Quantitative Techniques for Competition and Antitrust Analysis_16 pptx

The impact of a vertical merger on own and rivals’ costs.In effect, a vertical merger followed by input foreclosure by the vertically grated firm would mean that 1 TomTom’s competitors i

10.1.3.1 Territorial Restrictions

Territorial restrictions are normally used to reduce intrabrand competition stream When sales and service efforts by the retailer are important to the manufac-turer, they will want to ensure that retailers reap the rewards of their investments inservice quality By granting exclusive rights of sale in a given territory to a singleretailer or a group of retailers, the manufacturer can ensure that retailers in differentareas do not free ride on each other’s investment By eliminating the capacity toattract shoppers from other territories, the horizontal pricing externality, wherebyretailers hurt the manufacturer by lowering prices to steal customers from each other,

down-is also removed It down-is, of course, essential for thdown-is practice to have the desired effectthat arbitrage opportunities are eliminated That said, exclusive territories activelyprevent competition between retailers and therefore may clearly potentially haveimportant anticompetitive consequences Indeed, it is exactly the “horizontal pric-ing externality” which competition authorities usually fight very hard to protectprecisely because it results in low prices for consumers In an extreme case, with amonopoly manufacturer and a set of retailers who, absent the territorial restrictions,would otherwise compete, territorial restrictions could enforce a market divisionarrangement entirely equivalent to explicit collusion between the retailers To eval-uate such a policy we may need to evaluate the way in which consumers trade offpotentially higher prices for goods against any higher quality of service provided.Territorial restrictions as described above have been a tradition in the car sales sec-tor in Europe, where markets were traditionally defined as national, with a distribu-tion system characterized by exclusive dealing and geographic restrictions, including

a restriction by manufactures of cross-border sales In 2002, the European sion concluded that exclusive dealing agreements between car manufacturers andcar dealers as well as the exclusive sales territories granted by the manufacturers tothe dealers were not justified on grounds of efficiency as the consumers were “notgetting a fair share of the resulting benefits.”12The European Commission issuednew conditions for the sale of cars in the European Union notably imposing theright of exclusive dealers to sell to operators outside of the manufacturer’s officialnetwork.13

Commis-10.1.3.2 Resale Price Maintenance: Minimum Price

An alternative way to induce retailers to provide the level of services, sales effort, oradvertisement that is optimal for the manufacturer is to establish a minimum salesprice thereby restricting price competition Firms could, for example, simply refuse

12 “Commission adopts comprehensive reform of competition rules for car sales and servicing” IP/02/1073 of 17/07/2002 available at http://europa.eu/rapid/pressReleasesAction.do?reference=IP/02/ 1073&format=HTML&aged=0&language=EN&guiLanguage=en.

13 Commission Regulation (EC) no 1400/2002 of July 31, 2002 on the application of Article 81(3) of the Treaty to categories of vertical agreements and concerted practices in the motor vehicle sector.

to sell to retailers who charge a retail price below an established minimum Forinstance, in the second half of the 1990s, music recording companies in the UnitedStates allegedly notified music retailers that, if they advertised music CDs at a priceless than a stipulated amount, the recording companies would withdraw the financialsupport for advertisement and sales that they usually granted to those retailers Thoseadvertisement and promotion payments were an important source of revenue forretailers and the rule was alleged as a de facto establishment of a price floor, triggering

a multidistrict class action by purchasers of prerecorded music against the musicmajors.14Because resale price maintenance can also facilitate collusive agreementsand was, until the Leegin15 decision, a per se offense in the United States, therecording companies allegedly tried to circumvent the antitrust statutes by penalizingadvertised prices as opposed to sales prices Recording companies claimed thatsome electronic and mass merchant stores were setting low prices of popular CDs toattract customers into their stores, thereby undercutting specialized music stores thatprovided services such as listening stations, in store advice, and promotional events.Those services resulted in an increase of music sales that was allegedly essential tothe music business The case was finally settled in 2003 for $143 million and thepractice of establishing a minimum advertised price was terminated.16It remains auseful illustration of an alleged attempt by manufacturers to prevent aggressive pricecompetition at the retail level that appears to have been dramatically decreasing theprovision of services and sales effort Of course, the per se status of RPM duringthat case meant that it was not necessary to show that consumers suffered harm as aresult of the practice Following Leegin, that will no longer be the case in the UnitedStates and as a result such cases will probably be far harder to prosecute in eitherpublic or private antitrust spheres

In Europe, in general the attitude to RPM is less permissive than the new legalregime in the United States Specifically, in the EU, RPM is currently treated as ahardcore pricing restriction that is illegal unless the parties bring forward substanti-ated claims of efficiencies, that is, there is a “rebuttable presumption of illegality.”17One stated reason is that RPM is often associated with increasing prices RPM canalso be used as a facilitating device to cartelize retailers’ prices because it effectivelysets final retailer prices across retailers by way of a contract What may look like avertical contract may on occasion be a device for horizontal price fixing with neg-ative consequences for final consumers The challenge for antitrust agencies under

14In re Compact Disc Minimum Advertised Price Antitrust Litigation, M.D.L no 1361 (U.S.D.C Me).

