Resale of International Services: How Big an Impact?
by Henry Ergas
Counselor for Structural Policy
Organization for Economic Cooperation and Development
The views expressed in this paper are those of the author and do not necessarily represent those of the OECD.
The persistence of high charges for international calls has led to increased interest in policy options which might increase competition on international telephony and thereby bring call prices down. Allowing firms which lease capacity to resell some o r all of that capacity to final customers is one such option.
At present, most countries - including those which have otherwise liberalized entry into international service provision - effectively prohibit resale:
the countries where services are operated as a monopoly on the grounds that resale would undermine the exclusive privileges of the monopolist;
and those which allow competition on the grounds that resale would be open to abuse by foreign monopolists, who could set up as resellers - either in their own market or at the liberalized foreign end - and use their market power in the home country to distort competitive outcome (for example, by shifting their outgoing traffic from the switched network onto leases, thereby reducing their liability for settlement payments to the foreign carrier - a practice known as 'accounting rate by-pass').
However, this situation is changing, largely as a result of three factors:
Given these pressures, this note examines some of the simple economics of international resale. My primary concern is with inter-continental traffic; but the approach would not differ greatly in analyzing flows within a region. I will seek to highlight some of the factors which affect the economic viability of resale operation, and the assess their likely impact on the major inter-continental streams.
1. Simple Resale
It is useful to begin by considering what is commonly referred to as 'simple resale': that is, a situation in which a service provider (as against a facilities-based carrier) has leased capacity between two countries and uses that capacity to offer swi tched service in one or both directions between those countries.
To offer this service, say from the service provider's home country to the terminating end, the service provider must: collect traffic at home and end transport it to its international gateway; carry out the call switching and management associated wit h setting up, supervising, and closing down the call; transport the call to the foreign end; secure distribution of the call at the foreign end to the called party; undertake billing and other aspects of sales and marketing. Each of these components has a n associated cost, which in some cases depends on the charges set by carriers (notably call collection at the home end, transport between the ends, and foreign delivery or termination), in others on the costs of equipment (notably for multiplexing and com pression, switching, call management and billing), and in yet others on the service provider's strategic choices (notably marketing and general overhead).
It is relatively easy to estimate the broad outline of these cost components:
It is reasonable to assume that - at current world prices - the switching, DCME and billing system capacity required to handle some 200,000 minutes/erlangs/year on 16 circuits (derived by a compression factor of four on a 512k link) would costs around $1 million, translating into an annual cost (assuming fifteen percent real rate of return) of around $375,000. This implies a capital cost per minute of just under $0.12 at each end.
Operating costs per minute are unlikely to be below $0.20-0.25 per call, taking account of limited economics of scale in billing and administration, and higher invoicing and receivables costs associated with the more price-sensitive parts of the market .
Total direct costs would therefore be of around $0.50 per minute before charges for leased transport and for domestic and foreign end reticulation.
Now, effective leased charges for inter-continental streams are unlikely to fall below $36,000 a year (amortizing installation over 5 years) per derived voice circuit, giving a transport cost per call minute of around $0.18. If domestic reticulation we re charged at both ends at the lowest US WATS rates, this would mean a total transport charge of just under $0.50. Even on this unrealistic assumption about termination fees, the total reseller cost per call minute would therefore be of around $1; assumin g - as seems more likely - a foreign end reticulation charge in the order of the average OECD area charge for the longest distance domestic call minute (some $0.40), the total reseller cost per minute would be just under $1.20.
There is clearly some scale sensitivity in these results. By and large, it is reasonable to assume a cost-elasticity for direct costs and inter-continental transport of 0.85, so that a shift to 2Mbit capacity utilization would reduce average cost by a round 8 percent. The reseller might also be able to cut costs by only collecting traffic at locations near its gateways or from customers with direct (ie leased) connections to those gateways. This would allow it to escape domestic reticulation costs, bu t at the expense of economics of scale.
Overall, therefore, the reseller is unlikely to have costs below $1.00 per call minute. These costs can then be compared to the costs and charges of the incumbent carriers.
The direct costs (excluding settlements) to a large, efficiently-managed, incumbent carrier of switched inter-continental traffic can be estimated as $0.20 cents per outgoing call minute. Assuming efficient location of gateways, and high-capacity utili zation on domestic trunks, domestic reticulation charges should not exceed $0.08 per call minute, giving a total cost before settlements of $0.28. With an average settlement of $0.80, this gives an outgoing call minute cost of just over $1.00. However, fr om this average outgoing settlement must be subtracted the settlement payments received on incoming calls (which outgoing calls can be regarded as 'buying'). These payments can be assumed to reduce the effective settlement rate on each outgoing minute to no more than $0.50 (and probably closer to $0.45), bringing the carrier's per minute cost down to less than $0.80.
