MIT economist Heidi Williams won a MacArthur genius grant last year and her work has been cited in US Supreme Court briefs. She isn’t (yet) well-known outside the economics profession, but she and her collaborators have done fascinating and important work on the economics of innovation incentives — how to foster the right kind of scientific and medical research; flaws and opportunities in the design of patents and the protection of intellectual property more generally; when and how the public sector should supplement the private sector.
We talk about all this and more in our longform conversation, which I greatly enjoyed. But as I noted in the podcast, some of Heidi’s work is quite technical, and below we provide a brief layman’s guide to the papers we discussed in addition to excerpts.
The first paragraph of the paper is worth quoting in full — broken up here for online readability:
Innovation is a key driver of economic growth, but competitive markets may under-incentivize innovation because of the public-good nature of ideas (Nelson 1959; Arrow 1962).
Intellectual property (IP) rights, such as patents and copyrights, aim to incentivize innovation by allowing firms to capture a higher share of the social returns to their research investments. Traditional evaluations of the effectiveness of IP have focused on whether the prospect of obtaining IP rights stimulates the development of new technologies.
However, in many markets technological change is cumulative, in the sense that product development results from several steps of invention and research. In markets in which innovation is cumulative, the overall effectiveness of IP in promoting innovation also depends on a second, less studied question: do IP rights on existing technologies hinder subsequent innovation?
The contribution of this paper is to provide empirical evidence on this second question by investigating how one form of IP on the human genome influenced subsequent scientific research and product development.
Two important points should be understood about this paper in considering its methods and findings.
First, it investigates the impact of a specific kind of intellectual property protection* — but not the impact of patents, which is addressed in a subsequent paper (number 2 below).
Second, it studies the impact of intellectual property rights for existing technologies on subsequent innovation that builds on these same technologies. It does not study the impact of intellectual property on the invention of new technologies, though this too is studied in a later paper (in a different context — number 3 below).
The Human Genome Project was publicly funded and started sequencing genes in 1990. Because the Project operated under the so-called Bermuda Rules, each of its discoveries was immediately placed in the public realm after it was made. When genes are sequenced, the innovation that follows can include published work by academics; knowledge actually learned about the genes, for instance about the links between certain genes and diseases; and diagnostic tests developed to investigate such links.
A private company called Celera was founded in 1999, when it too began sequencing the human genome. By 2001, Celera had completed the sequencing of some genes that the Human Genome Project would not sequence until 2003.
For that two-year period between 2001 and 2003, the innovation that followed on the genes sequenced by Celera can be compared against the innovation that followed on the genes that were sequenced by the Human Genome Project during the same two-year period.
If the pace of innovation during that window of time was slower for the genes sequenced by Celera, which were protected by intellectual property rights, than for the unprotected and publicly available genes sequenced by the Human Genome Project, then that would suggest that the intellectual property rights were slowing the pace of follow-on innovation.
That’s what the paper did find. But it also contained many nuances that we discussed in the podcast, including the fascinating ways in which the restrictions apply.
* Celera had sought patents for the genes it had sequenced, but generally the patents were not granted. Williams isn’t sure why but suggests one possibility: for a patent to be granted, the invention must satisfy a utility requirement that it have a known benefit. Celera had sequenced the genes but perhaps didn’t yet know how the knowledge would be applied in practice.
Whatever the case, the company’s lawyers sought a different kind of intellectual property protection that Williams describes as “a combination of a shrink wrap license and a contract”.
The shrink wrap is similar to the protections of a product like Microsoft Office: the data of the sequenced genes could be bought from Celera but not be distributed further. This allowed Celera to discriminate on price, charging academics less for the data than it charged pharmaceutical companies. Academics and pharmaceuticals also had to negotiate with Celera before using the data they had purchased for commercial applications, needing to first arrive at an agreement for how future revenues would be shared.
We investigate whether patents on human genes have affected follow-on scientific research and product development.
Using administrative data on successful and unsuccessful patent applications submitted to the US Patent and Trademark Office, we link the exact gene sequences claimed in each application with data measuring follow-on scientific research and commercial investments.
Using this data, we document novel evidence of selection into patenting: patented genes appear more valuable — prior to being patented — than non-patented genes.
This evidence of selection motivates two quasi-experimental approaches, both of which suggest that on average gene patents have had no effect on follow-on innovation.
Emphasis mine. The “no effect” conclusion works in both directions: patents on genes were neither a hindrance nor a spur to follow-on innovation.
The paper’s conclusion also includes a helpful passage summarising the combined lessons of this paper and the prior study (explained above) of non-patent types of intellectual property protections on human genes:
Taken together with the prior literature, our evidence suggests two conclusions.
First, for the case of human genes, the traditional patent trade-off of ex ante incentives versus deadweight loss may be sufficient to analyze optimal patent policy design, because any effects of patents on follow-on innovation appear to be quantitatively small.