15Leegin Creative Products Inc v PSKS Inc., Supreme Court of the United States, June 28, 2007.

16District of Maine, In re Compact Disc Minimum Advertised Price Antitrust Litigation, M.D.L.

Docket no 1361 Litigation Decision and Order on Notice, Settlement Proposals, Class Certifications, and Attorney Fees.

17 RPM is considered an “object” restriction under Article 81 of the EC treaty (and its U.K embodiment, the Competition Act, 1998) As a result RPM is viewed as harmful by object—that is, very likely to be anticompetitive so that a harmful effect may be presumed.

a rule-of-reason approach is to tell apart the efficient use of RPM from its potentialfar less benign use.

10.1.3.3 Exclusive Dealing

Exclusive dealing, also called single branding, occurs when the upstream firmrequires or induces the retailer to only sell its brand There can be an explicit require-ment for exclusive dealing or alternatively such an outcome can be generated via

a carefully designed pricing structure For example, we may de facto see exclusivedealing if advantageous rebates are granted only to those retailers who purchase alltheir products from a single provider, the result will effectively enforce an exclusivedealing arrangement There are, of course, many entirely valid substantial reasonsthat an upstream firm may want this type of contractual arrangement One clear moti-vation may be a desire to protect their own investment in advertisement and quality

by preventing retailers from steering consumers who arrive in the store to priced, less well-known rival products once the consumer is in the store Retailersmight have an incentive to do so if rival products were able to provide retailerswith higher margins despite lower sales prices, for example, because few advertis-ing costs were incurred Thus exclusive dealing may solve a horizontal externalityacross producers within the retailer A related example may arise if a manufacturerinvests in its distribution channels to increase the level of service and promotionalactivity It is possible that a retailer might choose to use these skills or resources topromote products from other producers The contract in that case provides a mech-anism for the retailer to credibly commit not to engage in such activity and therebyprovide the manufacturer with incentives for promotion to the potential benefit ofboth firms and also quite probably consumers On the other hand, exclusive dealingmay also provide a mechanism for foreclosure.18

lower-10.1.3.4 Tying and Bundling

Tying and bundling are also ways in which manufacturers can condition the sions of retailers downstream These practices consist of conditioning the sale of agood on the sale of another, usually complementary, good This can be done, forexample, through a contractual obligation or because the pricing structure renders

deci-it unprofdeci-itable to purchase the two goods separately An example might be aircraftengines and aircraft instrumentation

There is a large body of literature analyzing the reasons for tying and bundling.The explanations range from quality concerns to price discrimination and of coursesimple transaction cost advantages Bundling can also be motivated by potentialeconomies of scale and scope in production or distribution that allows the firm to

18 See, in particular, the foreclosure models discussed by Salop and Scheffman (1983), Comanor and Frech (1985), Schwartz (1987), Mathewson and Winter (1987), Rasmusen et al (1991), Bernheim and Whinston (1998), Segal and Whinston (2000), and Simpson and Wickelgren (2007).

lower prices and increase sales by bundling the sales of several products In theory,tying complementary goods can increase the incentives to lower prices since thefirm will benefit from the increase in the demand of the initial product and also thetied product In the case of metering and price discrimination, the outcome is lessclear and will vary across customers.

Although tying can have many nonexclusionary motivations, it can neverthelessalso lead to intended or unintended foreclosure on the market Whinston (1990)and Nalebuff (1999) among others analyze the incentives to tie and present conclu-sions for stylized examples involving assumptions about the consumer valuation ofthe tying product, the link between the valuations of the tied and tying products,and the nature of competition in the tied market (see also Bakos and Brynjolfsson1998; Carlton and Waldman 2002) Whinston (1990) illustrates that a monopolistcan, under some circumstances, profitably foreclose a market by tying and therebycommit to a low price for the bundle His results also show that even though tyingcomplementary products is less “costly” for the firm, the incentives to tie are alsoless obvious unless some particular conditions are fulfilled Nalebuff (1999) showsthat with heterogeneous preferences, bundling for complementary products can beprofitable and foreclosure can be achieved In a dynamic market characterized byinnovation tying can also be used to weaken or foreclose potential competitors (Choi2004; Carlton and Waldman 2002)

