This post is a foundational overview of characteristics of network industries. Much of the terminology will deserve deeper discussion, but we have to start somewhere.
In his book The Economics of Network Industries, Professor Oz Shy lists four characteristics of network industries.
The main characteristics of these markets which distinguish them from the market for grain, dairy products, apples, and treasury bonds are:
- Complementarity, compatibility and standards
- Consumption externalities [network effects]
- Switching costs and lock-in
- Significant economies of scale in production
In this essay, I’ll quote from Dr. Shy in explaining each of these characteristics. I’ll also offer a few thoughts as to how these characteristics apply to healthcare. More specifically, I’ll discuss physician adoption of EHRs (electronic health records) and patient adoption of PHRSs (personal health record systems).
Why a PHRS instead of a plain old PHR? Think of a PHRS as a PHR data repository platform bundled with multiple high-value applications. For a more detailed explanation, read here.
Let’s look at the characteristics of network industries one at a time.
1. Complementarity, compatibility and standards
While these terms are related, they’re not the same thing.
Computers are not useful without having monitors attached, or without having software installed…. Complementarity means that consumers in these markets are shopping for systems (e.g, computers and software, cameras and film, music players and cassettes) rather than individual products….[Commentary: Dr. Shy’s book was written in 2001 — how much has changed!]
On the technical side, the next question to ask would be how complements are produced. In order to produce complementary products they must be compatible. The CD album must have the same specification as CD players, or otherwise it can’t be played…..Trains must fit on the tracks, and software must be workable with a given operating system….
This means that complementary products must operate on the same standard. This creates the problem of coordination as how firms agree on the same standard.
To date, we have little complementarity, compatibility and standards in health IT. In fact, it’s just the opposite — most companies have been building proprietary business models and IT with the intent to lock customers in to a specific technology.
BUT, physicians and patients value complementarity, compatibility and standards — and a primary thrust of the HITECH Act legislation is to bring these characteristics to healthcare.
Physicians are looking for EHR systems; an EHR is an interconnected bundle of components such as a data repository, decision support capabilities, order entry, results management, etc.; physicians also value their EHR being connected to their practice management system. Today’s monolithic definition of EMR 1.0 is giving way to a more flexible, modular way of thinking about EHR 2.0 and clinical groupware.
PHRSs have already been moving toward complementarity, compatibility and standards. While most PHRs were initially tethered (to an employer, health plan, hospital, etc.), the entry of Google Health and Microsoft HealthVault is changing the dynamics.
Today, tethered PHRs are DOA (had enough initials?). Google and Microsoft have set the defacto industry standard by offering open, interoperable PHR platforms…and when you add applications from partners — you get a full blown PHRS.
So…we’re making some progress, but there’s still a wide chasm to cross.
2. Externalities [network effects)
Some economists distinguish between externalities and network effects, but let’s put the distinction aside for now.
The reader should ask herself the following question: would I subscribe to telephone service knowing that nobody else subscribes to a telephone service? Answer should be: of course not!… Utility derived from the consumption of these goods is affected by the number of other people using similar or compatible products….
A good example of this behavior is the fax machine, which has been used since the 1950s by flight service stations to transmit weather maps every hour on the hour (this transmission of single page took about one hour at that time). However, fax machines remained a niche product until the mid-1980s.
Think of (direct) network effects as “my value comes from you using the product”. One fax machine is useless; two connects you and a friend; a broad network of fax machines connects you to the world. More and more value is created for you as more and more people own fax machines.
At some point the market reaches a critical mass (tipping point). At that point your thinking will be “I need a fax machine because I might be out of the communication loop if I don’t have one!”
As shown in the diagram below, a network effect in a market results in the size of the network growing rapidly when critical mass is reached.
The potential for network effects in healthcare and care management is huge. Physicians would value other physicians (and patients) being on the same network so that they could share data and coordinate patient care. Patients would like to communicate with their physicians and to coordinate care for themselves and their loved ones.
3. Switching costs and lock in
We’ll talk a lot more about network effects later in this series.
Dr. Shy offers an explanation and examples:
There are several types of switching costs that affect the degree of lock in. Shapiro and Varian (1999) provide a nice classification of the various lock-ins.
Contracts: Users are sometimes locked into contracts for service, supplying parts, and buying spare parts….
Training and learning: Consumers are trained to use products operating on a specific standard. Switching costs would include learning and training people, as well as lost productivity resulting from adopting a new system.
Data conversion: Each piece of software generates files that are saved using a particular digital format. Once a new software is introduced a conversion software may be needed in order to be able to use it. Notice that the resulting switching cost increases over time as the collection of data may grow over time.
Search cost: One reason why people do not switch very often is that they would like to avoid the cost of searching and shopping for new products.
Loyalty cost: Switching technology may result in losing some benefits such as preferred customers’ programs, for example, frequent-flyer mileage.
To users of EMRs and first generation PHRs — does any of this sound familiar?
Lock-in tactics have become a staple of healthcare business and IT strategy. For physicians, choosing an EHR vendor is a major decision because it is very difficult to transfer data from one vendor’s system to another. Kaiser has spent nearly $2 billion installing their EPIC IT system — think they’re going to switch to another vendor soon?
Lock-in and switching costs are becoming less of an issue for patient PHRs; Google Health and Microsoft HealthVault have pledged that their platforms will be able to exchange data — there’s not going to be a standards war.
Lock in and switching costs are not inherently evil, even though as consumers we might not like them. We don’t savor having to pay an early termination fee to switch from a Verizon Blackberry to an AT&T iPhone, but nobody dies because of it.
The problem in health IT today is that much of the lock-in and switching costs derives from the lack of complementarity, compatibility and standards (characteristic #1). Lack of interoperability in healthcare is different — lives are at stake.
4. Significant economies of scale
This characteristic is straightforward:
Software, or more generally any information has the highly noticeable production characteristic in which the production of the first copy involves a huge sunk cost (cost that cannot be recovered), whereas the second copy (third, fourth, and so on) costs almost nothing to reproduce…the average cost function declines sharply with the number of copies sold to consumers.
Consider two factors about today’s physician EMR market:
- Adoption is low (about 4% for comprehensive EMRs, 20% for less comprehensive ones)
- Prices are high
These factors are interrelated — a chicken-egg problem. EMR vendors have significant upfront R&D costs, and they’ve had to spread that cost over a small user base.
We should expect prices for EHRs to decline over time (perhaps rapidly) as economies of scale allows vendors to distribute fixed costs over a larger base of customers.
So, is healthcare a network industry?
The biggest stumbling block today is characteristic #1 — lack of complementarity, compatibility, and standards. BUT, physicians and patients would value these highly, and recall that a primary public policy objective of HITECH Act legislation is to create these characteristics in healthcare.
The other characteristics either exist or would be unleashed if #1 were present. The potential for network effects is high; switching costs and lock in are already characteristics of the market, but could be much lower; and healthcare software has significant economies of scale.
I’ll refer back to the title of this series — Healthcare Crosses the Chasm to the Network Economy. We’re not there yet, but we’re on the verge….