IN THE COMPETITION TRIBUNAL OF SOUTH AFRICA
Case No: 39/AM/MAY06
In the matter between:
PRIMEDIA LTD First Appellant
CAPRICORN CAPITAL PARTNERS (PTY) LTD Second Appellant
NEW AFRICA INVESTMENTS LTD Third Appellant
And
THE COMPETITION COMMISSION Respondent
AFRICAN MEDIA ENTERTAINMENT LTD Intervenor
Panel : D Lewis (Presiding Member), N Manoim (Tribunal Member), and Y Carrim
(Tribunal Member)
Heard on : 30 January 2008
Delivered on : 09 May 2008
REASONS
[1] This merger has been referred back to this panel of the Tribunal for reconsideration,
by the Competition Appeal Court (‘the Court’), following a review of our decision in February
2007(the ‘February decision’) to approve the merger without conditions. 1
Background
[2] A brief history of this merger is necessary to explain the context of the matter being
sent back to us on review. This merger is an intermediate merger, and thus, in terms of the
Competition Act, one that can be approved by the Competition Commission. 2 On 26 April
2006, the Competition Commission in its report found the merger would lead to adverse
unilateral anticompetitive effects, but thought that these effects could be remedied through
the imposition of behavioural conditions.
1 See decision of the CAC, African Media and Entertainment Limited v David Lewis N.O. and six others. Case
No: 68/CAC/Mar07.
2 Section 14(1) (b).
1
[3] Unhappy with the way these conditions had been framed, Primedia brought the
matter to us by way of consideration in terms of section 16(2) (b) of the Act. Primedia asked
us to approve the merger without conditions or alternatively, to approve the merger subject
to a revised set of conditions that it proposed would remedy the competition harm, assuming
we found the merger to be anticompetitive.
[4] At the commencement of our hearing the Commission altered its position, and
decided to oppose the approval of the merger, contending that neither the conditions it had
originally proposed nor the new ones proposed by the merging parties cured the competition
harm identified. African Media Entertainment Limited (AME), a firm which owns interests in
radio stations, applied for and was allowed to intervene in our proceedings. Like the
Commission, its stance was that the merger be prohibited.
[5] We heard the matter during the course of 2006. Argument was heard on 18 January
2007 and we gave our decision in February 2007. We decided to approve the merger
without conditions.
[6] Following our February decision, AME took our decision on review to the Competition
Appeal Court. Although intervenors are not given the right to appeal decisions of the
Tribunal in respect of mergers, the Court decided that they nevertheless, if they have been
admitted as parties in the proceedings, enjoy the right to bring a review application. AME
was successful in reviewing our decision on a point of law, but the Court declined to decide
the matter on the merits, and referred it back to us for determination as to whether the
merger should be approved or not and if so, on what basis. 3
[7] Following the outcome of the review we invited all three parties (the Commission, the
merging parties and AME) to make further written submissions to us in the light of the CAC
decision. We heard oral argument based on these submissions on 31 January 2008.
Basis for the Court’s decision on review
Basis for the Court’s decision on review
[8] In our February decision we noted that the merger between Primedia and Nail, which
was the primary transaction which triggered the obligation to notify it as a merger in terms of
section 12(1) of the Act, brought in its wake, the acquisition of an indirect interest by
Primedia, in a company called Kaya FM (Pty) Ltd, which in turn owns a radio station
operating in Gauteng, known as Kaya FM. 4
[9] It remains common cause that there is no competition concern about the direct
3 See paragraphs 6162.
4 This interest continues to be held by Nail to avoid the exercise of preemptive
rights by the other shareholders as we explain more full in our February decision. In
terms of Section 12(1) (a) of the Act “… a merger occurs when one or more firms directly or
indirectly acquire or establish direct or indirect control over the whole or part of the business of another firm.”