On this analysis, reseller costs are significantly higher then carrier costs, even taking settlements into account. This implies that there is a welfare loss from resale, except to the extent to which it forces carriers to reduce their charges. How lik ely is it that this will happen?
If we concentrate on the countries where the provision of international service is liberalized - which after all are the only ones likely to allow resale - the answer must be 'only to a very limited extent.'
Taken first the effect of resale - which we will assume originates in one of these countries - on charges between countries with a liberal regime. The fact of the matter is that average collection rates between these countries are already at levels sim ilar to reseller costs, and are likely below these to the largest customers (which obtain volume discounts of some 10 percent off the standard rate). Thus, the call charge from the US to the UK is currently $1.20 a minute before taking into account of opt ional calling plans; that from Australia to the US (also before taking account of optional calling plans) around $1.30. It is difficult to see resellers having much impact on these charges.
The impacts might be greater on streams where the foreign end has a monopoly regime. Outgoing charges from the liberal countries to these countries are typically significantly higher than on streams between liberal countries; allowing resale could ther efore be thought to have some price-reducing effect. Whether this happens, however, will depend on the level of reseller costs.
In effect, leased rates are typically 30 percent or more higher on these streams than on the liberalized streams; and the reseller could expect to face high foreign reticulation charges. There is, in particular, no reason to think that the reticulatio n charges at the foreign end would be set lower than current settlement rates (assuming that the foreign monopolists do not believe that current liberal-end prices are above the level consistent with joint profit maximization). Assuming, for example, a cu rrent average per minute charge from the US to these countries of $1.80, resale might lower the charge to closer to $1.50 - not an insignificant reduction, but one which (given that the resellers have much higher costs) would lead to negligible gains in d omestic welfare.
In short, simple resale will have only a very modest effect on prices and welfare - far smaller that the effect of full, facilities based competition. Indeed, on some of the thickest streams, resellers might not even be viable at current charges, and h ence would not add significantly to the competitive disciplines bearing on the carriers.
What factors might alter this assessment? Four are worth examining.
A first on the cost side is the extent to which a reseller could reduce costs by using International Toll Free (ITF) services rather than operating its own facilities. As matters now stand this seems extremely unlikely. In effect, current ITF rates fro m liberal countries are around or just below off-peak collection charges. Even assuming minimum overheads and in-country reticulation, an ITF reseller would require charges some 5 to 10 percent above the peak time rate.
Second, resellers' costs might be affected by access to return traffic. Thus, were resellers to secure return traffic - say, in proportion to the traffic they originate - the settlement payments associated with that traffic would alter the effective co st to them of terminating traffic overseas. But this effect needs to be seen in its impact in both directions: were the foreign carriers to provide return traffic to the resellers, they (the foreign carriers) would presumably want to be no worse off than in the status quo ante. This implies that the resellers would have to pay the foreign carriers a per minute settlement no lower than the current settlement rate; so that (depending on the reseller reticulation charge and on the balance of traffic in the t wo directions) a resellers' costs might actually rise.
Third, reseller revenues might rise if the reseller offered a re-filing service: that is, took traffic from its country of origin to a final destination via a transit country. The extent of the opportunities for re-filing depends on the degree to which bilateral charges (say from the US to Australia) exceed the sum of charges on the triangular path (from the US to Australia via New Zealand) plus transit charges. These disparities do occur when the originating point is the current monopoly countries, bu t are insignificant as between the group of liberal countries. Resale arbitrage, to the extent to which it occurred, could therefore mainly affect charge distortions in the monopoly countries, in particular forcing some reduction in their charges on rela tively think, very long distance streams (for example, Europe to South America). Whether it does occur will depend both on regulation in the monopoly country, and on the extent to which these arbitrage opportunities, when aggregated up, exceed the minimum market scale required for entry to be viable.
Finally, the economic impacts of resale would be modified were the foreign-end monopolist to engage in one-way accounting rate-by-pass.
Consider a monopoly carrier in country A sending traffic to country B. The A carrier might deliver its outbound traffic to a B country reseller (which it might own) in exchange for a lower effective charge (for terminating traffic in B) than that built into the current accounting rate. This is equivalent to increasing the termination charge for traffic from B to A (since the B country's carriers pay as much as before for each minute they originate, but receive less for the total number of A minutes the y terminate); on most realistic assumptions, this will lead to a welfare loss, since the monopolist is unlikely to cut prices in the A market (its price levels in that market not usually being constrained by the settlement charge), while the prices in the B market will rise (or profits fall, but with a re-distribution mainly to foreigners). It is difficult to believe that this could be viewed as a desirable outcome of policy.
