the need for defense lawyers to document their negotiations on the actual plea estimate, and seek clarifying language - either written in the plea agreement or orally on the record at the guilty plea - that the government considers itself bound by the estimate – especially meat and potatoes issues like loss, weight, relevant conduct and role.

See Archives for all posts since September 2007.