Legal or Factual Hypothetical in the S 8 “But For” World?

Apotex Inc v Sanofi-Aventis Canada Inc 2014 FCA 68 Sharlow JA, Pelletier JA concurring, Mainville JA dissenting, affd 2015 SCC 20 SCC 35886 varg 2012 FC 553 Snider J

Teva Canada Ltd v Sanofi-Aventis Canada Inc 2014 FCA 67 Sharlow JA, Dawson JA concurring, Mainville JA dissenting varg 2012 FC 552 Snider J

1,187,087 & others / ramipril / ALTACE / NOC s 8

Pfizer Canada Inc v Teva Canada Ltd 2016 FCA 161 Stratas JA: Ryer, Gleason JJA varg 2014 FC 248 Zinn J

            1,248,540/ 2,199,778 / venlafaxine / EFFEXOR XR

In a post last week I revisited the question of how to construct the “but for” world for the purposes of s 8 damages when there are multiple claimants. I pointed out Ramapril / Apotex 2014 FCA 68 (and the companion Ramipril / Teva 2014 FCA 67), raised a puzzle as to the exact nature of the hypothetical used in constructing the “but for” world. A few days later, in reviewing Venlafaxine 2016 FCA 161, it occurs to me that there is a basic question as to whether the “but for” world is premised on a legal hypothetical or a factual hypothetical. On a legal hypothetical, we are to assume that some aspects of the PMNOC Regulations don’t exist for some purposes; on a factual hypothetical we are to assume that the patentee would have behaved differently in some way.

In Ramipril the FCA rejected Sanofi’s argument that there should be only one “but for” world in which none of the potential generic entrants would have been constrained by the PMNOC Regs. Sharlow JA summarized in Ramipril / Teva [145] by saying (my emphasis):

in the hypothetical world constructed for the purposes of determining section 8 damages, the NOC Regulations should not be assumed away except to the extent required by paragraph 8(1)(a) [now 8(2)], that is, for the purpose of determining the beginning of the section 8 liability period. For all other purposes, the NOC Regulations should be assumed to exist in the hypothetical world

This posits a legal hypothetical as the basis for the “but for” world: the Regs are assumed away, though only for the specific purpose of determining the beginning of the liability period.

In last week’s post I pointed out that this is hard to reconcile with Norfloxacin 2011 FCA 329 [75], a s 8 case where the FCA stated that “the Federal Court had to assess [the generic’s s 8] damages on the basis of a hypothetical question: what would have happened had [the patentee] not brought an application for prohibition?” This is a factual hypothetical; the Regs are exactly as they are in the real world, with no modification whatsoever, and the only difference is that the patentee acts differently than it did in the real world. Venlafaxine 2016 FCA 161 also explicitly poses a factual hypothetical: to assess the impact of lost sales, the court’s task is “to assess a hypothetical world where the defendant’s impugned conduct did not take place” [46] (and similarly [45]). This says nothing about the Regs; it is the patentee’s conduct that is different in the “but for” world. More specifically, in Venlafaxine the s 8 claim was triggered because it was ultimately determined that the patent at issue was not eligible for listing on the patent register [24], and “with the benefit of hindsight, it can be said [the patentee] should not have listed [the patent at issue] on the patent register . . . and should not have brought a prohibition application” [26]. The FCA held that the trial judge had properly proceeded on the basis that “the hypothetical world . . . was one where [the patentee] did not improperly list its [] Patent,” and it is for that reason that the generic would have received its NOC on the patent hold date [71]. That is, the impugned conduct was listing a patent that we now know was not legally eligible for listing: see also [26].

There is considerable logic to positing a factual hypothetical rather than a legal hypothetical. It avoids the problem of assuming that there is a world in which some parts of the Regs exist for some generics, but not for other generics. This is a a strange assumption, as a law that applies to some parties but not to others who are similarly situated would not be possible in the real world. Moreover, the legal hypothetical used in Ramipril is not even internally coherent; as I explained in last week’s post, the decision assumed that in the “but for” world the patentee would have brought a prohibition application against the claimant, that would have unfolded as it did, even though the prohibition application would have been moot from the outset if the statutory stay did not apply to the claimant (which would have therefore received its NOC). One might say that it is just a hypothetical to allow us to calculate damages, but it is a tortured hypothetical nonetheless. In general, the damages hypothetical is realistic—the negligent driver would have driven safely, the defective house would have been built to Code, the infringer would have competed with a non-infringing alternative. Indeed, a hypothetical that is not realistic is generally not permitted: in assessing s 8 damages, the generic has to show that it could and would in fact have supplied the market, so why should we assume that the patentee would have done something that in reality would have been quite irrational?

More generally, the usual damages hypothetical is that the wrongdoer would have acted legally instead of illegally. This general approach can be applied to s 8 by using a factual hypothetical, in which the patentee acted properly instead of (with hindsight) improperly. In a case such as Venlafaxine, in which the patent was not eligible for listing, it is reasonable to suppose that the patentee would not have listed it in the first place, if the patentee knew that its patent was not in fact eligible. If the patent is ultimately held to be invalid, then it is also reasonable to suppose that the patentee would not have listed it, had it been able to foresee that outcome. If the s 8 claim arises because the patent was eligible and valid, but the generic’s product was found not to infringe, then the obvious counter-factual is to suppose that the patentee would have listed it, but would not have brought a prohibition application against that particular claimant, knowing it was bound to lose. (I’m referring to prohibition applications because that is the regime under which those cases were decided, but the same logic could be applied under the new Regs.)

A factual hypothetical is also consistent with the broad holding in Ramipril, summarized in Venlafaxine [136], that “the regulatory barriers to entry, including the PMNOC Regulations, which all generic manufacturers face in the real world, also affect all generic manufacturers in the hypothetical world. Thus, in order to assess whether and when other generic manufacturers could have and would have entered the market in the hypothetical world, the Federal Court had to assess, among other things, whether regulatory barriers stood in their way.” A factual hypothetical assumes that the Regs themselves operate exactly as they do in the real world, without any modifications whatsoever to the Regs themselves; the difference in the “but for” world lies only in the factual hypothetical as to the patentee’s behaviour, plus whatever consequences might realistically flow from that, in the context of the NOC Regs as they actually exist.

