4.

3 COST BENCHMARKS FOR EXCLUSIONARY CONDUCT 193

Table 4.3 Average avoidable cost

Initial production 150 units
Expanded production following entry 200 units
Alleged exclusionary output 50 units
Variable costs (€)
Variable cost on initial units 8
Variable cost on additional units 10
Fixed costs(€)
Cost of plant that can be avoided (machines that don't need to run) 500
Cost benchmark(€)
Average variable cost 8.5
Marginal cost 10
Average avoidable cost 20

in the Intel rebates case (discussed in more detail in section 4.6) there was extensive
discussion of what were the relevant costs for AAC. 23
In its Guidance on Article 102, the Commission has proposed that prices below AAC
indicate that the dominant company is sacrificing profits in the short term and that an
equally efficient competitor cannot serve the targeted customers without incurring a
loss. 24 AAC therefore establish es the price floor for the provision of a good or service to a
targeted group of customers for a limited period of time. Avoidable cost makes sense in
m ost cases as it gets at the 'sacrifice' of predation: the company would have been better off
producing nothing and avoiding these costs than producing the units of output at a price
below AAC.

4.3.4 LONG-RUN IN CREMENTAL COST

LRIC is the ch ange in total costs resulting from the production of an increment in the
quantity of o utput, which can be the whole output of the product in question or just the
increm ental output associated with the exclusionary conduct. The Commission's
Guidance refers to LRIC as 'the average of all the (variable and fixed) costs th at a com-
pany incurs to produce a particular product:25 But if we want to understand the profit-
ability of the exclusionary conduct it can b e more appropriate to calculate the LRIC of
just the additional output associated with exclusion. You can see here that the incre-
ment m atters. Unlike AAC, LRIC includes all product-specific fixed costs even if those
costs were sunk before the period of pred atory pricing, ie, LRIC includes b oth recover-
able and sunk fixed costs.

" Tntel (Case COMP/C-3/37.990), Commission Decision of 13 May 2009 D(2009) 3726 fina l, [ 1036)- (1153 ].
" European Commission (2008), 'Guidance on the Commission's Enforcem ent Priorities in Applying Article
82 EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings; December, [26).
25
Ibid., at n 2.
194 ABUSE OF DOMINANCE

Continuing with our example, think of the dominant company responding to the
threat of entry-first it t akes out newspaper advertising to promote its product and
expands its factory space, and subsequently it lowers prices and expands output. The
advertising and factory expansion are associated with the alleged exclusionary output
and form part of LR IC. They do n ot form part of A AC-even if the additio nal 50 units
are never produced, these costs can no longer be avoided once incurred. In our example
the LRIC of the incremental output is €40-(€500 for plant+ €500 for advertising+
€500 for factory space+ 50 units at € 10 per unit), divided by 50 units. In our example it
is also easy to calculate the LRIC of the whole product (ie, all 200 units) rather than the
relevant increment, which in this case is €38.50- €2,000 for plant+ €2,000 for advertis-
ing + €2,000 for factory space + (150 units at €8 per unit + 50 units at €10 per unit),
divided by 200 total units. See Table 4.4. You can see that LRIC in this example comes
close to ATC. A main difference between the two (not reflected in our example) is that
ATC is based on current or past costs, whereas LRIC it is forward-looking and may
therefore incorporate any expected shifts in costs (eg, in telecoms the costs o f network
and equipment keep falling, so that the LRIC fo r a brand new telephony network may
be lower than the ATC of the current network). Another difference (which is reflected
in the example) is where the company produces multiple products and the ATC of a
product may include an allocated part of a company's overhead costs, while the LRIC of
that particular product does not.
Pricing below LRIC can be economically rational apart from any exclusionary effect.
As LRIC includes all product-specific sunk costs, a company pricing below that cost can

