Opinion

One provision of California’s Unfair Practices Act specifies that “It is unlawful for any person engaged in business within this State to sell any article or product at less than the cost thereof . . . for the purpose of injuring competitors or destroying competition.” (Bus. & Prof. Code, § 17043. ) The fundamental issue here is how to compute the cost of news paper advertising space in a major metropolitan market. A complicating factor is that, for all practical purposes, there are only two actors in that market.

An unbroken line of decisional authority holds that just what constitutes an economic actor’s “cost” is an issue of fact to be decided by the trier of fact. In this case, however, the trial court decided the cost issue as a matter of law in circumstances that virtually ensured a verdict for the plaintiff. We therefore shall reverse the judgment for plaintiff together with an ensuing order awarding attorney fees.

Background

The dispute between the parties, and the extensive record generated by that dispute, reveal a mountain of detail, not all of which warrants recapitulation here. For present purposes, the following abbreviated narrative will suffice;

For a number of years the City and County of San Francisco let an annual contract for its “official newspaper” publishing public notice of matters as required by state and municipal law. (See Gov. Code, § 6000 et seq.; S.F. Admin. Code, § 2.81.) The contract became an issue of bitter contention between two competitors for the contract, the San Francisco Examiner and the San Francisco Independent.

By 1993, the Examiner had lost out in the bidding process for this contract to the Independent for three years’ running. The last time out prior to the year in controversy here, the Independent’s winning bid was approximately 35 cents per line less than the Examiner’s. This time, the Examiner dropped its bid from $2.28 per line to $1.40 per line. The Independent’s final bid increased from $2.35 to $2.67 per line. After the Examiner won the contract, the Independent brought suit for compensatory and punitive damages, attorney fees, and a permanent injunction. Although various causes of action were alleged in the Independent’s complaint, its case against the Examiner came to trial on a single theory—that the Examiner’s winning bid was less than the costs incurred and thus in violation of the statutory prohibition against “below cost” sales.

From the outset, both sides recognized that the issue of cost was crucial. No sooner had the case been called for trial than both sides submitted in limine motions attacking and requesting exclusion of the other party’s “cost study evidence” or “cost allocation model.” Because the Examiner and the Independent agreed that “the fundamental legal dispute ... on the proper cost method should be resolved by the Court ‘before the presentation of any argument or evidence to the trier of fact,’ ” the court conducted an extensive evidentiary hearing pursuant to Evidence Code section 402. (See fn. 11, post.) Based on the evidence it had heard, the court ruled that the Examiner “produces two relevant products”—the actual “physical newspapers” and advertising space—for purposes of this action.” The Independent’s “cost allocation method” recognized the existence of the two products and allocated costs between them; the Examiner’s motion to exclude the Independent’s cost determination study was therefore denied. On the other hand, because the Examiner’s method did not allocate costs between these two products, the Independent’s motion to exclude evidence of this method was granted.

In line with this ruling, the Examiner’s expert, Dr. Crichfield, prepared a new study that allocated costs between newspapers and advertising. The Independent moved to exclude all evidence of this new study. Following a second Evidence Code section 402 hearing, the court denied the Independent’s motion and ruled that the study could be used at trial. Each side would be permitted, the trial court ruled, to produce its own cost determination study, despite difference in their methodologies. The Independent’s method of determining cost was based on a “revenue allocation,” the Examiner’s on allocation according to the amount of space dedicated to advertising, as opposed to news or editorial coverage.

Trial then commenced. The first witness was Earl Pierson, who testified about the newspaper business in general and the state of that business in San Francisco in the early 1990’s. The Independent’s next witness was its economic expert, Dr. McLeod. He testified that the Examiner’s winning bid of $1.40 was well below its actual per line cost of $1.82. The major purpose of Dr. McLeod’s testimony was to attack the validity of the Crichfield model. He accused Dr. Crichfield of employing “accounting tricks” that “distort” the real picture of the Examiner’s costs.

The Independent’s next witness was accounting expert Dr. Maher, who had reviewed McLeod’s findings and endorsed them. Dr. Maher specifically agreed with McLeod that the Examiner’s costs had to be determined using the “revenue basis of allocation.” He also criticized Crichfield’s model. Dr. Maher’s testimony ended the fourth day of trial.

Prior to the start of the fifth day of the trial, while discussing another matter with counsel in chambers, the court announced that “I have something to talk about.” The court then explained that it had reexamined the issue of the cost studies prepared by the parties’ experts and concluded that “[T]he only model that’s appropriate is a revenue allocation model for allocating costs in this case.” The Independent’s experts, Drs. McLeod and Maher, endorsed a revenue allocation model. The Examiner, by contrast, relied on Dr. Crichfield’s allocation of division between advertising and editorial/newsgathering functions according to the unit of measure of inches of the Examiner devoted to one or the other of these tasks. The court told counsel: “[T]he only model that’s appropriate is a revenue allocation model for allocating costs in this case. You cannot use this physical unit allocation model approach in this context. You can’t do it. [¶] . . . This is not a fully-allocated cost model, the one that Dr. Crichfield made. It just isn’t.” The court repeatedly called Dr. Crichfield’s approach “irrational.” The court denied the Examiner’s motion for a mistrial and trial proceeded for approximately six weeks.