While I think using a factual hypothetical is appealing, I doubt it is consistent with Venlafaxine or Ramipril. As noted, the court in Venlafaxine stated a factual hypothetical as the basis for constructing the “but for” world, but when it came to actually assessing damages, Zinn J in Venlafaxine FC stated that

[130] The plaintiff generic need not comply with the NOC Regulations in the but-for world as they relate to the patent(s) that were the subject of the Prohibition Application for the reasons provided above. However, any competing generic manufacturer must do so because in the real world it has not addressed the patents on the Patent Register.

This is a legal hypothetical: the Regulations do not apply to the claimants, but do apply to other generics. The FCA affirmed [136]. As I understand it—and to be clear, I have not reviewed the Venlafaxine FC decision closely on this point—Novopharm’s entry was delayed by NOC proceedings related to the patent that was improperly filed. This was true in the real world and Zinn J took it to be true in the hypothetical world as well [136]. But if we actually use the factual premise that the FCA stated was the basis for constructing the “but for” world—that Pfizer had not listed the patent that was improperly listed in the real world—then Novopharm’s entry would not have been delayed. That is not because of a legal hypothetical that the Regs would not have applied to Novapharm, but rather as a consequences of the way in which the factual hypothetical plays out. So, if I am reading the cases correctly, it seems that Zinn J stated a factual hypothetical as being the correct approach, but actually applied a different, legal, hypothetical—and the FCA affirmed both. Similarly, in Ramipril, as I understand it (again, I did not review the decisions in detail on this point), the patents at issue were declared invalid, which, to my mind, implies that the appropriate factual hypothetical is that Sanofi would not have listed the patents.

I wonder if the use of a legal hypothetical in Ramipril was a consequence of the way the case was argued. Recall that Pfizer had argued for a methodology that “assumes the NOC Regulations away for the purpose of constructing the hypothetical market” 2014 FCA 68 [157]. As Mainville JA put it, dissenting on this point and agreeing with Pfizer, “those Regulations should be disregarded not only with respect to the claimant generic drug manufacturer, but also with respect to any other generic drug manufacturer that is found, on a balance of probabilities, to also be a market participant” 2014 FCA 67 [98]. That is, Pfizer argued for a sweeping legal hypothetical in which the Regs were assumed away for all purposes; Sharlow JA rejected this in favour of a minimalist legal hypothetical in which the Regs are assumed away only to the extent necessary to establish the liability period start date. Given that Sanofi’s argument was framed around a legal hypothetical, it is not surprising that Sharlow JA chose a minimalist approach. With that said, Snider J at first instance seems to have relied on a factual hypothetical, that Sanofi would not have responded to Apotex’s NOA with an application for an order of prohibition (Ramipril / Apotex FC [6], [157]), which is a reasonable counter-factual if we suppose Sanofi knew it would lose, and Sharlow JA reversed on that point, though without clearly articulating the premise of her hypothetical.

Finally, there is also the statutory point relied on by Sharlow JA in Ramipril / Teva [145], quoted above, which is that 8(1)(a) (now 8(2)) of the Regulations themselves assume away the Regs for a specific purpose, namely for the purpose of determining the beginning of the s 8 liability period. The argument is that the Regs themselves mandate a legal hypothetical. But the Regs do not explicitly say that any part of the Regs are to be disregarded for the purposes of constructing the “but for” world. Rather, as Sharlow JA noted, it says they are to be disregarded for the purposes of determining the start date for the liability period. Para 8(1)(a) (and 8(2)) states that if the patentee brings an action and is unsuccessful, the patentee is liable to the generic “for any loss suffered after . . . the day. . . on which a notice of compliance would have been issued in the absence of these Regulations.” On its face, this says nothing about how the “but for” world is to be constructed, but only addresses the start date for the liability period. It seems to me that this provision is entirely consistent with awarding s 8 damages based on a but for world based on a factual hypothetical in which, eg the patentee never listed the ineligible patent.

I don’t see how to sort this out. It strikes me that adopting a factual hypothetical is consistent with the Regs, is consistent with the general approach to damages in other areas of the law, and would avoid some of the contortions required with the legal hypothetical. But it may be that, as a matter of precedent, that ship has sailed.

Nova v Dow: The Legal Background

Nova Chemicals Corp v Dow Chemical Co 2022 SCC 43 Rowe J: Wagner CJ, Moldaver, Karakatsanis, Brown, Martin, Kasirer and Jamal JJ concurring; Côté J dissenting affg Nova Chemicals Corporation v Dow Chemical Company 2020 FCA 141 Stratas JA: Near, Woods JJA affg Dow Chemical Co v Nova Chemicals Corp 2017 FC 350, 2017 FC 637 Fothergill J

2,160,705 / film-grade polymers / ELITE SURPASS

My previous post on Nova v Dow provided an overview of the problem and tried to identify the central intuition behind both Rowe J’s reasons for the majority, and Côté J dissent. This post reviews the legal background to the case.

Prior to 2004, the usual view was that an accounting required the infringer to disgorge the difference between its revenue and its costs, which is to say the accounting profits, or what Rowe J referred to as the “actual profits”[102]. Then, in Schmeiser 2004 SCC 34, Supreme Court held that the preferred method for calculating an accounting of profits is the “differential profit” approach, in which “[a] comparison is to be made between the defendant’s profit attributable to the invention and his profit had he used the best non-infringing option.” [102], citing my article on “The Innocent-User Problem” (2004) 20 CIPR 79 and Collette v Lasnier (1886), 13 SCR 563, 576. The defendant, Schmeiser, had grown patented herbicide resistant canola, but, so far as the evidence showed, he had not taken advantage of the herbicide resistance, but rather had treated it in the same way as conventional canola, so that his profits “were precisely what they would have been had [he] planted and harvested ordinary canola” [104]. On this evidence, Schmeiser “earned no profit from the invention and Monsanto [the patentee] is entitled to nothing on their claim of account” [105].