2
acquisition of Nail and so we need not go into this aspect. 5 The issue is the effect of the
indirect acquisition– where Primedia, via New Africa Investments Ltd (“Nail”), acquires the
right to vote Nail’s 24.9% stake in Kaya FM. Although Primedia will be able to vote the full
24,9%, its economic interest in Kaya FM is less extensive, amounting to at most 18,17% of
the station’s equity; the balance of the economic interest is held by Capricorn. 6
[10] When the matter came before us by way of reconsideration the issues were limited to
the consequences of the indirect acquisition by Primedia of a stake in Kaya FM through its
acquisition of control of Nail. Expressed more simply, the alleged significance of the merger
from a competition perspective is that although Nail’s stake in Kaya FM remains unchanged,
the ownership of Nail has changed. This is alleged to have competition consequences
because the Nail stake is now owned, and to the extent that this may prove material,
controlled, by a company that owns radio stations that may be considered to compete with
Kaya FM.
[11] Although radio stations compete for both listeners and advertising, it was common
cause in this hearing that any competition concerns that might arise, would be in the market
for advertising. The Commission and AME analysed the advertising markets implicated by
the merger, defined them narrowly, and found them to be highly concentrated and
composed of few firms, some with interlocking interests, and accordingly, recommended
prohibition. Primedia defined the markets much more widely and concluded that no
competition interest was at stake, but in the alternative, suggested that to the extent that it
was, the conditions that it had proposed would take care of the concerns without the need
for a prohibition. 7
[12] In our February decision we concluded that the stake being acquired by Primedia
indirectly, as a result of the Nail transaction, did not amount to it being able to control Kaya
FM either solely or jointly. We stated:
“Given this conclusion, it is unnecessary for us to consider, whether, if it had
acquired control, this would lead to an anticompetitive outcome.” 8
[13] We did not for this purpose conduct a substantive analysis ( i.e. the analysis
5 The direct acquisition involves Capricorn and Primedia acquiring control of Nail. We are advised that the
shares will be registered in the name of Capricorn, but will be beneficially owned by Capricorn Capital Partners
Investment (Pty) Ltd a wholly owned subsidiary of Capricorn. Although Primedia and Capricorn will each own
50% of the rest of Nail, the arrangements in respect Kaya FM are an exception, and the split here is 73% to
Primedia ( hence 18,17% effectively) and the balance to Capricorn. The agreement states that Primedia has the
right to manage this interest. ( Clause 12) We make the assumption that this gives Primedia the right to vote this
interest without the need to obtain the consent of Capricorn. (See merging parties, bundle page 3 paragraphs 5
and 8).
6 See page 64 of the merging parties’ Economic report.
7 These conditions which are set out in our February decision ( see Appendix 2) are designed to ensure that
Nail’s right to appoint a director to the Board of Kaya FM is constrained in a manner to ensure that the director
appointed is an independent director in a manner understood in company law. Primedia however was unwilling to
waive the right Nail has to appoint a director in terms of the Kaya FM shareholders agreement.
8 See Tribunal February decision paragraph 80.
3
contemplated in section 12A of the Act) save for analysing the possibility of joint control,
where certain assumptions about the market needed to be made to test various theories of
control.
[14] We were criticised by the Court on review for this approach, and for conflating the
issue of control, a jurisdictional issue, with the substantive enquiry mandated by section 12
A, the section that deals with the evaluation of a merger. As the Court expressed it:
“The question for the Tribunal was all about control. Manifestly control per se
is relevant to determining whether a merger exists and thus whether the
Tribunal has jurisdiction to examine the transactions in terms of the factors set
out in section 12 of the Act. Once a merger exists, the Tribunal must focus its
enquiry into whether the merger is likely to substantially prevent or lessen
competition. Again the nature and scope of control which fourth respondent
[Primedia] could exercise over Kaya is an important consideration in this part of the
enquiry. But alone it is insufficient. The mandated enquiry had to be undertaken
within the broader context of the market and the dynamics within such a market.” 9
(Our emphasis)
[15] The Court goes on later to refer to an article by two United States academics and
states: 10
“Significantly for the purpose of this dispute, O’Brien and Salop
comprehensively show that the previous assumption developed by Areeda
and Turner that “a noncontrolling interest has no intrinsic effect [threat] to
competition at all’ was incorrect...” 11
[16] The Court goes on to cite O’Brien and Salop’s article again with approval, in
particular their thesis that different forms of partial ownership can have different effects
depending on “ specific financial interests, corporate governance and market power ” 12
Since the Court has placed so much emphasis on this article, and by implication adopted its
analysis, we will first examine what the authors state in relation to partial acquisitions in the