This literature contrasts sharply with the position taken by the Chicago school,who heavily critiqued what they called “leverage theory.” Proponents of that viewargued that if a firm had a monopoly in one good but faced competition in a sec-ond complementary product and consumers desired the goods in fixed proportions,then a “one-monopoly-profit” argument holds Specifically, the monopolist in thefirst product need not monopolize the second market to extract monopoly profits.Two recent empirical papers consider variants of this debate Chevalier and ScottMorton (2008) consider a horizontal version of the one-monopoly-profit argumentarising from tying casket sales and funeral services together and their results favorthe one-monopoly-profit argument However, Genakos et al (2006) find evidencethat incentives for foreclosure exist empirically, looking at Microsoft’s incentive toleverage its monopoly in the “client operating system market” (aka regular Win-dows) to the “server operating system market” (aka Windows for network servers)

In particular, they find that an incentive to leverage market power can exist providedperfect price discrimination is not possible for a monopolist If so, then leveragecan become a method that can help a monopolist to extract more rents from themonopoly market.19

19 In support of their theory the authors report that in 1997 Microsoft’s Chairman Bill Gates wrote in

an internal memo: “What we’re trying to do is to use our server control to do new protocols and lock out Sun and Oracle specifically the symmetry that we have between the client operating system and the server operating system is a huge advantage for us.”

10.1.3.5 Refusals to Deal

There exist cases of straight refusal to deal whereby a firm upstream simply refuses

to supply a firm downstream that wants its output as an input The legal treatment

of this type of conduct varies under different jurisdictions Increasingly, the goal ofprotecting the incentives for innovation and large upfront investments is balancedagainst the benefits that an access to the input would generate through a more intensecompetition downstream In general, refusal to deal is unlikely to be regarded as

a problematic action unless the upstream firm has some degree of market power.One important source of upstream market power may arise from the fact that a firmoperates an essential facility A deepwater port is an example of something that may

be considered to be essential facility A country may, for instance, have only one

or two deepwater ports suitable for handling large cargo vessels Entry, building anew port, is fairly obviously costly and may be impossible depending on geography,while transport costs for goods within a country may mean a given port owner hassubstantial market power The difficulty for antitrust authorities is that there willalso be cases in which the firm stops supplying a downstream firm with which it waspreviously trading for perfectly legitimate reasons The termination of a relationshipwith a firm downstream may occur because the supplier thinks the intermediate firm

is not keeping up with quality standards or otherwise not fulfilling aspects of anexplicit or implicit contract It may also be the result of changes in market conditionsthat affect the incentives of the firm upstream, for instance, changes in costs meaningthat marginal units become loss-making Clearly, even a dominant firm should not

be forced to sell goods at a loss if economic efficiency is our aim Thus, an antitrustauthority must attempt to distinguish “legitimate” refusals to deal from illegitimateones Quantification of the effect of refusals to deal are generally quite difficult andinvolve comparing the outcome of a world where a business exists with one wherethe business does not exist Perhaps as a result, to date, the assessment of refusal todeal cases has therefore been primarily qualitative in nature

10.1.4 Effects of Vertical Restraints on Market Outcomes

To summarize this first section of the chapter, the theoretical effect of verticalrestraints on consumer welfare is, in many cases, ambiguous On the one hand thereare numerous potential motivations for vertical restraints that are entirely innocentand unlikely to cause legitimate concern to antitrust authorities On the other handvertical restraints may also facilitate outcomes that should be of concern to agenciesseeking to make markets work well for consumers Vertical restraints can sometimes

be used as mechanisms to soften competition In extreme cases the incentive to pete on prices can be entirely eliminated by the existence of such restraints In othercases, vertical restraints may result in foreclosure of either inputs or customers.The bottom line is that economic theory does not allow general conclusions aboutwhether vertical restrains are “good” or “bad” for welfare In any given instance

com-the question is an empirical one, which means com-the competition authorities must firstattempt to determine the circumstances in which a vertical practice may be cause forconcern and should therefore be the object of scrutiny Second, they must attempt

to evaluate whether or not the vertical restraint should be banned or restricted forthe benefit of consumers

Because theoretical predictions are not always, or even usually, clear about thenet effect of vertical contractual arrangements on either total or consumer welfare,there have been a limited but perhaps increasing number of attempts to empiricallyassess the effect of vertical practices in both the case and academic literatures In thenext section, we present a number of different methodologies that have been used

to try to assess the effect of vertical restraints In looking at each example we willstrive to illustrate both the benefits and limitations of such exercises

In the first section of this chapter we established that the motivations for verticalrestraints are many and also that the theoretical predictions regarding the effect ofthese practices on welfare are often ambiguous Sometimes a vertical restraint willsolve an important externality problem to the benefit of both firms and consumers