The differential profit approach was then elaborated by the Federal Court of Appeal in a series of decisions. The first, Rivett 2010 FCA 207, confirmed that the differential profit approach was indeed generally applicable in the context of an accounting, dismissing an argument that it was confined to the particular facts of Schmeiser. Next, in Lovastatin Damages 2015 FCA 171, the Court of Appeal held that the differential profit approach was also applicable in the context of damages: “if a court may consider a defendant’s resort to a non-infringing alternative when calculating the infringer’s profit, there is no reason in principle to ignore such conduct when calculating the patentee’s lost sales” [60]. To apply the differential profit test, it was therefore necessary to reconstruct the hypothetical or “but for” world in which the infringer had pursued a non-infringing course of action [45], [48]. The Court of Appeal in Lovastatin Damages endorsed a “could and would” test for determining what non-infringing alternative the infringer would have pursued had it not infringed: [49]–[55]. The construction of the “but for” world and the nature of the “could and would” test was further elaborated and refined in a series of decisions involving both damages and accounting of profits: see Venlafaxine 2016 FCA 161 (s 8 damages); Perindopril Accounting #1 2017 FCA 23; Cefaclor Damages 2018 FCA 217; Perindopril Accounting #2 2020 FCA 60.

The effect of these decisions was that the basic approach to either damages or an accounting was the same: both involved the construction of a “but for” world in which the infringer pursued the alternative course of action established by the “could and would” test. Once the “but for” world was constructed, the quantum of damages was the difference between the patentee’s profit in the actual world and its profit in the “but for” world; the quantum of an accounting was the difference between the infringer’s profit in the actual world and its profit in the “but for” world.

In my “Innocent-User” article that was cited by the SCC in Schmeiser, my basic argument was that the differential profit approach should be adopted because it was “simply the application of ‘but for’ causation to an accounting of profits”: 91–93. When the SCC in Schmeiser cited my article as support for the differential profit, I had assumed that it was persuaded by the reasoning, and I understood the subsequent cases through that lens. The general tenor of the Court of Appeal jurisprudence, with its fact-intensive inquiry into what the infringer could and would have done in the hypothetical world “but for” the infringement, appeared to directly reflect the principle of “but for” causation. So, Lovastatin Damages 2015 FCA 171 explicitly stated that its approach was based on “but for” causation: see [45], [52], and throughout, referring to the “but for” world. Cefaclor Damages 2018 FCA 217 [23], [94] was similarly explicit. Perindopril Accounting #1 2017 FCA 23 [61] and Perindopril Accounting #2 2020 FCA 60 [50] both relied on Cadbury Schweppes [1999] 1 SCR 142 [73], which, in the next paragraph, [74], explicitly endorsed “but for” causation as the guiding principle in as the correct approach to the remedy of equitable compensation (the equitable equivalent of damages).

However, on the facts, the non-infringing alternatives endorsed by the Court were arguably market substitutes for the infringing product. In Rivett, like Schmeiser, the alternative to patented seed was conventional seed. In both Lovastatin Damages and Cefaclor Damages, the patents related to a process for producing a drug, and the infringer argued it would have been able to make the same drug using a non-infringing process: see here and here. In the Perindopril Accounting cases, a substantial amount of infringing perindopril had been manufactured by the infringer in Canada and exported to markets such as the UK and Australia, and the alternative at issue was to source perindopril for those markets from other jurisdictions where it was not patent protected: see here. (Venlafaxine was a s 8 case, so the product was venlafaxine itself, and the nature of the non-infringing alternative was not at issue: see here.)

These cases are all entirely consistent with “but for” causation; after all, if a generic had decided to enter the market for a drug, it is not very surprising that next best thing to using an infringing process would be to use an unpatented process to make the same drug. At the same time, the cases were also consistent on the facts with the view that the non-infringing alternative must be a market substitute for the infringing process. That is how Dow characterized those cases—as only allowing the use of the differential profits based on “true market substitutes for the patented invention” (Dow Reply FM050 [6]—relying on Collette v Lasnier (1886) 13 SCR 563, 576, Rivett 2009 FC 317 [56], Lovastatin Damages 2015 FCA 171 [73], Perindopril Accounting #1 2017 FCA 23 [41]–[42], Cefaclor Damages 2018 FCA 217 [54], and Perindopril Accounting #2 2020 FCA 60 [46]–[48], in addition to the FCA decision in Nova v Dow: see Dow Factum FM020 [90]–[93] and Dow Reply FM050 [8]–[11].

A key passage relied on by Dow was the statement in Lovastatin Damages [73] that when considering the effect of legitimate competition from a non-infringing alternative, the court must consider whether the alleged non-infringing alternative is “a true substitute and thus a real alternative.” Elaborating on this, in Cefaclor Damages 2018 FCA 217 the FCA stated (emphasis added by Dow FM050 [10]):

[54] Armed with these principles, I now turn to the very first issue to be considered by any court when determining whether an NIA defence is available: “Is the alleged NIA a true alternative to the inventions at issue?” In non-pharmaceutical cases, this is a very important question that usually turns on whether the product at issue would be considered a true substitute by the consumer. However, in pharmaceutical cases where generic products are bioequivalents of the original product, this aspect is not an issue.

I don’t see this as signaling a departure from “but for” causation. These were both damages cases. In damages cases the patentee is claiming for lost sales. The infringer is arguing that it would have made those sales even without infringing. The only way the infringer can take sales from the patentee without infringing is to sell a market substitute. As the FCA put it in Lovastatin Damages, “in cases where, in the ‘but for’ world, the infringer could and would have made and sold a non-infringing alternative, these sales may well reduce the patent owner’s sales” [49]. So, if Dow in this case had sued for damages on the basis of sales of food wrap plastic lost to Nova’s infringing competition, then, on a “but for” approach, it would indeed be entirely irrelevant whether Nova could have made $300m in the crate and pail market; the question is whether Nova would have been able to take food wrap sales from Dow without infringing. It follows that on a “but for” approach, in a damages case, the infringer must indeed show that its product would have been a market substitute. But this does not imply that a non-infringing substitute must be a market substitute in an accounting case; the market substitute requirement is not a basic principle, but is only a reflection of how “but for” causation plays out in the damages context.

