This en banc hearing arises from a retrial on the issue of damages ordered by this Court in Austin v. Loftsgaarden, 675 F.2d 168 (8th Cir.1982) (Austin I). In Austin I, we affirmed the jury's verdict that Loftsgaarden and three of his closely-held corporations (hereinafter Loftsgaarden) violated state and federal securities laws, but held that the district court erred in ruling that the plaintiff-investors' rescissionary damages could not be reduced by the "tax shelter"1 benefits they received as a result of their investments. Thereafter, we denied plaintiffs' petition for rehearing or rehearing en banc.

On remand, the district court deducted plaintiffs' tax benefits from their damage awards and both sides appeal. After oral argument before a panel of this Court, the panel issued an order setting two issues for en banc consideration: First, whether Austin I should be reconsidered as it relates to the offset of tax benefits from rescissionary damages awarded in a private securities fraud action involving an investment that was structured and marketed as a tax shelter; and second, assuming that Austin I was properly decided, what is the proper measure of plaintiffs' rescissionary damages?

The facts in this case are set forth in detail in Austin I, 675 F.2d at 173-76. We briefly review them here.

Plaintiffs Austin, Anderson, Neumann and Randall, invested over $152,000 in Alotel Associates (hereinafter Associates), a limited partnership organized by defendant B.J. Loftsgaarden to help finance the building and operation of a Ramada Inn in Rochester, Minnesota. Loftsgaarden's offering memorandum claimed that purchasers of the limited partnership interests would receive substantial tax benefits during the first three years of their investments and thereafter would share in the hotel's projected profits.

The Rochester Ramada opened several months behind schedule in 1974 at a construction cost substantially in excess of that projected. Thereafter, the hotel incurred significant operating losses and Loftsgaarden requested the limited partners to extend to Associates several large loans to prevent insolvency. The limited partners hired a lawyer and an accountant to investigate Loftsgaarden's stewardship of the project, which revealed that Loftsgaarden had knowingly committed several frauds on the limited partners. Although Loftsgaarden resigned as general partner and the limited partners made additional loans,2 the partnership ultimately defaulted on its obligations and the hotel was foreclosed upon.

The four plaintiffs3 then filed this securities fraud action against Loftsgaarden. The jury found that Loftsgaarden had violated "section 10(b) of the Securities Exchange Act of 1934, 15 U.S.C. Sec. 78j(b); Rule 10b-5, 17 C.F.R. Sec. 240.10b-5; the antifraud provision of the Minnesota Securities Act, Minn.Stat. Secs. 80A.01, et seq., 80A.23; and [had committed] common law fraud. The jury rendered an advisory verdict--in which the district court concurred--that defendants were liable for violating Sec. 12(2) of the Securities Act of 1933. 15 U.S.C. Sec. 771(2)." Austin I, 675 F.2d at 172 (footnotes omitted). The court applied a rescissionary measure of damages in the amount of the consideration each plaintiff paid for his limited partnership unit or units, prejudgment interest from the date of purchase, and attorneys' fees for plaintiffs Randall and Neumann for a total judgment of $273,720.

We affirmed the findings of liability, but vacated the award of damages and remanded to the district court, stating "the court committed reversible error in refusing to allow proof of any economic benefits received by plaintiffs on account of the investment and in failing to instruct the jury that the damage award must be reduced by any value shown to have been received by plaintiffs." Austin I, 675 F.2d at 181.

On remand, the district court awarded to plaintiffs damages in the amount of their consideration paid, with eight percent interest thereon, less tax benefits received, for a total amount as follows: Anderson, $35,172; Austin, $31,277; Neumann, $39,371; and Randall, $31,569. It also granted attorneys' fees under Minn.Stat. Sec. 80A.23, subd. 2 as follows: Anderson, $0; Austin, $0; Neumann, $29,000; and Randall, $20,000. In addition, the court granted costs of $5,321.83 to Austin, Anderson and Neumann and $2,955.49 to Randall. We now consider the two issues raised before the Court en banc.

DISCUSSION.

I. Should Austin I be reconsidered as it relates to the offset of tax benefits from rescissionary damages awarded in a private securities fraud action involving an investment structured and marketed as a "tax shelter"?