On other occasions, it is exactly the externalities that drive good outcomes forconsumers and so removing them via a vertical restraint generates poor outcomesfor consumers An example is when a vertical restraint acts to remove the horizontalpricing externality between firms, the one that usually means that competition leads

to low prices and high-quality goods

Sometimes the ultimate goal of the practice is outright foreclosure and the resultmay be higher prices and lower output with no concomitant efficiency gains Onother occasions, the two effects will cumulate Empirical analysis is a way to try

to determine the effects of vertical restraints on consumer welfare in a particularcase Unfortunately, precisely because many of the effects we are trying to isolateare difficult to measure, it is particularly difficult to undertake an effects-basedanalysis of vertical restraints Empirical strategies that have been used to determinethe effects of vertical arrangements include regression analysis, particularly fixed-effects regressions, natural experiments, and event studies Each is familiar fromelsewhere in the book However, it is important to note that such methods can onlypotentially help solve identification issues when there are data available on the

situation with and without the practice Ex ante analysis requires the construction

and estimation of a structural model, which is very difficult to do without makingstringent and quite possibly unrealistic assumptions about firm behavior We discusseach of the available strategies in the rest of this chapter Before doing so we brieflydiscuss informal and semiformal quantitative methods for evaluating the incentivefor foreclosure

10.2.1 Informal and Semiformal Analysis of Incentives

Informal quantitative analysis can sometimes be insightful for evaluating the tive for foreclosure An example of such an analysis was provided involving a mergerbetween English, Welsh and Scottish Railway Holdings (EWS) and Marcroft Engi-neering (Marcroft) EWS is a freight haulier on the railways, whereas Marcroft was

incen-a provider of rincen-ailwincen-ay mincen-aintenincen-ance services mincen-ainly serving the rincen-ail freight industry.The United Kingdom’s Office of Fair Trading (OFT) in its phase I investigation con-sidered whether EWS would have an incentive to foreclose access to the Marcroftmaintenance depots since by doing so it could potentially harm its competitors inthe downstream rail haulage market To evaluate this option the OFT considered(1) the potential returns in the downstream market to foreclosure and also (2) thecost of foreclosing access to the Marcroft maintenance facilities The OFT decisiondocument notes from company accounts and information provided by third partiesthat both volume and profit margin are lower in the upstream maintenance marketthan they are in the downstream freight haulage market.20Assuming the margins

do not change, a rough calculation of the incentive to foreclose would involve anevaluation of loss in upstream profits from maintenance

ProfitMaintenanceD MarginM
VolumeMagainst the gain from higher downstream profits from haulage

ProfitHaulageD MarginH
VolumeH:Obviously, even these simplified expressions involve changes in volume rather thanthe levels of volume, but the OFT may have believed that the expected changes

in volume would be reflective of the overall levels of each activity Thus, sinceMarginH> MarginMand if

VolumeH> VolumeM H)
VolumeH>
VolumeM:

Then

ProfitHaulageD MarginH
VolumeH>
ProfitMaintenance

D MarginM
VolumeM:Obviously, such a rough calculation involves some very strong assumptions as areappropriate for an authority exploring whether there is the potential justificationfor further investigation In the end the OFT decided to refer the merger to theU.K Phase II merger body, the Competition Commission, but decided that since

it had also found potential horizontal problems with the merger, it did not need tocome to a final view on the potential vertical concerns In the end the CC accepted

20 See, in particular, paragraph 43, OFT decision document available at www.oft.gov.uk/shared oft/ mergers ea02/2006/railway.pdf.

undertakings from the company involving the divestment of part of the Marcroftmaintenance business.21

In 2008, the European Commission investigated a vertical merger involving theupstream market for the databases that allow the construction of navigable digitalmaps (NDMs) and the downstream market involving various electronic navigationdevices.22Specifically, TomTom, a producer of personal navigation devices (PNDs),proposed a merger with the navigable digital map database provider TeleAtlas.23Public documents do not allow a complete reconstruction of the calculations per-formed to evaluate the total foreclosure story considered by the Commission, butnonetheless exploring the example is instructive

The vertical arithmetic approach suggests that to evaluate the plausibility of either

a total or partial foreclosure theory of harm, the competition agency should evaluatethe loss of profit upstream and the potential gain in profit downstream Doing soallows an evaluation of the incentive to engage in foreclosure Under a total fore-closure strategy, the vertically integrated firm will lose profits upstream because itstops selling to the “merchant market”—those firms competing with its downstreamsubsidiary That means those rivals will face higher costs because, according to thetheory of harm, they will have to buy from rival upstream suppliers who no longerface competition in supply and so will raise prices In the TomTom–TeleAtlas casethis total foreclosure theory of harm amounts to TeleAtlas deciding to stop compet-ing for the custom of rival PND companies that need navigable maps to build theirnavigation devices As a result, TeleAtlas’s rival Navteq would face a reduction ofcompetition and be able to increase prices (or more generally follow some otherstrategy such as reduce quality)

21 See www.competition-commission.org.uk/inquiries/ref2006/marcroft/index.htm for further details.

22 We will not divert our discussion with a detailed evaluation of the various market definitions However,

we do pause to note there that the Commission came to the view that: (1) Upstream there was demand-side substitution between the navigable digital maps provided by TeleAtlas and Navteq However, there was

no demand-side substitution between navigable maps and more “basic” digital maps which could not be used for real-time navigation while driving your car Moreover, the Commission received estimates that

it would take something like 1,000–2,000 people five to ten years to upgrade a basic map to the quality of

a navigable map Thus there was neither demand nor supply substitution Geographic markets upstream were left ambiguous as they were judged not to affect the conclusion of the analysis (2) Downstream, the Commission noted that there were various forms of navigation devices: personal navigation devices (in the form of a handheld device that you could put in your car), maps on personal digital assistants,

“in-dash” navigation devices (navigation devices built into car dashboards), and GPS enabled mobile phones The Commission after looking at the evidence decided that PNDs constituted a downstream market in itself.

23 See Case no COMP/M.4854 TomTom/TeleAtlas At around the same time Nokia (the mobile phone producing company) merged with TeleAtlas’s main rival navigable map producer, Navteq (See Case no COMP/M.4942 Nokia/Navteq.) Both mergers were ultimately cleared The analysis undertaken in these mergers was widely seen as testing the European Commission’s latest vertical (nonhorizontal) merger guidelines, which were adopted in November 2007 The new set of guidelines was developed partly in response to criticisms from the Court of First Instance following the European Commission’s controver- sial decision to block the proposed merger between GE and Honeywell (See Case no COMP/M.2220 General Electric/Honeywell.)

Figure 10.1. The impact of a vertical merger on own and rivals’ costs.

In effect, a vertical merger followed by input foreclosure by the vertically grated firm would mean that (1) TomTom’s competitors in the downstream marketwould face higher input costs following the merger while (2) TomTom itself, as thedownstream division of a vertically integrated firm, would be able to reduce its costs

inte-if vertical integration has aided the reduction or avoidance of double tion In the case of TomTom–TeleAtlas we know from the Commission’s decisiondocument that pre-merger upstream gross margins were high, approximately 85%.The reason is that developing a digital map involves a great deal of essentially fixed-cost investment while the resulting database can subsequently be duplicated at lowmarginal cost That means that if pre-merger vertical contracting was not able tosolve the double marginalization problem, then TomTom’s (TT’s) marginal costscould decline considerably post-merger At the same time, rival downstream firmswould, according to the theory of harm, face higher input costs as they would nowsuffer from a lack of competition upstream

marginaliza-Figure 10.1 presents the impact of vertical integration when it (1) reduces ble marginalization for the merged firm and (2) increases marginal costs for rivalsbecause the merged firm follows a foreclosure strategy The former effect shifts TT’sreaction function downward while the latter effect shifts the rival’s reaction functionrightward to reflect an increase in input costs (for any given price of TT, rivals willnow choose to charge a higher price)

dou-Figure 10.1 shows that the impact of such a change on downstream competitiveoutcomes can involve lower prices for the vertically integrating firm as well as poten-tially higher prices from its downstream rival(s) Naturally, the aggregate welfareimpact of such a change will depend on the relative magnitudes of the consequentprofit and consumer surplus gains and losses It is this observation that inducesmany agencies to choose a framework for vertical merger analysis which includes

an analysis of ability, incentive, and consumer harm That said, those competition

agencies whose statutory framework do not immediately “net-off” consumer plus gains and losses will note that some customers have lost out under a merger

Figure 10.2. The impact of a vertical merger on own and rivals’ costs (2).

that led to this outcome, even if ultimately overall consumer welfare is higher, haps because the vertically integrating firm has a far larger share of the market whobenefit from lower prices post-merger

per-Figure 10.2 shows that when the integrating firm benefits from removing ble marginalization are small relative to the magnitude of the effect of increasedcosts suffered by rivals in the downstream market, the outcome will tend to involvethe prices charged by all firms increasing Such an outcome would clearly beunambiguously bad for consumers, all else equal

dou-Taking some data from the TomTom–TeleAtlas case, according to the case ments, there were unit sales of 10.8 million NDMs for PNDs with an average sellingprice of€14.6 Moreover, pre-merger TeleAtlas sold their database to TomTom andalso to other downstream producers, who accounted for between 10 and 30% ofthe downstream market,24 that is, between 10:8  0:1 D 1:1 and 10:8  0:3 D 3:2million units TomTom’s downstream rivals include Garmin, Mio-Tech & Navman,Medion, and My Guide Since the case tells us that gross margins were 85%, aforeclosure strategy would involve sacrificing profits (or at least a “contribution”

docu-to upstream fixed costs) of between €14:6  1:1  0:85 D €13:7 million and

24 Only a range is available in the public decision document.

such an exercise could be performed, given the analysis presented in figures 10.1and 10.2 Rather than present the full calculations, we present a rough back-of-an-envelope calculation, looking directly at the change in profits associated with apotential change in market shares and margins.