9 Paragraph 51 of CAC decision, supra.
10 See Daniel O’Brien and Steven Salop, “Competitive effects of Partial Ownership: Financial interest and
Corporate Control” , Antitrust Law Journal, Volume 67, 2000, page 559.
11 Paragraph 53 of CAC decision, supra.
12 Paragraph 58 of CAC decision, supra.
4
context of the present case and then apply two of their scenarios to the facts of this case as
a form of section 12A substantive analysis. In so doing we will be examining whether under
a unilateral effects scenario, the acquisition of a noncontrolling stake in a rival firm may
have anticompetitive effects or whether under a coordinated effects analysis, the acquisition
would strengthen an existing coordination or increase the likelihood that the firms would co
ordinate.
[17] As will become clear during our analysis, we have proceeded on the assumption that
Primedia is not able to control Kaya FM. That was our finding in our February decision, and
although we have been urged that we are not bound to hold to that decision, we find no
reason to depart from it. What is novel, and was not available to us at the time of our
February decision, is a recent United Kingdom case 13 in which the question of what
amounts to control was considered; here the acquirer’s shareholding was at a similar level to
that of Primedia in this case, and yet the UK Competition Commission found that a merger
had been established, because the acquirer had acquired ‘material influence’ over the
target. The notion of ‘material influence’ is to be found in section 12(2)(g) of our Act, one of
the provisions which sets out when a person is deemed to control another firm. Because of
the factual and legal similarities to the present case, we have considered whether, in the
light of the UK approach, we should reconsider our approach to the question of control.
[18] Although we find the reasoning of the U.K.Commission in that case persuasive, we
show why the facts on which it based its decision, are not present in this case, and hence
we see no reason to alter our February conclusion that Primedia will not be able to control
Kaya FM. This discussion follows after our discussion of the substantive issues.
The O’Brien Salop thesis
The O’Brien Salop thesis
[19] In their article the authors are considering the competitive effects of partial ownership
transactions. Because under US law, as we examine later in the postscript, merger review
can apply to transactions even where control is not established, when US writers consider
‘partial stock acquisitions’ they use this term to apply to a range of transactions that fall short
of a transaction in which the ‘whole’ of the stock is purchased, but included in the spectrum
of partial ownership are ones our law would consider as conferring control and others not. 14
The authors thus take the spectrum of partial ownership possibilities, which in control terms
range from acquiring full control to having no control, and examine the different
anticompetitive outcomes that they may be associated with them.
13 See a report sent to Secretary of State, prepared by the Competition Commission which is titled “Acquisition
by British Sky Broadcasting Group PLC of 17.9 per cent of shares in ITV PLC. See ://www.competition
commission.org.uk/rep_pub/reports/2007/535itv.htm
14 It seems that the term ‘partial’ derives from the language of section 7 of the Clayton Act, the United States
primary merger review statute that forbids the acquisition of the “whole or any part of “….the stock or assets of a
corporation.