The Independent’s expert, Dr. McLeod, testified at length about the revenue allocation methodology utilized in his cost study. Dr. McLeod’s testimony will be discussed in more detail post, but its general import was to allocate cost according to revenue generated by those costs. His bottom line was that the Examiner’s genuine cost per line was $1.92, far above the $1.40 bid.

In accordance with the court’s most recent ruling, the Examiner’s expert, Dr. Crichfield, was not allowed to testify about his cost model, but he did testify as to methodological defects he perceived in Dr. McLeod’s model. After both Crichfield and McLeod had testified, and shortly before the case was sent to the jury, the court granted the Independent’s motion to strike portions of Crichfield’s testimony that were critical of McLeod’s methods and conclusions; the theory was that allowing Crichfield to criticize McLeod’s methods and conclusions would permit the Examiner to circumvent the court’s ruling excluding substantive affirmative testimony from Crichfield. During the course of discussing instructions, the court granted the Independent’s motion for a directed verdict on the Examiner’s “meeting competition” affirmative defense. (See § 17050, subd. (d).)

The court instructed the jury in effect that the only relevant evidence on the issue of cost had come from the Independent, a point the court reiterated when expressly asked by the jury. Three days after the case was sent to them the jury sent a note to the court asking “[W]hat do we do when we have come to an impasse?” The court questioned the jury, determined that there was not a true deadlock, and sent them back to continue the deliberations. Three days later the jury, by a vote of nine to three, returned a “general” (i.e., liability only) verdict in favor of the Independent.

After a brief damages phase, the jury returned a unanimous “special” verdict for the Independent in the amount of $348,981—the amount requested—together with interest. After this amount was trebled (§ 17082), and prejudgment interest was added, a judgment for $1,149,636 was entered. Almost twice this amount—$2,230,051—was awarded for costs and attorney fees. The Examiner’s motions for a new trial and for judgment not withstanding the verdict were denied. The Examiner appeals from the judgment, the attorney fee order, and the order denying its motion for judgment notwithstanding the verdict.

Discussion

I. Determination of Cost Under the Unfair Practices Act Is an Issue of Fact to Be Determined by the Jury.

A. Applicable Law.

The purpose of the Unfair Practices Act (UPA) is “to safeguard the public against the creation or perpetuation of monopolies and to foster and

encourage competition, by prohibiting unfair, dishonest, deceptive, destructive, fraudulent and discriminatory practices by which fair and honest competition is destroyed or prevented.” (§ 17001.) It forbids most locality discriminations (§§ 17040-17041, 17049-17050), the use of loss leaders (§ 17044), gifts (§ 17043), secret rebates (§ 17045), boycotts (§ 17046), and “deceptive, untrue or misleading advertising” (§§ 17200, 17500). It also prohibits the sale of goods and services below cost (§§ 17043, 17048.5, 17049). This last was the violation the Independent alleged it had suffered at the hands of the Examiner.

Proof that a defendant sold or distributed articles or products below cost will be “presumptive evidence of the purpose or intent to injure competitors or destroy competition” (§ 17071). The UPA provides definitions for “cost” in various settings (e.g., production and distribution) (§§ 17026, 17029). A number of UPA provisions specify what sort of evidence may be used to demonstrate cost (§§ 17027, 17071.5-17077).

This case involved production, for which the UPA defines cost to include “the cost of raw materials, labor and all overhead expenses” (§ 17026). The separate definition of overhead expenses is given as “labor (including salaries of executives and officers), rent, interest on borrowed capital, depreciation, selling cost, maintenance of equipment, delivery costs, credit losses, all types of licenses, taxes, insurance and advertising” (§ 17029). These statutes embody California’s fully allocated cost standard, that is, a fair allocation of all fixed or variable costs associated with production of the article or product. (See Turnbull & Turnbull v. ARA Transportation, Inc. (1990) 219 Cal.App.3d 811, 820 [ 268 Cal.Rptr. 856 ]; G.H.I.I. v. MTS, Inc. (1983) 147 Cal.App.3d 256, 275 [195 Cal.Rptr, 211, 41 A.L.R.4th 653 ].) Fixed costs are those that do not vary with changes in output, while variable costs are those that do vary with changes in output. “[F]ully allocated cost has been equated with average total cost, which ‘reflects that portion of the firm’s total costs—both fixed and variable—attributable on an average basis to each unit of output.’ ” (Turnbull & Turnbull v. ARA Transportation, Inc., supra at p. 820.) Thus, cost has been described as the initial expense of producing the article together with “its share of the load of carrying on the business through which it is sold.” (People v. Kahn (1936) 19 Cal.App.2d Supp. 758, 765 [ 60 P.2d 596 ].) Cost has also been treated as “the burden” imposed upon the producer’s “overall cost of doing business.” (Turnbull & Turnbull v. ARA Transportation, Inc., supra, at p. 822.)

Determination of the defendant’s cost has always been treated as an issue of fact. One of the earliest decisions construing the UPA found our Supreme Court stating that “. . . any difficulty in computing cost is a factual one” (People v. Pay Less Drug Store (1944) 25 Cal.2d 108, 118 [ 153 P.2d 9 ]; accord, Turnbull & Turnbull v. ARA Transportation, Inc., supra, 219 Cal.App.3d 811, 820 ; People v. Kahn, supra, 19 Cal.App.2d Supp. 758, 765 ). That principle remains true. (See Western Union Financial Services, Inc. v. First Data Corp. (1993) 20 Cal.App.4th 1530, 1539 [ 25 Cal.Rptr.2d 341 ].)