Second, our evidence together with the evidence from Williams (2013) on how a non-patent form of intellectual property on the human genome affected follow-on innovation suggests a somewhat nuanced conclusion: while patent protection on human genes does not appear to have hindered follow-on innovation, an alternative non-patent form of intellectual property – which was used by a private firm after its gene patent applications were largely unsuccessful in obtaining patent grants – induced substantial declines in follow-on scientific research and product development.
This pattern of evidence suggests that changes to patent policy must carefully consider what strategies firms will use to protect their discoveries in the absence of patents, and that an understanding of the relative costs and benefits of patent protection compared to those outside options is needed in order to evaluate the welfare effects of patent policy changes.
An important paper — and the answer to the question posed in the title is yes. From the opening:
Over the last five years, eight new drugs have been approved to treat lung cancer, the leading cause of US cancer deaths. All eight drugs targeted patients with the most advanced form of lung cancer, and were approved on the basis of evidence that the drugs generated incremental improvements in survival. A well-known example is Genentech’s drug Avastin, which was estimated to extend the life of late-stage lung cancer patients from 10.3 months to 12.3 months.
In contrast, no drug has ever been approved to prevent lung cancer, and only six drugs have ever been approved to prevent any type of cancer.
The paper looks at the economic incentives for pharmaceutical companies to invest in late-stage cancer treatments versus preventive and early-stage treatments. A particular problem is the design of patents, whose protections in some industries begin at the time of discovery rather than at the time of the first commercial sale.
But early-stage treatments for cancer can take a while to arrive at the market, in part because clinical trials required by the Food & Drug Administration naturally take longer than they would for late-stage cancers, especially given that the traditional qualification for success is lower mortality.
Thus the period of exclusivity for early-stage treatments is shorter, potentially incentivising pharma companies to focus more on later-stage treatments.
The findings of the paper have fascinating implications both for patent design and for the role of the public sector. We discuss them on the podcast, but two more excerpts from the paper itself are helpful.
First, some data:
Using this data, we document that, consistent with our conjectured distortion, patient groups with longer commercialization lags (as proxied by higher survival rates) tend to have lower levels of R&D investment. Panel A of Figure 1 gives a sense of this basic pattern using stage-level data.
On average, metastatic cancer patients have a five-year survival rate of approximately 10 percent, and have nearly 12,000 clinical trials in our data. In contrast, localized cancer patients have a five-year survival rate of approximately 70 percent, and have just over 6,000 clinical trials in our data.
This pattern is even more stark if we contrast recurrent cancers (advanced cancers with very poor survival prospects) and cancer prevention: fewer than 500 trials in our data aim to prevent cancer, whereas recurrent cancers have more than 17,000 trials. A rough adjustment for market size—looking at the number of clinical trials per life-year lost from cancer—does little to change this basic pattern.
The second excerpt is about publicly vs privately funded research:
To illustrate, consider two examples of clinical trials for prostate cancer treatments, both published in the New England Journal of Medicine in 2011.
A first study, de Bono et al. (2011), analyzed a treatment for metastatic prostate cancer (an advanced stage of prostate cancer with a five-year survival rate on the order of 20 percent). The study tracked patient survival for a median time of 12.8 months, and estimated statistically significant improvements in survival (a gain of 3.9 months of life).
A second study, Jones et al. (2011), analyzed a treatment for localized prostate cancer (an early stage of prostate cancer with a five-year survival rate on the order of 80 percent). The study tracked patient survival for a median time of 9.1 years, estimating statistically significant improvements in survival. As expected, this stark difference in patient follow-up times translates into a large difference in clinical trial length: 3 years for the metastatic patient trial versus 18 years for the localized patient trial.
Consistent with the idea that commercialization lags differentially reduce private R&D incentives, the study of metastatic cancer patients was funded by a private firm (Cougar Biotechnology) whereas the study of localized cancer patients was funded by the National Cancer Institute.
The paper also discusses the potential use of “surrogate endpoints” — essentially an effect of the drug on the body that suggests an improvement in mortality without actually having to run a full-length clinical trial. So if a drug lowers blood pressure, it is likely to also be decreasing cardiovascular disease mortality. But the use of surrogate endpoints is complicated for reasons that Heidi gives on the podcast.
We present a simple graphical framework to illustrate the potential welfare gains from a “top-up” health insurance policy requiring patients to pay the incremental price for more expensive treatment options.
We apply this framework to breast cancer treatments, where lumpectomy with radiation therapy is more expensive than mastectomy but generates similar average health benefits.
We estimate the relative demand for lumpectomy using variation in distance to the nearest radiation facility, and estimate that the “top-up” policy increases social welfare by $700–2,500 per patient relative to two common alternatives. We briefly discuss additional tradeoffs that arise from an ex ante perspective.
