Thursday, September 02, 2010

Another question for AAB

In posing some (excellent) questions to the Alliance for Affordable Broadband Computerworld states;

I think we can all agree right now that everyone wants “a competitive National Fibre Backhaul Network (NFBN) platform”, and it is something all parties have put forward plans on.

I'd like to put up my hand as one person who doesn't agree. The basis for my disagreement is the meaning of the word "competitive". The simplest economic justification for competition policy is economic efficiency - this is defined using a standard known as Pareto optimality and refers to a case where "no one can be made better off without making anyone worse off". For the purpose of policy analysis this is usually accompanied by a "compensation principle" which adds the rider "or those who are made better off could compensate anyone made worse off".

But at its simplest the justification is based on a theory that in a competitive market prices are set to equal marginal cost - that is the technical meaning of a "competitive price". One of the core assumptions is that the supply curve "behaves" - most specifically that it is upward sloping. The whole theory struggles once you have production technologies with increasing returns to scale - the marginal cost is always below average cost so marginal cost pricing is loss making. Its more practical limitation is that the high fixed (and sunk) costs means there is a dramatic divergence between short-run marginal cost and long-run marginal cost. In an actually competitive market once the upfront investment is made each competitors short run profit is maximised by pricing closer to short-run marginal cost - which is even more loss making than long-run marginal cost.

The second feature of the production function is that costs are "sub-additive" - in other words all the output can be produced more efficiently by one provider rather than two. This means it is counter-intuitive to duplicate infrastructure on the belief that "competition" will drive down prices.

he example cited by Senator Conroy earlier in the year was to compare pricing per Mbps of capacity on the (monopoly) Perth to Port Headland route to the (competitive) Perth to Sydney route. The analysis completely ignored the fact that the cost to build is flat rate per kilometre, whereas the revenue available depends upon the number of "communicators" at each end of the link.

And ultimately this is the problem for regional Australia. We don't make backhaul cheaper to a country town by duplicating the existing fibre. Worse any investment in duplication ultimately denies other towns access to fibre. Just ask those towns still connected by radio backhaul what their priority is?

Now people familiar with policy debates will think this sounds similar to the arguments from the 1980s about how competition would drive up the prices of long distance calls on thin routes but down on thick routes. The actual outcome has been the elimination of all price discrimination on "national calls" including the erosion of the peak/off-peak distinction. The reason for this distinction is the way calls are sold - telephone calls are sold as a "bundle" - you buy the right to ring everyone not the ability to call different places. The nature of the competition for the bundle has seen the ability to maintain price discrimination break-down - the providers need to average their costs. This does mean that providers make losses on some actual phone calls - a position the accountants always find troubling, the economists fully understand and the marketers simply don't understand.

The difficulty with rural towns is that not very provider has to go there to offer a service. Most importantly the "competitive providers" can choose where they offer service. They do get involved in the "cherry-picking" we were always threatened would be a consequence of competition.

Now my analysis of the backhaul market is open to critique - especially by one of the Cynical Seven who built a business selling backhaul, and another for whom it is one of his few remaining assets. These competing investments have found customers and have in cases been profitable (but don't tell NextGen and IP1 that). The reasons for that vary - not least the fact that the primary provider does undertake a degree of cross-subsidy and doesn't always have all the capacity required - but primarily bcause it is a reluctant wholesaler that is not trusted by its customers. The second reason is that purchasers have also used the competing providers as a way to provide diversity.

However, even diversity is better provided by including all the traffic in a single well designed network.

All this means that the country is not better served by a "competitive backhaul network" if "competitive" means actual cost based prices. The country is probably better served by a structurally separated monopoly backhaul provider that charges the same "price per bit" at every point of interconnection. This is the proposal that was put forward by Axia and that no one seems to really understand.

Now that monopoly needs to be constrained to a "cost based" price - and that's where the loss of the dynamics of competition cuts in. Structural separation alone does not guarantee this, and a regulator suffers from the fact that the quantity actually purchased has a huge impact on what the price needs to be. There are indeed market mimicking mechanisms around this....but now is not the time for this.

Is it too much to ask that we actually check all the assumptions?

Novae Meridianae Demetae Dexter delenda est

No comments: