[Epistemic status: Uncertain, especially on the accuracy of the economic studies cited]
Sarah Kliff of Vox replies to my piece from last week. She writes:
Earlier this week I wrote about how a lack of drug price regulation in the United States allows pharmaceutical companies — including EpiPen’s manufacturer, Mylan — to charge exceptionally high prices for their products.
Scott Alexander at the blog Slate Star Codex argues that I’ve got it all wrong: The problem isn’t a lack of price regulation. Instead, its too much regulation, which has prevented generic competitors from entering the market and has left EpiPen’s price so high. […]
Alexander and I are both essentially pointing out two examples of how the United States has created a regulatory system that is incredibly favorable to pharmaceutical companies. We’ve set up a system that makes it incredibly easy for drug companies to score high profits and charge exceptionally high prices for their products.
One way it’s favorable is that we let drug companies pick their own prices — in this way the United States is exceptional, as the vast majority of developed companies regulate their drug prices. Another way we’ve created a favorable regulatory environment, as Alexander writes, is by allowing roadblocks to stand in the way of generic drug makers who want to enter.
More generics can help America’s drug spending problem. But they can’t solve it.
Greater use of generic drugs is widely accepted as a way to drive down drug spending. The FDA has found that drug prices decline to 55 percent of the brand-name price when two generic manufacturers begin making a product. Right now, the United States already uses a lot of generic drugs. In fact, about 90 percent of drugs prescribed in the country right now are generic. Brand name drugs are the reason that America has higher per-capita drug spending than other countries. Brand-name drugs make up just 10 percent of prescriptions filled in the United States, but account for 72 percent of drug spending.
Drugs that are under patent are the true source of high American drug costs. EpiPen is, in a way, a bit of a red herring […]
Harvoni, a [patented] pill that cures Hepatitis C, costs $32,114 here — and $22,554 in the United Kingdom. Less red tape around generic drug competition wouldn’t really change that fact. As long as we’re going to have patented drugs, letting drug manufacturers set their own prices will remain a key driver of America’s higher drug spending.
(the above is some excerpts stitched together to provide a taste; you should really read the whole thing)
First of all, thanks to Ms. Kliff and Vox for responding to me; it’s always neat to get featured in real news sources as if I were a real writer or something instead of just a guy with a blog. Additional thanks for a very measured and charitable tone despite my own tendency toward snarkiness. I worry that I am being unfair by acting snarky to somebody who is writing for a professional publication and so is not allowed to be snarky back; if so, I’m sorry and will try to control myself as best I can, which admittedly is not very good.
That having been said, I’m super against all of this and think it’s totally wrong.
I would kind of like to complain about Vox calling EpiPen a “red herring” when they were the ones who brought it up, but I think the problem is deeper than that. Discussing generic drug costs is completely different from discussing brand name drug costs, and the two issues have very different arguments around them. To transition fluidly from one to the other, saying we need more price controls for generics and then backing up your argument by saying that you were mostly thinking of brand-names after all – elides a distinction which is the heart of this entire subject.
Generic drugs are overpriced because we’re morons who can’t come up with a decent regulatory regime. Brand-name drugs are overpriced because of a deliberate decision to overprice them to encourage research.
The economic argument goes: the more profitable new drugs are, the more incentive a company has to make them. If we didn’t reward pharmaceutical companies for inventing new drugs, then they wouldn’t go through the $2.5 billion, ten-year hassle of seeking FDA approval with no guarantee of success. The way we reward them is by giving them a twenty-year monopoly when they can charge lots of money without anybody telling them not to.
(this isn’t quite right. The law says a twenty-year monopoly, but it’s dated from the time the drug is invented. Since it takes ten years to go through the FDA, it’s effectively more like a ten-year monopoly on actually selling the drug)
The reason I usually limit my griping about pharmaceutical overpricing to generics and avoid brand-names is that while high generic pricing is inexcusable, high brand-name pricing is debatably useful. Some people would say that the benefit of encouraging more drug development is worth the cost of higher prices, other people would say that it isn’t. I didn’t want to wade into this complicated debate.