A “top-up” policy involves the health insurance provider paying up to the cost of a basic treatment, leaving the patient with the option to choose a more expensive treatment and pay the difference between the two.
To generalise a bit, right now the health insurance plans available in the United States tend to be generous for those who have them, covering most of the cost of even expensive treaments — but many Americans don’t have insurance. In the UK, everyone has health insurance, but it only covers certain treatments, and the patient must pay the full cost of any treatment not covered.
A “top-up” approach is therefore a middle path between the generous but exclusive insurance plans available in the US and the universal but (relative to the US) ungenerous coverage available in the UK.
The design of this study is very clever, though Heidi is careful to emphasise that its conclusions are limited in scope.
The paper notes that for breast cancer treatment, the cost of performing a mastectomy (removal of the entire breast) is lower than the cost of performing a lumpectomy (removal of the tumour within the breast) plus radiation. Health insurance in the US covers nearly the full cost of both. The empirical evidence shows that health outcomes are roughly the same for patients of either treatment.
All things equal, patients tend to opt for the lumpectomy, possibly for cosmetic and psychological reasons. But things are not all equal, and the cleverness of the paper is in estimating the relative willingness of patients to pay for the lumpectomy vs the mastectomy by looking at an important difference between the two.
Unlike the mastectomy, the lumpectomy requires about two dozen radiation treatments after the surgery. The time spent commuting to the radiation facility and undergoing radiation can be quantified, and it represents an incremental cost that is not (indeed cannot) be covered by health insurance. And it therefore reflects the increase in cost that patients are willing to absorb for the treatment they prefer — like in a “top-up” insurance regime.
From the paper:
We estimate, for example, that the efficient “top-up” policy—in which patients pay $10,000 on the margin for a lumpectomy—increases the lumpectomy rate by 15–25 percentage points relative to the UK-style “no top-up” regime, and decreases the lumpectomy rate by 35–40 percentage points relative to the US-style “full coverage” regime.
In other words, under the UK-style regime, insurance wouldn’t cover any of the cost of the lumpectomy — neither the surgery nor the $10,000 worth of time. It would only cover the full cost of a mastectomy. But under a system in which the surgery is covered but the patient has to pay for the $10,000 of time, about 15-25 per cent more patients would indeed choose the lumpectomy.
Conversely, if the US-style system fully applied to breast cancer treatment as it does to other treatments, the patient would expect the full cost of both the surgery and the time to be reimbursed. If only the surgery is covered, then 35-40 per cent fewer patients would opt for the lumpectomy.
The United States has higher infant mortality than peer countries.
In this paper, we combine microdata from the United States with similar data from four European countries to investigate this US infant mortality disadvantage. The US disadvantage persists after adjusting for potential differential reporting of births near the threshold of viability.
While the importance of birth weight varies across comparison countries, relative to all comparison countries the United States has similar neonatal (<1 month) mortality but higher postneonatal (1–12 months) mortality.
We document similar patterns across census divisions within the United States. The postneonatal mortality disadvantage is driven by poor birth outcomes among lower socioeconomic status individuals.
A chart:
And a passage from the paper:
We then take a slightly more parametric approach and define a high socioeconomic status group in each country based on education and other demographic characteristics. Infant mortality rates for mothers in our advantaged group are statistically indistinguishable in the United States and elsewhere. However, there are very large differences across countries in infant mortality rates for mothers outside of this group. We see similar patterns across census regions within the United States.
Effectively, either across countries or across regions within the United States, we see that the observed geographic variation in postneonatal mortality is heavily driven by variation in health gradients across socioeconomic groups. Notably, when we look at neonatal mortality we do not draw the same conclusions, suggesting that the inequalities we observe emerge especially strongly during the postneonatal period.
The idea here is that patients might use more healthcare (by total cost of the care provided) in some geographic locations than in others, but with similar health outcomes. How much of the excess cost is mainly down to the nature of the patients in that area — for instance if it is an area with a less healthy population that requires more care — and how much is the result of differences in how medical care is practiced in differing areas?
From the abstract:
We study the drivers of geographic variation in US health care utilization, using an empirical strategy that exploits migration of Medicare patients to separate the role of demand and supply factors. Our approach allows us to account for demand differences driven by both observable and unobservable patient characteristics.
Within our sample of over-65 Medicare beneficiaries, we find that 40-50 percent of geographic variation in utilization is attributable to demand-side factors, including health and preferences, with the remainder due to place-specific supply factors.
And as further explained in the paper:
In this paper, we exploit patient migration to separate variation due to patient characteristics such as health or preferences from variation due to place-specific variables such as doctors’ incentives and beliefs, endowments of physical or human capital, and hospital market structure. As a shorthand, we refer to the former as “demand” factors and the latter as “supply” factors.
Copyright
The Financial Times Limited . All rights reserved. Please don't copy articles from FT.com and redistribute by email or post to the web.