An issue of fact is one where the evidence introduced will support a decision on either side, that is to say, reasonable minds could fairly differ as to the answer to the question posed. (E.g., Hamilton v. Pacific Elec. Ry. Co. (1939) 12 Cal.2d 598, 603 [ 86 P.2d 829 ]; Greyhound Lines, Inc. v. Superior Court (1970) 3 Cal.App.3d 356, 260 [ 83 Cal.Rptr. 343 ].) An issue of fact can become an issue of law when reasonable minds can draw only one conclusion from the evidence. (E.g., People v. Wild (1976) 60 Cal.App.3d 829, 832 [ 131 Cal.Rptr. 713 ]; Wozniak v. Peninsula Hospital (1969) 1 Cal.App.3d 716, 725 [ 82 Cal.Rptr. 84 ].) An issue of fact cannot be taken from a jury by the trial court and treated as an issue of law unless only one conclusion is legally deducible and any other conclusion cannot command the support of substantial evidence that will survive appellate review. (E.g., Seneris v. Haas (1955) 45 Cal.2d 811, 821 [ 291 P.2d 915 , 53 A.L.R.2d 124 ]; Serian Brothers, Inc. v. Agri-Sun Nursery (1994) 25 Cal.App.4th 306, 311-312 [ 30 Cal.Rptr.2d 382 ].)

Several reported decisions are helpful in demonstrating the minimal quantum of evidence necessary to qualify the issue of cost as one which must be determined by the trier of fact.

Tri-Q, Inc. v. Sta-Hi Corp. (1965) 63 Cal.2d 199 [ 45 Cal.Rptr. 878 , 404 P.2d 486 ] involved competing manufacturers of equipment and machinery used in the printing of newspapers. Tri-Q sued Sta-Hi for selling a specific product below cost, in violation of section 17043. During the relevant time period Sta-Hi had no formal cost accounting system in place. Sta-Hi’s president testified that he fixed the price of the product using a rough estimate of the labor and materials involved. The opposing evidence was in the form of an accounting report showing that the “range of estimated probable unit costs” was above the prices charged by Sta-Hi. The trial court, acting as the trier of fact, found that Sta-Hi had not sold its products below cost. This determination was upheld by the Supreme Court, which held that “Since there was substantial support in the record for the trial court’s resolution of the issue as to whether the product was sold below cost, the finding of fact with respect thereto cannot be disturbed on appeal.” (Id. at pp. 203-207.)

At issue in Western Union Financial Services, Inc. v. First Data Corp., supra, 20 Cal.App.4th 1530 was the cost of money orders. Relying on what is described as “a single document created by a nonmanagement First Data employee,” Western Union presented expert testimony that First Data charged $9 for a money order whose cost could range from $20.80 to as much as $26.60. First Data’s evidence was that its cost per transaction was $16.26, but it realized revenue of about $17.31 per transaction. The trial court, acting as the trier of fact, found that First Data was making a profit on the transactions and therefore was not selling its products below cost. (Id. at pp. 1533-1535.) The Court of Appeal held that “. . . there is a factual dispute about the manner in which costs were computed . . . and what that means is that we will not disturb the trial court’s findings.” (Id. at p. 1539, citations omitted.)

For present purposes, Turnbull & Turnbull v. ARA Transportation, Inc., supra, 219 Cal.App.3d 811 , is perhaps most instructive. Turnbull and ARA operated bus transportation services and were competing bidders for annual contracts to bus children within the San Joaquin County school system. After repeatedly losing out to ARA, Turnbull brought suit, claiming that ARA’s bids were below its costs. ARA’s routine destruction of documentation meant that Grundig, Turnbull’s economic expert, was unable to recreate precise cost figures. “Grundig used four different methods in calculating whether ARA’s bid was below its costs. The first method involved comparing the revenue per mile received by ARA on the San Joaquin County contract with the revenue per mile it received on other bus transportation contracts in other counties, after having made adjustments for cost differentials between the counties. The second method compared the revenue per mile ARA received on the San Joaquin County contract with the revenue per mile it received on its Delta College contract in San Joaquin County. In the third method, Grundig calculated ARA’s costs based on two national cost surveys regarding operating costs of private passenger vans. Grundig’s fourth method entailed an analysis of ARA’s operating statements and an allocation of ARA’s operating expenses per mile traveled. ARA’s total operating costs for the year were divided by the total miles traveled to yield an average cost per mile. This figure was then multiplied by the contract mileage to determine the contract cost. In each year in question, ARA’s bid was below its costs of providing the contract services.” (Id. at p. 816.)

