Prospects for Deregulation in Telecommunications Mildly Revised Version, May 30, 1997 Joseph Farrell Professor of Economics, UC Berkeley and Departing Chief Economist, FCC 1. Overview On May 9, I gave a speech at the FCC on prospects for deregulation in telecommunications. This document is a re- organized and otherwise slightly revised version of that speech. It's also my valediction to the FCC as I hand over to my distinguished successor, Michael Riordan, who arrives on June 2. The FCC and state regulators have spent a great deal of time and energy in the last fifteen months (and before), restructuring regulation to make it more compatible with competition. To summarize very briefly, regulators have formulated policies for "arbitration" of interconnection between carriers if voluntary negotiations fail, with the goal of facilitating competition through allowing entrants to share incumbents' networks; they have moved towards restructuring "universal service" subsidies to make them more compatible with competition; and they have initiated changes in how incumbents charge for their networks and services that bring entry incentives more closely in line with efficiency. Yet we must all be aware that these initiatives, while important steps towards the possibility of a de-regulated competitive environment in telecom, are not themselves deregulation. How and when, then, do we actually get to deregulate? I don't have the answers, but I want to do what I can to bring the questions to the top of everyone's agenda. Unfortunately, the process of deregulation will be controversial, contentious, and complex. The level of carrier-to-carrier cooperation demanded by telecommunications, and the forces that absent sharing and cooperation would create a natural monopoly, mean that some forms of regulation promote rather than interfere with competition. Also, "the state of competition" in telecommunications is a highly multi- dimensional beast. Finally, (in my view unfortunately) the history of regulation has created what amount to political constraints on deregulation. Thus deregulating will be a complex problem. My incomplete and, I stress, tentative ideas today are based on the following organizing principles: A. Deregulation is justified even under quite imperfect competition. In assessing how imperfect, one should weigh any short- term (or for that matter long-term) price or other market power inefficiencies that deregulation would allow, against the likely benefits of deregulation, broadly considered. In this balancing, we should remember that deregulation may cause firms to improve efficiency in ways that are hard for anyone to imagine ex ante, so, although this advice is almost impossible to take, we should weigh the unspecified as well as the relatively predictable benefits. (In abstract principle, indeed, the more predictable benefits might be achieved through regulatory reform rather than deregulation.) Although telecommunications is unique in many ways, I also think it is salutary to contemplate analogies from other industries, to avoid being overly influenced by the telecom-specific tradition of pervasive regulation. In a society in which we choose not to regulate Microsoft's prices, we should keep asking ourselves "is this regulation really necessary?" Sometimes the answer will be "yes," but it's important to ask the question. B. This balancing should consider not only current competitive conditions, but also the effects on the growth of competition of adopting a particular rule for when to deregulate. If deregulation will enhance incentives for efficient entry, then in order to get the greatest benefits, it may be necessary to commit in advance to a rule that will seem somewhat "too deregulatory" in each instant application. C. While the early growth of efficient competition will be greatly enhanced by the rights of entrants to share the incumbent's network, it will be important to have policies for deregulating this sharing too. In assessing such policies, our views will be colored by which of two appealing interpretations of regulated sharing we adopt: should sharing be seen as an acceptable, even desirable long-term solution, or should it be seen only as a stepping-stone to facilities-based competition and (relatively) full deregulation? D. Despite our progress in reform of "universal service" subsidies, one serious barrier to deregulation will be the culture of entitlement to broad subsidies that encrusts our telecommunications policy. It will therefore be crucial to reduce the scope of that culture and those subsidies. One likely strategy may be to start by deregulating "new" services, to wall them off from the culture of entitlement. Again, proper consideration of long-run effects may imply a rule that would seem somewhat "too deregulatory" when examined narrowly in any particular application. I stress once more that these "principles" are tentative, and I mean them to start a lively public discussion rather than to be enshrined. Below are some halting steps towards exploring their implications and possible implementation. Writing this revised version, even more than the original speech, makes it very clear that there are a vast number of dimensions to explore. The real bottom line is meant to be an exhortation to the telecom policy community to explore some of these, rather than focus narrowly on current FCC actions and other "news." 2. Some Thoughts About Balancing (Principle A) Why is deregulation desirable? If regulation helps keep prices close to cost, and stops users being gouged by firms with market power, isn't that good? Does that mean that deregulation should just be a matter of "delete needless rules" -- eliminating regulations that are redundant because competition is a tighter constraint, or about as tight a constraint, on price gouging? In short, no. I