Download Jean Tirole and more Slides Public Policy in PDF only on Docsity!
Market Failures and Public Policy
Prize Lecture, December 8, 2014
by Jean Tirole
Toulouse School of Economics, France.
1. INTRODUCTION
Economists have long extolled the virtues of markets. Unfettered competition protects consumers from the political influence of lobbies and forces producers to deliver products and services at cost. Alas, competition is rarely perfect, mar- kets fail, and market power—the firms’ ability to raise price substantially above cost or to offer low quality^1 —must be kept in check. Industrial organization studies the exercise and control of market power. To this purpose, it builds models that capture the essence of the situation. The predictions of the model can then be tested econometrically and possibly in the lab or the field. In the end, the reasonableness of, and robustness to modeling as- sumptions and the quality of empirical fit determine how confident economists are in making recommendations to public decision-makers for intervention, and to companies for the design of their business model.
- (^) This lecture is dedicated to the memory of Jean-Jacques Laffont. It is of course unlike any lecture I had ever given. It is filled with emotion, intellectual indebtedness and very fond memories. The lecture is also very unfair to the community of researchers who have developed industrial organization economics in its modern form. The lecture indeed is in no way a survey, even on its very limited subset of topics, and does not attempt to rec- ognize contributions. Its purpose rather is to use examples drawn from my own research to illustrate the approach and use of theoretical industrial organization. This should how- ever not obscure the fact that modern industrial organization is the outcome of a collec- tive undertaking by a (still vibrant) community of talented researchers.
508 The Nobel Prizes
Industrial organization has a long tradition: first theoretical, with the work of French “engineer-economists” Antoine Augustin Cournot (1838) and Jules Dupuit (1844); then policy-oriented with the enactment of the Sherman Act (1890) and subsequent legislation; then descriptive with the studies of the Har- vard school (“Structure-Conduct-Performance”) comforting and refining the antitrust drive; and finally skeptical with the Chicago school. The Chicago school correctly pointed out the lack of underlying theoretical doctrine and went on to cast doubt on the whole edifice. However, it did not develop an alternative anti- trust doctrine, perhaps because it was broadly suspicious of regulation. By the late seventies and early eighties, the antitrust and regulation doctrine was in shambles and had to be rebuilt. The modern intellectual corpus that then emerged has been very much a collective effort, involving not only me, but also my closest collaborators on the topic^2 and the many scholars whose own work and discussions with me have deeply influenced my thinking. My being un- der the spotlight owes more to their contribution than to my own talent. But I claim credit for having been in the right places at the right times and in having learned from fabulous colleagues and students, in the area for which the prize was awarded and in other fields as well. With sometimes a bit of luck, as when my MIT fellow classmate and fellow Eric Maskin advisee, Drew Fudenberg told me about an interesting field (I ac- tually did not know what “industrial organization” meant.. .). Having already taken my generals, I then sat in fascinating lectures given by Paul Joskow and Dick Schmalensee, and started fruitful collaboration with Drew. A stroke of good fortune indeed, as the required tools, game theory and information economics, were witnessing a series of breakthroughs. On the policy front, there was widespread recognition that old-style public utility regulation, which by and large insured public utilities against poor cost performance, led to inflated cost and poor customer satisfaction, and so reforms were called for. To crown it all, institutional change favored the use of economic reasoning. Where disputes were settled and regulations designed opaquely in the minister’s office, transparent processes run by independent agencies were put in place. For instance, competition authorities and regulatory agencies sprang up in Europe, which used economic reasoning. This most fortunate conjunction of circumstances led to a new paradigm. As was emphasized in the committee’s scientific background report, this para- digm is rich and complex. First, counting the number of firms or their market shares provides only a rough indication as to whether the market is competitive.