ARA provided no testimony or documentation concerning projected or actual costs for performing the contracts. One ARA financial officer testified that “. . . ARA was concerned with making a profit on each contract and would not bid under cost. Jack Gottsman, the president of ARA’s transportation group, testified .... ARA was concerned that each contract make a profit and bids were reviewed to make sure ARA was going into a contract on a profitable basis, [¶] Robert Griffiths, the area controller who was in charge of the bidding process . . . , testified concerning ARA’s method of allocating costs. It appears that variable costs were allocated per bus, per day. Fixed or overhead costs were allocated on the basis of bus days and revenue generated by each contract. The overhead would first be allocated among the various divisions based on the number of bus days generated by each division. Then each division would allocate the overhead per contract in proportion to the percentage of revenue generated by each contract. According to Griffiths, the San Joaquin County bids covered ARA’s costs and generated a profit.” (219 Cal.App.3d at pp. 818-819.)

The jury returned a verdict for Turnbull. ARA appealed, contending inter alia that the UPA was unconstitutional “because the method for determining whether a service is sold below cost is arbitrary and irrational.” (Turnbull & Turnbull v. ARA Transportation, Inc, supra, 219 Cal.App.3d 811, 819 .) In the course of rejecting this claim, the Court of Appeal stated: “ARA claims the fully allocated cost method is arbitrary because there are many ways of allocating costs (for example, in the present case per mile, per bus or per student) each of which may result in significantly different cost profiles. While we agree there are many ways of fully allocating costs, the possibilities are not without limitation. To be legally acceptable, the allocation of indirect or fixed overhead costs to a particular product or service must be reasonably related to the burden such product or service imposes on the overall cost of doing business. [Citations.] Moreover, a defendant is free to demonstrate to the trier of fact that its fully allocated cost, using another reasonable allocation method other than that used by the plaintiff, is actually lower than the plaintiff alleges and lower than defendant’s sales or bid price. [Citation.] ARA did not do so. It merely presented witnesses who testified ARA always bid above cost or always sought to make a profit. In fact, ARA did not even demonstrate it sold its services above its average variable cost or long-run incremental cost, the methods it espouses. As such, defendant failed to demonstrate that the use of the fully allocated cost method was arbitrary or irrational as applied to the facts of this case.” {Id. at. pp. 822-823, italics added.)

Several conclusions are obvious. First, the concept of cost may appear simple, but can often prove deceptively hard to grasp in the real world. (See People v. Kahn, supra, 19 Cal.App.2d Supp. 758, 765 [“It must be conceded that in many cases it is going to be extremely difficult to determine what cost of an article is.”]; Transamerica Computer Co., Inc. v. IBM Corp. (9th Cir. 1983) 698 F.2d 1377, 1387 [noting “the uncertainty and imprecision inherent in determining ‘costs’ ”].) Authorities in the field argue passionately over the many possible models and approaches. (See, e.g., 3 Areeda & Hovenkamp, Antitrust Law (1996 rev. ed.) §§ 722-724d, 739-742, pp. 221-244, 367-426.) Second, and as a corollary, the determination of cost is best approached on a case-by-case basis. The provisions of the UPA and various judicial glosses reflect an appreciation that no single formulation will ever be broad enough to encompass all possibilities. The principle may be fixed and constant, but the application is prey to the infinite permutations of the question—what did it cost a particular defendant to produce or sell a given article at a specific point in time? Third, California appears to have adopted a very expansive approach to the evidence that may be used to establish- cost; no formula has been expressly sustained or denounced.

With this background, we turn to the Examiner’s first contention—that the trial court committed reversible error when it precluded the Examiner from presenting its version of how it calculated cost for purposes of bidding for the City’s contract.

B. Application of Law in This Case.

The parties continue the intransigence they displayed throughout the trial, each insisting that its economic model is the only acceptable method for determining cost, and that the other side’s model is worthless and erroneous as a: matter of law. The Examiner insists not only must the judgment be reversed, there must be no retrial, and it is entitled to judgment notwithstanding the verdict. The Independent is equally adamant that the trial court was correct in excluding evidence of Crichfield’s analysis because “revenue allocation the only rational way to determine the Examiner’s cost.” (Original italics.)

The revenue allocation adopted by Dr. McLeod, and to a lesser extent by Dr. Maher, appears to be an allocation approach used if the standard cost allocation (see § 17026 [“cost of raw materials, labor and all overhead expenses”]) either will not work or is inappropriate for some reason. McLeod believed it was inappropriate in this case to use the standard cost allocation (§ 17026 [“cost of raw materials, labor, and all overhead expenses”]) because there is not a common physical unit of measurement, and because “you have two products that are intertwined and therefore there is no rational way of simply allocating the costs.” (The latter premise obviously ties in to the trial court’s “two-product” ruling; see fn. 4 and accompanying text, ante.) Using the revenue allocation method, McLeod assigns costs to the two products “[b]ased on an approximation of the value received from the sales of those two products.”

Dr. Crichfield’s opinion was that revenue allocation was not appropriate because there was a common physical unit of measurement—the inches of space devoted to the advertising and “editorial” (i.e., news gathering and reporting) functions, which could be used to allocate the cost of production of the Examiner. He allocated 55 percent of the Examiner’s total costs to the editorial side, and 45 percent to advertising.

The difference in result between the two approaches is gaping. Working from the conceded figure of approximately $60.5 million for the Examiner’s total costs, McLeod allocated only $11,228,000 to “non-advertising revenues,” resulting in $49,250,000 of costs allocated to advertising; that figure was then divided by the number of “advertising inches” to produce a “cost per classified ad line” of $1.85. Dr. Crichfield found $23.8 million of advertising costs divided by the same number of advertising inches to produce a per line cost of 89 cents for advertising. Where Crichfield assigned 45 percent of the Examiner’s total costs to the “advertising product,” McLeod allocated almost 85 percent.