believe, on the contrary, that deregulation should be undertaken even where we expect some adverse short-run (indeed, even long-run) price impact. Of course, this is a matter of degree and judgment, but we should keep in mind that even where regulation holds prices nearer efficient levels than would result from deregulation, it likely has other, undesirable consequences. This merits some explication, because economists have long been pushing regulators to bring prices closer to costs. Am I suddenly saying this is unimportant? Not at all. If we assume that prices are to be regulated, that's the right principle to aim at (although certainly paying very close attention to cost incentives as well as price/cost comparisons). But economists' other message to regulators, which probably is more important if the two conflict, is that we should encourage competition, even quite imperfect competition, and allow market forces to work even if the result in the short term is prices considerably different from costs. Unregulated competition is typically better than regulation in many ways that have little to do with simple price levels. It's worth mentioning some of these ways, because if deregulation at a particular point in time would let prices rise somewhat, we must balance the bad effects of such a price rise (bad, that is, if price does not start out below cost!) against the good effects of deregulation. I won't try to give you a complete list, but here are a few. Ü Unregulated competition does subtler things on prices than bring price levels to cost in the long run. It allows firms to cover common costs in creative and flexible ways. More generally, it lets firms experiment to find how customers prefer to pay the costs they incur. (Regulators can use economic principles to predict what pricing structures should be efficient, but in the end efficiency should be measured by what customers actually want, not what we predict they will want.) And it lets prices -- both as consumption signals and as investment signals -- move at least somewhat in tandem with the first-best ideal, which, to oversimplify somewhat, is short-runmarginal cost when there is plenty of capacity, and capacity-filling price, perhaps well above cost, when there is not. (Peak-load pricing is an example of this kind of pricing.) This is much subtler pricing behavior than "keep prices near long-run cost", which is probably the best that regulators can realistically do (trying to implement the first-best ideal would likely be terribly demanding in terms of information, and extremely subject to manipulation). The two coincide in long- run equilibrium, but transitory differences are likely to be important in industries in which investment and capacity costs are important and demand is somewhat unpredictable. Pricing at long-run cost pays forinvestment, but doesn't give the sharp signals "invest all-out in capacity" or "don't invest in capacity", with their high- powered incentives, that the unregulated market can give. Ü Incentives to minimize costs are surely much better under price caps than under old-fashioned regulation, but even price caps without explicit sharing are not the same as having an unlimited upside and a downside limited only by bankruptcy. Moreover, even price cap regulation may create problems of quality choice or of non-price discrimination that are absent under unregulated competition. Ü Both by improving incentives and by removing administrative/legal hurdles, unregulated competition (or even unregulated monopoly) encourages more rapid innovation. The incentive effects on regulated firms come from (probably justified) skepticism about a regulated firm's ability to keep the possible large rewards of successful innovation; although this can be mitigated if regulated firms can treat development expenses as cost, the resulting incentives are at best diluted. As to the hurdles and "may I?" problems created by well-meaning regulation, the long delays in introducing cellular and other wireless services are a painful reminder. Ü The smart and highly-paid lobbyists, and the smart and under-paid regulators, could do something more productive, and legislators could turn their attention to our society's serious problems rather than argue endlessly about who will or won't pay a few bucks more or less a month. With that encomium for deregulated competition, let me bring up (as suggested in principle A) the natural big question: Why not go all the way now? An immediate deregulatory option Most of what I'll say today suggests a relatively gradual path to deregulation. But some people argue that we should simply deregulate everything except call termination, accept that some prices would almost certainly rise dramatically as the regulated local near- monopolists become unregulated local near- monopolists, and watch facilities-based competitors build out as fast as they can, to try to get in on the big profits before they're competed away, and in the process of course compete them away. There is a certain appeal to this picture, perhaps especially for economists. I think this appeal to economists is only partly because economists are more alert to the benefits of deregulation; it's also partly because economists by and large are fairly prosperous. Less prosperous people, and those less committed to the merits of deregulation, might take a less philosophical attitude to a short-run doubling of their phone bills, even if it speeds innovation and reduces costs and prices in the longer run. I therefore suspect that full-scale deregulation now is not an option. Moreover, the argument as stated forgets that the right to share the incumbent's network facilitates even a primarily bypass-oriented strategy by non-incumbents: more on this below under Principle C. However, I firmly believe that we should keep this