I don’t read the accounting cases as establishing a market substitute rule either. In Perindopril Accounting #1 and #2, the FCA cited the passage from Lovastatin Damages [73], as the approach to follow “in assessing the impact of legitimate competition from a defendant marketing a non-infringing alternative product” [#2 48]. In Perindopril, the infringer argued that it would have made and sold a non-infringing market substitute, just as in Lovastatin, so it is reasonable to apply the same approach. This doesn’t imply that the infringer is restricted as a matter of principle to arguing that it would have sold a market substitute.

The remaining cases cited by Dow are, in my view, even weaker support for a “market substitute” rule. The SCC in Collette v Lasnier (1886) 13 SCR 563, 576 said it was an error for the differential profit to be assessed by using an outdated alternative, which was no longer used, instead of with “the latest precedent and best known mode.” I find it impossible to read this as mandating a comparison only with a market substitute. The “but for” alternative will often be a market substitute, and the remark was really aimed at the error of using an outdated comparator. Rivett was an accounting case, but I read [56] as addressed at a different point entirely, holding (correctly, in my view) that the non-infringing alternative cannot be that the infringer would have taken a licence from the patentee, or an accounting would reduce to a reasonable royalty.

In my view, the best reading of the cases as a whole is that they endorse “but for” causation. With that said, it is true that on the facts they are consistent with the idea that the non-infringing alternative has to be a market substitute for the infringing product, and there are a few comments that can be read as supporting that position.

It is interesting to note how Stratas JA treated these decisions in his decision for the majority in Nova v Dow. He rejected “but for” causation in the context of an accounting, but he does not seem to have accepted the market substitute test, nor does he appear to have read the prior cases in the same way as Dow. Stratas JA never referred to the “true alternative” language from Lovastatin Damages and he never cited [73]; on the contrary, he cited Lovastatin Damages for the proposition that “‘but for’, hypothetical reasoning applies when courts award compensatory damages for patent infringement” [45] and he accordingly affirmed the use of “but for” reasoning in the damages context [45], [67], [76]–[76]. He didn’t cite Cefaclor Damages or Perindopril Accounting #2 at all. He cited Perindopril Accounting #1 2017 FCA 23 extensively, as illustrating the correct approach on the facts [58]–[62], and for stating the requirement of a causal link, but he did not cite it for the specific approach, and indeed he found it necessary to doubt the decision’s apparent endorsement of “but for” causation, as being obiter: [79]. He did rely extensively on Rivett, though it is of course not binding authority.

This was the context for the SCC decision in Nova v Dow, in which Rowe J affirmed Stratas JA’s decision, though without specifically approving his reasoning. Stratas JA’s decision is not entirely moot, as the SCC did not address the question of whether the approach it took to an accounting in Nova v Dow also extends to the damages context. Dow’s argument implies that it does, while Stratas JA’s decision implies it does not.

Rowe J set out a three-step test to “conceptualize” an accounting of profits [15]:

Step 1: Calculate the actual profits earned by selling the infringing product — i.e., revenue minus (full or differential) costs.

Step 2: Determine whether there is a non-infringing option that can help isolate the profits causally attributable to the invention from the portion of the infringer’s profits not causally attributable to the invention — i.e., differential profits. It is at this step that judges should apply the principles of causation. . . .

Step 3: If there is a non-infringing option, subtract the profits the infringer could have made had it used the non-infringing option from its actual profits, to determine the amount to be disgorged.

Rowe J noted that the principal issue on appeal was Step 2 [16]. There are two key issues at Step 2: (1) the nature of the causation requirement; and (2) the nature of the non-infringing option, or “NIO.” The issues are related: if we know the causation concept, we can decide what constitutes an appropriate NIO; and if it is clear what constitutes an appropriate NIO, then we can infer the causation concept. Unfortunately, the decision provides very little guidance on either of these key points. Rowe J held that the causation requirement and the nature of the NIO are both questions of facts, with no strict rules: [3], [15], [67], [70]. Subsequent posts, and my case comment, will attempt to unpack these concepts.

Section 8 Damages with Multiple Claimants

Apotex Inc v Sanofi-Aventis 2014 FCA 68 Sharlow JA, Pelletier JA concurring, Mainville JA dissenting, affd 2015 SCC 20 varg 2012 FC 553 Snider J

1,187,087 & others / ramipril / ALTACE / NOC

Dr Reddy’s Laboratories Ltd v Janssen Inc 2022 FC 1672 Southcott J

Apotex Inc v Janssen Inc 2022 FC 1473 Southcott J

2,661,422 / abiraterone acetate & prednisone / ZYTIGA / NOC

If a generic that is kept off the market as a result of a statutory stay under s 7(1)(d) of the PM(NOC) Regs prevails in the subsequent proceedings, it is entitled under s 8 to damages for its loss during period that it was ‘wrongly’ (in hindsight) kept off the market. The idea is broadly that the statutory stay is the equivalent of an automatic interlocutory injunction, and s 8 damages are equivalent to the undertaking in damages that is normally required of a party seeking an interlocutory injunction.

If more than one generic has been kept off the market for the same drug during the same period, there is a difficult question of how to construct the “but for” world for the purpose of assessing damages, and specifically, whether there should be a single “but for” world applicable to all the generics. This is important, because in reality, there is essentially a fixed market for generic products, and the more generics in the market in the “but for” world, the smaller the market share for the s 8 claimant. If there is in principle only one “but for” applicable to all the potential s 8 claimants, then the patentee’s total liability is capped by the size of the generic product market. But if the “but for” worlds for the different claimants are completely independent, then Generic A might get 100% of the generic market in its s 8 action on the view that Generic B would have been kept out of the market by the stay; and at the same time, Generic B might also get 100% of the market in its s 8 action on the view that Generic A would have been kept out of the market by the stay.

This question was addressed by the FCA in Ramipril 2014 FCA 68 varg 2012 FC 553 affd 2015 SCC 20, which held that the second option, with independent “but for” worlds, is the law. We now have a couple of procedural motions involving Janssen as the patentee, and Apotex, Riva and Teva as the s 8 claimants, that should flesh out the Ramipril approach. This case will be very interesting to follow; depending on how the facts play out, it might well give rise to the counter-intuitive result that Janssen will be liable to the various generics for an amount that represents two or three times the total generic market. That would make for an interesting test of the holding in Ramipril.