Table 4.4 Long-run incremental cost

Initial production 150 units
Expanded production following entry 200 units
Alleged exclusionary output 50 units
Variable costs(€)
Variable cost on initial units 8
Variable cost on additional units 10
Fixed costs (€)
Cost of plant that can be avoided (machines that don't need to run) 500 for 50 units;
2,000 for a.II units
Cost of brand advertising (sunk) 500 for 50 units;
2,000 for all units
Common cost of factory attributable to this product 500 for 50 units;
2,000 for all units
Cost benchmark(€)
Average variable cost 8.5
Marginal cost 10
Average avoidable cost 20
Long-run incremental cost 40 for 50 units
38.5 for all units
 4.3 COST BENCHMAR K S FOR EXCLUSIONARY CONDUCT 195

still generate a p ositive cash fl ow (ie, it would cover its variable costs and hence make a
contribution to the recovery of its already-sunk fixed costs). Such sales, which a LRIC
or ATC standard might condem n as exclusionary, would therefore be po tentially profit-
able and hence might reflect commercially rational b eh avio ur. In its Guidance, the
European Commission h as therefore proposed that prices ab ove AAC but b elow LRIC
indicate that an equally efficient competitor could be foreclosed from the market- there
is n o presumption of foreclosu re as is the case with prices below AAC.26 LRIC serves in
m ost cases as a price floor above which concerns about exclusionar y b elow-cost pricing
are unlikely to materialize.
LRIC has b een deem ed to b e of p articular relevance to n etwork industries (as is
also reflected in price and access regulation in these industries-see Chapter 9). In its
1998 n otice on th e application of th e competition rules to the telecommunications
industry, the Commission indicated that LRIC could be an appropriate cost floor for
predation cases:
In the case of the provision of telecommunications services, a price whkh equates to the
variable cost of a service may be substantially lower than the price the operator needs in
order to cover the cost of providing the service. To apply the AKZO test to prices which are
to be applied over time by an operator, and which will form the basis of that operator's deci-
sions to invest, the costs considered should include the total costs which are incremental to
the provision of the service. In analysing the situation, consideration will have to be given
to the appropriate time frame over which costs should be analysed. In most cases, there is
reason to believe that neither the very short nor very long run are appropriate.

In these circumstances, the Commission will often need to examine the average incremen-
tal costs of providing a service, and may need to examine average incremental costs over a
longer period than one year. 27

In D eutsche Post (2001), the Commission fo und the incumbent postal operator in
Germany guilty of predatory pricing (and of cross-subsidy and granting fidelity reb ates).
It determined that:

[A)ny service provided by the beneficiary of a monopoly in open competition has to cover
at least the additional or incremental cost incurred in branching out into the competitive
sector. The Commission considers that any cost coverage below this level is predatory pric-
ing which falls foul of Article 82 of the EC Treaty [now Article 102 TFEU].2•

4.3.5 AVERAGE TOTAL COS T

ATC is the sum of fixed and variable costs, divided by total output. In our example the
dom inan t company m anufactures several products. It h as allocated a share of the cost

26
European Commission (2008), 'Guid an ce on the Commissio n's Enforcement Priorities in Applying Article
82 EC Treaty to Abusive Exclusio nary Conduct by Dominant Undertakings; December, [26).
27
European Commissio n ( 1998), 'Notice o n the application of competition rules to access agreements in the
telecommunicatio ns sector-Framework, Relevant Markets and Principles' [1998) OJ C265, [114]- (1 15).
28
Deutsche Post AG (Case COMP/35. 14 1) Commission Decision 2001/354/EC, [2001 ) O J 1125/27.
 ABUSE OF DOMINANCE

of its overheads t o the product which is subject to the exclusion ary conduct claim. The
overheads would include a share of investment in company-wide branding and head
office costs. Thus we have €7,000 total fixed costs for this product, €1,000 of w hich is the
allocated overhead. The €1,000 is n ot included in the LRIC because it is not increm ental to
producing the product in question. So the ATC is €43.50-(€7,000 total fixed cost+ (150
units at €8 per unit+ 50 units at €10 per unit)), divided by 200 total units. See Table 4.5.
Prices above ATC are n ot exclusionary under the as-efficient competitor test since
they allow any efficient competitor to m ake a profit. There are two possible challenges
to this conclusion. One is in the case of structured or targeted discounts, where prices
can exceed ATC if m easured against all units sold to a customer, but b elow cost (even
below AAC) if measured against the 'contestable' units on ly. We explain this p oint in
section 4.6. The second challenge is where the dominant company enjoys positive net-
work effects or econ omies of scope that a competitor cannot currently replicate. In this
case, while the competitor's costs are higher than those of the dominant company, a
competition authority may take a dynamic view on costs, ie, whether in the absence of
the alleged exclusionary conduct or when competition is m ore established, the rival
could overcome its disadvantage and become 'as efficient' -see sections 4.2 and 4. 7 on
the concept of a 'hypothetical reason ably efficient' competitor that has been used in
place of the 'as-efficient' competitor.
What is the conclusion in this example? The dominant company's initial price of €45
covered ATC. The n ew price of €25 is above the AAC of €20 but below the LRIC of €40
(or €38.50 if the LRIC is m easured for the w hole product). It is therefore capable of