immediate approach in mind, because at some point it will be the time for this approach -- and as you will see at the end of this talk, I think it may already be time for a limited version. Perhaps even more importantly, if we keep the immediate option in mind then we are closer to a world in which the default is deregulation, in marked contrast to the world of telecommunications we know. 3. Incentives for Entry (Principle B) One element of the argument sketched above for immediate deregulation is the resulting strong incentive for entry. One might naturally ask whether this incentive can be achieved equally well but with less by way of gouging meanwhile, if deregulation follows rather than precedes facilities-based competition. In general, economists have argued that incentives for entry depend on entrants' expectations of incumbents' post-entry prices, not (directly) on what happens before entry. This would suggest that pre-entry regulation of the incumbent would be irrelevant for entry incentives. This view is somewhat oversimplified. In particular, it ignores the possibility -- probably important in telecommunications -- that entrants may be customers (notably long-distance companies, who are LEC customers in access services), who may enter so as not to continue paying high "pre-entry" prices. It also may be optimistic about how promptly deregulation would in fact occur after competitive entry. Nevertheless, if regulators commit strongly and publicly to prompt (partial or full) deregulation in response to the arrival of competition, then incentives for such entry may be much enhanced. So would be incentives for incumbents to refrain from trying to block entry. Finally, the same would apply not only to deregulatory responses but also to other responses that would slow the regulator-enforced downwards trend in telecom prices, such as a drastic cut in the price-cap X-factor. Of course, "the arrival of competition" is unlikely to be a sharply defined event. Rather, competitive conditions will improve gradually and at different speeds in different places and for different kinds of customer. Thus, decisions must be made about how much competition for how large a fraction of customers warrants how much deregulation in response. These will not be easy decisions. The upcoming "Flexibility Order" in the FCC's access reform docket will have to address some of these questions. Others must be addressed in the context of section 271 applications, where the Department of Justice has stressed irreversibility of competitive conditions. 4. Deregulatory Benefits of Entrants' Rights to Share Incumbent's Network; and Deregulating Those Rights (Principle C) In this (longer) section I briefly explore two large and important topics. First, what are the direct deregulatory benefits of such sharing rights? And second, when and how might we deregulate such sharing? Clearly there is a tension between these two discussions. I describe two views of the role of regulated sharing, one of which stresses the benefits and the other of which stresses eventual deregulation. 4A: Retail deregulation based on wholesale regulation Smoothly functioning wholesale regulation (sharing the incumbent's network, including the so-called platform -- note that service resale does not in itself help here) permits and indeed almost demands retail deregulation. If multiple providers can compete for a customer's business and promptly supply it at a reasonable overall cost, even if they do so by leasing the incumbent's facilities, then it would seem that prompt deregulation of all charges to the provider's end-user will be appropriate. If a carrier tries to charge too much overall to the end-user then another will undercut, and by hypothesis this can happen quickly. If a carrier tries to charge a reasonable amount overall but in an inefficient manner, then another carrier can offer a more profitable alternative pricing package that is also better for the end-user. There should be no need for regulators to resolve the difficult issue of "how" end-users want to pay the cost of service -- how much in flat charges, how much in usage charges, how much for special features, etc. Indeed, if regulators continue to regulate the incumbent's retail prices, and don't happen to replicate the solution that the incumbent and the customer jointly find most beneficial, it puts the incumbent at an artificial competitive disadvantage. Thus, while there are obvious risks in premature deregulation of incumbents, there are also risks in waiting too long. Getting the timing just right will be a great challenge for regulators -- the FCC and the states alike. Long and variable lags will cause problems. I worry that the information flow and the organizational structures are not optimized for this kind of challenge, and I'd like to suggest we should rapidly try to improve them, although I don't today have specific suggestions for how. Another major problem will be that in general provisioning of the platform or network elements will not go suddenly from "slow" to "immediate," so there may be an intermediate period in which either or both of the risks from wrong timing will be inevitable. What about access charges in this world of platform competition? It seems logical that in such a world, originating access charges can be deregulated to the extent that they flow through causally to the end-user: that is, to the extent that if a LEC charges more to originate traffic to an IXC, the LEC's own customer pays the extra. This effectively turns them at the margin into an end-user charge. However, there are at least two potential problems with this. First, today in the long-distance industry firms do not vary long-distance charges according to the originating access charges imposed by the calling party's LEC. If a LEC takes that as given, then it will