A different perspective on reverse settlements

Before I begin my initial post, I want to thank Holly for inviting me to post on this blog.

I want to take up reverse settlements in litigation over pharmaceutical patents.  Circuits are divided on how to treat these settlements under antitrust law (Elhauge & Krueger, Texas L. Rev., 91:283, 285, 2012).  The Supreme Court has decided to take this the topic up this term; it will hear oral arguments in Federal Trade Commission v. Watson Pharmaceuticals on March 25, 2013.  However, this is a topic about which I believe the legal literature has lagged substantially behind the health economics literature.  As a result, I think the conventional (legal) views of such settlements get the economics of pharmaceutical patents and innovation wrong.  (That does not mean they are getting the law wrong. Although the law in this area is highly unsettled, the goal of the law may not coincide with economic prudence.  I am commenting primarily about economic prudence.)

Before I address a topic concerning drug companies, let me make a disclosure.  I do not receive any funding or support from drug companies, either generally, for research on reverse settlements, or for this post .  You can find more details about my past and present funding in the footnotes.[1]

To set the stage for my arguments, let me describe the conventional view of reverse settlements among most legal scholars – and the DOJ and FTC.  A pioneer drug company gets a patent from the patent and trademark office (PTO).  That does not ensure that the patent is valid.  But as a result of the patent the pioneer has a monopoly and consumers pay a high price for the drug, with a loss to consumer and total welfare.  Any given generic drug company has no incentive to challenge the patent.  Challenges are costly.  Moreover, once the patent is invalidated, other generic companies will enter, driving economic profits to zero.  The Hatch-Waxman Act addressed this problem by giving the first generic company to successfully challenge a patent 180 days of exclusivity before other generics can enter the market.  This would allow the generic to enjoy 180 days of duopoly profits with the pioneer.  In other words, Hatch-Waxman took some potential consumer surplus (from after patent invalidation) and offered it to the generic company to get it to help eliminate invalid patents.

The problem with the Hatch-Waxman solution is that the first generic and the pioneer can collude to hurt the consumer.  Note that the pioneer’s monopoly profits (if the patent were left in place) are greater than the duopoly profits that the pioneer and first generic each get during the period of exclusivity (if the patent were invalidated).  The pioneer can offer the first generic a part of this wedge to settle its challenge in the pioneer’s favor.  This side payment – called a reverse settlement – increases the period of time during which the pioneer retains its profits.  In theory, these settlements undo the Hatch-Waxman solution.  Most scholars – including my friends Scott Hemphill (on leave from Columbia) and Einer Elhauge – argue that this hurts consumers.[2]

I think that there are many missteps in this argument.  In this post I highlight one.  I will try to highlight others in future posts.

First, the pioneer drug company’s monopoly may not reduce *static* welfare.  The conventional argument for a patent is that it encourages innovation (dynamic inefficiency).  The conventional argument against a patent is that it causes high (monopoly) prices that price some consumer out of the market (static deadweight loss or inefficiency).  Patents are given a finite duration to balance the dynamic efficiency (a benefit) and static inefficiency (a loss).  Recent empirical research in health economics suggest, however, that drug patents may not price any consumers out of the market.  A key prediction of the conventional model of patents is that, when a drug goes off patent, generic companies should enter, prices should fall, and quantity of drugs sold should rise.  However, this prediction is rejected by the data.  While drug prices fall after patent expiration, drug sales do not rise!  See the figure 1 below.  Since quantity does not rise, that strongly suggests no consumers are prevented from buying drugs due to drug patents.  There does not appear to be static welfare loss!

Figure 1: Drug quantity does not rise after patent expiration.

Source: Figure 5 from Lakdawalla, Philipson, and Wang, “Intellectual Property and Marketing,” Journal of Law & Economics, forthcoming.

A natural question is why drug quantity does not rise after patent expiration.  (We don’t need to answer this to cast doubt on static welfare loss, but it does illuminate the gap between legal thinking about reverse settlements and health economic thinking on such settlements.)  There are two possible explanations.  The simplest explanation is that pioneer drug companies advertise their drugs and this advertisement raises consumption.  So even at monopoly prices, patients consume the same amount of drugs they would at competitive prices.  When patents expire, pioneer drug companies stop advertising because they do not appropriate the benefits of the ads; ads increase both pioneer and generic sales.  See the figure 2 below.  The decline in ads offset the increased consumption due to the price decline after patent expiration.[3]

Figure 2: Advertising falls after patent expiration.