2. Carrier Resale
In short, simple resale is unlikely to add much to effective competition in the liberal markets. If allowed, it could serve to reduce price distortions built into stream-by-stream pricing of the monopoly markets; the net effect of this is likely to be greatest on fairly thin streams (since this is where the price distortions seem to be most acute).
Greater effects may arise from carrier resale: that is, from situations where a carrier in one country uses capacity to which it has access to sell switched services into a country other than its country of origin.
Perhaps the most important form of carrier resale is the resale of 'Country District' service. This is a service in which a carrier from country A acquires what is in effect a toll-free line in country B for the purpose of connecting customers in B to an operator assisted service in A. Customer billing is done by the A carrier; and the call is treated for settlement purposes as an outgoing A call, so that A carrier pays B carrier the settlement rate per minute of traffic. However, the call is usually e xcluded from the return traffic formulas, so that the A carrier may also lose a fraction of the in-payments from B.
Originally intended to facilitate reverse charge and bill-to-third-number calls, Country District service is only provided by carriers, and on terms which are generally only available to carriers. Resale of Country District services occurs when the A c arrier actively promotes the use of this connection by customers who would under other circumstances have used the B carrier to (i) place calls to country A; and (ii) using A as a transit point, place calls to other destinations. Three factors make this e conomically viable, especially between the liberal countries (which would here be the A country) and monopoly regimes (the B country).
First as Table I - drawn from a recent report by the FCC2 illustrates, charges from B to A (and other destinations) now greatly exceed those from A to B (and other destinations), even when the A charge s are for operator assisted service. Given this charge disparity, customers have a strong incentive to be billed at the A rates, even if this involves some inconvenience in terms of longer call set-up time.
Second, thanks to the reductions achieved in accounting rates, settlement rates are still well below A country call charges, especially for operator assisted service. The sum of the direct cost of handling a Country District call plus the settlement ch arge and the loss in return traffic is consequently below the collection rate on virtually all calls. The A carrier therefore has a strong incentive to secure this traffic, especially if the overall traffic it receives from B is low relative to the traffi c it sends to B (so that the proportional loss in in-payment occurs over a smaller base).
Third, the costs of engaging in carrier resale and refiling are coming down as a result of technical change. In particular:
service digitalization and extensive use of common-channel and DTMF signaling makes it possible to automate the provision of Country Direct service - including functions such as account verification, call re-direction and billed number screening - and to undertake complex transit routings with no discernible service degradation;
while at the same time, the rise of global credit card systems provides a basis for low-cost charging, reducing the loss on receivables to normal commercial limits.
The main economic effect of Country District resale is to drive B country charges towards the A country operator assisted rates on direct lines, and towards the sum of A country OAS charges and the A country switched charge on indirect links (ie the ch arge on links from B to C via A would be pushed towards the sum of the OAS charge A-B and the switched rate from A-C). Given that the average difference between these charges (that is, between OAS rates in liberal countries and switched rates in monopoly countries) is around 50 per cent, and that full automation will allow OAS rates in the liberal countries to fall further, the impact on price differentials could be large. In practice, however, B country carriers are not likely to fully adjust their charg es to take account of Country District resale, given that - at least initially - this is not likely to affect a large segment of the market. As a result, the A carriers should be able to secure, at least temporarily, sizable rents from arbitrage.
There will also be effects on international settlements. Country District resale, though unambiguously welfare enhancing, raises outpayments from A to B countries - highlighting once again that a reduction in total outpayments is not a sensible goal o f policy. At the same time, the B carriers might react by either seeking an increase in the settlement rate or obstructing even further its reduction. There is consequently no reason to believe that Country District resale would lead to lower settlement r ates, at least in the short to medium term.
3. Summary and Conclusions
This note examines the impact of resale of international capacity on call charges and settlements. The following main points are made:
Note 1: See, for a particularly strong statement of this view, Telecommunications Resale in Australia , Document submitted by the Australian Delegation to the tenth session of the OECD Working Party on Telecommunications and Information Services Policies, Paris 30 November 1992. Back.
Note 2: See "Calling Prices for International Message Telephone Service Between the United States and Other Countries" A report of the Common Carrier Bureau to the Federal Communications Commission, Washington DC, October 199 2. Back.
Columbia University Working Paper Server Project