have every incentive to increase its originating access charges, since the increased charge is in effect socialized and paid by all long-distance customers. If the averaging were only a contingent custom, then we could at least hope that any LEC who charged much too much would find the custom changed. But, depending on interpretation, section 254(g) of the Act may make this hard. Thus it is not self-evident that originating access charges could be deregulated if we want to preserve "equal-access" competition. Of course, the surplus from high originating access charges will be part of the competitive reward to winning the customer, so it need not weaken "full-line" or "one-stop shopping" competition. Moreover, any individual long-distance company can compete for the customer and offer low long-distance prices as part of the package, so increases in "originating access" in this sense need not lead to high long-distance charges. Thus it may be that the only net effect would be that "equal-access" competition was displaced, not supplemented, by one-stop-shopping competition. As one-stop competition develops, how important should equal-access competition remain? Terminating access charges, like other termination, may be a different matter, because carriers may compete to sign up customers and then be able, to some extent and perhaps to a great extent, to gouge originating carriers. If the originating carrier passes any "gouging" through to the originating caller, this may be no problem -- it's like making your phone number a 900 number (a tempting option for dealing with telemarketers, perhaps). Similarly, termination charges assessed (as today in CMRS in this country) on the call recipient are an end-user charge and could be deregulated in line with other such charges. But shifting to these arrangements in wireline would demand institutional and perhaps technological changes. Whatever the details of the treatment of access, etc., the broad picture remains that retail could be deregulated in reliance on wholesale regulation. How valuable is this form of (semi-) deregulation? I don't think we know yet, but tend to believe that it is unlikely to be a fully acceptable solution for the long term, although we should not abandon it hastily. When might we reasonably abandon it, and deregulate the "wholesale" aspects of the incumbent's business? This important question has received too little attention. 4B: Sharing and Wholesale Deregulation: Two Views and Two Possible Triggers 4B(1): Two Views of Network-Sharing There are two appealing interpretations of the point of regulatory sharing provisions, and which interpretation you favor will color your views of de-regulatory proposals. Perhaps the two are best described by starting from the observation that competition requires that the incumbent's bottleneck facilities be bypassed or shared. In the first interpretation, regulation should strive to facilitate competition that involves efficient sharing by competitors of any or all parts of the incumbent's network, because there is no point bypassing the bottleneck if it can more efficiently be shared. In this interpretation, it's quite possible and quite OK if (for some parts of the network, in some areas) there is never bypass: it will prove that sharing works just fine, and save society the very large costs of bypass. If there is bypass, in this first interpretation, it should be driven not by the need to bypass merely in order to compete, but by the need to bypass in order to do things that sharing cannot or will not do. Thus bypass will be driven by the urge to innovate in ways that the incumbent's network cannot allow, and by the inefficiencies of regulated sharing, such as provisioning problems that regulators can't cure, or overpricing, or underpricing if it makes the incumbent's network fall behind the technological frontier, or by non-incumbents' simple reluctance to rely on a competitor, notwithstanding regulatory protection. This interpretation in its strong form takes no position on whether bypass or sharing, or what mixture of the two, is the desirable end result; it aims to let "the market" decide (element by element, and area by area) between unregulated bypass and regulated sharing. Moreover, to the extent that bypass will be the solution in the long run, it lets the market decide when, without meanwhile closing off all competition. In particular, we might not want competitors to overbuild now if the costs of overbuilding are falling very rapidly. (Some have suggested that this is the case in cable telephony or in fixed-wireless local loop service.) In this perspective, if competitors delay buildout because they hope to share the incumbent's facilities at least for a while, that is perfectly fine. In the second interpretation, sharing is not a solution but a stepping-stone, because only bypass will enable (more or less) full deregulation. In this view, the rationale of the sharing rules is that bypass might never happen, or might be slower to arrive, if it had to happen all at once, so sharing is encouraged as a transition or as a fill-in to facilities-based competition or bypass. This interpretation rests on a grand judgment that bypass and (more) deregulation, not sharing, is the desirable end result. In this view, if the right to share delays or discourages bypass it is a bad thing, because only bypass can bring true deregulation. This second interpretation is a more radical departure from the old "natural monopoly" theory, which supposed that bypass was generally undesirable. And, of course, since bypass will indeed permit much more deregulation than sharing, it may seem more appealing to those who think regulation is most inefficient -- although this reaction may be too simple, in that the more inefficient is regulation, the stronger the incentives to bypass even under the first interpretation. I think we should not irrevocably choose between these two interpretations a priori, but rather should watch how competition with regulated sharing evolves. Does it come to be an effective means of having vigorous competition without inefficient bypass? Does it constitute a crucial stepping-stone to efficient bypass? Or, does it tame competition through an incumbent firm's discretionary gift of unenforceable cooperation? Does it produce inefficient rent- seeking by entrants more interested in getting underpriced leases than in efficient competition? Whatever the answers, the tension between the two interpretations will infuse our thinking about deregulation of the terms of sharing the incumbent's network. 