First note that if subscript “0” denotes pre-merger and subscript “1” denotes merger prices, quantities, and costs, then we can write the impact on downstreamprofits as

so would mean its sales would grow from q0 D 0:4  10:8 D 4:31 million to

q1 D 0:45  10:8 D 4:85 million, a growth of 0.53 million customers per year Inthe downstream market, gross margins were reported to be between 0 and 50%, sosuppose 25% while average final price pre-merger was p0D€200 For simplicity,suppose that downstream gross margins did not change pre- and post-merger whileprices did fall by some amount because some of the reduction in database inputcosts was passed through to final consumers With a gross margin of 85% upstreamand a pre-merger average selling price upstream of€14.6 per unit, the reduction inTomTom’s marginal cost may be as large as 0:85  14:6 D€12:4 Thus final priceswould be between€200 and €187.6, depending on the extent of the pass-through

of the cost reduction to final customers Supposing pass-though were 50%, then wewould have p1D€193:8 and hence the increase in downstream profits would be

Obviously, we have made a lot of assumptions in this rough calculation and a morecareful calculation would get closer to real numbers for the potential upstream lossesand downstream gains from foreclosure For now we note that while our back-of-the-envelope calculation does not make a clear case for or against the incentive for total

foreclosure, such a calculation can help us to explore which alternate assumptions(e.g., about captured market share, downstream margins and how they are likely

to change, and pass-through rates) that would provide grounds for either concern

or reassurance Naturally, in such an evaluation it will be important to think aboutthe realities of the market place For example, one downstream PND producer,Garmin, had signed a fairly long-term contract with Navteq in 2007, due to expire

in 2015, although even then there was an option to extend until 2019 Such a contractmeant that for at least seven years Navteq would not be able to increase prices forits database to a major downstream producer Such a fact is clearly important inevaluating the likely profitability of a downstream foreclosure strategy premised onthe idea that Navteq would have the incentive and ability to be able to raise the price

of its database to TomTom’s downstream rivals

Before closing this discussion of foreclosure strategies, we make three furthercomments First, we note the important contribution from Hart and Tirole (1990).Their paper suggests that a total foreclosure strategy may not ultimately be an equi-librium in a static game since a firm attempting to foreclose downstream competitorsmay actually be better off (given the strategies of rivals) by deviating from the fore-closure strategy by selling into the merchant market Thus, Hart and Tirole suggestthat the commitment not to sell to the merchant market is not a credible one Ordover

et al (1992) disagreed and, more generally, the role of reputation and credibility

of commitments may be best studied within a repeated game context Whatever theappropriate scope of these theoretical concerns, the experimental evidence appears tosuggests that the commitment problem may not always be overwhelming (Normann

et al 2001; Normann 2007)

Second, we note that within the context of a static model, partial foreclosuremodels, where the firm may find it optimal to raise prices to the merchant marketand thus partially foreclose it, are not subject to the credibility concerns They may,however, result in fewer occasions where the actual economic effects of foreclosureare harmful to consumers

And, finally, we note that in the TomTom–TeleAtlas case there were potentiallyimportant efficiency benefits from the merger, namely that cars driving around withTomTom PNDs could actually send information back to TeleAtlas about wheredrivers could and could not drive The parties argued that, in the future, allowingsuch information transfer from cars back to the database will reduce the cost ofcollecting the detailed up-to-date information required for generating navigablemaps

10.2.2 Regression Analysis of Vertical Integration

In this section we illustrate the use of regression analysis in the vertical mergercontext We begin by looking at the kind of data set that may be available from a

“natural experiment.”

10.2.2.1 Estimating the Effects of Vertical Integration in

the Retail Gasoline Market

An empirical attempt to determine the impact of vertical arrangements on consumerretail prices in the gasoline market in the United States can be found in Hastings(2004) That paper looks at the sale in California of Thrifty, a chain of independentgas stations, to ARCO, a large U.S oil company which is vertically integrated Thesale occurred in March 1997 after the 75-year-old owner of the Thrifty gas stationsdecided to retire There was a 60-day waiting period after which all Thrifty stationsfell under the control of ARCO Thrifty stations were just branded as ARCO andplaced under new contracts Some of the gas stations became company operated andothers were leased to dealers who operated them under the ARCO brand There was

no remodeling, expansion, or other investment in the gas stations The rebrandingprocess was completed in September 1997

