Lesson 5 of 8
Network effects and limits
A network effect is an incumbent advantage, not a law. When aspartame's European patent expired in 1987 the price fell 60 percent and the entrant lost money.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- Network effects sit alongside precommitment contracts, licences and patents, and learning-curve benefits as incumbent advantages that deter entry.
- Most claimed network effects are local: the network that matters to a user is the one in their city or their trade, and a national total says nothing about it.
- A legal barrier and a network effect fail differently, and the aspartame case shows what happens the year the legal one ends.
A network effect is the claim that each additional user makes the product more valuable to every other user. Where it holds, it is the strongest barrier there is. Where it is asserted without a test, it is the most convenient thing an investor can believe.
Four incumbent advantages that deter entry
| Advantage | How it deters an entrant | How it ends |
|---|---|---|
| Precommitment contracts | The business is already committed to the incumbent for years | The contract expires, or the input stops being scarce |
| Licences and patents | Entry is unlawful or infringing | The term runs out, on a date that is public |
| Learning-curve benefits | The incumbent's cost falls with accumulated experience | The learning is embodied in equipment anyone can buy |
| Network effects | The entrant's product is worth less because fewer people use it | The network fragments, or stops being the reason to use it |
The question that separates a real one from a claimed one
Is the network the user cares about the same as the network the company reports. For a payments business connecting buyers and sellers in one city, the user's network is that city, and a company with ten million users spread thinly across a hundred cities has no network effect anywhere.
The test is whether a new user's value depends on total users or on users in some smaller set. If it is the smaller set, then the relevant share is the share of that set, and the barrier lives or dies locally.
What happens when a legal barrier ends
Mauboussin, Callahan and Majd give the case. Monsanto, through NutraSweet, and Holland Sweetener Company were two producers of the sweetener aspartame. After the patent on aspartame expired in Europe in 1987, Holland entered the market to compete. The competition drove the price of aspartame down 60 percent, and Holland lost money.
Two lessons sit in that one paragraph. A patent has an expiry date, which makes its ending the most predictable moat failure there is. And entry does not require the entrant to succeed: the incumbent's economics were damaged by a competitor who was losing money doing it.
The advantage that never existed
The hardest case is the network effect that was really something else. A product that grew quickly and holds its users may be held by habit, by switching cost, or by nothing but the absence of a serious alternative so far.
The distinguishing evidence is what happens to a user who leaves. Under a real network effect, a departing user loses access to other users and pays for it. Under a claimed one, they lose nothing except the inconvenience of change, and the barrier is a switching cost that a competitor can subsidise away.
Check yourself
What distinguishes a real network effect from a claimed one?
- The number of users
- Whether a new user's value depends on total users or only on users in a smaller relevant set
- Whether the company says it has one
Show the answer
Answer: B.
If the value depends on a local set, the relevant share is share of that set, and a large national total describes no barrier at all.What happened after the aspartame patent expired in Europe in 1987?
- The incumbent's price held and the entrant withdrew
- Holland Sweetener Company entered, the price fell 60 percent, and Holland lost money
- The two producers merged
Show the answer
Answer: B.
An entrant does not have to succeed to damage an incumbent's economics. The price fell by 60 percent while the entrant was losing money.What evidence separates a network effect from a switching cost?
- The growth rate of users
- What a departing user loses: access to other users under a network effect, only the inconvenience of change under a switching cost
- Whether the product is free
Show the answer
Answer: B.
A switching cost can be subsidised away by a competitor. Lost access to other users cannot be, until the competitor has the users.Sources
- Mauboussin, Callahan and Majd, Measuring the Moat, Credit Suisse, 1 November 2016: incumbent advantages including precommitment contracts, licences and patents, learning-curve benefits and network effects, and the aspartame case following the expiry of the European patent in 1987.
What this leaves out
- The aspartame case is reported as that paper sets it out. This academy has not read the underlying sources it cites.
- Both tables use invented round figures and are labelled illustrative.
- Nothing here is advice about any security.