Publicly Funding Drug Development


Spotted something like this in the wild, with a much bigger savings estimate - see this post.


It's important to realize that high, uncontrolled prices for on-patent drugs in the USA is a deliberate policy choice. We incentivize private industry to invest a pile of money in developing new medicines, in return they get the exclusive right to sell it in the USA for 5 to 20 years (20 years after they apply for the patent, it takes a few years to be granted).

The rest of the world gets a partial subsidy from this; these medicines are subject to price controls in many places. (That's why importing drugs from Canada is an incoherent policy at scale; they're cheaper in Canada because of price controls - that would just be free-riding on their controls, we could just set our own instead, though then where would the R&D money come from?)

I tend to be skeptical of government-granted monopolies to the private sector, though. Competition makes the private sector good at the things it does. Setting up a monopoly has a tendency to get us a public-private partnership with the worst of both - the limited accountability of government, combined with the rent-seeking of the private sector. Consider cable utilities for an illustration. When I see such public-private partnerships, my first question is "would this be better as direct public provision?" (Consider municipal broadband, in the places it's allowed).

Public Provision

The government spends a lot of money on buying on-patent drugs (also on generic drugs) through e.g. Medicare, Medicaid, and the VA. So in the spirit of doing an energy calculation before starting a big project, how much more would it cost for the government to pay for R&D directly?

The cost of R&D: $102 billion in 2018, per the industry's own lobbying group PhRMA. I think we can assume that's not an under-estimate; that would be lobbying malpractice.

The idea: if the US government put up this $102 billion a year, and got rid of pharmaceutical patents (making all drugs, in effect, generic), how much would the US government save, and how much would the US private sector save?

Transition to that state would take some time, and I'll put together some ideas for that in a future post.

Current Government Drug Spending

Existing government-funded drug research like the NIH isn't being counted here, because (I assume) it's also not counted in our $102 billion target.

Table 4 in the CMS National Health Expenditures dataset lists prescription drug spending (not broken down by name-brand/generic) by source for 2019, in billions of dollars:

TotalOut of PocketPrivate Health Insurance MedicareMedicaidOther GovtOther Private
369.753.7164.6 104.631.4114.3


For rough numbers, we can estimate from this FDA paper that generic drug prices would be 40% of the name-brand prices (this is a median based on Medicare prices for drugs with initial generic entry 2015-2017; with four competitors, the median price is 20% of name-brand price).

We will fudge this to 50% for a conservative savings estimate.

As for the fraction of spending on name-brand drugs vs. drugs that are already generic, this Statista report puts this at 80% name-brand. CMS fine-grained data on Medicare Part D and Medicaid drug spending agrees.

We will fudge this to 75% for a conservative estimate.

Put together, we get a savings of 37.5% of total expenditures by making all drugs generic.

Government itself

The government spent $147 billion in 2019, so it would save $55 billion.

Private sector

The private sector spent $222.7 billion in 2019, so it would save $84 billion.


This is an overall net US savings of $37 billion anually, as a $47 billion increase in government spending and an $84 billion decrease in private spending.

Some savings would also accrue in the rest of the world, figuring out how much is left as an exercise for the reader.

There are also no doubt lots of non-fiscal pluses and minuses to this approach (e.g. trading off "me-too" drugs for political dysfunction around funding); looking at those (and a transition) will be a future post.