Source: Figure 8 from Lakdawalla, Philipson, and Wang, “Intellectual Property and Marketing,” Journal of Law & Economics, forthcoming

A second explanation, which I like better, is that consumers do not face the monopoly price of patented.  Specifically, most consumers do not pay for drugs out of pocket.  Instead they have health insurance that covers the cost of drugs.  Health insurance does not charge consumers the full price of drugs; rather it charges a copay, e.g., $10.  This copay is close to re competitive price of drugs.  As a result, health insurance permits consumer to purchase the same amount of a patented drugs as they would if it were not patented (and not insured).[4]  The table below nicely illustrates this point.  It shows that for drugs largely covered by insurance, patent expiration is associated with no increase in drug quantity.  For drugs less well covered by insurance, there is, indeed, an increase in drug quantity. Since we now have an individual mandate that requires nearly everyone to buy insurance (or provides them government insurance), we don’t have to worry about the effect of patents on less-well insured drugs: nearly all drugs will be insured going forward.

Table 1: Patent expiration raises quantity only for drugs that are not well-insured

Source: Lakdawalla and Sood (2012).  Bold numbers indicate significantly different from zero.

Interestingly, while health insurance lowers the price of patented drugs that consumers face, it does not (have to) lower the price that pioneer drug companies receive.  The insurance company can pay the pioneer a monopoly price and charge the consumer a near-competitive price.  The reason is that insurance is a like a two-part pricing scheme where consumers are charge a large entry price (the premium) for the right to buy individual units of drugs at a low marginal price (the copay) (Lakdawalla and Sood, J. Public Economics, 2006).  In this manner, health insurance eliminates the tradeoff between dynamic efficiency and static efficiency that plague patents in non-pharmaceutical markets.

A skeptical reader might ask: if consumers consume a competitive quantity of patented drugs, but the insurer pays monopoly prices for those drugs, won’t that increase insurance premiums?  And won’t that premium increase reduce insurance consumption?  In other words, isn’t the static deadweight loss of monopoly drug prices just shifted from the drug market to the health insurance market.  There is certainly an increase in the price of premiums due to drug patents.  However, there are three reasons to think this does not result in static inefficiency.  First, employer-sponsored health insurance is tax subsidized.  This subsidy reduces the price that consumers face for insurance and thus the extent to which drug patents reduce consumption of insurance.  Second, under the Affordable Care Act, nearly all individuals are required to buy insurance.  This mandate in effect eliminates any change in quantity of insurance (at least at the extensive margin) due to changes in insurance premiums.  Thus, it also mitigates the deadweight loss from drug patents.[5]

In my next post, I will address the question of whether drug patents, even if they do not reduce total welfare, reduce consumer welfare in particular.  (Antitrust law seems more concerned with consumer surplus than total welfare.)


[1] The one time I received money from a drug company was Pfizer in 2006; specifically I received a $35,000 grant to study a statistical question – how to estimate heterogeneity in treatment effects using only data from parallel arm trial.  The grant did not pay for my research time; it only covered the cost of a programmer.  Because I work at a law school, I am in a hard-money environment so I do not depend on research funding for my salary.  I finished the work for the grant within a year.  The work – with Bart Hamilton of Washington University – was not published.  Currently I receive funding mainly internally from the University of Chicago, but none for this project.  I thank the Microsoft Fund and the Samuel J. Kersten Faculty Fund at the University of Chicago for financial support.  I am working on a large health insurance experiment in India; but that project is primarily funded by the Department for International Development in the UK.

[2] Indeed, Einer, in his 2012 Texas Law Review article on reverse settlements notes that this problem – collusion between patent holder and patent infringer – afflicts all patent cases (see fn. 3).  Thus the arguments made against reverse settlements are arguments against all patent settlements.  I would go further and say that, in theory, settlement collusion is an antitrust problem in any litigation that affects market structure.  But that is a digression.

[3] A skeptic might argue that different people consumer drugs due to advertising and due to low prices.  Perhaps, but we have no evidence of this.  It is a great paper topic for a young health economist.

[4] Of course generic drugs are also covered by insurance, and sometimes at a lower copay.  But the price elasticity is sufficiently low that we do not observe a big jump in quantity upon patent expiration.

[5] A super-skeptical reader might note that the nature of the tax subsidy for health insurance implies that the static inefficiency from drug patents is felt not in the drug or insurance market, but through taxes.  The tax subsidy is proportional to the price of insurance.  Moreover, income taxes introduce distortions in labor markets proportional to the tax subsidies they must finance.  However, the size of these distortions may not be very large.  Drugs are only a fraction – roughly 10-15% – of all health spending; spending on patented drugs is only a portion of that.  Thus the premium increase due to patented drug prices is not large relative to the overall price of health insurance.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.