I blogged Ramipril on this issue when it came out, but looking back, I’m not sure my post accurately described the holding, and in any event, it was confusing even to me, so I’ll try again before turning to the cases at hand.

In Ramipril, Sanofi was the patentee and the s 8 claimant was Apotex. Two other generics, Teva and Riva had filed ANDS and served NOAs on Sanofi in respect of the same drug during the relevant period, and were also seeking s 8 damages in separate actions [FC 8], [FC 154], [FC 163]. Sanofi argued that there should be one “but for” world applicable to all the generics, arguing that if the but for worlds are all separate, then the patentee might be liable to each generic in separate s 8 actions for the entire generic market. Snider J, at first instance, rejected the “one world” approach. She held that the NOC Regs are assumed to exist and operate against all generics in the “but for” world in exactly the same way as in the real world. For Snider J, the only twist in the “but for” world is that we should assume that Sanofi did not bring an application for prohibition against the s 8 claimant, namely Apotex, in response to its NOA: [FC 6], [FC 157]. On this point she relied on Norfloxacin 2011 FCA 329 [75], a s 8 case where the FCA stated that “the Federal Court had to assess [the generic’s s 8] damages on the basis of a hypothetical question: what would have happened had [the patentee] not brought an application for prohibition?” That is a minimally counter-factual scenario that is entirely consistent with the Regs existing and operating as normal. Sanofi is assumed to resist entry by the other generics, much as it did in the real world—subject to whatever consequences flow from the fact that it did not oppose Apotex’s entry.

On appeal, Sharlow JA for the majority, largely agreed with Snider J’s approach [159]. The key question on appeal was whether Snider J was right to reject Sanofi’s “one world” approach. Sharlow JA unequivocally affirmed Snider J on this point, holding that the NOC Regs should be assumed to be effective against the other generics [159], [161], [171]. Consequently, “the behaviour of competing generic drug manufacturers must be determined on the basis that the NOC Regulations exist, and each generic drug manufacturer will conduct itself accordingly” [162]. In general, this means that so far as the other generics are concerned, the NOC process will proceed in the “but for” world in the same way as it proceeded in the actual world: “it seems to me that Riva and Teva would have behaved in the hypothetical world just as they did in the real world, which was to seek summary dismissal as soon as they considered they had a fair chance of success” [187]. This assumption holds unless there is some specific legal basis in the “but for” world that would support an earlier entry date [182].

The difficulty with Sharlow JA’s approach (and Snider J’s approach) is that is may lead to the patentee being liable for damages that are greater than the size of the market. Mainville JA, dissenting, gave the example where there are two generics that each served an NOA and were each subject to the stay in the real world. In the subsequent s 8 action, each would be entitled to 100% of the generic market (less any authorized generic), during the two years of the stay [108]. Sharlow JA did not disagree with this assessment: “that inconsistency is inherent in the scheme of section 8 of the NOC Regulations. If that is a problem that requires a remedy, the remedy lies with Parliament or the Governor-in-Council, not this Court” [164]. (Snider J suggested that in an egregious case, an adjustment might be made pursuant to s 8(5): [FC 138].)

One point puzzles me about Ramipril. In the real world, Apotex, Riva and Teva all served NOAs on Sanofi, and Sanofi brought prohibition applications against all of them [FC 34]. The application against Apotex was heard first and dismissed. The applications against Riva and Teva were then also dismissed as being an abuse of process, as was established law under the Regs at the time, so that Riva and Teva piggy-backed off Apotex’s success [181]. In the s 8 action, Apotex argued that if no application for prohibition was brought against Apotex in the “but for” world, that non-existent application could not have been dismissed, and Riva and Teva could not have piggy-backed off that dismissal. That means that in the “but for” world, Riva and Teva would have entered later than in the actual world. Snider J accepted this argument in principle, though she held that it didn’t make any difference on the facts (for reasons I don’t entirely understand) [159]. Sharlow JA reversed on this point, saying “it appears to me that in the hypothetical world as well as in the real world, the prohibition applications against Apotex would have been dismissed just as they were in the real world” [186].

Here is the puzzle: if the prohibition application was dismissed in the hypothetical world, that means it must have been brought in the hypothetical world. But if the prohibition application was brought in the hypothetical world, what exactly is the difference between the hypothetical world and the real world? What is the counter-factual premise for constructing the “but for” world? So far as I can see, Sharlow JA simply doesn’t address this question at all; her decision is focused on the effect of the Regs against the other generics and whether Snider J was right to reject the “one world” approach. But given that Sharlow JA expressly held that the prohibition applications against Apotex, the s 8 claimant, were dismissed in the hypothetical world, they must first have been brought in the hypothetical world. The idea seems to be that in the “but for” world, Apotex would not have been subject to the statutory stay, but everything else is the same. That doesn’t really make sense to me – if the stay doesn’t operate, Apotex has its NOC, and the prohibition application is moot. Further, I don’t really see how to reconcile this with the Norfloxacin FCA decision that Snider J relied on: it was not mentioned at all by either Sharlow JA or Mainville JA in his dissent. In any event, that’s the puzzle; I don’t have a solution.

Ramipril was decided under the old NOC Regs, but the same puzzle arises, with a twist. The proceeding is now an action. If Generic A serves the NOA and the patentee responds by bringing an action, and Generic A is successful and the patent is declared invalid, then all other generics on patent hold can enter the market, even if they had not served an NOA: s 7(5)(a). If, in a s 8 action under the new Regs, we follow the logic of Snider J and Norfloxacin, and assume that the patentee never served the NOA, and the action never happened, then all the other generics will be assumed to be kept off the market until the expiry of the patent, even though we know it’s invalid. On the other hand, if we follow the logic of Sharlow JA, and assume that the patent was declared invalid, then it seems to me that we have to assume that all the other generics would enter the market immediately (subject to normal lead times). Presumably this will all be sorted out when the cases at hand go to trial.