4B(2): Two Possible Triggers for Wholesale Deregulation Although allowing regulated sharing of the incumbent's network has a variety of good properties, it is not without problems and will demand continuing regulation. It therefore seems very important to ask when and how we can deregulate such sharing. Let's start with two observations. First, when a bottleneck is bypassed, it's not a bottleneck any more, although duopoly (two competing providers) is distinctly imperfect competition. Second, while pricing an incumbent's network at long-run forward- looking cost, correctly implemented (including full economic depreciation), pays for the incumbent's investments, it does not provide high-powered incentives for incumbents to invest or innovate. Thus, investment incentives are probably better if regulators do not require unbundling of an incumbent's network innovations immediately on cost-based terms. If I may play lawyer for a moment, I note that Section 10 of the Communications Act, as amended, authorizes the Commission to forbear from applying statutory provisions in light of competitive conditions, but this does not apply to section 251 "until fully implemented." However, section 251(d)(2) tells the Commission, in choosing what network elements should be unbundled, to consider, at a minimum, whether unbundled access to "proprietary" network elements is "necessary," and whether failure to "provide access" would impair competitors' ability to provide service. I won't pretend to be lawyer enough to untangle all this, but just note that there seems to be scope, as there ought to be, to consider the competitive implications of requiring or not requiring unbundling. With those ideas in mind, let's think about two situations that might perhaps warrant deregulation of the sharing of an incumbent's facilities: bypass by a competitor, and bypass or upgrade/overbuild by the incumbent itself. Bypass by competitor Suppose a competitor bypasses -- say -- the incumbent's loop to my house. Then that element is no longer a true bottleneck: there is an alternative way to get calls the last mile to and from my home. Should this trigger deregulation of sharing of the incumbent's loop? Here are some arguments suggesting a positive answer, based on competitive analysis once bypass is achieved: -- There is now a second firm that can compete to serve me without leasing the incumbent's loop. Thus, even if leasing does not work at all without regulation, competition (even rapid competition) is not entirely foreclosed. -- A third firm that needs a leased loop can plausibly get one without being wholly at the mercy of an incumbent monopolist. It has two possible sources from which to lease a loop. Duopoly is very different from perfect competition, but it is also very different from monopoly. -- If I the customer can be served using only one loop, then one of the two loopy firms likely has an idle loop to my house, so the third firm might be able to get a really good deal (although the entrant would be unlikely to build if it thought that it would end up leasing to a third party at below long-run cost). -- The bypass is some evidence that further bypass would be perfectly possible, so it's less likely that the third firm is somehow subtly blocked (for instance through control of rights of way) from doing its own bypass. What about ex ante incentive effects of a policy of deregulating after bypass? Such a policy would reward the incumbent for cooperating (for instance, in rights of way, in co- location, or in number portability) with bypass. This is clearly good under the pro-bypass interpretation of the Act. It has good aspects under the sharing interpretation too, although this is a little less clear, since it might also give the incumbent incentives to "encourage bypass" by hampering sharing. Turning to non-incumbents' incentives, -- the entrant has stronger incentives to bypass when it knows that, should it do so, the incumbent will no longer have to lease its loop at a regulated rate: the "post-entry pricing" effect I mentioned earlier. -- With multiple (potential) entrants the rule might even create something of a race to bypass, since building a third loop will presumably not be very attractive. Is this good? Clearly it is if we want to drive bypass; it is less clear how to evaluate it if we want to preserve efficient sharing but also efficient bypass. For these incentive effects to work at full strength, deregulation would have to come promptly upon bypass, raising the same timing questions mentioned earlier. Naturally, implementation of any such policy contains challenges. For instance, -- does the possibility of CMRS bypass count? One might argue that Congress evidently thought not in February 1996 (it instructed regulators to stand ready to regulate sharing if voluntary negotiations fail), I suppose because wireless prices (and quality) are still not really competitive with wireline. But then how much must wireless capacity and/or prices change before it should count? -- Does the existence of coaxial cable that potentially can be, but for the most