The Healthcare Marketplace and Product Design
Unique considerations for health companies
Kenneth Arrow and the Unique Marketplace
In his 1963 paper, Arrow outlined what makes healthcare different than other industries. It’s been called “the article that launched a thousand studies,” but that would be understating it: Arrow framed the entire field of health economics, and thus the debates surrounding one of the largest sectors of every economy on the planet.
Arrow specifies five characteristics of the healthcare market, none of which are totally unique to medical care but none of which are all present in any other market:
Demand is weird. It’s basically random, it is super high value, and it has uniquely strong emotional associations.
Ethics are super important. There’s no way to test it before you use it despite the emotional risk.1 The result: an unusual ethical obligation among physicians.
Supply is also weird. You’re dealing with uncertain treatment effects, possible complications, and highly specialized knowledge.
Supply is artificially scarce. Arrow focuses on licenses, though many industries have them, and you also see lots of patents.
Unusual pricing strategies; in particular, Arrow focused on extreme price discrimination. This one has only gotten truer over time! Now we have self-insurance, HMOs, Medicare/Medicaid, HSAs…
But what’s at the root of all this? At its core, healthcare has a unique nexus of information uncertainty and high value. This central insight has been incredibly influential. It’s also the reason that Arrow thought single-payer systems are the best healthcare systems yet devised. And keep in mind that Arrow’s article is about healthcare, not biotech, so think hospitals, not pharma; your mileage may vary when applying this insight to even adjacent markets. You can see the effects in the structure of the market, as Arrow notes the importance of trust and delegation in the healthcare market. It’s kind of surprising (since when do economists talk about emotions like that?) but also quite elegant (it flows naturally from his observation that healthcare is both inscrutable and imbued with strong moral foundations).
The Arrow Redux
There’s a lot that we can add to this idea that wasn’t in Arrow’s original article. I can think of three big things.
The first is about patient data. Here’s the reality: neither patient nor provider has full data visibility, nor can they.
Take the provider side. Even small providers may interact with dozens of employees and contractors of various levels of understanding, with informal or quick decisions that may never be recorded and therefore propagated throughout the system written in incantations only comprehensible to other specialists. Much uncertainty also comes from other providers. The story of the modern healthcare system, after all, is one of triage of specialists and trips between institutions, with biopsy samples that can travel to five different entities and opaque health records that almost no hospital will export to HIPAA-compliance. The result: no health professional has visibility of a patient’s full record, and even if they did it might be incomprehensible.
Then take the patient side. Remember, the whole point of Arrow’s article is to point out that patients don’t understand their health because it’s such a specialized field. Patients may be unable to communicate their health or know what to share from their providers. They may forget to report information or not have a good, real-time way to report information to their providers. Plus, patients don’t even know all of their health data because they need to go to a provider to get a test in the first place, may not know what to ask for, or may not know how to describe their symptoms (quick, what’s the difference between a “sharp” pain and a “shooting” pain?). In short, providers can’t always get all the information they need from patients to actually provide care.
The second would be the subtlety that much of the lack of visibility is imposed by the market itself. While CMS regulations now require paymasters to be made public, the prices paid to insurance companies are not, making the public hospital prices essentially useless. Prices are intentionally kept secret in order to allow for price discrimination by provider and to avoid public outrage. Similarly, administrated self-insurance, which has significantly increased under the ACA, inherently siloes this information further, partially to minimize the administrative burden on employees and partially to avoid being gouged by providers. Government regulations complete this circle of secrecy. HIPAA requires providers to communicate the “minimum necessary” amount of health information; trade secret law incentivizes biotech companies and hospital systems alike to reveal as little as possible to their competitors;2 drug policies push significant swathes of treatment underground. Especially now that everything is so complex and the dependencies are so deep, the stakes are so high that people don’t want to actualize any risk. Now, my European friends are ready to scream at me: this just applies to the US! Healthcare is a human right! Bernie 2016! Wrong. This applies to the healthcare market in every country. Other countries have just responded to this by imposing a national HMO.
It’s also intriguing to see this market-imposed silo effect with health startups — it doesn’t often interact much with the rest of the startup ecosystem. Healthcare, and especially biotech, is way more specialized than other hard science sectors! There was no specialized aerospace VC to fund Planet Labs (DFJ), SpaceX (Founders Fund), or Rocket Lab (DCVC). In fact, regular VCs fund hard science companies frequently, like rocket companies, chip companies, oil and gas companies, legal tech companies, and fintech companies. But biotech is relatively isolated. Andreessen Horowitz’s first specialist fund was a biotech fund, led by Vijay Pande. And the league tablesfor biotech contain mostly unfamiliar names. Ask anyone in the space why. They’ll tell you it’s because biotech is otherwise just too damn hard.
The third is that the fact that everything is mediated by trust inverts the traditional strength of network effects. Generally, there are two types of network effects. The first is demand-side network effects, where the network acts on the demand side by making the product more useful as the number of users increase. Think Facebook, and how it is hard to disrupt because all of your friends are already there. The second is supply-side network effects, where the network acts on the supply side by creating some kind of supplier advantage, like liquidity or economies of scale, that make it advantageous for others to join the thicker market. Think Uber, which uses driver liquidity to drive demand, thus bringing even more drivers on. Traditionally, demand-side effects are thought of as stronger than supply-side effects. The first is that supply-side effects can usually be bought, so they can be blitzscaled. The second is that it is almost impossible to spontaneously generate demand that can make people move, and similarly make people stay when their utility is lower due to the smaller network.
In healthcare, this effect is inverted. Supply-side effects are stronger than demand-side effects. The reason is trust. It is exceedingly rare that consumers ever actually make the purchasing decision on their own. Every decision is mediated by trust and highly individualized, so it is quite rare for a healthcare decision to be made due to the presence of other users. Additionally, because so much of demand is mediated by trust, a trusted provider can instantly create an island of demand even if there were a network effect, allowing other products to start and subsist. There are some exceptions, but they’re either fitness products or rare products with network effects among providers, like Epic. On the supply side, however, the uncertainty and high switching costs make supplier networks quite strong. Besides, because demand is mediated through trusted intermediaries, it can be extremely expensive to overcome existing supplier relationships, regulatory requirements, or plain sufficiency requirements.
So You Want to Get Into Healthcare
So, say you want to build a healthcare product that isn’t biotech. Essentially, if you want to build something new, you have four major options to deal with the core information uncertainty problem:
Reduce uncertainty, thus putting more control in the hands of consumers or providers, whatever the case may be
Reduce consumer control, thus abstracting away uncertainty
Find a niche exploiting uncertainty
Build trust yourself
There are ways to reduce uncertainty by improving integration between providers, like with Ciitizen; giving patients and providers more avenues to connect, like with OneMedical; or to streamline a healthcare process, like with Pillpack.
When building a product, founders can look closely at who makes the care decision to see if there is a rare opportunity for a true demand-side network effect. Most of the time, though, the network effects will be small, supply-side, or nonexistent.
Above all, remember that the healthcare industry is extremely complex, highly regulated, and above all build on trust. Customer trust is important for many companies and industries, from Airbnb to Demisto. But with healthcare, it’s a bust without it.
Though this is true of many service industries.
A lot of this uncertainty is in supply chains, which are highly complex and mostly kept secret. Furthermore, the science is some of the least understood, and therefore highly stochastic, making it almost impossible to disrupt.