Title: Margin squeeze and predatory prices by multiproduct firms
1Margin squeeze and predatory prices by
multiproduct firms
- Encore, The Hague, 3 April 2008
- Dr Andrea Coscelli
2Overview
- Margin squeeze cases with multiproduct firms
- Some additional considerations on the application
of predatory pricing tests to multiproduct firms
3Rule of reason vs. simple rules
- Discussion of possible margin squeeze or
predatory pricing cases involving multiproduct
firms requires a complex economic analysis under
a rule of reason approach - There are no simple rules and a number of
alternative approaches need to be tested to
assess whether in fact there is a problem (and
how significant it is) - This creates uncertainty for the dominant firms
involved and problems for the Courts undertaking
judicial reviews
4Key elements (1 of 2)
What is a margin squeeze?
Competitive market
Bottleneck market
Before margin squeeze
After margin squeeze
5Key elements (2 of 2)
- For the margin squeeze test to be passed
- Retail Price Wholesale Price Downstream Costs
Competitive Return - Key modelling inputs
- Downstream (retail) prices
- Upstream (wholesale) prices
- Downstream (retail) costs including a competitive
rate of return for the downstream activity
6Key precedents in broadband and broadcastingAll
involving multiproduct firms
- CFI in Wanadoo
- EC in Telefonica, Wanadoo and Deutsche Telekom
- Oftel/Ofcom in BT vs. Freeserve (2003), BT
Together (2004) and BT IPStream (2004)
investigations - OPTA and NMa Price Squeeze Guidelines (2001)
- ACCC in Telstra ADSL and 2005 guidelines
- OFT in BSkyB (2002)
- EC and Italian Telecom Regulator in Sky Italia
7Key modelling assumptions
- There are at least 10 significant methodological
aspects to be considered when modelling a
specific margin squeeze case - A number of alternative assumptions can be made
at each step when choosing the methodology - Choice driven by features of the specific market
being investigated, case precedent, available
data and (at least partially) policy
considerations of the relevant agency
81. Downstream cost standard EEO or REO?
- Equally efficient operator (EEO)
- Pros uses incumbents own downstream costs
favours efficient entrants can be modified to
approach the REO standard - Cons economies of scale/scope ? small scale
entrants might be excluded - Reasonably efficient operator (REO)
- Pros uses downstream costs of reasonably
efficient entrant, so can facilitate entry on
small scale - Cons entrants data not readily available,
additional assumptions must be made incumbent
may set wholesale prices to extract surplus
generated by more efficient entrants (assuming
incumbent can price discriminate) - Features of the retail competition (e.g., do all
players offer similar bundles/compete across all
segments?) will play a role in the choice of the
cost standard to use
92. Downstream cost standard AVC, LRIC or FAC?
- Average variable costs (AVC)
- Pros correct cost standard for assessing short
term price points minimal data requirements
approaches LRIC as review time period increases
used as the cost standard for antitrust predation
cases - Cons not appropriate for assessments of long
term price points results affected by short term
variations in costs ignores common costs - Long run incremental costs (LRIC)
- Pros correct cost standard for assessing long
term price points can be adapted to include
efficient replacement costs of assets (forward
looking LRIC FL-LRIC) or a contribution to
common costs (LRIC) - Cons greater data requirements than FAC
limits the incumbents short term pricing
flexibility relative to AVC ignores common
costs which may inhibit entry when the incumbent
enjoys economies of scope (but it might be
adjusted to include a contribution to common
costs)
102. Downstream cost standard AVC, LRIC or FAC?
- Fully allocated costs (FAC)