5
[20] The one position on that spectrum of partial ownership of significance to the present
case is the authors’ view that acquisitions by one firm of a noncontrolling interest in another
can lead to possible anticompetitive effects. The traditional view on this, expressed by
Areeda and Turner, was that unlike a complete merger, a noncontrolling interest constitutes
no intrinsic threat to competition. 15 Areeda and Turner also expressed the view that these
effects would in any event be impossible to measure. O’ Brien and Salop challenge this
conventional view and respond to the Areeda and Turner thesis, by arguing that there are
circumstances where partial acquisitions change the acquiring firm’s incentives, even when
the acquiring firm will have no control over the target’s actions. They go on to suggest that in
horizontal mergers these effects may be anticompetitive and that it is possible to provide, as
they term it, a toolkit to measure these effects. They refer to two such methodologies; one a
modified version of the familiar Herfindahl Hirschman Index (MHHI) and the other, based on
the work of Carl Shapiro, what they term price pressure analysis or PPI. 16
[21] O’Brien and Salop set out to distinguish the consequences of acquiring control from
those of acquiring a passive or what they term ‘silent’ financial interest. If a firm acquires
control of another, this may have competition implications in respect of the acquired or to
use our Act’s language, the ‘target’ firm. If an acquiring firm only acquires a passive financial
interest in a rival, the target firm’s incentives may be unaffected, but it may alter the
incentives of the acquiring firm. 17
[22] O’Brien and Salop examine several forms of partial ownership ranging from the
acquiring firm having a silent financial interest in the target firm, to the acquirer enjoying
unrestricted control over the target, what they term ‘total control’. (Note that the latter
unrestricted control over the target, what they term ‘total control’. (Note that the latter
concept is to be distinguished from a full merger where full ownership is aligned with full
control. In total control a firm acquires only a partial financial interest in the target, but gets
full control). In between these two scenarios – silent financial interest and full control the
authors argue there are a range of partial control scenarios. The only scenario of partial
control that equates to the legal and factual circumstances of the present case is what they
term ‘Coasian joint control’.
[23] We will apply two of these concepts to the facts of this merger; silent or passive
financial interest and Coasian joint control. We first explain what is meant by these terms
and their possible competition implications in a transaction giving rise to partial ownership in
a rival, and later, we go on to apply the concepts to the facts of this case.
Passive financial interest
[24] Here the acquiring firm acquires a stake in the target but cannot control it. However,
a transaction of this kind can give rise to anticompetitive effects under certain conditions, not
15 In fairness to Areeda and Turner they appear to have been misunderstood on this point. They do not say that
noncontrolling acquisitions constitute no threat to competition , rather they are saying that they do not pose an
‘inherent’ threat, because unlike an acquisition of control they do not lead to the replacing of two independent
firms with one controlling mind.9 (See Areeda and Turner , “Antitrust Law” , 1203 (d) (1980).
16 See O’Brien, supra, page 563.
17 See O’Brien, supra, page 571.
6
because the acquirer can control the target or change its incentives, but because it can
control its own pricing behaviour having regard to the effect that will have on the target. Thus
the authors posit a situation where a firm acquires a 20 % interest in a rival firm that it cannot
control. If pre transaction the acquirer raises its own prices by 10% that may prove
unprofitable as the increase in profits may be offset by a greater loss of business to rivals.
However, post transaction, if some of that lost business is diverted to a rival in which it has a
partial financial interest, the lost business may be regained by the share of profit in the rival.
In other words, under certain conditions, an acquiring firm may find that raising its own prices
would be profitable if it acquires a passive financial interest in a rival and so it might be
incentivised to do so post merger, despite the fact that it would not have found it profitable to
do so without the transaction.
[25] In this sense then the acquisition may be said to be linked to an anticompetitive
outcome that is transaction specific and would not have occurred but for the transaction.
