Externalities and Network Effects
Network externalities, also known as the network-effect, are often important to consider when analysing technology investments. They are economic concepts that are useful in predicting the future structure of particular markets, especially technology markets where network effects are often prominent. Since this is something that I have a particular interest in, I decided to attempt to articulate and explain these concepts in this blog as a way to structure my thoughts.
To begin with, it is worth understanding the economic concept of externality. If we consider any voluntary exchange between parties, we can assume that there will be a benefit to every party that is directly involved. Otherwise they wouldn’t have voluntarily entered the exchange. However, there are often times when third parties to the exchange either benefit or incur a cost even if they are not directly involved. These benefits and costs borne by parties unrelated to an exchange are called externalities. For example, if I pay a security guard to keep an eye on my premises, then both the security guard and I will see the arrangement as beneficial. However, my neighbours may also gain a benefit of safety even though they aren’t part of the arrangement and haven’t paid for the services. This is an example of a positive externality. Another example would be if I leased my premises to the local rock-band for band practice. In this case both parties will see the arrangement as beneficial, though the neighbours may bear a cost of noise pollution late into the evenings without any compensation. This is an example of a negative externality. These positive and negative externalities are not the rare exception but occur with most economic transactions.
A network externality is a specific type of externality that occurs when a new user of a good or service has a benefit or cost to the other users of a good or service. Again, there can be both positive and negative network externalities. A classic example of a positive network externality is the fax machine: when a new user purchases a fax machine it makes everyone else’s fax machine more valuable. An example of a negative network externality is traffic congestion: the more people using the road, the less value of the road to each user.
Positive network externalities are often termed the network effect. The more people who adopt a particular technology, the more valuable the technology is for the other users. For example, the more people who have a telephone, the more valuable telephones are to everyone else. Another similar example is email: the more people who have access to email, the more valuable it is to everyone else.
Related to the network effect is Metcalfe’s law. Robert Metcalfe, the inventor of Ethernet and the founder of 3Com, came up with a concept that quantifies the network effect. He noted that a network with ‘n’ unique members has n(n-1)/2 possible unique connections, and therefore as the network grows the number of possible connections grows in the order of n squared. This insight is the basis of Metcalfe’s law that states that the value of the network is proportional to the square of the number of members within the network. In reality, this probably over-estimates the value of the network effect, though it captures the concept well.
With the examples I have given so far of the network effect, there is a single group of users that benefit as the number of users increases. For example, with fax machines or with email, the users of these technologies benefit as the number of users increases. Similarly a web-site such as Facebook has strong network effects with the value of the network increasing for all members as the network grows.
There are also networks that have more than one group of users; a two-sided network has two distinct user groups. It is common with two-sided networks, that each group of users gains a benefit when the number of users in the other group increases. For example, if we consider Microsoft Windows, the benefit of the software increases for end-users as the number of applications for the platform increases. For application developers, the value of the platform increases as the number of end-users increases. That is, adoption by one set of users drives adoption by the other set of users. Therefore, the platform often has to attract both types of users to gain the benefits of network effects.
With most two-sided networks, the different groups of users have very different requirements and needs, and one group of user may be harder to attract than another. Using the Microsoft Windows example again, the group of end-users have very different requirements and needs to the group of application developers. One set of users is often more numerous or more profitable than the other, so there are many cases where the profit-maximising strategy is to lose money on one set of users in order to drive adoption of the other set of users. Continuing with the Microsoft example, I have noticed that they have continued to reduce the price of their professional development tools. They have even started to provide free developer tools to students and universities, as well as providing free versions of their developer tools that they call their express-editions. I suspect that this is in recognition of the two-sided network effect and that it is critical that applications are developed for Windows to ensure adoption of Windows by end-users.
In some cases, all of the benefits from these valuable networks are realised by the end-users. For example, email has network effects since the value of email increases as the number of users increases, and these network benefits are realised by the end-users only. In other cases, some of the value of the network is accrued to a single company. For example, consider the value of the network effects of Microsoft Windows, Trademe (EBay), or Facebook. The end-users gain benefits as the number of users in the network increases, but the companies that own these technologies accrue some of the benefit. In the case of Microsoft this is the increase in license pricing they can change; in the case of Trademe (EBay) it is the increase in auction success fees they can charge; and in the case of Facebook it is the increase in advertising they can sell. Therefore, these network effects can be very valuable for these companies.
In summary, a lot of new technologies have positive network externalities, referred to as network effects. When they exist they often follow Metcalfe’s law, with the value of the network being in proportion to the number of members squared. So when considering a new technology it is often important to consider whether there are any potential network effects and what this will mean for the future market structure.