Hastings uses panel data of retail prices at the station level for four months in theLos Angeles and San Diego Metropolitan Statistical Areas (MSAs) The data coverthe months of February, June, October, and December 1997 so that the data provideinformation on a range of markets before and after the sale Hastings (2004) assumesthat the geographic market definition is one mile along a surface street or freewayaround the petrol station The sale of Thrifty to ARCO was arguably an event thatwas largely independent of market conditions given the owner’s desire to retire, and

it generated an increase in vertical integration in some local retail gasoline marketswhile in other markets the ownership structure remained unchanged Specifically,there were 669 stations in the price sample and 99 of them had a Thrifty within onemile and therefore saw the structure of vertical ownership in the market as a result ofthe acquisition change with an increase in the level of vertical integration of the retailgas market The data set appears therefore to provide a nice exogenous movement inthe extent of vertical integration in some markets whose effects on prices we should

be able to trace Moreover, the fact that some markets were unaffected means that

we also have a “control” sample of markets which may allow us to control for anyother factors changing over the time period of the study Doing so mean Hastings canuse a difference-in-differences approach to identification, comparing the change inprices (before and after) the merger in markets which were affected with the change

in markets which were not

The data show that the average retail prices at gas stations that competed withThrifty increased after the sale relative to the retail prices at gas stations that werenot affected by the sale This is shown in figure 10.3 Gas stations which have aThrifty station within one mile had prices 3 cents lower than the rest of the gasstations before the sale After the acquisition was completed in September 1997, theaverage price for those same gas stations was 2 cents higher than the average price

of the gas stations unaffected by the sale Hastings reports that a similar result wasobtained for the San Diego area and that she found no price difference among the

December February June October December

Competed with

a Thrifty

1.15

All other stations1.25

1.351.45

1.151.501.401.301.20

Figure 10.3. Los Angeles gas prices: treatment and control Source: Hastings (2004).

gas stations that converted to company-owned gas stations and those that becamedealer operated These results suggest that vertical integration and the disappearance

of an independent retailer is correlated with, perhaps even causes, higher prices inthis particular market.25

Before we conclude that the vertical merger causes higher prices, Hastings firstnotes that a simple descriptive analysis such as that provided in figure 10.3 ignoresthe effect of changing conditions in the market It could be that demand or costsincreased in those areas where Thrifty was present, confounding the effect of thechange in ownership structure To control for this, Hastings runs the followingfixed-effects regression:

pjt D  C ˛j C ıCity;tC cjtC  zjtC "jt;where  is a constant, ˛j is a station-specific effect, ıCity;t are a set of city/quarterdummies, cjt is an indicator of whether the station becomes a company operatedstation (as distinct from a dealer-operated station under lease), zjtis an indicator ofwhether or not the station competes with an independent gas station, and "jtis theerror term The fixed effects control for potential omitted variables that determineprices and they turn out to be quite important in the regression, indicating thatthere are in fact many unobserved determinants of the price at the local level Theresults of three variants of the regression are provided in table 10.1 The estimates incolumn 3 suggest that there is a 5 cent increase in retail prices when there is no longer

an independent station in the market There is no additional statistically significanteffect of becoming a fully integrated company-operated gas station compared withbecoming a dealer with a contractual relationship with the upstream company

25 While the direction of these results is not in contention, Taylor et al (2007) argue that in their closely related data set the actual price difference appears to be considerably smaller, approximately 1 cent per gallon before and after the transaction, a net effect of 2 cents per gallon rather than Hastings’s difference

of 5 cents per gallon Since the FTC paper is a relatively new one, the reader may find that Professor Hastings has subsequently responded to their paper Whatever the resolution of that debate, the ideas behind Hastings’s approach remain of substantial interest.

Table 10.1. Fixed-effects estimation of the effect of Thrifty’s acquisition.

(0.0421) (0.0415) (0.0287)Company operated 0.1080
0.0033
0.0033

Hausman test for random effects:

Hausman’s M value: 622.296 Prob > M : 0.000

aStandard errors in parentheses Source: Table II, Hastings (2004).

One area of particular concern arises from the fact that the merger results in achange of an unbranded product into a branded product and that may explain theprice rise, once again independent of any effects of the vertical integration on prices.The potential importance of branding is fully explored in the paper In an attempt toaddress this concern, Hastings breaks up the “treated group,” those gas stations thatwere formerly competing with an independent station, into gas stations with a strongbrand presence in California (Chevron, Shell, or Unocal), gas stations with mediumbrand presence (Exxon, Mobil, or Texaco), and gas stations with low brand presence(Beakon, Circle K, or Citgo) The effect of the disappearance of an independentcompetitor is stronger on those gas stations that have a lower brand presence and issmallest on the gas stations with established brands That suggests that the brandingeffect may indeed be rather important in driving the price increase Inasmuch asthe structure of vertical ownership is changing the degree of product differentiation

downstream, we could go so far as to argue that the increase in gasoline retail pricesafter the sale of Thrifty is a vertical effect However, the empirical exercise leaves

us with a distinctly more subtle question than the one we began with Despite theapparently extremely clean natural experiment in the data, to evaluate the impact

of the change in vertical structure on consumer welfare, in this case, because thevertically integrated firm already had a downstream brand we must evaluate whetherthe increase in branding (probably appropriately considered an aspect of productand/or service quality) is sufficiently valued to justify the price increase that weidentified We examine methods capable of evaluating such trade-offs in the nextsection