Each side can and does point to flaws in the other’s model. This, we believe, illuminates the fundamental point missed by both parties and by the trial court. The hyperbole obscures the fact that each approach has its relative merits and demerits. Neither is entitled to prevail as a matter of law. Neither is so strong that it can banish the other. In short, both approaches are sufficiently reasonable that both ought to have been put before the jury.

When the Examiner condemns McLeod for using “an irrational ‘revenue allocation’ methodology that is not sanctioned by California law,” it overstates its case. The revenue allocation method may lack an express sanction in California, but so does the Examiner’s approach. And while no reported California decision explicitly sanctions or endorses any approach to determining cost, the revenue allocation method was mentioned in Turnbull & Turnbull without a hint that it was inappropriate. Moreover, the Examiner never suggested that the revenue allocation approach was new, unscientific, or unaccepted in the fields of economics and cost accounting. (Cf. In re IBM Peripheral EDP Devices, etc. (N.D.Cal. 1979) 481 F.Supp. 965, 998 [“Revenue apportionment is a generally accepted and commonly used method of . . . cost allocation.”].) Certainly with the record so undeveloped on this point, we cannot hold as a matter of law that the revenue allocation approach could not be used.

On the other hand, the advantage of measuring the cost of an item backwards from the revenue generated by that item is not immediately self-obvious and can seem incompatible with common sense. This approach appears to have no connection with the traditional tests of “initial cost,” or the item’s “share of the load of carrying on the business through which it is sold” (People v. Kahn, supra, 19 Cal.App.2d Supp. 758, 765 ), or “the burden . . . impose[d] on the overall cost of doing business.” (Turnbull & Turnbull v. ARA Transportation, Inc., supra, 219 Cal.App.3d 811, 822 .) It further appears that there are distinct logical flaws in the approach as developed in this case.

According to Dr. McLeod, the revenue allocation approach should be used only in a joint cost situation where there is no common unit of measurement. In this situation, however, as the Examiner repeatedly urged, there was a common unit of measurement, specifically the literal inches of the Examiner. It is not irrational to divide the Examiner by the inches of space devoted to the editorial and advertising products. On the other hand, as Dr. McLeod testified, one may buy X number of inches of advertising space, but “You can’t go to the newsstand and say, ‘I’d like to buy 25 inches of news copy today.’ ” Moreover, as the Examiner points out in its opening brief, McLeod’s model treats more than $23 million of the Examiner’s total costs, those for “editorial costs” plus ink and newsprint, as joint costs when in fact they are obviously and easily segregated as part of the “editorial product.” They would therefore not be subject to the revenue allocation percentages and would almost certainly require a recalculation of those percentages. This suggests that Dr. McLeod’s analysis was perhaps too ambitious in its scope and its application of deductive revenue allocation; in effect it made all Costs joint costs. These flaws are not, however, of sufficient magnitude to condemn the revenue allocation approach per se and make it inadmissible as a matter of law. These and other imperfections would go to the weight the jury ought to give McLeod’s testimony, but they do not make it wholly illegitimate.

We conclude that neither of the parties’ cost models was irrational or insufficiently allocated to comply with the UPA’s requirement. Neither model is so persuasive that it alone would attract all reasonable minds, allowing the issue to be resolved as a question of law. (See Seneris v. Haas, supra, 45 Cal.2d 811, 821 ; Serian Brothers, Inc. v. Agri-Sun Nursery, supra, 25 Cal.App.4th 306, 311-312 .) The issue of the Examiner’s cost remained a question of fact on which the parties could introduce conflicting evidence. It was therefore error to exclude evidence of the Examiner’s model from being presented to the jury. (E.g., Western Union Financial Services, Inc. v. First Data Corp., supra, 20 Cal.App.4th 1530, 1539 ; People v. Kahn, supra, 19 Cal.App.2d Supp. 758, 765 .)

C. Taking Issue of Determination of Cost From Jury Requires Reversal.

The error clearly requires reversal because of its importance and its manifold consequences. Cost was the issue of this litigation as it went to trial. Dr. Crichfield’s testimony would therefore be crucial for the Examiner. Although both sides in opening statements made reference to Crichfield’s anticipated testimony, the comments of counsel for the Independent were far more numerous and pointed. The preemptive attacks on Crichfield continued virtually from the start of the actual jury trial, with the Independent’s second and third witnesses (Drs. McLeod and Maher) spending almost as much time attacking the as-yet-unheard Crichfield conclusions as explaining their own. The court’s sua sponte overturning of its ruling could not help but cause confusion and consternation for the jury. The thing they had heard so much about would now not be produced, and they were told to forget all about it. For the Examiner, whose counsel had featured Crichfield prominently in his opening statement, the inability to call that witness was unquestionably damaging. (Cf. People v. Corona (1978) 80 Cal.App.3d 684, 725-726 [ 145 Cal.Rptr. 894 ] [defense counsel dealt “devastating blow” to client when he failed to present evidence promised in opening statement].) The jury did finally get to hear Dr. Crichfield, but he was only allowed to present negative testimony, which is to say, testimony about the defects in the Independent’s case. Even then, a good part of his testimony was stricken. It is hardly surprising that the jury initially had trouble remembering what part of Crichfield’s testimony was available to them. (See fn. 5, ante.) Counsel for the Independent told the jury that the only relevant evidence on the dispositive issue of cost favored the Independent, and the trial court twice told the jury that the Independent’s cost model was the sole legal method for determining cost. (Ibid.) This was tantamount to the court directing a verdict against the Examiner. In light of the six days it took the jury to return a nine-to-three verdict in favor of the Independent, there is more than a reasonable probability of a more favorable result; the error thus compels reversal. (Cal. Const., art. VI, § 13; People v. Watson (1956) 46 Cal.2d 818, 836 [ 299 P.2d 243 ]; Code Civ. Proc., § 475.)