With that review of Ramipril in mind, we can turn to the cases at hand. Both arise out of the same facts. Apotex, Dr. Reddy’s and Pharmascience all sought to launch a generic version of Janssen’s abiraterone acetate product. The parties agreed to have the subsequent NOC action (under the new Regs) heard at a common trial [6], with the result that the 422 patent was held to be invalid: Janssen v Apotex 2021 FC 7 affd 2022 FCA 184 (blogged here and here). The three generics then brought separate actions for s 8 damages.

In Dr Reddy’s v Janssen 2022 FC 1672, Dr Reddy’s brought a motion to determine a question of law prior to trial, pursuant to Rule 220(1)(a). Relying on Ramipril, Dr Reddy’s argued that, since, in the real world, Janssen brought an action in response to Apotex’s NOA, and Apotex was therefore kept out of the market by a statutory stay, Janssen cannot turn around and say that Apotex would also have entered the market in the “but for” world when Janssen is defending against Dr Reddy’s s 8 claim [21], [23]. Janssen acknowledged that “there is a legal presumption in a section 8 proceeding that a patentee would have taken the same steps in the but-for world that it did in the real world, it is open to the patentee (or any party) to lead evidence that events in the but-for world would have unfolded differently than they did in the real world” [25]. My reading of Ramipril is that Janssen is right, though it will be interesting to see exactly what kinds of arguments along these lines might be successful. With the “one world” rule shot down in Ramipril, patentees will now focus on trying to establish specific counter-factuals to show that particular generics would have entered, despite the NOC hurdles.

In the result, Southcott J declined to answer the question, on the view that this is an important question, likely to have a considerable impact on damages [47], and “the Court could benefit from the full factual context provided by a trial in considering the parties’ legal arguments surrounding a question that may have an impact upon not only this litigation but also related and future litigation” [50]. That strikes me as a sound position. As Sharlow JA essentially acknowledged, the Ramipril holding sets up some potentially paradoxical scenarios. It’s not clear at this point what kind of arguments will be ruled out as a matter of law, and which will just fail (or succeed) on the facts, so it does seem better to address the issue in a fully developed factual context.

In the other decision at hand, Apotex v Janssen 2022 FC 1473, Janssen sought to have four common issues heard together under Rule 105(a). The issues all relate to the nature of the “but for” world, namely the size of the market, whether the plaintiff generics would enter the market, whether the non-party generics would enter the market, and how Janssen would market its products [8]. [26]. The last point goes to the idea that Janssen would reduce its marketing of ZYTIGA and promote an alternative product, which is presumably still patented [26].

An initial issue on the motion was the test for consolidation under Rule 105(a). The purpose of Rule 105(a) is “to avoid a multiplicity of proceedings and promote an expeditious and inexpensive determination of those proceedings” [12]. One issue is of course whether the issues actually are common. It was also uncontroversial that prejudice to responding parties weighs against granting the motion. The contentious issue was whether the moving party, Janssen, also has to show it would be prejudiced. That is, is it enough that the trial of common issues would promote an expeditious and inexpensive determination and that no responding party would be unduly prejudiced, or is it additionally necessary to show that the moving party would be prejudiced without a trial of common issues?

In Ramipril 2009 FC 1285 [11] Snider J held that the moving party must show it would be prejudiced, relying on Apotex v Wellcome (1993) 51 CPR(3d) 480 (FCTD); Mon-Oil (1989) 26 CPR(3d) 379 (FCTD) and Fruit of the Loom (1984), 79 CPR(2d) 274. There was some dispute over whether the FCA decision in Bayer 2020 FCA 86 had indirectly signalled a departure from these principles: Southcott J held it had not, in my view correctly [22]. (That’s not to say that Bayer affirmed this view of the test, but only that it did not address the question directly enough to disturb the prior caselaw.)

So, the law is reasonably well settled. Is it sound? As a matter of principle, if a consolidation can secure a just result at less cost, without any prejudice to a responding party, it would seem that the interest in an expeditious and inexpensive determination of the matter is a good reason for ordering a consolidation, even if there is no specific prejudice to the moving party. The argument to the contrary is that “a genuine onus rests on an applicant seeking to interfere with a plaintiff’s right to pursue a lawful cause of action”: Fruit of the Loom 278. Against that, it might be said that in the absence of any prejudice to the responding party, the desirability of conserving the resources of the justice system should be a good answer to that concern. A consolidated trial doesn’t just affect the parties; by freeing up judicial resources it benefits litigants in entirely unrelated matters.

As a practical matter, I’m not sure it really matters that much whether the moving party has to show it will be prejudiced. If the responding parties are opposing the motion, they have some reason that can normally be characterized as prejudice. If the responding parties are prejudiced, then the ultimate question will be whether that prejudice is outweighed by the efficiencies of a common trial. So, in this case, while Soutcott J held that Janssen had to show prejudice, it was not determinative: he would have refused the motion under either test [24]. The same seems to be true of several of the other cited cases (though I didn’t review them all in detail).

To my mind, the more important question is not whether the moving party has to show prejudice, but as to what kinds of prejudice to the responding parties are cognizable. In particular, a key question is whether the litigation advantages that the responding parties might gain through separate trials–what Janssen referred to as “tactical” advantages—should be weighed against efficiency benefits to the judicial system. Southcott J held that such litigation advantages are indeed cognizable, so that their loss constitutes a prejudce to the responding party: [50]. For example, in this case, Southcott J gave significant weight to Apotex’s position that “Apotex does not wish to have non-party lawyers cross-examining adverse witnesses in Apotex’s s.8 damages case” [44]. So long as these kinds of advantages are cognizable, the responding parties will normally be able to show some kind of prejudice, and I doubt the issue of prejudice to the moving party will often play a major role.

I’m torn on the question of whether litigation advantages should be cognizable. On the one hand, our adversarial justice system depends on the parties being able to run their own case. As Southcott J put it,

[50] A party’s interest in planning and controlling its approach to litigation in which it is involved, and therefore avoiding an adverse impact upon that approach caused by others who would not normally be party to that litigation, strikes me as engaging an aspect of the administration of justice more fundamental than a mere tactical advantage.

