Opinion by Chief Judge WALLACE; Dissent by Judge REINHARDT.
WALLACE, Chief Judge:Vargas challenges the sentence imposed for his securities fraud conviction. The district court had jurisdiction pursuant to 18 U.S.C. § 3231. We have jurisdiction pursuant to 18 U.S.C. § 3742 and 28 U.S.C. § 1291. We affirm.
I
In 1985, Vargas began working for Reserve Financial Group, which was formed in 1984 by John Anthony Genetti for the purpose of creating, selling, and marketing bogus municipal bonds. Genetti masterminded the scheme in which Vargas and three others participated. Vargas falsely represented to prospective investors that the securities were backed by governmental agencies, that they were triple-A rated, and completely liquid, when in fact the securities did not even exist.
In October 1992, Vargas pleaded guilty to two counts of securities fraud, in violation of 15 U.S.C. §§ 77a and 78a, and to one count of submitting a false loan application to a federally insured institution, in violation of 18 U.S.C. § 1014. The two counts of securities fraud charged Vargas with sales of bogus bonds on November 4 and November 19, 1987. The third count charged him with submitting a false loan application on or around October 15, 1991. Although Vargas and the government had stipulated to an offense level which was based on application of the 1991 Guidelines, the district court realized during the course of an extensive sentencing hearing that the 1987 Guidelines would have to be applied to the securities fraud counts due to ex post facto concerns. Convinced that the base offense level provided by section 2F1.1 of the 1987 Guidelines did not adequately reflect Vargas’s culpability, the district court departed upward four levels on the basis of the extraordinary amount of loss caused by the scheme.
*825In an earlier memorandum disposition, we concluded that the district court had the legal authority to depart upward on the ground that the amount of loss was substantially greater than the maximum amount of $5 million provided for in loss table to section 2F1.1, and on the ground that application note 10 expressly authorized such departures. See United States v. Vargas, No. 93-50236, 1994 WL 622987 (9th Cir. Nov. 9, 1994) (unpublished disposition). However, the district court did not make adequate findings concerning the amount of loss or provide a reasoned explanation for the extent of the departure. We therefore remanded to the district judge to provide him an opportunity to make such findings and to provide a reasoned explanation of the extent of the departure.
On remand, the district court found that the amount of loss was at least $20 million and again departed upward four levels. See Supplemental Findings of Fact and Conclusions of Law, United States v. Vargas, No. CR92-483(a) (C.D.Cal. Feb. 23, 1995). The district court enumerated several justifications for the extent of the departure. First, the district court justified the four level departure by reference to the subsequently enacted Guidelines, which impose stiffer penalties for amounts greater than $5 million. However, as explained in our previous memorandum disposition, reliance on the subsequently enacted Guidelines is erroneous because the subsequently enacted version of the loss table in section 2F1.1 alters rather then merely clarifies preexisting law. See United States v. Martinez, 946 F.2d 100, 102 (9th Cir.1991) (Martinez).
The district court also reasoned that “[d]e-parting upward one level for each $5,000,000 increase in loss reasonably addresses the gravity of the harm not otherwise accounted for in the applicable version of the guidelines.” It went on to explain that its departure was proportionate, on a linear basis, given the amount of loss involved.
II
The district court’s factual finding that the amount of the loss exceeded $20 million is not clearly erroneous and is sufficiently precise to enable us to determine whether the extent of the departure was reasonable. The question that remains is whether the departure was reasonable in light of the “structure, standards and policies of the Act and Guidelines.” See United States v. Lira-Barraza, 941 F.2d 745, 746-51 (9th Cir.1991) (en banc).
The only possible problem with the district court’s basic approach of adding one offense level for each additional $5 million in loss is that the loss table in section 2F1.1 does not follow a simple linear pattern. Rather, the loss table in the applicable 1987 version of section 2F1.1 roughly adds one offense level each time the amount of loss doubles. If the district court were required to extrapolate from the loss table, a departure of only two levels would be appropriate for a loss of $20 million.