10.2.2.2 Estimating the Effects of Vertical Integration in the Market for Cable TV

Another interesting attempt to empirically measure the effect of vertical tion examined the U.S cable television industry and is provided by Chipty (2001).The paper looks at the effect of vertical integration between programming and dis-tribution services in cable television However, the essence of the paper uses twomethodologies which may each generally be useful for assessing the way in whichconsumers trade off the various combinations of price and quality which may arisefrom vertical contracting For example, a number of the models we examined inthe first part of the paper found that vertical contracting arrangements may result

integra-in higher prices but also perhaps greater service provision Such a defense wouldprobably be easy for any company whose vertical restriction was in fact anticom-petitive to at least allege Chipty (2001) provides us with two approaches to assessthis argument First, she uses an approach which is familiar from earlier chapters,specifically she examines a reduced-form regressions of equilibrium outcomes, inthis case a measure of quality and a measure of price, on demand and cost variablestogether with variables that capture the extent of vertical integration In doing soshe hopes to capture the independent effect of vertical integration on equilibriumoutcomes Second, she suggests a method for at least telling whether consumers do

in fact sufficiently value the services being provided to make consumers better off

in vertically integrated markets all else equal Each of these pieces of evidence areprovided from an industry where there was at least anecdotal evidence that verti-cal integration of cable system companies and content providers was resulting inrefusals by cable operators to carry rival programming services

Before discussing these two methods, it is worthwhile spending a few momentsproviding a little background on the industry To that end, the vertical structure ofthe cable television industry in the United States is broadly as follows Producercompanies such as Paramount or Universal sell their media productions (films, TVshows) to program service providers such as HBO or AMC, which in turn sell theprogram content to cable system operators Those cable operators are typically local

monopolies in their markets.26They provide the final consumers with different sets

of packaged channels at given prices Chipty considers the vertical integration ofservice providers such as HBO (upstream) with cable system operators such asComcast or TCI (downstream)

One very nice feature of this market for empirical work is that there are lots ofdistinct local markets for cable providers Moreover, those local markets exhibitdifferent degrees of vertical integration between the program services and the cablesystem operators

Chipty (2001) wishes to investigate the actual effect of vertical integration onforeclosure, a difficult problem in a context where there is no one–zero classificationfor markets which have been “foreclosed.” Instead of attempting to construct such

a variable and perform a regression analysis, she proposes to study the way thatobserved market outcomes at the retail level vary across vertically integrated andnonvertically integrated markets In doing so she hopes to consider in the round theeffect of vertical integration on outcomes and ultimately consumer welfare She uses

1991 data from the Television and Cable Factbook The database comprises 11,039cable franchises in the United States that are operated by 1,919 cable systems, whichare in turn owned by 340 cable system operators, which may own more than onecable system brand The data provide information on the structure of ownership, thechannel capacity, the number of homes with access to cable in the franchise area, thecable system’s program offer, the price, and the quantity of subscribers There arealso data on 133 program services (excluding pay-per-view and satellite) includingeight premium services such as movies, one general entertainment, and two sportprogramming services Data on the demographics of the market such as populationsize, fraction over 65 years old, or household size were taken from the 1988 Cityand County Data Book and USA Counties 1994

Vertical integration occurs in this context when the cable operator owns any part

of a program service that serves the franchise area In principle, vertical integrationcould lead to foreclosure and harm to consumers from increases in the prices ofthe final good On the other hand, it could increase product quality since there is ahigher incentive by the cable provider to offer premium channels as they will alsoget the profits from the sale of that more expensive content

The data used in Chipty (2001) show that, on average, a cable system provideroffers fifteen basic service channels and slightly more than three premium contentchannels Chipty uses the term “Basic system” to mean a cable system that is inte-grated with a basic content provider Such vertically integrated basic systems offertwenty basic channels and four premium channels on average Cable systems inte-grated with premium content providers (which Chipty calls premium systems) offernineteen basic channels and slightly fewer than three premium channels A sim-ple look at these descriptive statistics therefore suggests that integration with basic

26 More recently, some companies have undertaken a process of “overbuilding” cable franchise areas

so that there are duopolies in some market areas.