A decision by the Canadian Radio-television and Telecommunications Commission (CRTC) on wholesale Internet billing became headline news early this year. In late January, the CRTC affirmed its position (originally established in May 2010) that the incumbent local exchange carriers (ILECs) and the cable carriers (cablecos) could institute wholesale per gigabyte overage charges (i.e., usage-based billing or UBB) to independent Internet service providers (ISPs) that mirrored their retail overage charges (though at a 15 percent discount). The electronic protest wave that ensued, featuring a very successful and controversial online petition and mass emails, caught the attention of the government. The Chair of the CRTC, Konrad von Finckenstein, was summoned to appear before the House Industry, Science, and Technology Committee to explain; a motion was passed by that same committee to hold hearings and study the billing practices of certain Internet service providers; and appropriately enough, a tweet by the Prime Minister expressed concern and ordered a review, while the Industry Minister Tony Clement also used Twitter to order a review and indicate that wholesale UBB would not be permitted. The CRTC’s response is yet another public notice on wholesale usage-based billing, with a public hearing scheduled for July 2011.
The furor over wholesale usage-based billing is the fallout from two relatively unnoticed decisions by the CRTC. It is these decisions that created the circumstances under which an obscure and technical tariff issue ends up being front-page news. In my view these two decisions reflect, and are driven by, a fundamentally misguided view and understanding of competition in broadband markets in Canada. It is these decisions that a government interested in promoting and encouraging efficient competition and increasing consumer choice and welfare should have concerns about, not the UBB decision. Indeed the policy decision to ban wholesale UBB is likely inconsistent with insuring that Canadians continue to have world-class broadband service.
In the rest of this article I do three things. First, I explain the role of the two previous decisions that mandate access to the broadband networks of the incumbent providers, with a focus on why mandated access is not just ineffective in creating competition, but harmful for the availability of high-speed networks to Canadians. Second, I explain why UBB is not necessarily a result of market power, but consistent with controlling the costs of congestion on high-speed broadband networks, insuring high quality of service and minimizing network costs. Third, I discuss how the government’s decision to become a champion of the populist revolt undercuts the legitimacy of the CRTC, thereby increasing regulatory risk, which imposes costs on all Canadians by raising the cost and reducing investment in high-speed networks.
In its decision on essential facilities in 2008, the CRTC mandated that both the ILECs and the cablecos were required to continue to offer aggregate high-speed Internet access as a wholesale service. These services provide independent ISPs with both access to a customer’s location and transport to a single point of interface with the network of an ILEC or a single point of interface in a geographic area with the network of a cableco. Given mandated access to access and transport at regulated, cost-based rates, the investment requirements for an independent ISP to provide high-speed Internet service to their customers are servers, a connection from the interface points with the incumbents’ networks to their servers and access to the Internet. The CRTC determined that withdrawal of mandated access to both access and transport facilities would result in a substantial lessening of competition, i.e., that mandated access was having a substantial disciplining effect on the ability of the ILECs and cablecos to exercise market power.
The CRTC confirmed its view of the importance of mandated access to aggregated high-speed access in its matching speed decision released in August of 2010. In this decision it imposed a condition on the ILECs and cablecos requiring that they must provide wholesale options that match all the speeds they provide to their retail customers. The CRTC determined that if equivalent wholesale access was not mandated, the competitive constraint of the competitors would erode as the incumbents upgraded the speed of their retail products.
The cablecos and ILECs have been using usage-based billing for a number of years for their retail residential consumers of high-speed Internet access. The retail-rates of the ILECs and cablecos for high-speed Internet service are unregulated. UBB at retail is understood to be an important tool to manage congestion on networks and efficiently raise revenues to cover the fixed and sunk costs of the network (see below). The adoption by the incumbent carriers of UBB at retail " when coupled with flat-rate, unlimited usage, wholesale tariffs " creates a profitable business model for independent ISPs. The opportunity that retail UBB coupled with flat-rate wholesale creates for independent ISPs to engage in regulatory arbitrage is why they, and their high-use customers, oppose the introduction of UBB at wholesale.
Regulatory arbitrage arises when wholesale UBB is banned because users subject to overage charges when they are retail consumers of the incumbent networks can avoid these charges by instead using an independent ISP that simply resells the incumbent’s service. The customer gets essentially the same service from the same network, but the absence of usage-based charges at wholesale provides a margin to be split between the independent ISP (as profits) and the customer (as lower prices).