The starting point for our analysis is that the petition for rehearing or rehearing en banc in Austin I was denied, 675 F.2d at 168, and that, accordingly, its holdings are now law of the case. As we stated in Otten v. Stonewall Ins. Co., 538 F.2d 210 (8th Cir.1976):

Law of the case principles have been stated with clarity and consistency in this court over a long period of time. "This court has repeatedly held that the decision on former appeal is the 'law of the case' on a question presented in that former appeal, unless the evidence introduced at the subsequent trial is substantially different from that considered on the first appeal, and must be followed in all subsequent proceedings in such case in both district and appellate courts unless that decision is clearly erroneous and works manifest injustice. * * * While this rule of practice is not a limit of power, it is nevertheless a salutary one, and should be departed from only after careful consideration of situations arising in specific cases."

Id. at 212 (citations omitted).

The appellees contend that the damages holding in Austin I is "clearly erroneous" and works a "manifest injustice" and thus should be reversed. After a careful review of Austin I, the relevant cases, the arguments of the parties and the Tax Division of the Justice Department and the Securities and Exchange Commission, appearing as amicus curiae, we abide by our previous decision. In doing so, we note that since our opinion in Austin I, the United States Court of Appeals for the Second Circuit has substantially adopted the views that we expressed in that case. Salcer v. Envicon Equities Corp., 744 F.2d 935 (2d Cir.1984).4

Our decision in Austin I was based on the principle that: "[D]amages for securities fraud are determined in accordance with the extent to which false and misleading information actually harmed the complaining party." Austin I, 675 F.2d at 180 (citation omitted). We noted that for this reason, the courts typically use the out-of-pocket measure of damages,5 but that where a rescissionary measure of damages is applied, the "actual damages principle requires that a rescissionary * * * award be reduced by any value received as a result of the fraudulent transaction." Austin I, 625 F.2d at 181 (citation omitted). We held that this principle applied not only to the 1933 Act claims and the claim under Minn.Stat. Sec. 80A.23(2),6 (both of which explicitly limit damages to the "actual damages" sustained), but also to the common law fraud claim7 and the claim under section 12(2) of the 1933 Act. Although the 1933 Act does not explicitly limit rescissionary damages to "actual damages" sustained, we noted that:

Under Sec. 12(2), the defendant guilty of prospectus fraud shall be liable to plaintiff, who may sue either at law or in equity in any court of competent jurisdiction, to recover the consideration paid for such security with interest thereon, less the amount of any income received thereon, upon the tender of such security, or for damages if he no longer owns the security.

Id. at 181 (emphasis in original).

Because we found that section 12(2) implicitly incorporates the "actual damages" principle and that the words "income received" should be construed as including the tax benefits bargained for and received in the special case of a tax shelter investment, we concluded that the plaintiffs' damages must be reduced by the tax benefits they received.

The plaintiffs now contend that we should reconsider both of the above findings of Austin I. First, they argue that the "actual damages" language in section 28(a) of the 1934 Act does not justify subtracting tax benefits from awards under section 12(2) of the 1933 Act because section 28(a) applies only to the 1934 Act and the words "actual damages" do not appear anywhere within the 1933 Act. We reject this contention. Although the words "actual damages" do not appear in the 1933 Act, the courts have nonetheless applied the actual damages principle of section 28(a) to bar punitive damages8 under section 17(a) of the 1933 Act, Globus v. Law Research Serv., Inc., 418 F.2d 1276, 1278, 1283-86 (2d Cir.1969), cert. denied, 397 U.S. 913, 90 S. Ct. 913, 25 L. Ed. 2d 93 (1970), and have construed the rescission and restitution remedy provided by section 12(2) as substantially equivalent to the damages permitted under section 28(a). Cf. Affiliated Ute Citizens v. United States, 406 U.S. 128, 155, 92 S. Ct. 1456, 1473, 31 L. Ed. 2d 741 (1972). The goal of rescission under section 12(2) is to return the parties to the status quo ante, "and hence a plaintiff can recover no more than his or her 'net economic loss,' " i.e., "actual damages." Salcer, 744 F.2d at 940; Austin I, 675 F.2d at 181; Rolf v. Blyth, Eastman Dillon & Co., 570 F.2d 38, 49 n. 22 (2d Cir.), cert. denied, 439 U.S. 1039, 99 S. Ct. 642, 58 L. Ed. 2d 698 (1978).

