Fixed Costs and Sunk Costs

 Nova Chemicals Corporation v Dow Chemicals Company 2020 FCA 141 Stratas JA: Near JA / Woods JJA dissenting, aff’g 2017 FC 350, 2017 FC 637 Fothergill J

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

Yesterday’s post discussed Stratas JA’s holding that in an accounting of profits, an infringer may always deduct some portion of its fixed costs [162]. That holding was based on Stratas JA’s rejection of “but for” causation, which does not make allowance for fixed costs unless the infringer had an outside opportunity. I am of the view that “but for” causation, and the differential profit approach from Schmeiser 2004 SCC 34 [102], which follows from it, is sound in principle. Nonetheless, I acknowledge that a real inequity may arise if a deduction for fixed costs is prohibited in cases in which the infringer did not have any outside opportunity. This is not only a moral inequity; it is a serious concern in terms of innovation policy. In this post, I will argue that the ultimate reason for this injustice is not the treatment of fixed costs, but rather the treatment of sunk costs. Consequently, the solution must look beyond an accounting to broader remedial principles, especially principles relating to injunctive relief.

This post is based on the following remark:

[160] Denying the deduction of fixed costs generates a distorted picture of the infringer’s profits. It may be the case that an infringer has minimal variable costs but very high overhead costs such that the product is not, in fact, profitable. The incremental approach. . . could force that infringer to disgorge “profits” from an unprofitable product.

This statement is somewhat obscure and I treated it cursorily in yesterday’s post, noting that there is nothing wrong to requiring the infringer to disgorge profits from an unprofitable product if it would have been even less profitable but for the infringement.

But the phrase alluding to the “distorted picture of the infringer’s profits” suggests a different point. I’m not sure whether it reflects what Stratas JA had in mind, so I didn’t want to treat it at length yesterday, but it’s interesting enough that I decided to devote a post to it. The basic idea is that sunk costs are real costs that must be incurred to make an infringing product, and if no allowance is made for sunk costs the infringer may be required to disgorge more than its actual profits.

Consider the following example. Suppose Infringer A has a very sophisticated plant, which cost $100m, allowing it to make products with low raw material costs. Infringer B has a low tech plant, which cost only $1m; it can make any product that A can make, but with much higher raw material costs. Each plant has a 10 year life-span. For any given product, both make the same overall profit at any given price if both overhead (depreciation and capital costs) and raw material costs are taken into account. Both make one infringing product and sell into the same market for the same price, namely the price necessary for each to make an adequate overall profit. Both are sued for infringement by the same patentee. If only raw material costs can be deducted, A will be required to disgorge far more than B. This seems wrong: both compete equally well with the patentee, both sell at the same price, both have the same overall profit, and both have the same overall costs. The only difference is that A has high fixed costs and low incremental costs, while for B it’s the other way around. It seems arbitrary that this should have a substantial effect on the monetary remedy.

As discussed yesterday, under both the differential profit approach and the Dart Industries approach an allowance is made for fixed costs only if the infringer had some non-infringing opportunity. But the problem in this example arises simply because A and B have the same overall costs of making the infringing goods; this has nothing to do with whether either had an outside opportunity. If there is no outside opportunity, both “but for” causation and Dart Industries would require A to disgorge far more than B. In contrast, if we apply Stratas JA’s approach to this example, all of the fixed costs could be deducted (since the plant is solely used to make the infringing product), so both A and B would disgorge the same amount because their total costs are the same.

Perhaps it is this kind of problem that Stratas JA had in mind. If it was, I agree there is a real and important problem here, but I will suggest that Stratas JA’s approach does not solve it. The problem is driven by sunk costs, not fixed costs; consequently it arises no matter what approach we take to the deduction of fixed costs.

Sunk costs refer to the costs the infringer has incurred that are specific to implementing the infringing technology and which cannot be salvaged or redeployed for other purposes: see generally Cotter, Hovenkamp & Siebrasse, Demystifying Patent Holdup (2020) 76 Wash & Lee L Rev 1501. Fixed costs, in contrast, are costs which do not vary along with output. Sunk costs are normally fixed costs, but fixed costs are not necessarily sunk: for example, in a plant capable of making both infringing and non-infringing goods, the cost of lighting is fixed but not sunk.

To see why the problem is driven by sunk costs, suppose that both plants can only make the infringing product. This means the capital cost of each plant is a sunk cost, and the corresponding cost of capital and depreciation is also a fixed cost. The patentee becomes aware of both plants before production has even begun, and obtains separate quia timet injunctions prohibiting both A and B from making infringing products. Because the plants can’t make any other products, they are mothballed and the investors lose everything: A’s investors lose $100m, while B’s investors lose $1m. This difference in the treatment of A and B clearly does not turn on the deductibility of fixed costs in an accounting; there has been no accounting, as neither plant ever went into operation.

