How do you measure competition? I ask because on yesterday's PM programme Lynsey Hooper claimed (15 min in) that Amazon's broadcasting of some Premier League games next season means that "competition is now fierce."

I'm not sure this is right. Amazon will broadcast all ten Boxing Day games, This means it is a monopoly provider of live coverage then: if you want to watch a live match  on that day, you've no choice but to pay Amazon. Similarly, today Sky and BT are monopoly providers of live coverage of particular games. If you get cheesed off with the inane chunterings of Robbie Savage or Michael Owen you can't switch to watching the game on Sky. Nor, in many cases, are games substitutes for each other: West Ham vs Bournemouth is no substutite for the Manc derby, for example.

Yes, broadcasters are competing. But they do so via bidding wars for the right to particular monopolies - much like rail franchises. That's great for Premier League clubs and players who get higher revenues.But it's not obvious that customers get a good deal. It's quite possible that by the time you've paid Amazon, BT and Sky you are paying more to watch football than you would if Sky broadcast all games*.

The point here is one that competition economists have known for some time - that you cannot necessarily measure competition by the number of firms in an industry: the Herfindahl index, for example, can be a bad measure.

One reason for this can be that industries are actually hard to define. If we think of Premier League broadcasting as an industry, there'll be five firms in it from August 2019; three live broadcasters, plus BBC showing highlights and BBC and TalkSport offering live radio commentary. But if we think instead of live TV broadcasting of the Liverpool vs Arsenal game, there's only one provider.

This can matter. It doesn't much matter if there's monopoly in one industry if that industry faces competition from another. if the tinned pilchard industry wre a monopoly we wouldn't worry if it faced competition from sellers of tinned sardines or sild, but we would if it didn't. Similarly, if fans were happy to wait to see only highlights of a game, monopoly provision of live coverage would not be so bad. More seriously, rail monopolies would be more tolerable if buses were a decent substitute for rail, or if it were easy for commuters to respond to bad service by moving nearer to work.

A single provider can be acceptable if the market is contestable - that is, if there are potential entrants or substitutes. Equally, more providers needn't be evidence than customers get a better deal. For example, a market of (say) ten sellers in which switching is difficult can be less competitive than one of (say) five providers where it is easy.

But if you cannot measure competition by the number of firms, how can you?

One alternative, proposed (pdf) by Jan Boone, is profits elasticity. The idea here is that if profits fall sharply in response to a rise in marginal costs then the industry is competitive but if they don't then it isn't. A competitive market is one which punishes inefficiency.

Even this, however, has problems. What if marginal costs don't change much? What if there's uncertainty about whether a firm's cost and price rises are permanent or not, which deters potential entrants from entering the market? Or what if there are threshold effects such that customers will tolerate small price rises but not larger ones?

The point here is simple. Competition is hard to define and measure. For this reason, I fear, the word can be misused in popular discourse, with "competition" being used to justify what is in fact a near-monopoly.

* One could argue that multiple broadcasters have led to an improvement in the quality of its coverage, but most of this, I suspect, came when Sky first entered the market in 1992.