But I guess now I have to. So. Lit review time. I searched the economics literature for studies, models, and arguments used to calculate whether price regulation would decrease drug development, and if so, how the benefits and risks balanced out. Here’s what I’ve got:
1. Golec & Vernon (2006) say that as a result of European drug price regulation, “EU consumers enjoyed much lower pharmaceutical price inflation, however, at a cost of 46 fewer new medicines introduced by EU firms.”
2. Eger and Mahlich (2014) find that among pharmaceutical companies, “a higher presence in Europe is associated with lower R&D investments. The results can be interpreted as further evidence of the deteriorating effect of regulation on firm’s incentives to invest in R&D.”
3. Kutyavina (2010) finds that “brand-name pharmaceutical firms characterized by large R&D expenditures decreased their R&D efforts post 1993 threat [to regulate drug prices] relative to firms that did not engage in as much innovative R&D”.
4. Acemoglu and Linn (2004) find that “We find a large effect of potential market size on the entry of nongeneric drugs and new molecular entities”, which I think is supposed to generalize to mean that the more money they expect to make the more research they do. I will count this as half a study since the connection is not explicit.
5. Danzon & Epstein (2008) analyze price regulations and new drugs invented in 15 countries and 12 drug classes, and find that “If price regulation reduces drug prices, it contributes to launch delay in the home country.
6. Troyer & Krasnikov (2002) find that “the empirical relationship between pharmaceutical industry revenues and pharmaceutical industry innovation is estimated, allowing for an exploration of the impact of the Medicaid rebate program [which regulated drug prices somewhat]. Using the empirical results, the opportunity cost of the Medicaid rebate program is found to be as high as four new drug approvals annually. Given the increased interest in a Medicare drug benefit, regulators should be aware of the hidden cost of price regulation for pharmaceuticals.”
7. Vernon (2005) finds that “I simulate how a new policy regulating pharmaceutical prices in the US will affect R&D investment. I find that such a policy will lead to a decline in industry R&D by between 23.4% and 32.7%. This prediction, however, is accompanied by several caveats.”
8. Golec, Hegde, and Vernon (2009) find that “Results show that the HSA [a bill to regulate drug spending in the US] had significant negative effects on stock prices and firm-level R&D spending. Conservatively, the HSA reduced R&D spending by about $1 billion even though it never became law.”
9. Santerre and Vernon (2006) use drug demand data to simulate various regulatory regimes, and find that a certain price regulation policy they test, continued over twenty years, would have cost gains of $472 billion (!) but also “have led to 198 new drugs being brought to the US market” (!!). They note that “Therefore, the average social opportunity cost per drug developed during this period was approximately $2.4 billion. Research on the value of pharmaceuticals suggests that the social benefits of a new drug are far greater than this estimate. Hence, drug price controls could do more harm than good.”
10. Keyhani, Carpenter, et al (2010) find that “The United States accounted for 42% of prescription drug spending and 40% of the total GDP among innovator countries and was responsible for the development of 43.7% of the NMEs [ie new drugs invented]. The United Kingdom, Switzerland, and a few other countries innovated proportionally more than their contribution to GDP or prescription drug spending, whereas Japan, South Korea, and a few other countries innovated less…higher prescription drug spending in the US does not disproportionately privilege domestic innovation, and many countries with drug price regulation were significant contributors to pharmaceutical innovation.” This study does not attempt to address the effects of price regulation, only to say that European countries seem to do pretty well at innovation despite price regulation, which is suggestive that price regulation does not hurt drug innovation but not really scientific evidence for it. I’m going to count this one as half a study too.
So by my count, there are eight-and-a-half studies concluding that price regulation would hurt new drug innovation, and one-half of a study concluding that it wouldn’t. I’ve tried to eliminate all the studies sponsored by the pharmaceutical industry from this list, but I might have missed some, and I am always skeptical of anything that says anything the pharmaceutical industry approves of even I can’t trace the money directly.
One source I do trust is RAND, a think tank which is generally well-respected and pretty objective (despite the name, they are not associated with Ayn Rand or Rand Paul). In Regulating Drug Prices: US Policy Alternatives In A Global Context, they write:
U.S. consumers spend roughly twice as much on drugs as their European counterparts….Pressure is building in U.S. policy circles for the federal government to take action to regulate the cost of drugs. At the same time, there is debate about the pros and cons of doing so…To shed light on this debate, a team of RAND researchers examined the impact of drug price regulation…The results showed that:
— Globally, the regulation of pharmaceutical prices has increased in recent years.