II.-IV.

V. Conclusion.

The judgment and attorney fee order are reversed. The order denying the motion for judgment not withstanding the verdict is affirmed. Appellants shall recover their costs.

Hanlonj P. J., and Poché, J., concurred.

A petition for a rehearing was denied September 20, 1999, and the opinion was modified to read as printed above. Appellants’ petition for review by the Supreme Court was denied December 15, 1999.

Statutory references are to this code unless otherwise indicated. The Examiner is an evening paper owned by the Hearst Corporation. The Examiner and the San Francisco Chronicle, a non-Hearst morning paper, produce a joint Sunday edition; they share facilities and noneditorial business functions pursuant to a joint operating agreement authorized by the Newspaper Preservation Act ( 15 U.S.C. §§ 1801-1804 ; 28 C.F.R. §§ 48.1-48.16 (1998)). The overlap between the two papers is the responsibility of the San Francisco Newspaper Printing Company, Inc. That entity and the Hearst Corporation were named as defendants and answered the complaint at issue here. For purposes of simplicity, however, these defendants will be collectively referred to as “the Examiner.” The Independent is owned by Pan Asia Venture Capital Corporation, the plaintiff in this litigation. In order to maintain the symmetry of this dispute between newspapers, plaintiff will be referred to as “the Independent.” The Independent bid $2.35 per line, while the Examiner bid $2.28. The Examiner concedes, however, that technical factors such as differing type size and line count means that “the dollar amounts of the respective bids ... are not directly comparable. To derive directly comparable bids, the dollar amount of the Independent’s bid must be reduced by approximately 15% . . . which reduces the Independent’s bid of $2.35 per line to approximately $2.00 per line for purposes of comparison.” In its opening brief, the Examiner states: “Although Defendants continue to believe that the Examiner can be viewed as constituting a single ‘product’ with multiple revenue streams—a newspaper containing both editorial material and paid advertisements—the distinction is academic in the circumstances of this case. Whether the Examiner is viewed as a single, integrated product, or as two distinct products—a newspaper and advertising, Defendants did not engage in below-cost pricing. Nor would Defendants be found to have engaged in below-cost pricing if one separately calculated the costs of classified advertisements. For that reason, Defendants’ arguments in this brief are addressed to other rulings of the trial court, and assume—solely for purposes of this appeal, and arguendo—that there are indeed two ‘products’: a daily ‘newspaper product’ which consumers purchase from a newsstand or by subscription, and an ‘advertising product.’ Defendants reserve the right to contend in the future, should the occasion arise, that their ‘product’ or ‘products’ should be defined differently, [¶] It should be noted, however, that Plaintiff’s two-product theory—which is a fundamental premise of Plaintiff’s claim in this case—is inconsistent with Plaintiff’s rationale for the ‘revenue allocation’ methodology that the trial court approved.” The Independent treated this as a “concession” that the Examiner was not contesting the trial court’s ruling that two products were at issue. The Examiner responded in its reply brief that no such concession had been made; it also made the first developed argument attacking the two-product ruling as a grounds for reversal. Three days after the Examiner’s reply brief was filed, a number of publishers filed an amicus curiae brief supporting the Examiner, in which they passionately contend that the trial court’s ruling on this issue was error. This mutual desire of the Examiner and amici curiae to now litigate the two-product ruling on appeal will not be indulged. As did the Independent, we too read the language in the Examiner’s opening brief as a concession that it will not be contesting the correctness of the trial court’s two-product ruling. The operative language of the footnote—“the distinction is academic in the circumstances of this case,” “Defendants’ arguments in this brief are addressed to other rulings,” “assume— solely for purposes of this appeal. . . —that there are indeed two ‘products,’ ” “Defendants reserve the right to contend in the future, should the occasion arise” (italics added)—cannot be reasonably read as other than a concession. Nor does the reference to page 32 at the end of the footnote assist the Examiner. Page 32 finds the Examiner stating: “If, as Plaintiff urged and the trial court ruled, the ‘newspaper product’ and the ‘advertising product’ are indeed separate products for purposes of the UPA, then the cost of each such ‘product’ must be measured separately and accurately.” A footnote at this point goes on: “Conversely, if Plaintiff persists in arguing that the pricing of advertising and newspapers are inseparably intertwined, it will be necessary for the Court to examine the trial court’s ruling that they are, indeed, two separate ‘products’ for purposes of the UPA’s below-cost pricing statute.” These 10 lines do not constitute a developed argument with supporting authority and appropriate references to the record. The opening brief will therefore not be regarded as raising the issue of the two-product ruling for review. (See, e.g., Troensegaard v. Silvercrest Industries, Inc. (1985) 175 Cal.App.3d 218, 228 [ 220 Cal.Rptr. 712 ].) There being no reason why the issue could not have been developed in the opening brief, the Examiner’s attempt to raise it in the reply brief will not be indulged. (See, e.g., Scott v. CIBA Vision Corp. (1995) 38 Cal.App.4th 307, 322 [ 44 Cal.Rptr.2d 902 ]; Neighbours v. Buzz Oates Enterprises (1990) 217 Cal.App.3d 325, 335, fn. 8 [ 265 Cal.Rptr. 788 ].) The amicus curiae brief cannot remedy this defect because amici