510 The Nobel Prizes
The competition-policy issue is whether the upstream firm should give “equal” or “fair” access to all downstream suppliers (alternative train opera- tors, power producers, technology implementers, D 1 ,.. ., Dn in the figure) or whether it should be allowed to “foreclose” access to the bottleneck to all down- stream firms but one or a couple of affiliated entities or allies. Such fair access, so goes the argument, allows downstream firms to compete for the end users on a level playing field. Yet, as we will see, what constitutes “fair access” and “enabling of downstream competition” requires some thinking. Provided that bilateral negotiations between the upstream supplier and indi- vidual downstream firms are not precluded, downstream competition dissipates the profit that can be extracted from end-users. To see this intuitively, suppose that downstream competitors sell a homogeneous product with demand curve Q = D(p) or p = P(Q), that production upstream and downstream is costless, and that downstream firms transform 1 unit of input into 1 unit of output. The question is whether, by controlling the bottleneck, the upstream firm is able to capture the monopoly profit πm = max{QP(Q)} = QmP(Qm), where Qm^ and Pm are the monopoly quantity and price respectively. Let U and Di negotiate a quantity qi to be delivered by U and then put on the market by Di. Suppose that Di anticipates that the other downstream firms will bring Q–i = q 1 +... + qi–1 + qi+1 +... +qn to the market. Then the quantity qi that maximizes the sum of U and Di’s profits^6 is the best Cournot reaction RC(Q–i) to Q–i:
qi = R C^ ( Q (^) − i ) = arg max (^) { q (^) i P q ( (^) i + Q − i )}
In this example, the outcome of bilateral, private negotiations is therefore the Cournot equilibrium with n firms. The upstream firm behaves opportunistically
FIGURE 1.
Market Failures and Public Policy 511
and does not internalize the negative externality on other downstream firms when negotiating an increase in supply with downstream firm Di. Because the upstream profit is capped by the downstream profit, equal to the Cournot industry profit, fair access jeopardizes the ability of the upstream firm to profit from the essential facility: the upstream bottleneck owner is victim of its inability to commit not to flood the downstream market. The more competi- tive the downstream industry (the larger n is), the more the profit is destroyed and the more eager the upstream firm is to regain its market power (the up- stream monopolist makes zero profit in the limit of large n, and this even if it enjoys full bargaining power—i.e., appropriates the joint surplus—in bilateral negotiations). The more general message is clear: under unfettered bilateral ne- gotiations, downstream competition erodes the upstream firm’s market power. In practice therefore, the upstream firm often favors its downstream subsid- iary (D 1 in Figure 2) in a myriad of ways, for example by refusing to deal with rivals or to grant them a license, by charging prohibitive access prices, or by making its technology incompatible with the rivals’. If not vertically integrated, it may enter a “sweet deal” with a downstream firm to the same effect. In short, the upstream firm uses exclusivity to restore its market power. For example, a biotech company with a patent on a new drug will grant exclusive rights for the product approval, production and marketing stage to a single pharmaceutical company, either in-house (Sanofi for Genzyme) or external. A well-meaning antitrust authority might want to promote competition by requiring that the upstream firm give equal treatment to all downstream firms. This policy requires some transparency in contracting, i.e., all contracts must be made public. The equal access requirement will, however, involuntarily lead to
FIGURE 2.
Market Failures and Public Policy 513
The same reasoning underlies much of antitrust doctrine, which, follow- ing Schumpeter, does not view monopoly power as illegal, but frowns upon the further acquisition of market power through merger or abuse of a dominant position.^8 To be certain, finer information may be needed to assess the substantive merit of market power. Software markets are often dominated at any given point of time by a large firm benefiting from network externalities among users. Such network externalities may result from chance (users have just coordinated on the platform) or may have been created through investment. Similarly, it is of- ten not obvious whether a utility’s profit comes from its cost-reducing or de- mand-expanding effort or sheer luck. This brings me to the issue of regulatory information.
Handling a firm’s informational superiority
Regulators face a double asymmetry of information, called adverse selection and moral hazard respectively.
- First, regulated firms have superior knowledge about their environment: their technology, the cost of their inputs, the demand for their products and services.
- Second, they take actions that affect cost and demand: human resource management, strategic choices of plant capacity, R&D and brand image, quality control, risk management and so forth.
In its simplest version, the firm’s cost function can be written as:
C = f( β, e, q ) + ε,
where β is an efficiency parameter known only to the firm, e (possibly multidi- mensional) is a cost-reducing effort, q = (q 1 ,... ,qn) is the vector of outputs, and ε stands for exogenous uncertainty about the final realization of the cost. The effort e is also unobserved by the regulator and is costly to the firm. Unsurprisingly, authorities that neglect the asymmetry of information fail to deliver effective, cost-efficient regulation.^9 There are two broad principles here. The first is obvious. Authorities should attempt to reduce the asymmetry of information: by collecting data of course; but also by benchmarking the firm’s performance to that of similar firms operating in different markets; and finally by auctioning off the monopoly rights (as firms reveal information about indus- try cost when competing with each other).