— In most cases, regulation reduces pharmaceutical revenues.
— Regulatory approaches that reduce pharmaceutical revenues may generate modest consumer savings in the best cases, but risk much larger costs as decreased innovation leads to reductions in life expectancy.
In other words, such prices would be good in the short term as we get all the currently-existing drugs for very cheap:
Annual per capita spending in 2010 would fall for Americans age 55–59 by an amount in the range of $9,000 annually and for Europeans of the same age by an amount in the range of $400.
But bad in the long-term as pharmaceutical innovation declines and we have fewer interventions available to protect our health:
Life expectancy would fall by somewhere in the range of 0.2 years for Americans age 55–59 in 2010 and by approximately 0.1 years for Europeans in the same cohort and year. By 2060, this effect would increase for both Americans and Europeans to approximately 0.7 years.
Given the value they place on human life, they argue that this money-for-life-years trade is net negative:
Overall, as shown in Figure 1, the net value of price controls is positive in the short term (2010) for Americans age 55–59, producing approximately $1,100 in per capita savings, but negative for Europeans in the same cohort and year, who face increased costs in the range of $8,000. For both Americans and Europeans, price controls have higher per capita costs over the longer term: By 2060, reductions in life expectancy, after accounting for medical cost savings, would cost the equivalent of $51,000 and $54,000, to age 55–59 Americans and Europeans, respectively.
All of this sounds sort of boring and economics-y when you read it like this, and maybe your eyes are glazing over. So let me put this in context. In 2060 there will probably be 420 million Americans and 523 million Europeans. And suppose that whatever changes we make in drug regulations today last for one human lifespan, so that everybody has a chance to be 55-60. So about a billion people each losing about 0.7 years of their life equals 700 million life-years. Since some people live in countries outside the US and Europe  and they also benefit from First-World-invented medications, let’s round this up to about a billion life-years lost.
What was the worst thing that ever happened? One strong contender is Mao’s Great Leap Forward, in which ineffective agricultural reforms and very effective purges killed 45 million people. Most of these people were probably already adults, and lifespan in Mao’s China wasn’t too high, so let’s say that each death from the Great Leap Forward cost what would otherwise be twenty healthy life years. In that case, the worst thing that has ever happened until now cost 45 million * 20 = 900 million life-years.
Once again, RAND’s calculations plus my own Fermi estimate suggest that prescription drug price regulation would cost one billion life-years, which would very slightly edge out Communist China for the title of Worst Thing Ever.
Am I exaggerating or being facetious? I’m actually not sure.
Dammit, Jim I’m a doctor, not an economist. I’m not qualified to analyze any of those ten studies above beyond a quick check to see if they’re completely ridiculous. I’m not qualified to say if RAND is right or wrong to estimate a cost of 0.7 life-years, or whether I’m misusing their calculation to try to add up exactly how bad it would be. Maybe a real economist will look at this whole essay and say it’s really stupid. I don’t know.
The only thing I can say in my own defense is that I am acknowledging that the question exists. I am not at all sure that Vox has reached this level yet. They just wrote an article on price regulations for brand-name drugs which, first, mixed them liberally with generic drugs despite the different arguments around each, and second, didn’t mention anything about research or innovation. Call me overly demanding, but when you are proposing a policy which most economists think would decrease the rate of life-saving medical progress, and which by some calculations might edge out Mao’s China for the title of most disastrous and deadly thing in all of human history, then I feel like you should acknowledge, at least in a single sentence, that somebody has claimed, at least once, that the policy might have some slight downside. At least don’t act as if it’s the same issue as a different kind of drug regulation which doesn’t have that downside.
I think it’s an unfortunate omission to talk about the EpiPen cost increase as relating only to lack of price controls, and fail to mention the reason why this happens with EpiPen but never chairs or mugs. And I think it’s a further omission to talk about regulating brand-name medications but fail to mention that some people think it will backfire and impede innovation. While I appreciate the effort to say we’re both on the same team of reducing drug costs, I’m a little concerned about my teammates’ strategy here.