curiae are not allowed to expand the issues raised by the parties. (See, e.g., California Assn. for Safety Education v. Brown (1994) 30 Cal.App.4th 1264, 1274-1275 [ 36 Cal.Rptr.2d 404 ] and decisions cited.) Exceptions are allowed on points of jurisdiction or issues that will result in affirmance (see E. L. White, Inc. v. City of Huntington Beach (1978) 21 Cal.3d 497, 511 [ 146 Cal.Rptr. 614 , 579 P.2d 505 ]), but amici curiae are attempting to argue a nonjurisdictional evidentiary error they view as requiring reversal. The ordinary rule of party issue selection and preclusion therefore applies to this and the other evidentiary error claimed only by amici curiae. The Examiner will of course be able to ask the trial court to reconsider this ruling if there is a retrial. The jury was instructed that “the evidence in this case has been that the only method for allocating costs to these two products [i.e., the newspaper product and advertising space product] is the so-called ‘revenue method’ of allocating joint costs. The revenue method of allocating joint costs is a method based on the assumption that the ratio between the amounts of revenue earned by the products involved is the same as the ratio of the costs incurred in producing these products.” (Italics added.) Two days into their deliberations the jury asked the court “Is the revenue method the only legal method to consider in determining if the cost of advertising is below cost in this case?” The court responded that “The answer to that question is yes.” Counsel for the Independent had forcefully made the same point during his closing argument. The court drove the same point home by telling the jury what they could not consider: “During the trial of this action, there has been evidence of methods of cost allocation defendants used before [the date of their winning bid] to determine their costs. I instruct you that you are to consider those methods only as they may bear on the defendants’ state of mind, and not as methods which you may apply to determine the defendants’ advertising costs in this action.” One other point must be mentioned. The Examiner states in its brief that “[D]uring its deliberations the jury inquired as to which portions of Dr. Crichfield’s testimony were in evidence . . . but the trial court never clarified the issue.” The clear implication of this passage is that the trial court was, to use the mildest term, derelict in its duty. The Examiner’s statement is literally (rue but almost slanderous in its innuendo. What actually happened, and what clearly appears on page 7354 of the reporter’s transcript, is that the jury foreman told the court “The jury has decided we no longer need the information that we requested.” The court then verified “you are telling me that you no longer need this information; is that right?” The jury foreman answered, “That’s right.” Cost is to be measured as “the fair average cost of production over a reasonable time, rather than the cost of one item on a particular occasion.” (Johnson v. Farmer (1940) 41 Cal.App.2d 874, 882 [ 107 P.2d 959 ].) This is not a problem in this case, where both sides’ cost studies and models were based on annual production costs. Not even Dr. McLeod made this claim for his method. He candidly admitted that using revenue allocation was “a judgment call to some extent,” that it was not a “precise formula," but it was “the best you can come up with under the circumstances.” Dr. McLeod’s bottom-line per-line cost figure fluctuated from $1.82 to $1.92. As cogently expressed by amici curiae: “A physical-unit method of accounting is appropriate because it directly tracks the requirement that it be ‘reasonably related to the burden such product or service imposes on the overall cost of doing business.’ [Citation.] Unlike the revenue-allocation method, the physical-unit method actually has a close relationship to the working operations of the newspaper. Public notice advertising takes up physical space in the newspaper, thereby consuming newsprint and ink, is sold in physical units known as column-inches, and is delivered to readers in such physical form.” The difficulty with the revenue approach is perhaps best illustrated with a specific example: Dr. McLeod identified $15,863,613 as “editorial costs,” yet his approach allocates almost all of these costs to the “advertising product.” We reject the Independent’s argument that this issue amounts to invited error and was waived by the Examiner’s counsel when he asked the court to decide the reasonableness of the respective cost models as a matter of law. First, that request was made in connection with the first model submitted by Dr. Crichfield, which the court found was unallocated, resulting in its order that a new model be prepared. The Examiner’s request was not renewed with respect to the second Crichfield model, the one at issue here. The second Evidence Code section 402 hearing was conducted at the Independent’s request, based on its motion to exclude Dr. Crichfield’s second cost determination study. Second, the trial court obviously did not deem the Examiner’s request applicable to the second model, as evident from its initial decision that neither it nor the McLeod model was “unreasonable” and the choice between them would be left to the jury. As the Independent’s attorney told the jury in his opening statement: “[W]hat this case is all about is the allegation on the part of The Independent that when The Independent competed with The Examiner ... to get the public notice contract . . . The Examiner bid below cost, below their fully allocated cost.” E.g., “And so what they chose to do was to hire Dr. Crichfield. They will say that Dr. Crichfield is an independent accountant who gave them independent advice with regard to their costs. [¶] But the evidence will show that, in fact, Dr. Crichfield’s primary career role is the creation of litigation cost models, and that this cost model is