Coasian joint control
[26] In this scenario the respective managers of the acquiring firm and the target firm try
to maximize the profits of both. This scenario earns its appellation because it refers to the
Coase theorem which states that “in the absence of transaction costs, parties will negotiate
to the joint wealthmaximizing outcome, regardless of the allocation of legal rights.” 18 Since
under this scenario cooperation is a prerequisite, this raises the question of whether firms
have the incentives to exercise joint control over the target and further coordinate the
activities of the acquiring firm and the target. 19 As the authors explain:
“The transaction costs based on strategic behaviour are the familiar ones
“The transaction costs based on strategic behaviour are the familiar ones
discussed in the economic theory of collusion. Although the two firms have a
joint incentive to cooperate, they have unilateral incentives to deviate from the
cooperative outcome to maximise independent profits. For example, the
owner of the acquiring firm would have the incentive to ‘promise’ to raise its
prices up to the cooperative( merger equivalent) level, but then secretly cut
prices to increase its profits at the expense of the acquired [target] firm. After
all, the owner has a100% stake in the acquiring firm but only a partial financial
18 See O’Brien, supra, footnote 57.
19 The authors also posit another species of coordinated effect which they refer to as Oneway control. Here the
acquirer has enough power over the target to force the target to abide by the terms of the coordination and not to
cheat, but the acquirer does not consider itself so bound and acts in its own self interest taking into account it has
only a partial interest in the target. ( See O’Brien, supra page 583) On the facts of this case we have already
found in our February decision that Primedia would not have such power and hence the scenario of one way
control does not arise for further analysis.
7
interest in the acquired [target] firm. The managers of the acquired firm face a
similar situation. One of their shareholders, the owner of the acquiring firm
wants them to cooperate. But the other shareholders want it to secretly cheat
by cutting prices. Thus both types of transaction costs may prevent the
Coasian outcome from being achieved.” 20
[27] We will now go on to apply these two scenarios to a section 12A analysis of the facts
of the present case.
Market definition
[28] Both the passive investment scenarios and the Coasian joint control scenarios
require an analysis of the relevant market since they operate under the assumption that the
acquiring firm and the target firm are rivals. In order to determine if they are rivals one must
define the market in which they operate.
[29] In this case the evidence of the relevant market came largely from three competition
economists – each party had instructed their own economists – and two advertising
executives – one called by the merging parties and the other by the Commission.
[30] The experts were far apart on some of the core issues. Nevertheless let us first
consider what they agreed on before we examine what kept them apart.
[31] Radio markets are what are referred to in the literature as dual sided markets. Radio
stations compete on the one side of the market for listeners and on the other side for
advertising revenue. In this case the economists are all agreed that the competition
problems that the merger might create are not to be found in the market for listeners, since
all are free to air and, hence, there is no danger that post merger the costs of listening will
increase or the supply of airtime be reduced, but rather its effect on advertising rates. Kaya
FM, like all the Primedia stations is privately held and commercially driven and hence relies
on advertising revenue for its existence.
on advertising revenue for its existence.
[32] Although the market is dual sided there is a relationship between competition for
listeners and revenue from advertising. Advertisers’ willingness to pay a certain rate for a
slot depends on the value they perceive in the audience a station can offer it.
[33] For this reason one might expect a tendency for radio stations wanting to attract the
same listeners to become more homogenous with the market dictating the most successful
format for a particular type of audience. However, there is constraint on the ability of stations
20 See O’Brien, supra, page 582.
8
to optimise formats (the parties differ on how large that constraint is). 21 In order to
broadcast a station needs a licence issued by the Independent Communications Authority of
South Africa (“Icasa”) and in issuing licences, Icasa requires broadcasters to comply with
prescribed format obligations. In this case for instance, Highveld is required to play ‘adult
contemporary music’ whilst Kaya is licenced to play ‘adult African contemporary music’.
Radio stations further require frequencies on which to broadcast and these are also
allocated as part of their licence conditions. Frequencies may be national or restricted to a
particular geographic area. Highveld and Kaya FM are both restricted in terms of the
frequencies allocated, to broadcasting in an area that corresponds largely with the provincial
area of Gauteng. Within this area other stations broadcast. Some are regional in character
because of the frequency limitations imposed upon them (e.g. Jacaranda an adult
contemporary music station and Classic FM, an adult classical music station), while others
are licenced to broadcast nationally, but can also be received by listeners within the
Gauteng area.