We also reject plaintiffs' contention that tax savings are neither "income" nor are they "received" within the meaning of section 12(2). They cite United Housing Foundation, Inc. v. Forman, 421 U.S. 837, 95 S. Ct. 2051, 44 L. Ed. 2d 621 (1975) in support of their argument, but that case is inapposite. The plaintiffs in United Housing were tenants who were required to purchase refundable non-interest bearing shares in a nonprofit housing cooperative in order to rent a low-cost apartment. After a rent increase was imposed to meet costs, plaintiffs brought fraud claims under the Securities Act of 1933 and the Securities Exchange Act of 1934, alleging that they were misled in the purchase of the shares by misrepresentation in the cooperative's information bulletin. The Court looked to "economic reality," 421 U.S. at 848, 95 S.Ct. at 2058, in determining that the shares were not securities. It noted that the plaintiffs' purchase of the shares was motivated by a desire to obtain housing, not to make a profit in the sense of a "return on investment." The Court was not persuaded that the plaintiffs expected a profit on their non-interest bearing refundable shares (which were akin to a "security deposit") simply because they were able to reduce their taxes by deducting that portion of their monthly rental charges which were applied to interest on the mortgage. 421 U.S. at 855, 95 S.Ct. at 2062. The Court noted that "[e]ven if the tax deductions were considered profits, they would not be the type associated with a security investment since they do not result from the managerial efforts of others." Id. 421 U.S. at 856 n. 20, 95 S. Ct. at 2062 n. 20.

United Housing is not only distinguishable on its facts; if anything, its analysis tends to support our position in Austin I. Like the United Housing Court, the Austin I Court looked to "economic reality" in considering the tax aspects of a tax shelter investment.9 Additionally, unlike the marginal tax benefits referred to in United Housing, the tax benefits in the case at hand were a primary motivation behind the investment and resulted from Loftsgaarden's efforts in structuring and operating the partnership to take advantage of relatively complicated tax rules. See Salcer, 744 F.2d at 941 ("plaintiffs' argument that defendants are not entitled to 'credit' for the tax benefits because the latter were 'provided' by the government rather than by the defendants must be rejected * * * [T]hat the tax benefits were paid by the government rather than by the defendants does not lessen the defendants' role in securing them.").

In any event, Austin I did not hold that tax benefits received are a form of income in a strict accounting sense.10 Rather, we held that section 12(2)'s requirement that "income received" be deducted in determining rescissionary damages simply reflected that all economic benefits bargained for and received must be deducted from plaintiffs' damages because "the strictly compensatory nature of damages awarded in private securities fraud actions requires that such value be taken into account in determining whether and to what extent damages were inflicted upon plaintiffs." Austin I, 675 F.2d at 183.

The plaintiffs claim that their tax benefits have not been "received" must also be rejected because they have stipulated that they have received permanent tax benefits from their investments in Associates. Similarly, we reject the contention of the Tax Division and the SEC that plaintiffs' tax benefits "may" prove to be illusory because they "may" experience recapture upon sale or other disposition of the hotel. All of the plaintiffs in this case have already experienced all possible recapture when the hotel was foreclosed upon in 1978. We also reject, as without a factual basis, the Tax Division's claim that "it is far from clear that all the proper income recognition and recapture provisions were taken into account." Brief of Tax Division at 11. All of the plaintiffs' tax returns from the relevant tax years have already been audited and are now closed.

We also find without merit the Tax Division's claim that, under the "tax benefit rule," the plaintiffs, upon receiving rescissionary damages, will have to amend their prior tax returns to eliminate all tax benefits previously received as a result of their investments in Associates. The "tax benefit rule" does not require elimination of tax benefits already received. See Hayden v. McDonald, 742 F.2d at 440. Rather, it provides that a taxpayer who claims a deduction resulting in a tax benefit one year and who later obtains a recovery or repayment in a later year must include the recovery or repayment as ordinary income in the year of recovery.11 See Hillsboro National Bank v. Commissioner, 460 U.S. 370, 376-81, 103 S. Ct. 1134, 1140-42, 75 L. Ed. 2d 130, 142-44 (1983); Salcer, 744 F.2d at 943; 1 Mertens, Law of Federal Income Taxation, Sec. 7.34 at 7-115, Sec. 7.37 at 7-130. We need not address the contention of the Tax Division and the SEC that the tax consequences of the rescissionary award will "wash out" any prior tax savings, because we are not presented with such a situation in this case.