The incentive for the customer and independent ISP to engage in this behaviour is much more significant for high-volume users subject to greater overage charges. This provides the independent ISPs with considerable incentive to ”œcream skim,” that is to focus on competing away high-usage and high-revenue customers from the incumbent carriers.
When the incumbent carriers lose a customer to an independent ISP because there is no UBB at wholesale to match its retail UBB, it still provides the same network service, so its network costs do not change, but its revenues are reduced (though the incumbent carrier will benefit from the reduction in avoidable retail costs). To the extent these revenues were required for the incumbent carrier to cover its sunk network costs, the effect is to expropriate the incumbent carrier’s network investment. The reason is that it no longer earns sufficient gross profits to cover its sunk network costs. This is a form of regulatory holdup, which occurs when a firm makes an investment in sunk assets with the expectation that those expenditures will be recovered, but a change in regulatory policy reduces its expected revenue below total costs. The effect of banning UBB at wholesale on the incentives for all network carriers to continue to maintain their networks, let alone upgrade, is counterproductive.
In addition, the effect of significant losses to the incumbent carriers of high-volume customers harms all users because the costs of the incumbent networks will increase. The reason is that if the high-volume users are no longer subject to a cap on usage, they will have an incentive to use the incumbent networks more extensively. This will either push up the network costs of the incumbents if they maintain network quality, increase aggregate costs of congestion or provide incentives for network owners to substitute more heavy-handed measures to control congestion (e.g., traffic shaping). These effects will be amplified if, in response to the effects of regulatory arbitrage, the incumbents abandon UBB at retail.
The previous section argues that with retail UBB, a regulatory restriction banning wholesale UBB creates costs. The government and the independent ISPs argue that without restrictions on wholesale UBB the competitive constraint on the market power of the ILECs and cablecos in the provision of broadband service would be significantly reduced or eliminated. By this they appear to mean that the subsidy provided by regulatory arbitrage " at least in the short run while the incumbent carriers maintain UBB at retail " provides benefits to Canadian consumers in terms of increased competition. Let me strongly suggest that counting competitors is not an appropriate means to gauge the extent of competition, that all competitors are not equally important and that mandating access does not result in substantial benefits to Canadians; indeed the reality is quite the contrary.
The CRTC and the independent ISPs argue that mandated wholesale access creates competition that has a substantial impact on the market power of the ILECs and the cablecos. After all, in most geographic regions in Canada, in the absence of mandated resale, the options available to residential consumers would be service from the telecom incumbent or the cable carrier. Mandated access is a means, in their view, to discipline the market power of these duopolists, providing competitive alternatives to Canadians.
The allegation of market power is typically supported by reference to statistics that show that Canadian average broadband prices are much higher than in most other countries, that Canadian adoption rates are less and that we are falling farther behind. The jump in logic is that this relatively inferior performance must be attributable to market power.
Leonard Waverman and Kalyan Dasgupta have provided a detailed and exhaus-
tive analysis of the commonly used to assess broadband performance. They conclude that ”œsensible” metrics indicate as high if not higher penetration rates per household than the larger Western European nations (which all maintain mandated access policies) and higher average actual broadband speeds than these countries. The CRTC 2010 Communications Monitoring Report shows that Canadian prices for broadband service are typically less, our broadband penetration is the highest, and average measured speed the second highest of all countries considered. The CRTC data on prices for broadband services in Canada indicate an approximately 18 percent decline in their real price over the period 2002 to 2009. The CRTC data also point out the significance of the cable networks in Canada (and the United States). Globally the market share of cable in providing broadband service is 20 percent. In Canada it is over 55 percent.
More fundamentally, the allegation that a duopoly in broadband necessarily means such poor performance that mandated wholesale access is necessarily better is suspect. This assertion assumes that there are large benefits to mandated access and limited costs. With regard to the assertion regarding poor performance and benefits, four points suggest the opposite. Sometimes imperfect competition is better than regulatory intervention.