514 The Nobel Prizes
The second principle is that one size does not fit all: one should let the regu- lated firm make use of its information. Before we get to this, imagine that you are in charge of dealing with a contractor. Two familiar contracts will probably come to your mind:
- you can offer to fully reimburse the contractor’s cost, plus some set pay- ment over this cost; such a contract is called a “cost plus” contract if the taxpayer foots the bill or a “rate-of-return” contract if the cost and reward are derived from revenue from the users;
- or you can fix the total payment and tell the firm that this payment will cover its return as well as its cost, whatever the latter turns out to be; such a contract is called a “fixed price” or “price cap” contract.
The two contracts differ in the strength of incentives provided to the con- tractor: The cost-plus contract shelters the firm from fluctuations in its cost per- formance, while the fixed price contract makes the firm fully accountable for its cost performance. For example, in the case of a non-marketed good, the net return t for the firm is:
t = a – bC,
with b = 0 for a cost-plus contract,^10 = 1 for a fixed-price contract, and between 0 and 1 more generally. The slope b is called the “power of the incentive scheme” or “cost-sharing parameter.” The fixed price contract obviously elicits more cost-reducing effort from the firm. It however has the drawback of leaving substantial profit to the firm in lucky circumstances in which costs turn out to be particularly low or demand high, independently of any effort made by the firm. In the example above, the fixed fee a must be set sufficiently high so as to induce the firm to produce if its cost is high. Returning to the “one size does not fit all” idea, one can show that regulated firms should be confronted with a menu of options; to oversimplify, this menu might take the form of a choice between a fixed-price and a cost-plus contract. The firm then self-selects: an efficient firm will opt for being accountable for its cost, while an inefficient firm will opt for the protection of cost-plus.^11 Raising the power of incentives has been key to remedy the dismal cost per- formance of traditional regulation. However, theory and practice indicate some caveats regarding high-powered incentives:^12
516 The Nobel Prizes
is, if a price increase implies a substantial reduction in demand).^14 As the fixed production cost to be covered through markups increases, these markups also increase (i.e., θ increases). Thus, regulated prices should be “business-oriented,” similar to, but overall lower than those set by an unregulated monopoly (for whom the coefficient θ would be equal to 1.
pi − ci pi
θ η i^ (1)
Furthermore, and under some conditions,^15 the regulatory problem decom- poses: the trade-off between rents and cost-reducing effort should be addressed through the cost—or profit—sharing rule, and pricing should obey the Ramsey- Boiteux principle. This dichotomous result has practical implications, as we will see shortly. Yet regulators used to force regulated firms to set an economically very ineffi- cient price structure. Typically, utilities charged low prices on inelastic segments such as monthly subscription fees to be connected to the power or telephone network, and high prices on elastic consumption (e.g., long distance phone calls). They also charged high prices to business and low prices to residential consumers, while the former had more bypass opportunities. One justification for this was redistributive concerns; but these cross-subsidies also benefited the rich, and furthermore whether or not redistribution could be achieved by other, more efficient means (say, the income tax) was not discussed. This price tinkering away from Ramsey-Boiteux principles was also some- times motivated by the (correct) premise that regulators do not possess the information about cost and demand to fine-tune prices in a business-oriented fashion. This, however, ignored the possibility of making use of decentralized in- formation. The above-mentioned dichotomy between incentives and pricing opens the way for the use of business-oriented pricing. A price cap regulation, in which the firm must only comply with some cap on its weighted^16 average price, not only creates powerful incentives by making the firm accountable for its cost, but also sets the firm free to choose a business-oriented price structure. A special case of this idea arises when one of the “products” supplied by the monopoly is an intermediate input, that is, the provision of access for rivals to an essential facility. By imposing access prices at marginal cost (assuming they can measure it), regulators de facto bias the price structure and focus markups on those final segments for which the essential facility owner faces no competi- tion.^17 This is bound to be inefficient in general.
Market Failures and Public Policy 517
3. T WO-SIDED MARKETS
A particularly interesting choice of price structure arises in so-called two-sided markets. Two-sided “platforms” bring together multiple user-communities that want to interact with each other: gamers and game developers for videogames; users of operating systems and app developers for operating systems; “eyeballs” and advertisers for search and media platforms; cardholders and merchants for payment card transactions (Figure 4). The challenge for two-sided platforms is to find a viable business model that gets both sides on board. Regardless of their market power, whether they are Google or a free news- paper like Metro,
- they choose to allocate a lower burden to the side (say side i) whose pres- ence benefits most users on the other side. In equation (2), vj represents how much a side-j user values an extra user on side i. This willingness to pay for an interaction with an extra user on side i can be recouped by the platform through a price increase on side j; the platform’s real, or oppor- tunity cost is therefore the platform’s production cost c per interaction (perhaps equal to 0) minus vj,
- like ordinary businesses, they choose a lower burden for the side which has a relatively elastic demand (a high ηi in the formula).
pi − (^) ( c − v (^) j ) pi
η i^ (2)
Equation (2) often results in very skewed pricing patterns, with one side paying nothing (free search engine, portal, newspaper) or even being paid to
FIGURE 4.