part currently is not, used for telephone service, count? -- How should our answers depend on the duopoly behavior we observe in such cases? What if the two loop-owners manage to sustain near-monopoly prices? (In the long-distance industry, capacity is leased to at least some resellers at or near competitive prices, but there are more than two facilities-based carriers.) Again, much seems to depend on our balance between the two views of the Act's share-the-network provisions. If we aim to make sharing work as well as possible, we should be concerned that (third-firm) entrants be able to share an existing network at cost-based prices even in the long term. If, on the other hand, we view interim sharing as an aid to gradual development of facilities-based competition and deregulation, this may seem a wise place to consider deregulating, for two intertwined reasons: deregulation makes sense once there is bypass, and also, as argued above, such a policy encourages bypass. Bypass (overbuild) and divestiture by the incumbent My second possible trigger for discussion is bypass/upgrade by the incumbent. A year ago in a speech here [since published in Federal Communications Law Journal, Nov. 1996] I drew some analogies between advantages of incumbency that are traditionally viewed as sources of "natural monopoly" on the one hand, and intellectual property on the other. The interconnection provisions of the Act require incumbents to share those advantages, just as intellectual property policy requires patent- holders to share their intellectual property after the patent's term expires. And, just as in intellectual property policy, this sharing is only part of the right policy: reward to investment must also be given great weight. This perspective suggests that when an incumbent creates a "new" network element, perhaps the new one need not immediately be made available on regulated terms, provided the old one remains available to non-incumbents. I have in mind, for instance, the case where the incumbent builds new high-capacity loops to some of its subscribers, leaving the old copper in place (and in good shape); or where it installs a new switch and leaves the old one "co-located". How would we ensure that the old facility is indeed maintained and made available for competition? One possible answer, less regulatory than today's provisioning and maintenance requirements, is that the incumbent could divest the old facility -- either to a competitor or to an entrepreneur who could lease it to competitors. Such a policy would enhance an incumbent's incentive to make such investments, and to leave the old, potentially competing investment in place rather than rip it out. On the other hand, it may give us pause to note that if the improvement is important enough, it would re-create a kind of bottleneck market power, perhaps so great that the old element would hardly constrain the incumbent's pricing. This would be a big step away from efficient sharing. And if (for whatever reason) the incumbent has a strong first-mover advantage in such overbuilds, the policy might tend to re- create the "natural monopoly." Thus it would be worth exploring whether the exemption from unbundling might be temporary, like the patent- holder's exemption from sharing. Of course, the same tradeoffs apply as in intellectual property policy generally: the longer the period, the better the incentive for this innovation but the greater the potential market distortions during the exemption. And if truly workable competition arrives before the exemption period expires, then sharing presumably need not be imposed after all. Such an exemption policy would also affect non-incumbents' incentives. Leasing unbundled elements might become viewed more as a stepping-stone to innovative facilities-based competition, because a carrier who tries to rely permanently on the incumbent's facilities would risk being overbuilt out of business not only by other competitors but also by the incumbent. In other words, it may fit better with the "stepping- stone" interpretation than with the "efficient sharing" interpretation; the latter would lead us to worry somewhat more -- though not to the exclusion of concern about the incumbent's incentives -- about the effects on entrants' rights to share the overbuild. 5. Limiting the scope of subsidies (Principle D) In telecommunications, some end-users currently are charged below cost --- in some cases much below cost. This kind of entitlement creates competitive problems if those subsidies are funded by implicit cross-subsidies from other users who pay above cost, or if they are funded explicitly but not all competitors are equally able to receive the subsidy. As Professor Lawrence White has said, "cross-subsidies are the enemy of competition, because competition is the enemy of cross-subsidies." Congress addressed this problem in section 254 of the Act, instructing regulators to produce a system of explicit subsidies. This week the Commission adopted a plan for such a system. It will go a long way towards addressing the problem, compared to today's system of implicit subsidies. It will make it possible in principle for competition not to threaten desired subsidies, and will thus make it possible for those who desire the subsidies not to fight competition. Nevertheless, it is probably impossible for regulators to identify and make explicit allimplicit subsidies (and would be damaging to try). Even within a relatively small high-cost area such as a census block group, customers will differ greatly in how profitable they are to serve. Thus, as long as incumbents are subject to carrier-of-last-resort obligations, they will have colorable claims that competition is at least partly inefficient "cream-skimming." In an apparent attempt to resolve this problem, section 254(e) states that only