First, the duopoly analysis ignores the competitive constraint of wireless broadband networks. In most regions there is at least one broadband wireless provider. Indeed the CRTC reports that 96 percent of Canadians have access to a broadband wireless network, while in more heavily populated regions Canadians have a choice of broadband wireless networks. Second, the usual conclusion that follows from the 4 percent market share that the independent ISP resellers have in the provision of wireline broadband service would be that mandated access does not provide much of a competitive constraint on the market power of the ILECs and cablecos. Third, the competition that matters to Canadians is the race between ILECs and the cablecos to invest in so-called next-generation networks, i.e., higher-speed networks. The intense interplatform competition between the local ILEC and cablecos to provide all services (telephony, video and broadband) to a location creates incentives for them to make substantial investments in their networks. The facilities-based network carriers have averaged $5.3 billion per year in capital investment over the period 2005 to 2009 to upgrade their networks. In comparison the resellers have averaged $80 million per year, some 1.5 percent per year of the commitment by (primarily) the ILECs and cablecos.
Relative to the European experience, where there is not as extensive competition for the ILECs from cable networks and a tradition of mandated access, Waverman and Dasgupta document that Canada’s ILECs have higher capital intensity, have invested more per line and have upgraded more of their network (fibre to the node) than ILECs in the larger, more comparable countries in Europe. This despite the higher costs implied by longer distances than their European counterparts. They conclude that investment in next-generation networks in Canada compares ”œvery well” with most other Western countries.
Fourth, the extent to which the independent ISPs can compete is restricted, in terms of both price and product features and innovations, by their dependence on the networks of the ILECs. Their ability to compete over price depends on the extent to which their nonwholesale costs are important relative to the cost of wholesale access. Their ability to introduce product features and innovations associated with access is restricted by their dependence on the networks of the incumbent. The importance of banning UBB for wholesale services to the business model of the independent ISPs is in part explained by these limits.
Finally, the difficulty with mandating access is its effect on investment decisions. Incumbent networks will have less incentive to invest if the benefits of that investment must be shared with entrants. In theory it is possible to set efficient access prices that balance the trade-off between promoting competition on a network and investment in the network. In practice this is impossible and the danger, which appears to be borne out by practice, is that regulated prices will be too low, thus inefficiently favouring competition over investment. The empirical evidence suggests that investment in networks has been reduced in jurisdictions where access is mandated. Hence while the benefits from mandating access appear negligible, the costs can be significant in terms of a reduction in the extent of upgrading to next-generation networks and the pace of its rollout. Moreover, the benefits of retail UBB (as discussed in the next section) may be imperilled if restrictions on wholesale UBB result in the network carriers abandoning UBB at retail.
Retail usage-based billing as practised by Internet service providers in Canada is likely a response to a number of technological and cost characteristics associated with broadband networks. The following analysis indicates that it is consistent with efficient pricing and hence is not necessarily a result of the exercise of market power.
Broadband networks are characterized by economies of scale and scope and require significant capital investment that is sunk. In addition, broadband networks have elements that are often shared by users: total demand for the services of an element is the aggregate of the demand of all users. Congestion is particularly a problem in broadband networks because the shared network elements have capacity limits. When an element is congested, the effect is to delay processing of data, degrading the quality of service provided by the network to all users whose traffic uses that element. The costs of congestion include delays in information transmission and degradation in the quality of information. Delays in receiving time-sensitive information may decay the value of that information or, in the extreme, render the content requested useless. An example of degradation in the quality of information arises in particular with video playback, when congestion results in frozen playback while data is buffered or if some packets are lost (rendering a file damaged). Whether congestion results in delay or degradation there is typically also the additional time cost to users of waiting.
The capacity of a broadband network cannot be adjusted up and down to meet predictable variations in demand. The predictable variations in demand happen before capacity can be adjusted. For instance, capacity cannot be expanded to meet demand in the evening and then reduced when demand falls in the early morning hours. This occurs because installing capacity quickly is prohibitively costly, and since this capacity is sunk it is not possible to reduce capacity and recover expenditures on investment when it is not required.
These specific characteristics of broadband networks have important implications for pricing and investment. The first is the important advantage of using menu pricing to recover sunk capital costs. The second is the role of pricing in efficiently managing the peak-load problem and congestion on the network.