The plaintiffs next contend, with support from the SEC, that the collateral source rule requires that Section 12(2) awards not be reduced by tax benefits. This rule "prohibits taking into consideration benefits received from third parties as the result of wholly separate and distinct transactions, such as those received under insurance policies negotiated independently." Salcer, 744 F.2d at 941. We agree with the Salcer Court that the tax benefits of plaintiffs' investments are not a collateral source because "although paid by the government, [they] emanate directly from the tax shelter sought by the plaintiffs and provided by defendants, without which plaintiffs could not have realized any tax benefits. Benefits resulting directly from a transaction under attack must be credited toward the damage award." Id. at 941-42.

For similar reasons, we reject plaintiffs' claim that we should ignore their tax benefits because otherwise the defendants will unjustly "retain" the consideration previously paid by the plaintiffs. First of all, it does not appear that the defendant has "retained" any of plaintiffs' consideration; rather these monies were paid to the partnership and were invested in the project. Moreover, the purpose of the private civil remedies provided by the federal and relevant Minnesota securities laws is not to penalize defendants, but to compensate plaintiffs for any actual monetary loss. See id. at 941. This actual monetary loss cannot be determined without reference to tax benefits bargained for and received.

We also reject the similar argument of amici that deducting tax benefits from damages will deprive the government of just tax revenues. The government is entitled only to those revenues which are authorized under the tax laws. These laws create incentives for real estate projects to further public policy objectives, such as the creation of employment, enhancement of the real estate tax base and the local business climate. The Rochester Ramada project met all of the tax code criteria and thus generated the income tax benefits which the plaintiffs enjoyed. This case is therefore very similar to Salcer, where the Court stated:

Nor can we agree with the notion that our ruling makes "the government the banker for fraudulent tax shelter activity," Burgess v. Premier Corp., supra, 727 F.2d at 838. As Judge Broderick noted, there was no such activity in this case; the fraud, if any, pertained only to the investment aspects of the Greenspoint Project. The government got the residential development that it hoped to encourage by offering the tax benefits taken by the plaintiffs. Thus, it is "banking" precisely what it agreed to "bank." There is no suggestion that the project did not meet federal requirements as a viable housing development entitling its owners to tax benefits.

Id. at 941.

There is no legal justification for awarding a party damages to which he is not entitled merely to provide revenue for the government to tax.

Next, we reject plaintiffs' claim that the consideration of tax benefits in determining rescissionary damages in a tax shelter case creates unnecessary complexities. Our opinion in Austin I, 675 F.2d at 183, deals at length with this contention. We noted that "the complexity of evidence relating to plaintiffs' tax savings is not a viable reason for precluding such evidence in light of Norfolk & Western R. Co. v. Liepelt, 444 U.S. 490, 100 S. Ct. 755, 62 L. Ed. 2d 689 (1980)." Id. We cannot agree that the retrial of the damages issue on remand reflects such complexity that Austin I must now be reversed.

Finally, plaintiffs and the SEC12 contend that Austin I's tax benefit offset rule discourages private enforcement of the antifraud provisions of the federal securities laws in an area in which the Commission itself is unable to devote substantial resources.13 While it is unclear that this will be the result of the offset rule because most defrauded tax shelter investors will still be able to obtain at least some damages, we agree with the Salcer Court that it is for the legislature, not for the courts, to authorize the award of more than "actual damages" in order to encourage the filing of private damage actions. Salcer, at 941. See also e.g., Alyeska Pipeline Service Co. v. Wilderness Society, 421 U.S. 240, 95 S. Ct. 1612, 44 L. Ed. 2d 141 (1975) (shifting of attorneys' fees by court to create incentives is impermissible).

In sum, we affirm the holding of Austin I "that in a private securities fraud action involving an investment structured and marketed as a tax shelter, where a rescissory measure of damages is applied, evidence of any benefit derived by the plaintiff/investor via tax savings must be permitted." Austin I, 675 F.2d at 183-84.

II. How should the plaintiffs' rescissory damages be computed?

On remand, the district court first determined the amount of permanent tax benefits received by each of the plaintiffs as follows: Anderson, $29,615.00; Austin, $33,330.00; Neumann, $57,014.00; Randall, $36,404.00. The parties do not contest this finding; accordingly, we adopt these figures as the amount of tax benefits received by each plaintiff.

The court then determined that each plaintiff was entitled to prejudgment interest at the rate of eight percent simple interest14 under section 12(2) of the 1933 Act. The interest rate has been approved by the panel opinion announced today along with the en banc opinion.

The court then determined rescissory damages in accordance with the specific language of section 12(2): "consideration paid for such security with interest thereon less the amount of any income received thereon." Under the district court's amended determination, this computation produced the following results:

Plaintiff Austin
Consideration paid June 6, 1973:        $35,000
Simple interest at 8% to February 22,
1984:                                   $31,277
                                       --------
  Subtotal                              $64,610
Minus tax benefits:                    -$33,333
                                       --------
  Total damages:                        $31,177
Plaintiff Neumann
Consideration paid August 21, 1973:     $35,000
Plus consideration paid October 19,
1973:                                   $17,500
Simple interest at 8% to February 22,
1984:                                   $43,885
                                       --------
  Subtotal                              $96,385
Minus tax benefits:                    -$57,014
                                       --------
  Total damages:                        $39,371
Plaintiff Randall
Consideration paid November 13, 1973:   $35,000
Simple interest at 8% to February 22,
1984:                                   $28,770
Plus consideration paid 1975:            $2,512
Simple interest at 8% to February 22,
1984:                                    $1,691
                                       --------
  Subtotal                              $67,973
Minus tax benefits:                    -$36,404
                                       --------
  Total damages                         $31,569
Plaintiff Anderson
Consideration paid June 2, 1973:      $35,000
Plus interest at 8% to February 22,
1984:                                 $29,787
                                     --------
  Subtotal                            $64,787
Minus tax benefits:                  -$29,615
                                     --------
  Total damages:                      $35,172

Although the parties and amici suggest several variations from this damage computation formula, we believe they raise only three serious contentions. First, defendant Loftsgaarden argues that, under the section 12(2) damage computation formula, tax benefits are a return of "consideration paid" and thus must be deducted from consideration paid before adding on interest. As the district court noted, this course would drastically reduce the awards to Anderson and Austin and eliminate the awards to Neumann and Randall. We agree with the district court that this approach must be rejected. As we noted previously, Austin I did not indicate that tax benefits should be treated as a return of consideration. Instead, in holding that tax savings must be subtracted from damages we quoted section 12(2) and italicized the phrase, "less the amount of any income received thereon." Austin I, 675 F.2d at 181. Similarly, in Hayden v. McDonald, 742 F.2d at 441, we characterized Austin I as "implicitly construing the 'income' language in a federal securities statute to include tax considerations." At least one commentator has also characterized Austin I as "classifying tax benefits received by the plaintiff as 'income' within the meaning of section 12(2) of the 1933 Act." Note, Austin v. Loftsgaarden, etc., 16 Creighton L.Rev. 1140, 1151 (1983). Cf. S. Banoff, "To What Extent Will Benefits From Tax Shelters Be Permitted to Offset Rescission Damages?," 57 J.Tax'n 154, 156 (1980).

Additionally, whether or not tax benefits may be considered "income" in a strict accounting sense, tax benefits are far more dissimilar to "consideration." Moreover, although we find that section 12(2)'s language "income received thereon" may properly be construed as including tax benefits bargained for and received, we do not agree that tax benefits can be considered a "return" of consideration.

Finally, treating tax benefits as consideration returned would, even under Loftsgaarden's calculations, mean that the plaintiffs would not be compensated for their actual damages. For example, using Loftsgaarden's calculations of plaintiff Neumann's net interest cost (as we explain further below), he would receive no damages even though he suffered net economic loss of at least $992.15 Loftsgaarden no doubt prefers the zero judgment figure to the $992 figure, in part because this would lend substantial support to his argument before the panel that Neumann is not entitled to the $29,000 in attorneys' fees granted by the district court. In short, we cannot agree that the district court erred in refusing to treat the tax benefits as a return of consideration for purposes of determining the order of computation under the formula set forth in section 12(2).

