dissenting:
For forty years, the Pend Oreille County public utility has maintained a trespass on tribal and allotted trust lands of the Kalispel Indian Reservation in Northeastern Washington. The majority’s decision denies the Kalispel landowners the compensation and injunctive relief to which they are entitled and contravenes the deterrent purpose of the assessment of damages under the Federal Power Act. For these reasons, and because the district court clearly erred in assessing damages against the public utility and abused its discretion in granting an indefinite stay of injunction, I must dissent.
1. Factual and Procedural Background
In 1952, the Pend Oreille County Public Utility District No. 1 (“PUD”) obtained a license from the Federal Power Commission to construct and operate a hydroelectric dam. Since its operation began in 1955, the dam has caused portions of the nearby Kalispel Reservation, formerly flooded only in the spring, to be flooded year round. The district court concluded that the Kalispel Tribe was entitled to damages in trespass for its lost use of this submerged riparian land. United States v. Pend Oreille Pub. Util. Dist. No. 1, No. 80-116 (E.D. Wa. filed Aug. 10, 1987), aff'd, 926 F.2d 1502 (9th Cir.1991).
Under section 10(e) of the Federal Power Act, 16 U.S.C. § 803(e), licensees of hydro-power projects that use or occupy reservation land must pay a compensatory charge to the governing tribe. “[W]hen licenses are issued involving the use of ... tribal lands embraced within Indian reservations the [Federal Energy Regulatory] Commission shall, ... subject to the approval of the Indian tribe having jurisdiction of such lands[,] ... fix a reasonable annual charge for the use thereof ...” 16 U.S.C. § 803(e).
In 1980, the United States, acting as trustee of the Kalispel land, sued PUD in district court. In 1983, the district court held that PUD had trespassed on Kalispel land, and in 1987 held that damages were only equal to the land’s value for grazing. The issue of damages was appealed to this Court, which reversed and remanded. United States v. Pend Oreille Pub. Util. Dist. No. 1, 28 F.3d 1544, 1551 (9th Cir.1994)(hereinafter “Kalis-pel II ”). In doing so, the Kalispel II panel *616set out the “power site formula” by which the district court was to assess damages:
Although use of the value of the tribal lands as part of the power project in calculating the annual charge is not compelled by regulation or case law, we nevertheless conclude it is the most acceptable measure of damages for the Utility’s trespass.... [T]his is the formulation now applied by the Commission in fixing annual charges for use of tribal lands 'on all power projects.
Id. at 1551.- On remand, the district court calculated the annual charge, which the majority now affirms. However, the district court clearly erred in its application of the power site formula mandated by this Court in Kalispel II.
2. The District Court Incorrectly Applied the Power Site Formula
a. Standard of Review
Where, as here, the district court must answer a mixed question of law and fact, the standard of review is determined by whether the question is predominantly legal or factual. “If the application of the rule of law to the facts requires an inquiry that is ‘essentially factual,’ ... founded ‘on the application of the fact-finding tribunal’s experience with the mainsprings of human conduct’ ... [then it is] reviewable under the clearly erroneous standard.” United States v. McConney, 728 F.2d 1195, 1202 (9th Cir.1984), citing Commissioner v. Duberstein, 363 U.S. 278, 289, 80 S.Ct. 1190, 1198-99, 4 L.Ed.2d 1218 (1960). However, if the district court must “exercise judgment about the values that animate legal principles,” then review must be de novo. Id. Here, application of the power site formula is predominantly factual and therefore must be reviewed for clear error.
b. Application of the Formula
The power site formula consists of a three step, net benefits analysis. First, the cost of generating power at the site is compared to the cost of generating power at the next best alternative site. The difference equals the net benefits of generating power at the site in question. Second, the benefits are divided between the owner of the dam and the owner of the land. Each receives half the benefits because each is equally integral to the operation of the site. Portland Gen. Elec. Co., 12 FERC ¶ 63,055, 65,230 (1980), 1980 WL 24161, at *23 (hereinafter “PGE I”), citing Escondido Mutual Water Co., Opinion No. 39, 6 FERC ¶ 61,189, 61,387 (1979). If there are multiple landowners, the third step determines the percentage of the land belonging to each, so that each may receive a proportionate share of the benefits. PGE I, 12 FERC at ¶ 65,231,1980 WL at *25.
First, the district court failed to compare the cost of building and operating the instant project to the cost of generating the same amount of power at a hypothetical, next best alternative project. Instead, it compared the cost of power generation using the dam to the purchase price of energy available on the market, which is obviously lower than the price of alternative power generation. The damages were further diminished by the district judge’s application of the discounted purchase price which PUD receives as a public utility.