- Pros easy to implement from existing accounts
possibly the only option in markets not subject
to sector regulation reasonable measure when
common costs are a large proportion of the costs
promotes entry as FAC gt LRIC gt AVC (hence a
conservative standard) - Cons allocation of common costs is arbitrary
further limits the incumbents pricing
flexibility effectively implies a higher margin
requirement - Features of the retail competition (e.g., do all
players offer similar bundles/compete across all
segments?) will play a role in the choice of the
cost standard to use
112. Downstream cost standard AVC, LRIC or FAC?
- Precedents
- AVC not used in telecoms (but many precedents in
antitrust cases) - LRIC Oftel/Ofcom in BT Freeserve and BT
Together EC in Telefonica ACCC - FAC Ofcom in IPStream (due to data
availability) OFT in BSkyB (as common costs were
large) work for Sky Italia
123. Historic costs or DCF approach?
- Historic costs
- Pros easiest modelling approach as uses
available data can be adapted to current costs
can be adapted to economic depreciation (e.g. to
amortise subscriber acquisition costs (SACs)) - Cons straight line depreciation may allocate
costs over low utilisation in early years - Discounted cash flows (DCF)
- Pros explicitly specifies a path for cost
recovery over time preferable in new and
emerging markets with high SACs - Cons based on forecasts ? uncertainty could
conceal monopoly rents if long period is used - Precedents
- Historic costs EC in Wanadoo, Telefonica and DT
Ofcom in BT ISPs OFT in BSkyB - DCF Oftel in Freeserve EC in Telefonica ACCC
in Telstra ADSL - Cohort analysis used by Oftel in BT Freeserve
134. Time period for profitability assessment
- The relevant time period in margin squeeze
assessments is the economic life of key
investments - If SACs are key investments ? approximate average
subscriber life - Shorter period
- Pros data more readily available precludes
incorporation of monopoly rents - Cons insufficient time allowed for recovery of
SACs ? higher margin requirement than necessary - Longer period
- Pros allows appropriate time for recovery of
SACs - Cons data less readily available/reliable risks
incorporation of monopoly rents
144. Time period for profitability assessment
- Steady state assumption
- If market is in steady-state then time period is
much less relevant (use last 12 months of
management accounts as a reasonable proxy) - Precedents
- EC in Wanadoo (4 years) EC in Telefonica (5
years in DCF approach) Oftel/Ofcom in BT
Freeserve and BT IPStream (5 years) ACCC in
Telstra ADSL (3 years) OFT in BSkyB (SACs
depreciated over 10 years) - Features of retail competition not particularly
relevant for the choice (dynamic nature of retail
competition on the other hand is key)
155. Interconnection costs
- Include costs incurred by entrants (or by the
incumbent) in interconnecting with the incumbent,
but not incurred when the incumbent supplies
itself - Pros removes the incumbents cost advantage and
promotes entry - Cons the incumbent must set retail prices above
its integrated (wholesale and retail) costs of
supplying the service ? promotes inefficient
entry and higher prices for consumers - Precedents
- Regulators tend to include these costs (e.g.
Ofcom in BT IPStream and BT Together) - Features of retail competition not particularly
relevant for the choice
166. Level of aggregation how to deal with
variegated retail offers and retail discounts
- Level of aggregation options
- Test each product? Test each service (including
all products)? Or test the relevant market
(including all services supplied in the market)? - Test individual customers? Groups of
customers? Or all customers in the relevant
market - Higher level of aggregation ? greater
flexibility for recovery of common costs across
services/customers, but may inhibit entry on a
smaller scale/scope than the level of aggregation - Levels of aggregation depend on cost standard
- EEO standard aggregate over all services and
customers that the incumbent supplies in the
relevant market - REO standard aggregate over all
services/customers that a REO would be likely to
offer/target
176. Level of aggregation how to deal with
variegated retail offers and retail discounts
- Combinatorial tests proposed by some regulators
- Revenues from each service/customer incremental
costs of supplying that service/customer - Revenues from all services in a relevant market
sum of incremental and common costs of those
services - Precedents
- Regulators tend to prefer highly disaggregated
tests (e.g. OPTA in Price Squeeze Guidelines
Oftel/Ofcom test at service level) EC in
Deutsche Telekom separately tested access and
calls) - Sometimes greater aggregation occurs (e.g. EC in
DT aggregated over analogue, ISDN and ADSL
access EC in Telefonica conducted the test on
the aggregate mix of services in the relevant
market OFT in BSkyB and EC in Sky Italia
aggregated over basic and premium channels)
186. Level of aggregation how to deal with
variegated retail offers and retail discounts
- At the product level one would need to test
whether the revenues of each retail product cover
the incremental costs of that product - In practice especially in telecoms it is very
difficult to assess from the accounts what costs
are incremental to a specific product if the
definition of a product is very disaggregated
(e.g., for retail broadband we define a product
as a brand/speed/any additional specs (e.g.,
limit on download volumes)) - How to assess situations where a minority of
products make a negative contribution (possible
reasons include legacy products, new products
etc)? - Rationale for highly disaggregated tests
unconvincing (high implementation costs and
limited benefits)