The question before us is whether the district court was permitted to follow a linear approach in determining the extent of the departure, or whether the district court was required to extrapolate from the loss table and follow its rough exponential pattern. We hold that it was reasonable for the district court to add one offense level for each additional $5 million in loss. In the 1987 version of section 2F1.1, the Sentencing Commission chose to end the loss table at $5 million and stated that “[departures above the applicable guideline may be warranted if the loss substantially exceeds that amount.” U.S.S.G. § 2F1.1 (n.10) (1987). There is no basis for assuming that the Commission intended the district court to follow the pattern provided in the loss table when the loss substantially exceeded $5 million. If the Commission had intended such a result it could easily have said so. Indeed, that the Commission ended the loss table at $5 million and authorized the district court to depart upward if the loss substantially exceeded $5 million could have reflected the intent to give the district court greater discretion in calculating the sentence in fraud cases involving extraordinarily great losses.
We give no consideration to the 1991 amendment to section 2F1.1. As we have already pointed out, no weight can be accord*826ed subsequent amendments to the Guidelines in determining the reasonableness of a departure where the amendments alter rather than clarify existing law. Martinez, 946 F.2d at 102. The new version of section 2F1.1 alters the pattern contained in the loss table to increase substantially the offense levels for crimes involving large losses and has also replaced the old application note 10, which expressly authorized upward departures based on the amount of loss, with a new application note 10 allowing the district court to take other factors unrelated to the amount of loss into account. Because the subsequent amendments to the Guidelines alter rather than clarify the law, they are irrelevant to our analysis.
Resolving a similar problem involving an analogous Guideline provision, the First Circuit concluded that it was reasonable for a district court to depart upward two levels based on the embezzlement of $11 million, where the applicable loss table did not distinguish between crimes involving more than $5 million. See United States v. Harotunian, 920 F.2d 1040, 1045-46 (1st Cir.1990). Although the loss table involved in Harotunian was somewhat less predictable than the loss table involved here, and the First Circuit’s statements concerning the appropriate amount of deference afforded the district court’s decision to depart are phrased somewhat differently, the court expressly rejected “mechanical approaches to assessing the reasonableness of departure decisions.” Id. at 1045. We agree with the First Circuit that the reasonableness of departure decisions should not be determined by rigid mechanical approaches. As we explained in Lira-Barraza, “a reasonableness standard assumes a range of permissible sentences.” 941 F.2d at 751.
The district court is required in every case to provide “a reasoned explanation of the extent of the departure founded on the structure, standards and policies of the Act and Guidelines.” Id. When possible, it should do so by “analogizing to other Guidelines provisions.” United States v. Hicks, 997 F.2d 594, 599 (9th Cir.1993). However, it is neither possible nor necessary in every case for the district court to point to an analogous Guideline provision. Nor is the district court required in every case to extrapolate mechanically from the relevant provision. In this ease, the district court’s decision to increase the offense level by each additional $5 million in loss is consistent with the basic structure of section 2F1.1, which increases the offense level as the amount of loss increases. It also reflects the fundamental policy of section 2F1.1 to impose greater punishment on those who cause greater social harm and is consistent with application note 10, as there is no question that $20 million “substantially exceeds” $5 million. In light of the Commission’s decision to give the district court discretion to depart upward when the amount of loss substantially exceeds $5 million rather than to continue with the pattern contained in the loss table, we will not reverse the district court for following a reasonable path. In short, there is no reason to assume that the Commission wanted the district court to do exactly what the Commission told the district court it need not do. We conclude that the four level upward departure was reasonable.1
AFFIRMED.
. The dissent would require the district judge's separate linear schedule to be fused to the table in the Guidelines. See dissent at 827 n. 2. However, as the dissent recognizes, the district judge adhered to the Guidelines as required and used a "+11” level for “Amount of loss over $5 million.” Id. at 827 n. 1. Because the table did not calculate levels for losses over $5 million, the judge created a separate, linear schedule for amounts over $5 million and used that schedule to arrive at a “ +4” upward departure. Id. Just as the judge was not bound to the pattern in the 1987 table, he was permitted to create a separate departure schedule, so long as it was reasonable.