Second, the net benefits were not divided equally between the landowners and the utility company. Id. This division of benefits is intended to accommodate FERC’s policy objectives of encouraging hydropower development while also reasonably compensating the landowners. See Portland Gen. Elec. Co., 20 FERC ¶ 61,294 (1982), 1982 WL 40297 (hereinafter “PGE II ”); Portland Gen. Elec. Co., 12 FERC ¶ 63,055 (1980), 1980 WL 24161 (hereinafter “PGE I”). At trial, expert economists on both sides testified that the standard division of net benefits is fifty-fifty. Nonetheless, the district court attributed sixty percent of the net benefits to PUD and forty percent to the Kalispel Tribe. This unprecedented division was based on the location of the lands owned by each party. Because the Kalispel lands are located on the rim of the reservoir, the district court assumed them to be less valuable than PUD’s land under the dam. However, the only power at issue here is that generated by the flooded Kalispel land on the rim of the reservoir. Its value compared to land underneath the dam is irrelevant. Furthermore, considering the location of land in establishing its *617value for this purpose was expressly rejected in PGE I, 12 FERC ¶ 63,055 (1980), 1980 WL 24161 at *25.
Third, the district court failed to accurately determine what percentage of the project land belonged to the Kalispel Tribe. The district court rejected initial figures submitted by each side, then accepted a new figure offered by PUD, stating: “The Court has determined that the 8.1% figure most closely approaches a reasonable basis on which to determine damages.” However, the 8.1% figure was calculated incorrectly. PUD’s expert simply subtracted the surface area of the reservoir at 2028 feet from the surface area at 2035 feet. This calculation determines only the increase in the surface area of the reservoir after flooding, not the area of the lands submerged, which is sloped. Therefore, the district court’s conclusion that the Kalispel Tribe owned 8.1% is not accurate.
In general, the district court unjustifiably abandoned the mandated legal analysis in calculating damages. For example, the district judge described his method thus: “I will be honest with you, I am not going to use it because the result is so skewed that it is just, you know, out of sight.” Similarly, the district judge stated: “you cut a bunch of that stuff out and you get a figure or you can look at it and say this kind of — this passes the gut check or the smell test or whatever. I think this is the right result, now I will go back and justify it.” Working backward from a subjective belief about the amount of damages is particularly perilous where the damages are increased by forty years of unpaid interest. Such a method violates the fundamental fairness guaranteed by the Due Process Clause. U.S. CONST, amend. XIV, § 1; Dowling v. United States, 493 U.S. 342, 352-53, 110 S.Ct. 668, 674-75, 107 L.Ed.2d 708 (1989). In this case, forty years of unpaid interest worked to the detriment of property owners. The district court’s method contravenes the deterrent purpose served by accurately assessing trespass damages with interest.
The district court failed to correctly apply any of the three steps of the mandated net benefits analysis, and its decision was therefore clearly erroneous.
3. The District Court Abused its Discretion in Granting an Indefinite Stay of Injunction
We review the district court’s stay of injunction for abuse of discretion. Complaint of Paradise Holdings, Inc., 795 F.2d 756, 760 (9th Cir.1986). Here, the district court enjoined PUD from flooding the Kalispel land and simultaneously stayed that injunction. In Kalispel II, this Court ordered the following:
[T]he district court should consider whether to issue an injunction in light of ... the Utility’s and State’s assertion that a large part of the Pacific Northwest would suffer a power shortage if the Utility were enjoined from flooding ... If the district court decides to enjoin the flooding, it could fashion- a remedy to minimize the harm such an injunction might cause. The court could ... stay the order to permit the Utility to seek an amendment to its license ...
Kalispel II, 28 F.3d at 1552. In short, this Court directed the district court to consider the risk of a massive power outage as a basis for denying or staying the injunction. In its order, however, the district court stated: “No evidence was introduced at trial which would indicate that any major power shortage would occur in Pend Oreille County or the city of Seattle.” United States v. Pend Oreille Pub. Util. Dist. No. 1, No. 80-116 at 5 (E.D. Wa. filed Aug. 10, 1987), aff'd, 926 F.2d 1502 (9th Cir.1991). Nonetheless, the district court issued a stay of unlimited duration. It stated: “In order to allow [PUD] to apply to FERC for an amendment to its license to flood above 2028 to 2035 and to allow [PUD] to generate sufficient power to pay this judgment, this Court will stay the permanent injunction ...” Id. In so doing, the district court abused its discretion. This Court provided for a stay for the purpose of licensing if the region would suffer widespread loss of power without such a stay. The purpose of seeking a license, without more, does not justify a stay.
*618The district court’s second rationale also fails to support the stay. The idea that a trespass should be prolonged to assist the trespasser in the payment of damages stemming from the same trespass makes neither legal nor economic sense. Finally, a stay with no specific time limit is particularly inappropriate where, as here, the trespasser has used dilatory tactics1 to evade its obligation to compensate the landowner and cease its unlicensed use of the land.
4. Conclusion
The district court clearly erred in its application of the power site formula, abandoning the analysis mandated by this Court for each of the formula’s three steps. Further, the district court abused its discretion in granting an indefinite stay of injunction.
I DISSENT for the reason that we cannot condone the abandonment of legal principles and precedent in favor of subjective beliefs when adjudicating Indian property rights.
. As noted by this Court in Kalispel II, PUD attempted to relitigate the existence of a trespass, which had been found by the district court and affirmed on appeal. Kalispel II, 28 F.3d at 1547.