While on parole, Jerry Coleman led police in a high-speed chase with two other parolees in his car. The police saw two guns thrown from the car before they finally stopped Coleman and his companions. After Coleman was arrested for reckless driving, his parole was revoked.

In his parole hearing, the examining committee found Coleman guilty of reckless driving, and of association with individuals known to have criminal records. He was not found guilty of possessing a gun. The examiners calculated Coleman's offense severity rating as category one based on his conviction for reckless driving. Category one is the least severe offense severity rating. Based on his prior convictions, Coleman received a salient factor score of two. A salient factor score of zero is most severe; a score of four is least severe. Although the policy guidelines recommend parole between 12 and 16 months for individuals with offense severity ratings of one and salient factor scores of two, the panel decided to exceed the guidelines because Coleman had engaged police in a high speed chase. They recommended parole in 24 months. Commission regulations permit the panel to exceed recommended guidelines if the Commission can show good cause. See 18 U.S.C. Sec. 4206(c). The panel expressly noted that the two other parolees in Coleman's car--Jones and Tucker--would also be paroled after 24 months.

The examining panel's decision was reconsidered by the Regional Commissioner, who decided that 24 months was too lenient. He recommended that Coleman remain in prison for 40 months, until the end of his sentence. The regional commissioner did not mention Coleman's co-defendants' parole dates. He added, as reasons for exceeding the guidelines, that two other parolees were in the car, and that firearms were thrown from the vehicles. The National Commission affirmed the Regional Commission, explicitly stating that Coleman's treatment was not disparate with his co-defendants' because Coleman posed a worse risk and because Coleman drove the car.

No published regulation requires the Commission to consider co-defendants' treatment in making parole decisions. The Parole Commission's Rules and Procedures Manual in effect at that time does suggest, however, that the Commission should consider co-defendant treatment, and that disparity between similarly situated offenders should not exist without legitimate reasons. See United States Parole Commission Rules and Procedures Manual (1985) (Manual), Procedure 2.20-09. (explaining that " '[u]nwarranted codefendant disparity' refers to different parole decisions for similarly situated offenders where no legitimate reasons for the difference in decisions exists. It is to be remembered that different decisions for codefendants are not necessarily inappropriate."); Procedure 2.24-01(b) (stating that "an attempt should be made, where feasible, to ascertain the parole action taken relative to any codefendants.").

Coleman's co-defendant Jones was found guilty of a category three offense (more severe than Coleman's) because he had a gun. He was rated with a salient factor score of zero (also more severe than Coleman's) because he had four or more prior arrests. The guidelines recommend parole between 24 and 32 months for these scores. The Commission set Jones for parole after 24 months. Coleman contends that he received disparate treatment. Jones was released earlier despite his more severe ratings. Additionally, Jones was released as early as possible within his guidelines, while Coleman, who had better ratings, was forced to remain in prison beyond his recommended guidelines.

Coleman challenges his parole denial on two grounds. First, he argues that the Commission acted improperly by ignoring its own regulation requiring that it consider co-defendant parole disparity in making parole decisions, and that its failure to follow this rule is reversible error. Second, he argues that the Commission exceeded its statutory authority by rendering a decision outside the guidelines without demonstrating good cause. Prior to resolving these issues, we must decide whether this court has jurisdiction to review the Parole Commission's decision.

In Wallace v. Christensen, 802 F.2d 1539 (9th Cir.1986) (en banc), this Circuit held that federal courts can review Parole Commission decisions only within a very narrow scope. Courts may not review any decision involving the "exercise of judgment among a range of possible choices or options." Id. at 1552. They may review whether the Commission has acted outside its statutory authority or has violated the Constitution. Id. at 1552-54. We illustrated this distinction with several examples. First, if a plaintiff challenges the factors considered by the Commission in making a parole decision, a court can review whether the Commission considered factors that the statute requires it to consider. Id. at 1551. Presumably the same reasoning permits courts to review whether the Commission considered factors prohibited by statute. Courts may not, however, review the relevance the Commission attributes to the information it considers. Id. Second, if the plaintiff challenges the Commission's decision to exceed statutory guidelines, courts may review both whether the Commission exceeded the guidelines and whether the Commission stated good cause. Id. Courts may also review the Commission's decisions for constitutional violations, such as whether a decision is so arbitrary and capricious as to violate due process. Id. at 1552.

Wallace makes clear that this court has jurisdiction to review Coleman's second claim--that the Commission did not state good cause for exceeding the recommended guidelines. Id. at 1551. Whether this court may review Coleman's first claim--that the Commission failed to follow its own rule concerning co-defendant disparity--is not directly discussed by Wallace or by any other Ninth Circuit case. The district court held that it lacked jurisdiction to consider co-defendant disparity because variation among co-defendants "is a matter entirely within the discretion of the Commission." We agree with this conclusion, though not with the district court's reasoning.