Market Failures and Public Policy 519
pricing pattern (which is typical of two-sided markets), but the externality on non-contracting parties. The analysis reveals that the merchant fee should obey the following Pigov- ian principle: In the case of card payments, the merchant fee should be equal to the benefit that the merchant derives from a card payment.^22 The consumer, who decides on the payment method, then exerts no externality on the mer- chant. This principle is now the European Commission’s doctrine for regulating open systems Visa and MasterCard. In this realm as in many others, neither laissez-faire nor a shotgun regula- tory approach is warranted. Only sound economic analysis will do.
4. INTELLEC TUAL PROPERTY
The rule-of-reason approach to competition policy requires some confidence as to which, of efficiency and anticompetitive effects, dominates. In this respect, simple rules can greatly strengthen our confidence in policy choices. Consider intellectual property (IP), for which the shortage of data can be acute, with technologies not having yet hit the ground. Biotech and software technologies are often covered by a multiplicity of patents of varying impor- tance and owned by different owners. This “patent thicket” is conducive to “roy- alty stacking” (or “multiple marginalizations” in the parlance of economics). To understand royalty stacking, which was brilliantly formalized in 1838 by Antoine Augustin Cournot (1838) and more recently by Carl Shapiro (2001), it may be helpful to use an analogy (depicted in Figure 6) and return to medieval Europe, whose river transit was hampered by a multiplicity of tolls; for instance, there were 64 tolls on the Rhine River in the 14th century.^23 Each toll collector set his toll to maximize his revenue, oblivious of what this meant not only for the users but also for other toll collectors. Europe had to wait until the Congress of Vienna in 1815 and subsequent legislations to see the removal of toll-stacking.^24 High technologies are currently witnessing an evolution toward more af- fordable prices, similar to that for river traffic in the 19th century. New guide- lines have been set, so as to encourage the co-marketing of intellectual property through patent pools. Patent pools reduce the overall price of licensing comple- mentary patents, benefiting both intellectual property owners and technology users. Alas, patent pools and more generally co-marketing arrangements also may allow firms to raise price. For instance, the owners of two substitute patents (like the toll collectors on the two river branches in Figure 7) can raise licensing price
520 The Nobel Prizes
to the monopoly level by forming a patent pool (setting a collusive toll for down- stream access in the figure), akin to a cartel or a merger to monopoly. A flashback is useful again. A little known fact is that, prior to 1945, most high-tech industries of the time were run by patent pools.^25 But the worry about cartelization through joint marketing led to a hostile decision of the US Supreme Court in 1945 and the disappearance of pools until the recent revival of interest. But couldn’t competition authorities just ban bad (price-increasing) pools and allow good (price-decreasing) ones? They unfortunately do not possess the relevant information: There is often no long history of licensing, and fur- thermore the pattern of substitutability/complementarity changes with the uses made of the technology^26. However, simple regulations allow such sorting. First, “individual licens- ing” (the ability of individual owners to keep licensing their patents outside the pool, see Figure 8) re-creates competition when patent pools would otherwise have raised price. Patent pools with individual licensing therefore neutralize bad pools, while allowing good ones to achieve their price reduction. The reasoning is best illustrated in the very simple case of two substitute patents. The competitive price for the licenses is then equal to 0 (assuming away any licensing cost). A pool has the potential to raise the price up to the monop- oly price Pm, say. Suppose that the pool tries to sets a price P(≤Pm). Then each IP owner receives in dividends PD(P)/2, letting D(.) denote the demand function
FIGURE 6.
FIGURE 7.