Economies of scale and scope typically mean that if the network priced services at marginal cost, provision of network services would not be viable. By the very nature of their costs, firms that provide broadband service cannot price at short-run marginal cost, as firms in hypothetically perfectly competitive markets would. Instead they must price at levels in excess of shortrun marginal cost, thereby exercising ”œmarket power.” However, it would be wrong to expect that pricing above short-run marginal cost is necessarily an indication of the inefficient exercise of market power. Firms must be able to price so that they recover at least their average long-run cost of production. A trade-off between viability and efficiency arises when a firm is not viable if its services are priced at marginal cost. Raising prices above marginal costs increases the profits of the firm, but at the same time it harms consumers. In particular, as the price rises, consumers substitute other goods, other goods that are not as valuable an alternative.
An alternative is that firms have an incentive to raise revenues by instead using nonlinear pricing. A common form of nonlinear pricing is a two-part tariff, where consumers pay a fixed access fee and a per unit usage charge (i.e., there are two parts to the payment). The use of a two-part tariff is a more efficient means for the firm to recover its sunk costs and break even. It can leave the usage fee closer to marginal cost, thereby encouraging consumption, and use the access fee to extract revenue from consumers with relatively less of an effect on usage.
A menu of two-part tariffs, with a tariff targeted at each level of usage, often further reduces the inefficiency of raising the usage price above marginal cost. Such a menu is characterized by a trade-off between low access fees and high usage charges. This reduces the extent to which consumers with low valuations for the service opt not to purchase access, while at the same time reducing the incentives of high-valuation consumers to engage in personal arbitrage by masquerading as low-value consumers. The high-value consumers are high-value typically because they have a high willingness to pay for usage and as a result are heavy users. Charging a high usage fee with low access fees discourages them from avoiding a higher access fee since it reduces their usage. Hence the firm can charge higher access fees to them and lower usage fees. The effect of the first assists in recovery of fixed and sunk costs, the second contributes to expanding usage.
The firm, instead of discriminating on the basis of quantity, could discriminate on the basis of quality. Higher quality is provided at a higher price, lower quality for a lower price. Those with a high willingness to pay for quality are induced to pay the higher price: avoiding the higher price can be done only by opting for lower quality service, and the lower quality service, albeit at a lower price, is less attractive than higher quality and a higher price for those who value high quality.
One result of using such a menu of service options is that more of the burden of cost recovery is likely to fall on heavy users. Besides typically being more efficient than a uniform price, a menu is also likely to be viewed by many as more equitable. Those who benefit more from the network pay a greater share of its costs. Another important advantage is that under a menu approach it is more likely that those with relatively low willingness to pay will subscribe. The menu approach gives the service provider the flexibility to target an option to these users. If the service provider is constrained to offer ”œone size fits all” service, then it may be priced too high to attract lowvalue users.
Prices also play an important role in both managing congestion on a network and solving the peak-load problem. Efficient pricing of network usage requires that prices signal to users not just the avoidable private costs of the network operator, but also the costs of congestion. The costs of congestion are the harm imposed on other users of the network by the use of a subscriber. If the costs of congestion are greater than the costs of additional capacity, then it is efficient to expand capacity. If the costs of congestion are less than the costs of additional capacity, eliminating congestion is not efficient. Because capacity is constant through time and demand is variable, increasing capacity reduces congestion costs during peak periods, but at the cost of having excess capacity in off-peak periods. Installing less capacity lowers capacity costs but increases the costs of congestion in peak periods.
Congestion pricing that reflects instantaneous network conditions, or that reflects predictable daily variations in demand, is efficient only if the costs of implementation are less than the efficiency gains. If this is not the case then it may be possible to identify second-best proxies that are correlated with congestion, and an obvious candidate is aggregate volumes of a user. If there is sufficient positive correlation between aggregate usage and contribution to peak demand, then aggregate volumes may be a reasonable proxy.
The form of usage-based billing instituted by the major telecommunication and cable networks is consistent with efficient recovery of sunk network costs and congestion management. The plans are differentiated by bandwidth (maximum transfer rates), flat rate, monthly volume cap and overage charge. The lower bandwidth plans have lower flat rates, but their UBB components are less attractive: i.e., lower aggregate cap and a higher overage charge.
As expected there is a trade-off between speed and the flat fee: higherquality service is provided at a higher flat fee. Moreover the UBB components operate as theory would predict. Lower-speed plans have more stringent UBB restrictions, making them even less attractive to high-volume users. This means that the total highervolume users will pay (access fee and total usage charges) can be increased before the cost of their plan becomes sufficient to tempt them to substitute a lower-speed plan.