The district court cites for support Roberts v. Corrothers, 812 F.2d 1173 (9th Cir.1987). The district court's reliance on Roberts is inappropriate. In Roberts the court refused to review the claim that the Commission abused its discretion by giving different severity ratings to co-defendants. The court held that how the Commission weighed various factors in arriving at severity ratings was committed to the Commission's discretion. Id. at 1180. In this case, by contrast, Coleman argues that the Commission exceeded its authority by ignoring its own regulation requiring consideration of co-defendant disparity. Coleman does not argue, as in Roberts, that co-defendants must receive identical severity ratings, or even that co-defendants with identical severity ratings must receive identical treatment. Rather, he argues that by enacting regulations, the Commission has removed certain decisions from its discretion, and thereby made its failure to follow its own rules subject to review. The district court, therefore, improperly relied on Roberts in deciding that it lacked jurisdiction.

To determine whether this court can review the claim that the Commission violated its own rules, we must first determine whether the Commission is bound to follow its own internal regulations concerning co-defendant disparity. If the Commission is bound by these regulations, then ignoring them is not within the Commission's discretion, and violations of these internal rules would be reviewable. See Wallace, 802 F.2d at 1552. We conclude that the Parole Commission is not bound by these internal regulations because they are merely precatory. Accord, Lynch v. United States Parole Comm'n, 768 F.2d 491, 496-98 (2d Cir.1985). The manual does not require the Commission even to consider the status of co-defendants. It simply states that an attempt should be made to ascertain this status "where feasible." This language leads us to conclude that similar treatment for co-defendants is among the Manual's aspirations, not among its requirements. See United States v. Fifty-Three Eclectus Parrots, 685 F.2d 1131, 1136 (9th Cir.1982) (stating that agency pronouncements are binding only if they are substantive rules affecting individual rights and obligations, and that mere statements of policy, organization, procedure, or practice will not be binding).

Any decision to follow these guidelines is therefore properly within the Commission's discretion, and the exercise of that discretion is not reviewable by this court. As a result, we lack jurisdiction to consider Coleman's first claim.

The Commission decided to exceed the recommended guidelines for Coleman's parole for three reasons: (1) Coleman initiated a high-speed chase with police resulting in damage to vehicles; (2) Coleman drove a car with two other parolees; and (3) Coleman's companions were armed. A showing of good cause requires that the Commission advance in good faith reasons that are not " 'arbitrary, irrational, unreasonable, irrelevant or capricious.' " Wallace, 802 F.2d at 1551 (quoting Jt. H.R.-S. Conf.Rep. No. 838, 94th Cong. 2d. Sess. 19, 27, reprinted in 1976 U.S.Code Cong. & Admin.News 335, 351, 359). The district court found that the Commission had demonstrated good cause. Coleman argues that the Commission failed to show good cause for exceeding the recommended guidelines for two reasons.

First, Coleman argues that the high-speed chase may not be used as an aggravating factor because the Commission based his offense severity rating on the same chase. He argues that using the same action to justify a recommended parole date and to justify exceeding that date is irrational. The Circuits are split on this issue. The Second and Eighth Circuits have prohibited such double-counting as irrational. See Lynch, 768 F.2d at 496; Alessi v. Quinlan, 711 F.2d 497, 500 (2d Cir.1983); Briggs v. United States Parole Comm'n, 736 F.2d 446, 450 (8th Cir.1984). The Ninth Circuit has stated in dicta that it will permit this double-counting, at least in some circumstances. See Torres-Macias v. United States Parole Comm'n, 730 F.2d 1214, 1218 (9th Cir.1984); Reynolds v. McCall, 701 F.2d 810, 813 (9th Cir.1983). On other occasions, the Ninth Circuit has suggested disapproval for double-counting. See Walker v. United States, 816 F.2d 1313, 1316 (9th Cir.1987).

We need not reach the double-counting issue. The aggravating factor of engaging in a high speed chase to elude police officers was not used in setting Coleman's offense severity rating. His offense severity rating was based on reckless driving. Although the same behavior underlay both charges, his reckless driving was exacerbated by his attempt to elude the police.

Coleman also argues that the Commission acted irrationally by increasing his punishment based on his co-defendants' behavior. He claims that the presence of armed parolees in the car did not make his behavior more objectionable. However, Coleman knew that his passengers were parolees; he had been convicted with them. And he knew that he was not supposed to be in their company. See Manual Sec. 2.40(10) (stating that every parole is granted on condition that the parolee not associate with anyone having a criminal record). Violating the prohibition against associating with convicted criminals, together with engaging in a high speed chase to elude police, supports the finding of good cause to exceed the recommended guidelines.

We need not consider whether Coleman's co-defendant's possession of firearms could be considered good cause for exceeding statutory guidelines. Engaging in a high-speed chase and association with other parolees sufficiently support the Commission's finding of good cause. We therefore AFFIRM.