On the other hand, our system does ultimately aim to find the truth, and the idea that there might be different truths about the same world in different trials is problematic. Under the old NOC proceedings, the courts regularly expressed embarrassment about the prospect of different result re validity or infringement in the NOC proceeding and the subsequent action. This never struck me as such a problem, given the summary nature of the old NOC proceedings; there’s no reason for embarrassment if a prima facie view of the merits in an interlocutory injunction application turns out differently from the result in the subsequent trial. It seems to me that the embarrassment would be significantly greater in a case in which separate actions could mean that four different trials would find that four different generics would have had a majority of the market in the same but for world. But as just noted, in Ramipril the FCA effectively embraced that inconsistency, as being inherent in the s 8. Under the “one world” approach, in which the same but-for world would be used for all the s 8 claimants, a trial of common issues would have considerable appeal in order to avoid inconsistent results. But given that the FCA held, in effect, that there is no common truth, as a matter of law, there is nothing left to the argument that we need a common trial to discover a common truth. With nothing to balance against it, it is really just a matter of weighing any efficiencies against the desirability of allowing each party to control its own case.

On the facts, Southcott J gave significant weight to a variety of litigation concerns raised by the generics [44]–[45]; he also noted that because witnesses would have to testify to different but-for worlds in the common trial, the efficiencies in adjudication would not be as great as if the there was a truly common factual issue [51]; and, moreover, one of the generics, namely Pharmascience, would suffer a more uncontroversial kind of prejudice because its trial date was 17 months later than the other two: [55]. Consequently, he refused to order the consolidation.

Nova v Dow: The Intuition

Nova Chemicals Corp v Dow Chemical Co 2022 SCC 43 Rowe J: Wagner CJ, Moldaver, Karakatsanis, Brown, Martin, Kasirer and Jamal JJ concurring; Côté J dissenting affg Nova Chemicals Corporation v Dow Chemical Company 2020 FCA 141 Stratas JA: Near, Woods JJA affg Dow Chemical Co v Nova Chemicals Corp 2017 FC 350, 2017 FC 637 Fothergill J

2,160,705 / film-grade polymers / ELITE SURPASS

That has been a much longer blogging break than I’d anticipated back in December. As well as teaching a new course, I have been working on a case comment on Nova v Dow 2022 SCC 43, which has turned out to be very long. I will not try to summarize the argument in one post. Instead, I’ll post summaries of consecutive sections of the comment every Friday for a few weeks.

An accounting of profits is an equitable remedy which, in patent law, requires the infringer to disgorge profits made by the infringer through its use of the patented invention; this is in contrast with damages, which looks to the loss suffered by the patentee as a result of the infringement. In Nova v Dow the Supreme Court addressed the proper method of calculating an accounting of profits in the patent context.

In this post, I will describe what I take to be the intuition underlying the decision of both Rowe J’s decision for the eight-person majority, as well as Côté J's dissent. I am confident that I understand Côté J’s position, since I share it. But there is clearly a strong contrary intuition, which attracted eight members of the SCC as well as Stratas and Near JJA in the FCA, that I have had more difficulty grasping. That makes it all the more important to “steelman” Rowe J’s argument. With respect, Rowe J’s decision is not well reasoned at a technical level, but a strong intuition poorly expressed is a strong intuition nonetheless. To simply go through and point out shortcomings in the doctrine and policy would not persuade anyone who shared that intuition. So, I will first try to identify the best form of underlying intuition. I will then argue that the intuition is wrong, even in its best form, and the doctrinal shortcomings of the decision are not merely technical attacks, but rather reflect the defects of the underlying intuition.

Even the first step is an uncertain enterprise, as Rowe J never clearly articulated his driving intuition. While I disagreed strongly with the decision of Stratas JA, it was, in my view, a much stronger decision, which identified and addressed the key points more directly: the best discussion of what I believe to be the key intuition was in Stratas JA’s decision at [77]–[78]. But we can’t assume that Rowe J necessarily agreed with Stratas JA; while he affirmed the holding, Rowe J did not expressly approve Stratas JA’s reasoning. I think that I have identified the key intuition, but I’m not sure. I’d be very interested in any comments as to whether I have managed to identify the main intuition.

The essential facts are that Dow had a patent on specialized plastic film, used for items such as food packaging. Nova made and sold a competing film that was found to be infringing. Dow was awarded an accounting of Nova’s profits from sales of infringing film, with a quantum of $644 million. The major input to the infringing plastic is ethylene, an unpatented bulk commodity. Nova had a very efficient process for making ethylene—the “Alberta Advantage”—and so could make ethylene for far less than it would have cost to buy on the open market. On the facts, if Nova had not infringed, it would have used its ethylene to make bulk “pail and crate” plastic [FC 158]. Nova argued that because of the Alberta Advantage, it would have made approximately $300 million in profit on the sale of that plastic, despite the competitive nature of that market [FCA 187]. The question was whether Nova was entitled to deduct the amount it would have made in the pail and crate market, on the view that that part of the overall profit was caused by its efficient ethylene production process, and not by the infringement.

The background lies in what was traditionally known as the problem of apportionment. In some cases, as where a patent claims the active pharmaceutical ingredient of a drug, the patented technology contributes essentially all the value, and it seems evident that the value of the invention is its price less its cost. (Even here there are some refinements regarding cost of capital and fixed costs, which we can ignore for now.) But for many products it is clear that the patented technology contributes only a part of the value. When the product is a motor oil with a patented additive, a smartphone with a patented “bounce-back” feature, an iPad with patented 3G capability, or canola with a patented herbicide resistance gene, it is clear that the entire value of the product is not due to the patented technology alone.

In such cases, some part of the profit on the product as a whole must be “apportioned” to the patented invention. But how? One appealing answer is to compare the patented product with the non-infringing alternative. If the profit on motor oil with the additive is $1, and profit on motor oil without the additive is 95 cents, it seems clear enough that the profit attributable to the patented additive is 5 cents. If the iPad sells for $700 with 3G technology and $600 without it, the value of the patented 3G technology is $100. A similar logic applies to process patents: if a product sells for a dollar and the patented process reduces the cost of manufacture from eighty-five cents to eighty cents, it seems clear that only 5 cents of the twenty-cent profit is attributable to the patented process.

These examples suggest that the value of the invention is the difference between the profit on the infringing product, and the profit on the non-infringing alternative: the oil without the patented additive, or the end-product made by the unpatented process. How can we generalize this insight? What is the underlying principle? This was the key question facing the Supreme Court in Nova v Dow.

