News
Common cost allocation - anyone for Shapley Shubik?
(21 Nov 2008, BWCS Staff)Welcome to the regulatory blog from BWCS. At regular (or more likely irregular) intervals we will use this forum to muse about topics we find interesting or curious, or to comment randomly on events in the world of telecoms regulation. The posts will be informal and reflect the diverse views of different members of our team, not a BWCS party line. From time to time we may also post excerpts from other sources to stimulate debate or present a fresh angle we would like to share with our readers. If you want to comment on any of our posts, please feel free to email us at info@BWCS.com
Most people in telecoms regulation will be familiar with such gems as the Efficient Component Pricing Rule, Ramsey-Boiteux Pricing and Equal Proportionate Mark-Up as alternatives to the widely used incremental costing method of allocating common costs in telecommunications networks. However, I was surprised to find that a recent report commissioned by Comreg managed to review no fewer than 10 different methodologies, at least three of which I had never heard of before. Each method has its proponents as well as fierce enemies.
My own pet hate is Ramsey pricing, not just because it is a nightmare to implement in a cost model, but because it is one of those concepts economists tell me will maximise consumer welfare, while at the same time shamelessly robbing me of my consumer choice.
The method sets costs according to price elasticity, so that a higher share of costs is allocated to products and services that are highly price inelastic. According to economists Laffont and Tirole, "It would be absurd (on efficiency grounds) to charge high mark-ups on those services for which consumers are not willing to pay much above the marginal cost. Cost recovery should place a higher burden on those services with relatively inelastic demands." Which, in the old days, translated into raising costs for local access, a price inelastic service people depended on in the pre-mobile era, and reducing long distance charges. To put it more bluntly, Ramsey advocates would advise us to increase the price of insulin for diabetics and reduce the price of cream cakes. That's efficient pricing.
One of the more obscure allocation methods included in the Comreg review is Shapley-Shubik pricing. Originally formulated in the 1960s, the methodology has gone in and out of favour over the last forty years but seems to have made a come-back recently, especially in the Middle East.
The Shapley-Shubik method is based on game theory and has much in common with incremental costing, although it differs in some important aspects. It starts by allocating total costs to a single product, let's say POTS, which costs 500. Then PANS comes along and total costs go up to 800. So PANS has added 300 to the joint costs, but its stand-alone cost is 500, just like POTS. We can therefore directly attribute the 300 cost increase to PANS. If we were to change the order of arrival, whereby POTS was the second product added, this would equally increase total costs by 300, thus we can directly attribute 300 to POTS, too. What is left is the common cost, worth 200, which will be incurred no matter whether we provide a single service or both, and regardless of order. This common cost is split evenly between the two products, so each is allocated fully distributed costs of 400. It gets more complicated when high-speed PANS arrive, followed by NGPANS, but the principle remains the same regardless of the number of products or services within a product grouping.
The method has some merit. It allows full cost recovery and thus encourages investment in infrastructure. By being order-independent, it avoids cross-subsidisation between new and historical services and makes it more difficult to use common cost allocations to manipulate regulatory accounts - not that such a thing would ever be attempted, of course. And some Shapley-Shubik fans even claim it would save endless hours of negotation between the various stakeholders in regulatory meetings!
Why do we not see more Shapley-Shubik in practice? One problem with the method is that you need to know the stand-alone cost of each product. As this can't be observed in practice, we are back in the land of arbitrary cost allocation. Then you need to establish which services share common costs and thus make up a product grouping. This gets more complex the more services you add. Finally, the method is criticised by the Ramsey brigade for ignoring demand elasticities, but maybe that is not such a bad thing.
Back