really a cost model designed to defend The Examiner in a lawsuit. . . . There are some other accounting tricks . . . which Dr. McLeod [the Independent’s expert] will say makes this allocation study by Dr. Crichfield completely useless.” In its petition for rehearing, the Independent contends that we “mistakenly decided the ‘cost’ issue under Evidence Code section 403 instead of Evidence Code section 405.” We were not mistaken, because we decided no such issue, and the Independent’s belated presentation of this claim will not avert reversal. The Independent’s argument is apparently based on this statement from Turnbull: “To be legally acceptable, the allocation of . . . costs to a particular product or service must be reasonably related to the burden such product or service imposes on the overall cost of doing business.” (Turnbull & Turnbull v. ARA Transportation, Inc., supra, 219 Cal.App.3d 811, 822 .) The Independent reasons that whether Dr. Crichfield’s model complied with this standard is a preliminary fact, one that Evidence Code section 405 vests with the trial court to decide. Evidence Code section 403, by contrast, allows the jury to decide whether the preliminary fact exists if “the court finds that there is evidence sufficient to sustain a finding of the existence of the preliminary fact” (Evid. Code, § 403, subd. (a)). In fact, the latter statute requires the issue of preliminary fact determination to be submitted to the jury but “Section 405 requires the judge to determine the existence or nonexistence of disputed preliminary facts except in certain situations covered by Section[] 403 .... Section 405 deals with evidentiary rules designed to withhold evidence from the jury because it is too unreliable to be evaluated properly or because public policy requires its exclusion.” (See Assem. Com. on Judiciary, corns, to Evid. Code, §§ 403, 405, 29B West’s Ann. Cal. Evid. Code (1995 ed.) pp. 361, 374; Deering’s Ann. Cal. Evid. Code (1986) pp. 142, 150-151.) Whether the economic models’ compatibility with section 17026 and the Turnbull standard is properly handled under Evidence Code section 403 or Evidence Code section 405 is an intriguing question, but one we cannot, and need not, decide. There is undoubtedly a case to be made for viewing the admissibility of evidence dependent upon its compliance with an established legal standard as one involving both reliability and public policy, and therefore governed by Evidence Code section 405. On the other hand, it is not fanciful to view the problem as relevance, that is to say, a model that did not comport with Turnbull would not be relevant on the issue of cost. This approach would make the problem subject to Evidence Code section 403, subdivision (a)(1). Certainly this was the trial court’s initial view of the matter: In its written order deciding that the McLeod model could be used but Crichfield’s initial model could not (see ante at pp. 428-429), the court made these decisions “pursuant to Evidence Code section 403 (a)(1).” Although the court did not enter a written order when it ruled that Crichfield’s revised model was not “unreasonable in the context of . . . the Turnbull case” and not “unreasonable to the extent that it’s something that should be precluded from being presented to the Jury altogether,” it is difficult to believe the court would have shifted the statutory basis it had been employing up to that point. The blizzard of paper generated by the parties’ motions to exclude the other’s economic model demonstrates that neither party was entirely consistent in invoking either section 403 or 405 of the Evidence Code. Unfortunately, when the trial court reversed course it became impossible to determine if the court would have continued to treat the issue as an Evidence Code section 403 subject. It was because of this uncertainty that we said nothing more specific than that the court had conducted the pretrial hearings “pursuant to Evidence Code section 402.” (See ante, at pp. 428-429.) It is therefore inaccurate for the Independent to believe we “decided the ‘cost’ issue under Evidence Code section 403.” We did no such thing. The issue is also immaterial because.there is little practical consequence to which statute the trial court used. With or without special instructions (see Evid. Code, §§ 403, subd. (c), 405, subd. (b)(2)), the essential duty of the jury would have been pretty much the same— judge the evidence against the Turnbull standard given in the court’s instructions. With Turnbull as the benchmark, the jury would decide which party’s evidence it found to reflect the Examiner’s cost. If the jury had accepted Crichfield’s model, it would in effect have decided that his evidence was compatible with the preliminary fact of compliance with the Turnbull standard and the McLeod model was not. On the other hand, if the jury had accepted the McLeod/Maher testimony, it would have reached the opposite conclusions. Such speculation aside, the Independent’s contention is ultimately doomed because it does not matter which provision of the Evidence Code the trial court did or ought to have used. Even if (as the Independent assumes) the trial court did use Evidence Code section 405, the court erred when it determined that the Crichfield model was “irrational” and therefore presumably not fully allocated as required by section 17026 and Turnbull as a preliminary fact. Alternatively if the trial court did follow through on its initial ruling and continue to apply Evidence Code section 403 when it excluded Crichfield’s evidence, the court would also be in error because “there [was] evidence sufficient to sustain a finding of the existence of the preliminary fact” (Evid. Code, § 403, subd. (a)), i.e., that it was fully allocated in compliance with governing legal criteria. Both of these possibilities are encompassed by our analysis and our conclusion that reversible error occurred. See footnote, ante, page 424.