a carrier declared "eligible" under section 214(e) may receive the subsidy, and section 214(e) says that eligibility requires a carrier to offer service throughout a state-defined "service area." These two subsections protect these remaining implicit cross-subsidies by restricting the forms of competition that can receive subsidies, thereby discouraging other forms, perhaps including those most threatening to the remaining implicit subsidies. Some kinds of geographic specialization are permitted: those who serve only customers within a single state, for instance, are apparently eligible. But other kinds of specialization, such as service offered only to internet users, are ineligible. I see this as a compromise between allowing unrestricted competition and avoiding cream-skimming. As such, it inevitably compromises unrestricted competition to some degree. I am not arguing here that the subsections (e) are unwise. Although competitively troubling, they do try to address what could be a real problem, given the problems of carrier-of- last-resort status. Rather, my point is that such problems are difficult and their solutions will inevitably often be competitively troubling. Therefore I think it will be important in moving to deregulation that these broad subsidy entitlement programs be limited. Politically, however, some entitlements may be so entrenched that a strategy that relies on eliminating them will fail. Accordingly, I want to explore the idea that we can say at some point, "The subsidy buck stops here." By this I mean that in defining subsidized services, we should be rather slow to include "new" services, and this will aid in deregulating not only those services but also, as more people move to the deregulated sector because the offerings are more appealing, it will help even in limiting problems in the old, regulated sector. Moving from this rather vague notion to specifics, I'd like to propose that there is no need to subsidize second lines. A regulatory version of this proposal would involve moving the (still regulated) prices for second lines "to cost." I think that would be a very valuable step in the right direction. However, in part because many households are wired as a matter of course with two (sometimes more) lines even if only one is initially used, it is somewhat unclear what is the right concept of "cost" of the second line. To establish conceptually a limit on the scope of subsidy, perhaps it hardly matters psychologically what measure of cost is used. But here is a possible test of our willingness to change the presumption of regulation: perhaps we should actually deregulate the provision of second residential lines, establishing a conceptual limit not merely on subsidy but on regulation. This is a little like the "immediate deregulation option" in section 2 above, but much less threatening because it's limited to a part of the market where (a) there is more immediate competitive discipline, and (b) there is less reason to worry about harm to low-income or otherwise highly vulnerable consumers; moreover, I have in mind deregulating only end- user charges, so no issues of prices to competitors should arise. For several reasons, I doubt that the prices charged for such lines would go up very dramatically if deregulated (while first lines remain regulated). For starters, few households must have a second line. For some users, especially those who plan to use their second line for "overflow" conversation (that is, when more than one household member wants to make calls at once), wireless would be an acceptable, and in some ways better, substitute for a second wireline; at low usage levels, wireless is already not very expensive, and wireless companies already claim some success in marketing as alternatives to second lines. For other users, cable modems are or will be very attractive once the cable industry gets its telecom act together. Multi-line households will presumably be among the first to become attractive to competitors. And -- a slightly strange point, but let's be pragmatic -- "cheating" by putting a second line in a second name (thus getting the still-regulated first-line price) would often be easy. Given all this, do we really think that LECs would raise the price to a level where we should be much more concerned than we are about the price of perfume or of microprocessors? This "not much immediate impact" prediction (if widely shared) might make it easier to overcome the objections from those who want to keep their subsidized rates. It's also probably important to do it soon, while second lines are still generally viewed as something of a luxury, so that people feel at least a hint of shame in arguing that they must be subsidized. (Since this last statement has been misunderstood, let me clarify: I do not suggest there is anything wrong with having a second line, and I have had one myself for some time. It is a measure of how far we have traveled down the entitlements road that a proposal to relax regulatory limits on what firms can charge is misinterpreted, even by some business people, as a proposal to impose a Pigouvian tax.) Circling back to the issue raised earlier of economists' two messages for regulators, it is natural for an economist to note that allowing incumbents to increase the price of second lines, which probably have lower incremental costs than first lines and surely have higher demand elasticity, goes against the traditional advice to base prices on incremental costs and to respect the much-maligned wisdom of Frank Ramsey on markups and elasticities. Despite this, I think it is wise policy, because it is a feasible immediate step away from the culture of entitlement and towards the possibility of deregulation, which I believe is more important (as well as more likely to happen) than regulatory respect for Ramsey.