One view, taken by Côté J, Nova, and the interveners, is that the appropriate non-infringing option reflects the application of “but for” causation. Patent infringement is generally considered to be a species of tort, and damages in tort law are assessed as the difference between the plaintiff’s actual position, after the tort, and the position the plaintiff would have been in but for the tort: Athey v Leonati [1996] 3 SCR 458 [32]. Applying “but for” causation to an accounting of profits therefore means that the infringer is to be put in the position it would have been in had it not infringed, and the amount to be disgorged is the difference between the infringer’s actual position and the position the infringer would have been in but for the infringement. (For convenience, we may refer to this as a “but for” accounting.)

On this view, any similarities between the infringing acts and the “but for” world in which the infringer did not infringe are irrelevant in principle; what is fundamental is what the infringer would in fact have done but for the infringement. This is not to say that it is pure coincidence that the alternatives in the above examples are very similar to the infringing product. It is not very surprising that a motor oil manufacturer might choose to sell oil without an additive if it could not sell it with that additive; or that a canola farmer would plant conventional canola if they could not plant herbicide resistant canola; it is not surprising if a generic pharmaceutical company would use an unpatented process instead of a patented process to satisfy the market for a drug. So, if there is a very close market substitute, it is not uncommon that what the infringer would often have done had it not infringed would be to make and sell that substitute. But this is merely an empirical regularity, and not a fundamental principle; if, on the facts, the infringer’s acts in the “but for” world would have been very different from the infringing acts in the actual world, this is no cause for concern, and the analysis proceeds in exactly the same way. If the infringer would have withdrawn entirely from the market, then the entire profits are caused by the infringement (subject to refinements regarding fixed costs and capital costs); if the infringer would have made a substitute profitable product, that profit is deducted from the actual profit; if the infringer would have made an unrelated, but profitable product, that profit is deducted from the actual profit. On the “but for” causation, if Nova could establish that it would in fact have made $300m in the pail and crate market as a result of the Alberta Advantage, it should be allowed to deduct that amount from its actual profits, to arrive at the quantum to be disgorged.

The other view, taken by Rowe J for the majority, is that there is something special about certain non-infringing options such that a comparison with the non-infringing option reveals the value of the invention. While the argument is difficult to summarize, I believe the underlying intuition is that the infringer must disgorge profits reflecting the value of invention, and the value of the invention cannot depend on what happens in entirely unrelated markets. That requires rejecting “but for” causation.

To see why, suppose a firm called Mova sold infringing plastic in competition with Dow. In the actual (hypothetical) world, it made $900m in revenue from infringing sales, with $300m in costs, for a profit of $600m. It could have made a non-infringing plastic that was a perfect market substitute, so that it could have made exactly the same $900m in sales without infringing, but the non-infringing plastic costs $100m more to make because of higher energy costs. With the same revenue and $400m in costs, the profit from the non-infringing substitute would have been only $500m. That is less than $600m, but still a healthy profit. The value of the patented invention is that it allows a high-quality plastic to be made at a lower cost. A comparison between the profit on the infringing product and the profit on the non-infringing substitute reveals that value, which, in this example, is $100m. As we will see below, that is the amount to be disgorged on Rowe J’s approach.

Now consider how the “but for” causation plays out. Suppose that Mova could also have made non-infringing commodity grade plastic in the same plant, and if it had done so it would have made $450m in profit. While the plant was capable of making either non-infringing market substitute or commodity grade plastic, the former would have been more profitable than the commodity grade plastic, and, it is established that had Mova not infringed, it would have made and sold the market substitute plastic. The result, on a “but for” accounting, is that Mova would be required to disgorge the same amount as under Rowe J’s approach, namely $100m. To this point, the “but for” causation and Rowe J’s approach give the same result.

Now twist the hypothetical slightly and suppose that during the period that Mova was infringing, Crocs came back into fashion, and the market for commodity grade plastic tightened. With Crocs having taken the world by storm, Mova would have made $550m in the market for commodity grade plastic. That is still less than the profit on the infringing plastic, but more than the profit on commodity grade plastic, so in the “but for” world, Mova would have sold commodity plastic. On a “but for” accounting, the amount to be disgorged would be only $50m.

The result of applying “but for” causation to this hypothetical scenario seems very counter-intuitive. What actually happened is that Mova made and sold infringing food grade plastic. The value of the invention is $100m in saved costs. But because of some quirky fashion trend, probably started by an Instagram influencer, which caused fluctuations in a totally different market, Mova only has to disgorge $50m instead of $100m. But how can the vagaries of the fashion market possibly affect the value of an invention used for making food wrap? Surely the value of the invention is the same, regardless of whether Crocs are in fashion. To take a more extreme example, what if Mova could show that it could have made a profit in a tight market for Chinese tea? Should that profit be deducted? The principle is no different than the pail and crate example, but surely the price of tea in China has nothing to do with the profit attributable to Mova’s infringement of a patent on plastic food wrap. As we will see, Rowe J’s approach avoids these paradoxes. On his approach, the profit to be disgorged is $100m. This is the difference between the actual profit of $600m and the profit on the non-infringing market substitute which reveals the true value of the invention. Fashion trends in footware and the price of tea in China, play no role.

That, I believe, is the basic intuition behind Rowe J’s approach.

It is one thing to have an attractive intuition, it is another to turn an intuition into a generalizable principle. Subsequent posts, will show that Rowe J fails to provide any coherent methodology for identifying the non-infringing option that reveals the true value of the invention. This is because an invention does not have any inherent true value; and it turns out that the apparent paradoxes I have just outlined are not paradoxes after all.

Blogging Break Continues

My blogging break is turning out to be longer than anticipated. I have been unexpectedly tasked with teaching a course that I have never taught before, in an area – professional conduct – that is unrelated to any of my research interests, and I have had to focus on getting up to speed. I am also working on an in-depth analysis of Nova v Dow 2022 SCC 43. Between these two projects I have not had time to return to my regular blogging. I expect to return to my usual blogging schedule in a couple of weeks, starting with the cases that I have missed in the meantime.