522 The Nobel Prizes
Another information-free precept for the treatment of intellectual property originating from theoretical work is the suggestion that intellectual property owners commit to a cap on their licensing price before a standard is finalized. When a standard is designed, in many cases there are multiple routes to solving a given technological problem. Each one of these may be equally viable, but of- ten a standard setting body will choose only one avenue (to pursue the analogy, the public authority may have enabled traffic on the upper branch of the river by building a lock on that branch; or the presence of a major city on the upper branch may have made this branch a superior alternative, as in Figure 10). After the decision has been made, however, the chosen patent becomes a “standard- essential patent (SEP),” and the patent owner can ask for a high royalty even though another patent could have offered comparable value, had the technology been designed differently. To restrain firms from taking advantage of the fortuitous essentiality of their patents, an essentiality that resulted only from being included in a standard and not from technological merit, standard-setting bodies commonly require firms to commit in advance to license their patents on fair, reasonable and non-dis- criminatory (FRAND) terms. The problem with this approach is that FRAND commitments are very ambiguous: What exactly is a fair and reasonable rate? And in fact, large lawsuits regarding the meaning of the commitment proliferate all over the world. One would not build a house on a piece of land whose price is not known in advance. The same obtains for technologies. We have proposed that intellectual property owners commit to their licensing conditions prior to the final choice of standard and we tried to explain why this commitment requirement is unlikely to result from competition among standard-setting bodies.^30
FIGURE 10.
Market Failures and Public Policy 523
5. C ONCLUDING REMARKS
The theory of industrial organization has proved a very useful tool to think about one of the major challenges of our economies. It has fashioned antitrust and regulation. Recognizing that industries are different from each other and so one size does not fit all, it has patiently built a body of knowledge that has helped regulators to better understand market power and the effects of policy interven- tions, and firms to formulate their strategies. Industrial organization has gone a long way, but much work remains to be done. An especially gratifying aspect is that the field of industrial organization is currently thriving, with many top young researchers producing exciting work. Making this world a better world is the economist’s first mission. I believe that the entire community of industrial organization researchers has contrib- uted substantially to this mission. On behalf of this community, I was humbled, honored and grateful to be awarded the 2014 Sveriges Riksbank Prize in Eco- nomic Sciences in Memory of Alfred Nobel.
REFERENCES
- Aghion, P. and P. Bolton. (1987). “Contracts as a Barrier to Entry,” American Economic Review, 77(3): 388–401.
- Armstrong, M. (1998). “Network Interconnection in Telecommunications,” Economic Journal, 108 (448): 545–564.
- Armstrong, M. (2006). “Competition in Two-Sided Markets,” RAND Journal of Economics, 37(3): 668–69.
- Armstrong, M. and J. Vickers. (2000). “Multiproduct Price Regulation under Asymmetric Information,” Journal of Industrial Economics, 48: 137–160.
- Baron, D. and R. Myerson. (1982). “Regulating a Monopolist with Unknown Costs,” Econometrica, 50(4): 911–930.
- Boiteux, M. (1956). “Sur la gestion des Monopoles Publics astreints à l’équilibre budgétaire,” Econometrica, 24(1): 22–
- Bourguignon, H., R. Gomes and J. Tirole. (2014). “Shrouded Transaction Costs,” mimeo, TSE.
- Boutin, A. (2014). “Screening for Good Patent Pools through Price Caps on Individual Licenses,” mimeo, ULB.
- Caillaud, B. and B. Jullien. (2003). “Chicken & Egg: Competition among Intermediation Service Providers,” RAND Journal of Economics, 34(2): 309–328.
- Cournot, A.A. (1838). Recherches sur les Principes Mathématiques de la Théorie des Richesses.
- Dupuit, J. (1844). “De la mesure de l’utilité des travaux publics,” Annales des Ponts et Chaussées, Paris.
Market Failures and Public Policy 525
- Spaulding, R. M. (2011). “Revolutionary France and the Transformation of the Rhine,” Central European History, 44(02): 203–226.
- Whinston, M. (1990). “Tying, Foreclosure, and Exclusion,” American Economic Review, 80(4):837–859.
ENDNOTES
- Industrial organization economists have studied a variety of other market failures, involving information problems and a range of externalities (such as environmen- tal damages or banking failures) which arise even in the absence of market power; small banks (such as the Cajas and the Landesbanken recently in Europe) often fail, leaving the deposit insurance fund/taxpayer to pay for bailouts, and firms without market power often pollute. This lecture focuses on market power, which limits the diffusion of a product, service or technology to downstream firms and end-users.
- Chronologically, and very unfairly to the many others with whom I have had the chance to interact, Drew Fudenberg, Eric Maskin, Jean-Jacques Laffont—to whose memory this lecture is dedicated—as well as Patrick Rey, Jean-Charles Rochet, Paul Joskow, and Josh Lerner.
- They must also account for enforcement costs.