On the other hand, suppose that both plants are flexible and can readily turn to making non-infringing products. Both A and B had intended to make an infringing product, but were stopped by the quia timet injunction. Consequently, both produce non-infringing products instead. If the next best non-infringing product is the same for both, they will sell at the same price and will have the same overall profit once both fixed and incremental costs are considered. Presumably this will be less than they had originally intended, or they would have intended to make non-infringing goods from the outset, but they may both be profitable and in any event, they will both make, or lose, the same amount of money. Again, this has nothing to do with the deductibility of fixed costs in an accounting.

In the first example, A and B have different losses, while in the second example their losses are the same. The capital cost of the plant is a fixed cost in both cases, because it does not vary with the amount of product that is made; but in the first example the capital cost is a sunk cost, while in the second example it is not, because the plant can be redeployed. Thus it is the presence of sunk costs, not the presence of fixed costs, that determines whether A and B are differently affected. More particularly, it is injunctive relief in the presence of sunk costs that gives rise to the counter-intuitive differential treatment.

How does this relate to an accounting of profits? Suppose the patentee did not discover the infringement at the outset and by the time the trial was over both plants had reached the end of their 10-year life span and had already been shut down. As a result, no injunction was granted, but only an accounting. Under Stratas JA’s approach, the fixed costs are fully allocated to the infringement along with the incremental costs so both A and B will be required to disgorge the same amount. The quantum will be the same whether or not the plants were also capable of making non-infringing goods. On the other hand, using the differential profit approach which turns on “but for” causation, A and B will be put in the position they would have been in but for the infringement. In the first scenario, where the plants are only capable of making infringing products, this in effect means that the raw material costs can be deducted, because those costs would not have been incurred in the “but for” world, while fixed costs cannot be deducted, because they would have been incurred in any event. That means that A will be required to disgorge far more than B. This might suggest that the differential treatment of A and B turns on the deductibility of fixed costs. But another equivalent way of looking at this outcome is that A and B will both be in exactly the position they would have been in had the patentee had been granted a quia timet injunction; that, after all, is equivalent to the position in the “but for” world in which they had not infringed. On this view, the differential treatment of A and B stems from the effect of the injunction; using “but for” causation simply reflects the effect of the injunction in the context of an accounting.

More generally, if “but for” causation is used, it doesn’t matter whether the patentee obtains its injunction quia timet, or in Year 1or Year 9, or anywhere in between; the net loss to A and B, from the combined effect of the injunction and the accounting will be the same. In contrast, under Stratas JA’s approach A and B will be treated the same only in the special case where no injunction is granted; otherwise, the overall loss to each party will depend on exactly when the injunction is granted. If it is granted near the beginning of the useful life of the plant, the loss to each infringer will be very different; the later the injunction is granted, the more similarly they will be treated. It does seem wrong that A and B should be treated differently, but that problem is not solved by permitting deduction of fixed costs; they will still be treated differently, though the extent of the difference depends on exactly when the litigation occurs and how long it takes.

So, I would suggest that the fact that A and B are treated differently is indeed a problem, but it is a problem that turns on the grant of injunctive relief in the face of sunk costs and not on the treatment of fixed costs. The problem of sunk costs is not an obscure academic issue. On the contrary, it is perhaps the most pressing problem in the patent system today. If A is faced with an injunction, it will obviously be willing to pay a substantial amount to settle. More generally, a patentee, such as a patent assertion entity, can use injunctive relief to extract sunk costs from the infringer in subsequent licence negotiations. This is known as patent “holdup.” Patent holdup can be avoided if the patentee is able to discover the patent ex ante and decide with certainty whether the patent is valid and would be infringed. So, the problem of A losing all its money because it is faced with a quia timet injunction could be solved if A had discovered the patent at issue before building its plant. But in practice, discovering all relevant patents ex ante is impossible in many fields of technology, and even if the patent is found, assessing infringement and validity is always uncertain.