Loftsgaarden next contends that the district court erred in awarding prejudgment interest on the total amount of consideration paid by each plaintiff from the date paid until February 22, 1984, rather than awarding interest only on the amount of money which each plaintiff was "out-of-pocket" during each year of the investment. We agree. The goal of prejudgment interest is to compensate plaintiffs for the loss of the use of their money. See Cant v. A.C. Becker & Co., 384 F. Supp. 814 (N.D.Ill.1974); Chris-Craft Industries, Inc. v. Piper Aircrafter Corp., 384 F. Supp. 507, 527 (S.D.N.Y.1974), modified, 516 F.2d 172 (2d Cir.1975), rev'd on other grounds, 430 U.S. 1, 97 S. Ct. 926, 51 L. Ed. 2d 124 (1977). Because plaintiffs were not deprived of the use of the entire amount of their investments over the ten year period, prejudgment interest is due only on the amount of money they were out-of-pocket at any given time. Gerstle v. Gamble-Skogmo, Inc., 478 F.2d 1281, 1307 (2d Cir.1973) (prejudgment interest should be computed on the "net difference between value of assets given and value received"); Murphy v. Cady, 30 F. Supp. 466, 470 (D.Me.1939), aff'd, 113 F.2d 988 (1st Cir.), cert. denied, 311 U.S. 705, 61 S. Ct. 175, 85 L. Ed. 458 (1940) (reaching similar result by determining that plaintiff was entitled under section 12(2) "to recover the amount paid for [the security] with interest, less the amount received with interest").

Loftsgaarden has submitted what we find is a reasonably accurate determination of the net interest costs incurred by the plaintiffs taking into account the amount of money invested, and the tax benefits received by each plaintiff during each year of their investment in Associates. These calculations, prepared by John M. Carlson, Certified Public Accountant, are as follows:16

Next, both plaintiffs and defendants contend that in order for the plaintiffs to be truly made whole, and to be returned to the status quo ante, both the tax benefits received and the tax consequences of their recovery must be taken into account. Dicta in Hayden v. McDonald, 742 F.2d 423, 440 (8th Cir.1984) supports this contention, and, as the parties are in agreement, we agree that the tax consequences of plaintiffs' recovery must be considered. The parties disagree on how these tax consequences should be determined. We find Loftsgaarden's method to be reasonably accurate and much simpler than plaintiffs' formula and thus adopt Loftsgaarden's method. This method, which was also used in the "Reardon Report," which was cited in Hayden, takes into account the fact that, because each plaintiff will have to pay taxes on his award, the award must be calculated in an amount which each plaintiff would have to recover in order to "net out" their actual damage figure. We assume that each plaintiff is still in the fifty percent tax bracket and thus must receive twice the above amount of damages and net interest cost. Thus, in accordance with the panel's affirmance of the eight percent simple interest rate, each plaintiff is entitled to a damage award as follows: Anderson ($9,395 X 2 = $18,790); Austin ($3,833 X 2 = $7,666); Neumann ($992 X 2 = $1,984); Randall ($253 X 2 = $506).

In conclusion, the Court en banc affirms Austin I but vacates and modifies the district court's award of damages as set forth above.

There are three remaining issues which were initially heard by a panel of this Court. These issues are: (1) Whether the district court, on remand, erred in its interpretation of Austin I; (2) whether attorney fees should have been awarded; and (3) whether prejudgment interest was properly calculated. For convenience, the panel's opinion is incorporated herein:

Before LAY, Chief Judge, and ROSS and McMILLIAN, Circuit Judges.

Plaintiffs urge, on cross-appeal, that the district court should have allowed them to retry their case on the "out of pocket" damages theory. At the original trial the district court noted that "out of pocket" and rescissionary damages are essentially the same. However, plaintiffs hoped to avoid the Austin I mandate to consider tax benefits by changing to the "out of pocket" approach. On remand, the district court refused to allow the change in theories. We find no error. The plaintiffs tendered their shares to defendants shortly before the original trial and effectively elected the remedy of rescission. More importantly, plaintiffs should not be allowed to introduce a new theory some eight years later. The case is now in its second appeal, after two trials, both of which focused on rescissionary damages. We think it too late to change theories and find no abuse of discretion by either trial judge.

Plaintiffs also argue that Austin I applies only to investments "expressly structured and marketed as a tax shelter." Austin I, 675 F.2d at 183. According to plaintiffs no factual finding has ever been made that this particular investment was expressly structured and marketed as a tax shelter. The Austin I court defined a tax shelter as any investment which creates tax deductions through "artificial losses." Id. The court en banc has endorsed that definition. Austin II, slip op. at 3. The evidence shows plaintiffs were able to reduce their tax liability significantly in the first few years of the investment through accelerated depreciation and other deductions.

There are several contentions relating to attorney fees. Defendants contend Neumann and Randall failed to prove a cause of action which would support any award of fees and, even if they did, the amount awarded was excessive. Plaintiffs argue that the fees awarded were insufficient and, furthermore, that fees should have been awarded to Austin and Anderson.