And there’s another way we’re not quite on the same team. I’m on Team Left-Libertarian, which luckily is so confusing and contradictory that I can define it however I want. And today it means that while I’m not opposed to all regulation in principle, I at least get really scared when somebody pushes for regulation today and promises to check whether it will have bad consequences tomorrow. I think that’s how we got in this mess where the generics industry is so regulated that EpiPens cost hundreds of dollars, and even if Vox and I are on the same object-level team of Make Epi-Pens Cost Less, I worry we are not on the same meta-level team of Learn From The Fact That Epi-Pens Cost So Much And Worry That The Same Kind Of Thinking That Caused The Epi-Pen Problem Will Probably Cause Other Problems Too.
So do I have a solution to the high price of brand-name drugs? Well, I have a partial, unsatisfying solution. But first, a digression.
Vox gives the example of Harvoni costing $32,000 in the United States, but only $22,000 in the United Kingdom. This is supposed to be an example of the United Kingdom’s drug price regulation system working. I guess this is good, but you may notice that both numbers are really really high. There’s generic Harvoni available in India for $900. I can’t find how much it costs to manufacture, but reading between the lines and looking at some similar compounds, it’s probably about $100. So good work, Britain. You’re paying $22,000 instead of $32,000 for a $100 pill.
There are a couple of morals to this story. The first is that Vox’s claim that generics made by two competing companies cost 55% of the brand-name price isn’t the right statistic to use here. Look at their source and you find that as number of competing companies gets to 20, generics cost 5% of brand names. As number of competitors approaches infinity, drug cost should approach manufacturing cost, which can be very low – in the cast of Harvoni, less than one percent. This seems true in the case of modafinil, which I’ve talked about before; it costs about $25 per pill in the US and more like $2 per pill in more generic-friendly India.
So the second moral of the story is that almost all gains in prescription drug prices are to be found not in price regulation bringing prices down from $32,000 to $22,000, but in switching from monopoly brand-name drugs that cost $32,000 to heavily-competitive generic drugs that cost $100.
In a lot of cases, this is easier than you would think.
Pristiq is the brand-name of desvenlafaxine, a new antidepressant which is still brand-name only. Desvenlafaxine sounds a lot like venlafaxine – which is Effexor, an old antidepressant which is available in generic. In fact, desvenlafaxine is a tiny change to the venlafaxine molecule which may or may not have any interesting medical benefit over the original, and which was invented solely to have something whose patent hasn’t expired.
Wyeth, the company that makes Pristiq, says that it’s better than Effexor because it doesn’t have as many drug-drug interactions. But Effexor doesn’t really have clinically significant drug-drug interactions, and this seems to be them just saying random stuff and hoping people believe them. There are no good head-to-head studies comparing Pristiq to Effexor, but if you try to piece together a comparison from unrelated studies (not recommended, but we’ll do it anyway) Effexor actually seems better than its newer cousin. Even the data I took from drug rating databases shows patients preferring Effexor to Pristiq by quite a lot. Carlat Psychiatry, which is psychiatrists’ insider news site on pharmacology developments, has a blog post called Top Five Reasons To Forget About Pristiq. Most of the well-informed psychiatrists I know agree that Pristiq is a slightly worse version of an older antidepressant with no proven advantages.
A month’s supply of Effexor costs $20. A month’s supply of Pristiq costs $300. So let me amend the paragraph above. Pristiq is a slightly worse version of an older antidepressant with no proven advantages that also costs fifteen times as much.
It should come as no surprise to anyone familiar with the state of psychiatry that it is the second most-prescribed antidepressant in the USA, with three million prescriptions per month.
Why would this happen? The relevant study is called Pharmaceutical Industry-Sponsored Meals And Physician Prescribing Patterns For Medicare Beneficiaries, so you know it’s going to be good. It shows that doctors who often eat drug-company-sponsored free lunches are more than twice as likely to prescribe Pristiq as doctors who rarely eat such lunches. This matches my observations perfectly. Doctors prescribe Pristiq because they don’t know very much about antidepressants, but they attend free lunches by pharmaceutical companies who tell them that Pristiq is great, and they believe it. If this surprises you, be more cynical.