The problem is important because it can impede innovation by increasing risk in fields that are thick with patents, yet sunk costs investments are important for efficiency. If investors can’t be sure of finding all the relevant patents before incurring sunk costs, they will avoid incurring sunk costs. In our example, both plants could make the same product at the same overall cost. In that case, it was simply foolish for A to have invested $100m in a sophisticated plant thereby exposing itself to extra patent risk; A should have simply built the same plant as B. But the more general point is that if there is an undiscoverable risk of infringement with injunctive relief as the normal remedy, an investment with high sunk costs is riskier than an investment with low sunk costs. The uncertainty derived from the patent system adds a risk premium that increases with sunk costs. In many cases an investment with high sunk costs will be more efficient with lower overall costs. So, in a more realistic example, A’s plant can make the same product at a lower overall cost than B by economizing on raw materials. In that case, from a societal perspective the more efficient investment with high sunk costs is better, as it consumes fewer resources to produce the same goods. But the more efficient plant investment may not be optimal from the private perspective of the investor, because of the risk premium arising from the prospect of sunk costs holdup. Even if the sophisticated plant A is more efficient and can produce lower cost goods, the investors might still choose to build inefficient plant B, if the extra efficiency is more than offset by the extra risk. Consequently, the holdup problem created by the patent system may impede innovation rather than encouraging it.

Patent holdup is a major concern in many important industries. The decision of the US Supreme Court in its famous eBay decision, (2006) 547 US 388, greatly restricting the availability of injunctive relief, is essentially a response to holdup, as is the proposed amendment to the German patent act, providing for a proportionality test in awarding injunctive relief. There is a very large literature on the subject: for just two prominent examples, see Lemley & Shapiro, Patent Holdup and Royalty Stacking, 85 Tex L Rev 1991 (2007); Fed. Trade Comm’n, The Evolving IP Marketplace: Aligning Patent Notice and Remedies with Competition (2011). The recent book Biddle, Contreras, Love & Siebrasse (eds), Patent Remedies and Complex Products: Towards a Global Consensus (Cambridge University Press, 2019) is largely devoted to examining the various possible remedial responses to the problem of sunk costs and holdup; one of our primary recommendations is encouraging tailoring of injunctions, with stays and carve outs, to mitigate the holdup problem.

To return to the causation question, the premise of “but for” causation is that an accounting should put the infringer in the position it would have been in but for the infringement. In the discussion so far, we have implicitly assumed that the position the infringer is put back into is the position it was in immediately before the infringement began—which is to say, after the sunk costs have been incurred. If the infringer could instead be put back in the position it would have been in before it made its investment, then A and B would be treated the same. With some remedies that is possible, at least notionally. So, if the remedy at issue is reasonable royalty damages, as is now normally the case when dealing with a standard essential patent (SEP), the royalty is based on a hypothetical negotiation between a willing licensee and licensor. The traditional position is that the hypothetical negotiation takes place at the time of the first infringement. A number of commentators have argued that the negotiation should instead be assumed to take place at the time the sunk costs are incurred (or the time just before the standard in adopted, in the case of SEPs), precisely in order to avoid sunk costs holdup: see Patent Remedies and Complex Products at 28-30. As noted, denial of injunctive relief in US law in light of eBay also addresses the sunk costs problem. However, denial of injunctive relief is easier to justify when the patentee is a non-practising entity that would only have sought a licence. Denial of injunctive relief is less palatable as a response to sunk costs when the patentee is an operating company that seeks to exploit its market exclusivity directly.

The only way I can think of to put A in the position it would have been before it made its investment would be to require the patentee to pay A compensation as a condition of the injunction. This is actually the fourth rule proposed by Calabresi & Melamed, in their famous article, Property Rules, Liability Rules and Inalienability: One View of the Cathedral 85 Harv L Rev 1089, but it has never been widely accepted. Moreover, it is difficult to see how it would be helpful in the patent context. The ultimate problem is that it is difficult to avoid incurring sunk costs by identifying all relevant patents ex ante. Sunk costs represent real resources, and once the sunk costs are incurred they must be borne by someone. Requiring the infringer to bear the loss will minimize the problem by providing an incentive to identify the relevant patents ex ante, even if the problem is not eliminated. A theoretical alternative would be that the patentee would be required to pay compensation to the infringer as a condition of the injunction, but in return it would buy the plant. The idea is that then the costs would not be sunk, because the plant could be used for its sole purpose of making the patented goods, except now it would be making them for the patentee rather than the infringer. This suggestion is clearly not practical, as the patentee will not normally want the infringing plant, for example because it has enough capacity, or because every plant has different operating conditions that the operators are accustomed to. That means that the value of the plant to the patentee will very often be less than the sunk costs. Moreover, this example of an entire plant that can make only infringing goods is a simplification. In most cases the sunk costs are only one aspect of plant and the assets representing sunk costs cannot be separated out and sold independently. The point of this suggestion is not to say that the patentee should be required to compensate the infringer as a condition of injunctive relief; rather, the point is that once we identify the root cause of the problem, we can see how intractable it is.

In summary, the problem raised, or at least inspired, by Stratas JA’s remark is difficult and important, but it cannot be addressed simply by permitting deduction of fixed costs.