Table 4.5 Average total cost

initial production 150 units
Expanded production following entry 200 units
Alleged exclusionary output 50 units
Variable costs(€)
Variable cost on initial units 8
Variable cost on additional units 10
Fixed costs (€)
Cost of plant that can be avoided (machines that don't need to run ) 2,000
Cost of brand adve rtising (sunk) 2,000
Common cost of factory attributable to this product 2,000
Allocated overheads (not incremental to this product) 1,000
Cost benchmark(€)
Average variable cost 8.5
Marginal cost 10
Average avoidable cost 20
Long-run incremental cost 40 for 50 units;
38.5 for all units
Average total cost 43.5
 4.3 COS T B E N C HM A RK S FOR EX C LU SIONARY COND UC T 197

foreclosing an as-efficient entrant since prices are below the level of costs that this
entrant would have to incur to build up the same production capability. However, pric-
ing below LRIC but above AAC can be commercially ration al without intent to fore-
close. In addition , an efficient competitor that is already in the m arket can also price at
€25, cover its avoidable costs and m ake a €5 p er unit contribution to the recovery of its
already sunk fixed costs. Cost benchmarks are useful in this type of pricing abuse case,
but must be cons idered in conjunction with other evidence on market structure and
conduct, as we shall see in the remainder of this chapter.

4.3.6 PROFITABILITY T ESTS FOR BELOW- C OST PRICING

One sp ecific issue in predation and m argin squeeze is the allocation of costs over time-
the question being what costs are within the alleged period of below-cost pricing. Where
genuine investm ent is required in a particular year, it would be incorrect to assum e that
losses in that year represent below-cost pricing, since that investment is relevant to
revenues not only in that year but in future years. Th e treatm ent of costs sho uld reflect
these conditions. Rather than price-cost comparisons you can an alyse the profitability
of the product over a longer timeframe, using the techniques of profitability analysis
described in Chapter 3. Investments sho uld be am ortized or depreciated according to
their economic use-for example, operational costs (marketing or promotional costs)
that are front-loaded within a project lifetime, but m ay justifiably be recovered over a
number of years. The correct way to account fo r these is to amortize them over a period
linked to the underlying o utput they are supporting. Only then m ay any remain ing
observed losses be treated as an indication of below-cost pr icing.
A potential problem of profitability analysis in exclusion cases is that it does not
easily distinguish between future profits arising from normal com petition and futu re
profits that result from a successful predatory strategy. We com e back to this in section
4.4. We n ote that considering profitability in these cases is still more appropriate than a
simple price-cost comparison; it can provide useful insight into the drivers of the losses
that yo u observe in the early years (do they reflect a genuine investment to stimulate
subsequent dem and?), and the profits you observe in subsequent yea rs (do they d epend
on the exit of competitors?).
Another challenge is that th e results of the profitability test-or indeed of any price-
cost test- m ay be sensitive to specific assumptions or accounting approaches. In a
margin squeeze case relating to the broad caster BSkyB, the Office of Fair Trading
(OFT) included as a relevant cost the investment in supplying set-top boxes to n ew
subscribers, and amortized this cost over an average subscriber lifetime of ten years.29
However, alternative amortization perio ds migh t also have been valid- for exam ple,
based on the useful lifetime of a set-top box (wh ich is less than ten years due to the

29
Office of Fair Trading (2002), 'BSkyB Lnvestigation: Alleged In fringement of the Chapter II Prohibitio n:
OFT Decision, C A98/20/2002, 17 December, (463].