197. Bundled retail offers whether to test at the
level of the bundle or test each element
- For example, most telecom and pay TV incumbents
offer bundles of broadband, voice and pay TV
services - EEO standard test whether the bundle revenues
cover the bundle costs - A possible alternative approach by regulators
- A regulator might be inclined to test each
element of the bundle, e.g. test whether the
incremental revenue from broadband in the bundle
(attributing all the bundle discount to the
broadband service) covers the costs of a
broadband-only operator - No correct way in economics of pricing a single
product in a bundle (i.e. no correct way of
allocating the discount across products) - This would be a fairly extreme REO assumption
207. Bundled retail offers whether to test at the
level of the bundle or test each element
- The Spanish Telecommunications regulator (CMT)
for instance recently allowed bundled offers from
Telefonica that satisfy a margin squeeze test as
a whole and for which there is a sufficient
number of alternative bundled offers from other
market operators. - Replicability of the bundles by competing
operators is key. If efficient rivals can afford
bundle vs. bundle competition, there is no need
to test at a more disaggregated level. - Focus should be on assessing whether there are
any existing advantages by dominant companies
(for instance because of vertical integration)
that could not be replicated by their rivals. - Actual competition in the downstream markets is
key to choosing the approach to take
218. Retail product differentiation
- Implicit assumption in margin squeeze tests is
that incumbent and entrant retail products are
homogeneous - If retail products are differentiated, unclear
what is meant by an EEO or REO and product
differentiation must be taken into account (the
assessment needs to be linked to the view taken
on product aggregation for the tests)
229. Discounting at the wholesale level
- Issue What wholesale charge to use if volume
discounts are offered at wholesale level - Lowest wholesale charge (consistent with EEO
standard) - Pros Benefits consumers by passing on economies
of scale - Cons May inhibit entry as entrants need to match
the incumbents scale - Highest wholesale charge (consistent with REO
standard and usually preferred by regulators) - Pros Promotes entry by ensuring viability for
small scale entrants (with a view to better
outcomes for consumers in the long term) - Cons Encourages inefficient entry (and higher
prices for consumers in the short to medium term) - Product range less relevant here as this usually
relates to volume discounts
2310. Appropriate downstream margin
- Downstream margin
- ROT or WACC?
- ROT may be appropriate for downstream
distribution company with limited fixed assets
(OFT in BSkyB UK MMC in mobile phone charges
inquiry) - WACC widely used by regulators (e.g. Oftel/Ofcom
in Freeserve and IPStream AGCOM in Telecom
Italia cases EC in Telefonica) - Appropriate measure of WACC?
- Ideal would be to use WACC for a stand alone
retail operation but stand alone WACC difficult
to calculate for a division of an integrated
company - In practice regulators often use company wide
WACC as proxy for WACC of retail operation
2410. Appropriate downstream margin
- Downstream margin issues
- Incumbents or entrants WACC?
- The incumbents cost of capital may be lower than
entrants, as integration reduces risk an
adjustment may be necessary to reflect a REO
standard - Appropriate rate of return?
- BSkyB and Sky Italia 1.5 ROT
- Freeserve and IPStream BT Groups WACC of 13.5
- Telefonica WACC of 15.7
25Predatory pricing tests for multi-product firms
- Must ask whether the revenues from any
combinations of the firms products fall short of
the combined avoidable costs of the products.
These are the combinatorial cost tests originally
envisaged by Baumol (1996) Predation and the
Logic of the Average Variable Cost test, Journal
of Law Economics. - Avoidable costs refer to the costs that a firm
could have avoided but for the practice of
predatory pricing - This is the same cost test envisaged in the
Commissions Discussion Paper on Art. 82
(December 2005) - Structure to the test needs to be given by
features of downstream competition in the
specific market (e.g., this will drive the choice
of the unit of analysis)
26Predatory pricing tests for multi-product firms
- Example
- Airline sells two products 40 first class and
200 economy seats - Costs common to each product are fuel and pilot
salary 15,000 - AAC of each product is the food for each class of
passenger (none of the common fixed cost enters
the incremental cost of either product) - AAC First Class (40 x 20) 800
- AAC Economy (200 x 5) 1,000
- Suppose first class fare is 100 and economy fare
is 60 (so AAC are covered) - Combined incremental revenue (40 x 100) (200
x 60) 16,000 - Combined avoidable cost 800 1,000 15,000
16,800 - Predatory pricing test is failed but it this
the correct unit of analysis (a single flight
segment)?
27Predatory pricing tests for multi-product firms
- There is a general problem in applying predatory
pricing tests to multiproduct firms cost
accounting systems generate detailed product
specific information that is often inappropriate
for price/cost tests for predation - When the analysis of predatory pricing concerns
relate to only one of the products produced by a
multiproduct firm the common costs will not
normally be regarded as avoidable. (See the EC
Deutsche Post cases for instance) - When the allegation of predation concerns the
pricing of a bundle of products that constitutes
the relevant product market, the assessment will
be whether the price of the bundle is below
average avoidable costs for the bundle.