Monthly Archives: November 2016

Economics and coffee shops – The myth of ‘predatory pricing’


My local

“That’s the Starbucks cluster marketing model. It doesn’t matter if each of the individual stores makes a profit, they run them at a loss so as to distort the market so the independents go out of business”

So said a neighbour of mine about the supposedly high number of Costa coffee shops in the postcode we live in. This is a new outing for the old argument of ‘predatory pricing’. Happily, ‘predatory pricing’ is a myth.

The theory behind ‘predatory pricing’ is that the largest company in a market can reduce its prices to below cost of production and corner the market. To attract business, its competitors will have to match it for price. If their pockets are not as deep, they will go bust. Then, the big company, now a monopoly, can raise prices to whatever it likes, and make good the losses it ran up driving the competition to the wall.

This is a high risk strategy. If a big company, such as Costa, runs at a loss so as to drive the competition out of business, who is to say they will succeed? It might be that even with the low prices people just don’t like their coffee and keep on frequenting the independents in which case Costa will be running losses indefinitely and it will go bust.

In addition to the uncertainty over whether there will be a return at all to an investment in a firm engaged in ‘predatory pricing’, there is no knowing how long it will take to see one.

Then, think what happens when the big company achieves its goal, becomes a monopoly, and raises its prices way above cost of production to enjoy its profits. With a margin of price over cost of production it is now possible for an independent to enter the market and compete. The process repeats itself with the big company slashing prices again to below cost of production in an effort to drive the newcomer out of the market. Indeed, the theory of ‘predatory pricing’ means that, where barriers to entry are low (and there is a guy near my local tube station who sells coffee out of the back of a van), the big company will have to run at a loss permanently. This is clearly unsustainable.

So ‘predatory pricing’ is as much myth as theory and that, in fact, is what we see. Simply put, there aren’t examples of it. It doesn’t happen.

To see this, consider that one of the predictions of ‘predatory pricing’ theory is that the big players in a market will drive the smaller ones out. But in the UK the number of branches of Costa, for example, rose from 658 in 2010 to over 1,500 in December 2015. Over the same period however, the number of independent cafes in Britain rose from 7,000 to 19,000. If major coffee retailers are running their branches at a loss so as to drive independents out of business, the strategy is failing dismally and they are doomed.

There is a second prediction of ‘predatory pricing’ theory, that where a producer secures a monopoly it raises its prices. Again, we don’t see this happening in real life. Indeed, Costa’s prices are set centrally so a latte in a town with lots of independent competition will cost you the same as a latte in a town with none.*

The example of coffee shops is true more widely as well. ‘Predatory pricing’ doesn’t stack up in theory and, in consequence, it isn’t seen in real life. Stop worrying and have a coffee.

* There is a point to be made about Costa specifically here. The Starbucks ‘cluster model’ was based on the fact that each shop was owned by Starbucks. As a result, it could use the profits from one local shop to offset the losses of another, what is known as ‘cross-subsidisation’. That is not the case for Costa which operates a franchise model. The guy who owns the Costa by the train station might not be the same person who owns the Costa by the tube station and it might be a third person yet who owns the Costa in the Mall. In a business sense, these Costa owners are, in effect, as ‘independent’ as most coffee shops commonly described as such. There is no room between franchisees for cross-subsidisation so the ‘cluster’ argument does not apply.