- The discussion here follows Hart-Tirole (1990) and Rey-Tirole (2007), to which we refer for the “loose ends” left over by this intuitive discussion. See also Segal (1999) for a broader treatment, which covers applications beyond industrial organization.
- Antitrust practitioners are often reluctant to call a key patent an “essential facility”; for an essential facility is usually associated with an obligation to share. I nonethe- less include key patents in the essential facility category as it fits the criterion “some input that cannot be reproduced or bypassed by others at a low cost.” We will later discuss the obligation to share.
- This quantity maximizes the “cake” to be divided between the two parties; the shar- ing of this cake will involve a monetary transfer between them.
- Still another way of obtaining the monopoly profit is to offer the same fol- lowing two-part tariff to all downstream firms: T(q) = A + wq, where w, the wholesale price, satisfies: nqC(w) = Qm, the monopoly quantity, and qC(w) is the Cournot quantity given unit cost w. The fixed fee A is then set so as to capture the resulting Cournot profit of each downstream firm. None of the schemes allowing the upstream firm to reap monopoly profit is robust to additional, private contracting: for there is always money to be made by the upstream firm by contracting for a higher quantity with an individual down- stream firm as long as the latter’s quantity lies below the Cournot reaction curve (for zero marginal cost). Thus, to sustain individual quantities below qC(0), all contracts must indeed be public.
- For example, the antitrust authority may disapprove the handicapping of horizontal rivals, say, by making one’s technology incompatible when no protection of intel- lectual property is at stake, by locking in customers through long-term contracts that have limited efficiency justification (as in e.g., Aghion-Bolton 1987), or by
526 The Nobel Prizes
committing to tough behavior through irremediable tying (as in e.g., the model of horizontal foreclosure of Whinston 1990).
- To take two examples unrelated to market power, command-and-control in envi- ronmental regulation all over the world and judicial control of the business justifi- cation for layoffs in Southern Europe have backfired by imposing high costs on the industry and thereby casting doubts on the sustainability of these policies. The same principle holds for industrial organization.
- Define “effort” as being costly for the firm (what is done beyond intrinsic motivation or career concerns), and e = 0 as the minimum effort. Then the cost-plus contract elicits e = 0 and yields a β– independent rent for the firm. Hence, the rent can be made small without jeopardizing the firm’s participation.
- See, e.g., Laffont-Tirole (1986). Precursors to my work with Jean-Jacques Laffont include Loeb-Magat (1979), Baron-Myerson (1982) and Sappington (1982). Under some conditions, the optimal regulatory scheme lets the firm pick within a menu of linear incentive contracts t = a – b C, each characterized by a pair of fixed-payment and cost-sharing parameters (a, b).
- A broader list of caveats of high-powered incentives and their theoretical modeling can be found in Laffont-Tirole (1993).
- This is the so-called “ratchet effect” (see, e.g., Freixas et al. (1985) and Laffont-Tirole (1988)).
- Here we consider the case of independent demands; formulas must be adjusted to account for “cross-elasticities” in case of substitutes or complements.
- See Laffont-Tirole (1990). The dichotomy does not hold in the absence of cost or profit measurement (Baron-Myerson 1982) as then a single instrument (the vector of prices) must accomplish two goals: capture the firm’s rent and limit the departure from Ramsey-Boiteux principles.
- This, of course, is very much of an idealization. In theory, the weights should be equal to the anticipated quantities at the Ramsey-Boiteux prices (not those that result from the firm’s choice so as not to let it manipulate the weights); intuitively, an increase in price pi generates a reduction of consumer surplus equal to the quan- tity of good i; a price cap based on the anticipated quantities therefore creates just the right “internalization” of consumer surplus. The caveat is of course asymmet- ric information: the regulator does not know the Ramsey-Boiteux quantities. More research is needed on this front (see Armstrong-Vickers 2000 for a contribution in this direction). Empirically, though, the use of price caps has had a substantial rebalancing effect toward a Ramsey-Boiteux-oriented price structure.
- To see this, suppose that good i is supplied by a competitive downstream fringe at marginal cost cD. The regulated firm owns the bottleneck input that allows the downstream fringe to operate. The upstream marginal cost is cU. Let ci= cD + cU. If access is given at access price a = cD, then pi = ci is lower than in the Ramsey-Boiteux formula. The fixed cost must then be recouped entirely through (very high) prices on goods for which the regulated firm faces no competition. For more on such “one-way” access pricing, see Laffont-Tirole (1994). Here I will not discuss “two- way” access pricing, in which owners of competitive bottlenecks (e.g., mobile phone