I’m looking at the price of Pristiq in Canada, and it seems to range around $120 to $250. So if we instituted price regulations like Canada’s, we might lower the cost of Pristiq from $300 to $150. If we convinced doctors to prescribe Effexor instead, it would be $20, plus I really do believe Effexor is genuinely better.
Pristiq is far from alone in this. I don’t have good statistics, but I bet that at least half of brand-name prescriptions in the US are more like Pristiq (attempts to rip people off) than like Harvoni (genuinely wonderful breakthroughs in medical science).
So one of the best ways to deal with expensive brand-name drugs is to stop using expensive brand name drugs for no reason. Since I get to define what left-libertarianism means however I want, I will say that it is provisionally okay with banning pharmaceutical companies from buying doctors lunch, as long as there aren’t any studies concluding that this would kill more people than Communist China. There are probably lots of other ways to improve medical education and medical economics so that doctors are less easily bamboozled into prescribing these, but those can wait for other blog posts.
What about the genuinely novel brand-name drugs, the ones like Harvoni that really are better than anything that came before?
An optimistic answer: maybe after we stop spending our civilizational resources on Pristiq, we’ll have a little more money to afford them, and maybe we’ll be happy to subsdize the genuinely awesome pharmaceutical research that remains.
Another optimistic answer: once FDA regulatory requirements are loosened, there will be a wide selection of different brand-name drugs. For example, even when Prozac was a brand-name, pharma’s ability to inflate its price was limited by the existence of several very similar brand-name drugs, like Paxil and Zoloft. If there are twenty competing brand-name hepatitis wonder drugs – and I don’t think that’s outside the realm of what we can hope for – then I think that will tend to lower prices to cost. This would include the cost of research and licensing, and so still be pretty high, but as long as research is a real unavoidable cost it would probably be the best we can hope for.
(this would be a good time to bring up that chlorcyclizine costs fifty cents per pill and might work as well as Harvoni)
The pessimistic answer is that all we can do is ensure that the generics marketplace is fair and competitive. And then rich people can buy Harvoni now for $30,000, and poor people will have to wait ten years to buy Harvoni when it costs $100. Right now they’ll unfortunately have to figure out how to make do with the set of medications invented in 2006 and before.
And I know this is terrible, especially if someone has a disease with only one cure and it was invented after 2006. But think of it this way. This objection, rephrased, is that 2016 has more drugs available than 2006, and we want to maximize the number of new drugs available to the poorest patients. But if we try to do that by instituting price controls which decreases the rate of drug innovation long-term, then we end up decreasing the number of new drugs available to the poorest patients, exactly the opposite of what we thought we were doing.
Let me give an example. According to study (9) above, price controls would have caused about 200 fewer drugs to be approved over the period from 1980 – 2000. In fact about 600 drugs were approved during that period. So if they’re right, it would have cut the innovation rate by 1/3. That means that in Hypothetical Price Control World’s 2016, after 36 years of price controls, we would only have 24 of our years’ worth of drugs – ie, the drugs that we had in 2004 in our own world. But since drugs usually go off patent about ten years after approval, in fact we’ve genericized the drugs that we had in 2006 in our own world. So we have more drugs available just as generics than Hypothetical Price Control World has as generics and brand-names combined. If poor people can afford only generics or price-controlled brand names, our poor people are better off than Hypothetical Price Control World’s poor people (and our rich people are much better off than Hypothetical Price-Control World’s rich people). And as time goes on, our advantage over their world will only get bigger.
Maybe there are better ways to do this. Some people have talked about funding research via “prizes” rather than through an investment-and-profit model. Some people say we should fund it publicly through the NIH or something, which we already sort of do to a degree. Still other people say that we should abolish the FDA, cut the costs of drug development by an order of magnitude, and, um, see what happens. I don’t know about any of those things. I just feel like until you’re ready to set these up and have some idea that they work, do the thing that probably is going to result in people having the best access to the most life-saving drugs. Which right now looks like no price control.
Or maybe I am completely wrong about all of this. I am not an economist and have to take these studies at face value, and anything that touches pharmaceutical companies ends up being corrupted and full of lies. Maybe I myself am saying something very stupid that will end up killing more people than Communist China. If so I certainly hope that people who know more than I do will tell me why these calculations were wrong and how to look at this situation better.
But this is the level at which I think this discussion needs to be had.