First, the method of damage calculation determined by the court en banc makes it clear Neumann and Randall were damaged and proved a claim for relief. We find the award of fees to Neumann and Randall falls within the range of discretion of the district court. In weighing the arguments we find no abuse of discretion.

Second, plaintiffs cannot recover fees under federal securities statutes. The award to Neumann and Randall was based on Minn.Stat. Sec. 80A.23(2) (1984). This statute was not in effect before August 1, 1973. It is clear from the record that Anderson and Austin purchased their securities prior to that date. Under these circumstances the district court was correct in holding that Anderson and Austin were barred from recovery of attorney fees.

The issue of the method of calculating interest has been resolved by the court en banc. Plaintiffs have argued that the rate, eight percent, was too low. Defendants, on the other hand, argue no interest can be awarded to Neumann and Randall because they suffered no actual damage and thus failed to prove a cause of action. As noted previously, the court en banc's method of damage calculation shows Neumann and Randall did suffer damage and proved a cause of action. While the district court, on remand, was not bound by the original trial judge's use of eight percent as the rate of interest, plaintiffs point to no evidence in the record which would suggest a higher rate. Although we believe the rate of interest should be based on a finding of fact from evidentiary proof, in the absence of such proof we feel the district court did not err in applying an eight percent rate as a traditional and fair rate of interest.

The panel holds that the district court, on remand, correctly interpreted Austin I and did not err or abuse its discretion in the award of attorney fees and pre-judgment interest. In conclusion, we remand to the district court for entry of judgment as set forth in this opinion.

HEANEY, Circuit Judge, concurring and dissenting.

I concur in the majority's opinion except insofar as it overrules Hayden v. McDonald, 742 F.2d 423 (8th Cir.1984). I continue to believe that the issue is a doubtful one and that in cases of doubt we should give great weight to a decision of a United States District Court interpreting a statute of the state in which it sits. Moreover, I think it is bad policy to reach an issue that was not presented to us by the parties and which we need not decide.

LAY, Chief Judge, dissenting, joined by BRIGHT, Circuit Judge.

I incorporate my concurrence in Hayden v. McDonald, 742 F.2d 423, 441-42 (8th Cir.1984), and the reasoning set forth therein for urging that tax benefits of plaintiffs in a rescissionary damage action have no relationship to their present claim. The Department of Justice and the Securities and Exchange Commission have presented strong arguments against appellants' theory in amicae briefs.

The majority invokes the doctrine of law of the case. As Judge Learned Hand observed many years ago: "[I]t is now well settled that the 'law of the case' does not rigidly bind a court to its former decisions, but is only addressed to its good sense." Higgins v. California Prune & Apricot Grower, Inc., 3 F.2d 896, 898 (2d Cir.1924). A court should never be constrained by the weight of a prior erroneous decision. The purpose of an en banc hearing is to draw on the wisdom of the full court and rectify earlier errors. Reliance on precedent is a poor substitute for analysis.

The damages provisions of the federal statute, section 12(2) of the Securities Act of 1933, 15 U.S.C. Sec. 77 l (2), and the Minnesota statute, Minn.Stat. Sec. 80A.23(1), are quite similar:

Sec. 12(2): * * * recover the consideration paid for such security with interest thereon, less the amount of any income received thereon * * *.

Sec. 80A.23(1): * * * recover the consideration paid for the security together with interest at the legal rate, costs, and reasonable attorney's fees, less the amount of any income received on the securities.

The key words are "less the amount of any income received * * *." The majority concedes that the tax benefits received by plaintiffs are not really "income." Nonetheless, the court holds the benefits must be subtracted--notwithstanding the clear language of the statute--in order to give meaning to "actual income."

Loftsgaarden relies heavily on Salcer v. Envicon Equities Corp., 744 F.2d 935 (2d Cir.1984), cert. pending. The Salcer court was concerned the plaintiff might receive a windfall and, like the majority today, applied the tax benefits in favor of the defendant, transferring any possible windfall to the wrongdoer. The Ninth Circuit has held to the contrary, using the same reasoning I urge this court to adopt. Burgess v. Premier Corp., 727 F.2d 826, 837-38 (9th Cir.1984). The Burgess court found that "to simply subtract the tax benefits from damages would place an unfair burden on taxpayers generally. * * * Such a result leaves the government bearing the cost of defendants' fraud." Id. at 838. See also Western Federal Corp. v. Erickson, 739 F.2d 1439, 1444 (9th Cir.1984).