Given economies of scale and scope, we might have expected that an efficient (or profit maximizing) menu of pricing plans would feature decreasing marginal block pricing. The typical trade-off is to lower the marginal price to encourage usage by raising the fixed fee and/or charging higher prices in initial blocks. The increasing block prices in the broadband plans offered by Bell and Rogers suggest the importance of managing congestion. Increasing block prices are consistent with increasing social marginal cost attributable to congestion as usage increases.
The modern literature on regulation highlights two competing hypotheses regarding the exercise of a regulator’s coercive power. That power is a valuable resource that can be used to further private interests. The traditional hypothesis of George Stigler is that regulators will be captured by the firms they regulate, and regulation will be used to create and protect monopoly rents (e.g., by limiting entry and restricting price competition). A more recent focus has been on the use of regulatory power to hold up regulated firms. As explained above, regulatory holdup occurs when firms make sunk expenditures, usually in capital, which make them vulnerable to renegotiation ex post. The regulator can lower rates to a level that only covers avoidable costs, and the firm will still find it is better off to continue to provide services than shut down, even though its revenues are now insufficient to recover its capital costs. In anticipation of this possibility, firms will reduce their exposure by reducing investment and require higher costs of capital to compensate for this risk. In either case (regulatory capture or holdup) the use of the legal power of coercion distorts economic activity and is not consistent with economic efficiency. The political efficiency of the institution that exercises regulatory powers refers to the extent to which these incentives are restrained. The reason that we typically assign regulatory powers and functions to independent regulators " instead of those powers being exercised directly by the government " is that it is politically more efficient. There are a number of safeguards that reduce the possibility of either regulatory capture or regulatory holdup that can be implemented when regulatory powers are delegated to an independent regulatory agency. These safeguards include the following:
(a) Enabling legislation can direct the regulator to emphasize economic efficiency and competition.
(b) Regulatory processes can be implemented that subject interested parties’ positions to public scrutiny and evaluation. They provide a forum for a public debate that enables explicit challenge of facts and logic, thereby reducing the issues and problems caused by asymmetric information and strategic behaviour.
(c) Written decisions by the regulator can document the explicit alternatives considered and require the regulator to explain its rationale for the alternative adopted and why others were rejected.
(d) It is possible to provide incentives to regulators to be concerned about the long-run health of the industry. It is less likely that regulatory decisions will be driven by short-term political interests, and more likely that a focus on long-run benefits and costs can be maintained. This can be done by having overlapping terms for individual regulators, relatively long terms, and making it difficult for them to be replaced.
(e) Regulatory agencies can harness and develop the relevant expertise, historic awareness and background knowledge to understand, evaluate and adjudicate complex issues. This includes being able to critically evaluate the positions advanced by special interest groups.
(f) Regulatory processes can be designed to be transparent and allow for public participation, thereby contributing to accountability, understanding and legitimacy.
(g) The latitude of regulators can be restricted by allowing for appeals to courts.
The ability to design independent regulatory institutions that are less subject to the twin perils of regulatory capture and regulatory hold up is critical in a parliamentary democracy where Parliament is supreme and the government can change after an election. In these circumstances it is advantageous for governments to develop a reputation not to intervene and overturn regulatory decisions. Everyone is better off when governments are not responsible for regulatory decisions, like the form of wholesale tariffs for Internet service. The clear danger is that the government will respond to political pressures that result in holdup of the incumbent carriers, undercutting their incentives to invest in next-generation networks. While in the short run (some) consumers might benefit, in the long run Canadians will be harmed. The appropriate role of the government is not to legislate outcomes, but instead to design processes that result in efficient outcomes. Two important elements of this are a public vetting process of regulators to make sure they have the appropriate expertise and competence, while making sure that regulatory agencies are adequately funded.
In this article I have argued that there are important benefits of UBB and that without wholesale UBB those benefits are reduced. On the other hand exploiting the opportunity created by banning wholesale UBB appears to be an important part of the business model for independent ISPs. It is critical, however, to understand the relative lack of significance of the competitive constraint provided by the independent ISPs; the importance of interplatform competition between the incumbent network owners (ILECs and cablecos) for creating incentives for investment in faster networks; and the importance of minimizing regulatory risk for encouraging investment and reducing its cost.