Application of tax benefits to lessen damages in a rescissionary suit actually fails to comprehend the principle behind a "tax shelter." The investor receives no money from the tax benefit; the investor simply is allowed to deduct a certain amount from his ordinary income or apply a credit toward tax owed for the particular tax year. The actual value of the benefit depends on the income bracket the investor is in, which in turn depends on other deductions that may be available to the taxpayer that year. This fact alone demonstrates how speculative any benefit in tax savings may be. More significantly, the tax benefit is not a real or actual savings--it is usually only a deferral of tax. Ultimately, in a time of recapture when the investment is sold or otherwise disposed of, the taxpayer must return to the government tax benefits gained unless the investor suffers an actual loss. The court's conversion of the tax deferral to realized income by subtracting it from a plaintiff's damages provides a windfall to the defendant--the fraudulent party. The holding in this case violates the old principle that "[i]t is more appropriate to give the defrauded party the benefit even of windfalls than to let the fraudulent party keep them." Janigan v. Taylor, 344 F.2d 781, 786 (1st Cir.), cert. denied, 382 U.S. 879, 86 S. Ct. 163, 15 L. Ed. 2d 120 (1965), quoted in Myzel v. Fields, 386 F.2d 718, 747 (8th Cir.1967), cert. denied, 390 U.S. 951, 88 S. Ct. 1043, 19 L. Ed. 2d 1143 (1968).

In mitigation, the majority believes Loftsgaarden did not benefit from any of the consideration paid by plaintiffs. This is patently wrong. There was evidence at trial showing Loftsgaarden received over $100,000 profit from the investment scheme. Under the majority's approach, Loftsgaarden gains both his own profit from the scheme and the advantage of plaintiffs' speculative tax benefits. Such a result seriously undercuts the broad protective purpose of the Securities Act of 1933. See United States v. Naftalin, 441 U.S. 768, 774-777, 99 S. Ct. 2077, 2082-2083, 60 L. Ed. 2d 624 (1979).

The majority adopts the magnanimous suggestion in Loftsgaarden's brief that, in all fairness, plaintiffs' ultimate damages should be doubled in order to offset the tax consequences of recovery--assuming plaintiffs are in the fifty percent bracket. This serves to demonstrate how speculative and inaccurate the consideration of tax consequences can be in a suit involving rescissionary damages. A defendant should not have to pay damages based on the plaintiff's tax bracket in a given year. This is not a damage caused by the defendant. Their theory is that to make plaintiffs whole and restore the status quo ante we must account for the tax treatment of plaintiffs' recovery. However, the relevant statutes clearly require the court to restore the consideration paid to the defrauding party, with interest, less any income actually received from the investment. And tax benefits concededly are not income received from the investment.1 Consideration of tax benefits involves the court in the problems of recapture and net gains or losses. In turn, the court finds it must consider each plaintiff's income and deductions in the year of recovery to net out "actual damages" based on applicable tax rates. Rather than making up an artificial theory of damage by doubling the net gain, it is much simpler and more accurate to provide the plaintiff with restitution of his investment. The statutes provide a clear and simple formula. For example, Anderson invested $35,000 for a period of seven years. Loftsgaarden paid Anderson no interest or dividends and returned no principal. Anderson should recover $35,000 with interest for those seven years. Anderson will have to pay tax on the interest at ordinary income rates. In addition, he may be required to recapture past benefits. The tax transaction is between Anderson and the government and should not affect Loftsgaarden's liability. By this method a plaintiff recovers exactly what he lost: the principal, and interest he would have earned had that principal been invested in a legitimate security.2 This procedure is far more simple than requiring trial courts to undertake complicated and speculative evidentiary hearings on the mechanics of tax deferral plans and plaintiffs' tax status in given years of investment.

Today's holding provides the wrongdoer a surety by which to avoid liability, allowing the fraudulent party to use the United States Treasury for security in any fraudulent transaction. To the extent a defendant can, by his wrongful act, create a tax loss for the innocent victim, the defendant will be allowed to escape liability, shifting the burden of the loss onto the government. I do not believe section 12(2) or section 80A.23(1) was designed to accomplish this result. Accordingly, I dissent.