[34] This is where the consensus among the experts diverges.
[35] The case for the merging parties is that these constraints serve to so differentiate
stations that they do not conform to being ready substitutes, with the result that in the
advertising market, stations may compete with a host of imperfect radio substitutes and with
other forms of media such as print and television. The result of this analysis is to create a
market so narrow that no station has any competition or one so wide, that all stations
compete with one another and with other media platforms such as television or print, so that
the merger of two regional radio stations will confer limited market power opportunities on
the firms in question. The merging parties’ expert, Dr Theron, argued that she was not
the firms in question. The merging parties’ expert, Dr Theron, argued that she was not
contending for each station being a monopoly of its own. What she in fact meant is
somewhat confusing because in a supplementary report she repeated the assertion that all
radio stations might be in a separate market. 22 If that contention is correct it would make
each station a monopoly.
[36] The Commission’s market definition expressed in its expert report is “the LSM
610 audience for advertisers, in the Gauteng market for adult contemporary music stations,
broadcasting predominantly in English.” 23
[37] Based on this definition it says there are two possible markets conceptions of the
market here (i) a narrow market in which the only participants are Kaya FM, Highveld and
Jacaranda; (ii) a wider market in which it contemplates the possibility that Metro FM, YFM
21 The Commission would concede that a station cannot readily change its identity without causing confusion
and cites as an example a youth station seeking to rebrand itself to target middle aged listeners. See Tribunal
witness statement record page 234.
22 See Exhibit 10 page 2, Econex report entitled, “Response to Genesis and Johannesburg Economics”.
23 See Tribunal record, witness statements file, page 213. LSM stands for Living Standards Measure, an index
used by marketers to differentiate people according to their living standards, using criteria such as degree of
urbanization, ownership of cars and appliances, inter alia. Currently the LSM index is divided into ten segments
with LSM 1 at the bottom and LSM10 at the top. See Exhibit 11, page 9, a Kaya document that quotes the
SAARF.
9
and 5FM may be included, although it has reservations about this. 24
[38] The objection to the inclusion of SABC owned 5FM, and the privately owned
youth station YFM, is that they predominantly target youth, unlike Kaya and
Highveld. The objection to the inclusion of Metro FM is that it is a national
broadcaster, and including it would mean that one is not sufficiently sensitive to
advertisers’ reluctance to use stations for whom their would be wastage, if they were
interested in only targeting a Gauteng audience. The Commission also rules out
other language stations (all SABC’s so called African language stations) and highly
focussed stations (Classic FM). The Commission also does not include stations that
are not primarily music stations and for that reason does not include Primedia’s other
Gauteng flagship, 702 or SA FM the SABC English language station.
[39] The primary reason for the Commission excluding the stations that it does from the
relevant market is that advertisers will not pay for audiences who are not part of their target
market or, as the industry refers to it, ‘wastage’. The Commission suggests that the three
stations that comprise its narrow market, target an audience similarly situated, both
demographically and geographically, and they are thus the closest and most direct
competitors of one another. 25 Any other station, albeit that it may broadcast into the
Gauteng region, is less similar and leads to wastage either because of its geographic sweep
or demographic profile. In this the Commission relies heavily on its advertising expert
Gordon Muller.
[40] However this evidence that concerns over wastage make certain stations imperfect
substitutes and thus means they can be excluded from the relevant market is heavily
disputed by the merging parties’ advertising expert, Paul Wilkins, whose thesis is that all
stations are potential substitutes; it only depends on what you have to pay for your target
stations are potential substitutes; it only depends on what you have to pay for your target
audience. On his version, rates even as advertised, are in practice highly negotiable, and so
from the advertisers’ point of view, concerns that audiences might be wasted are in practice
not true; if the price is right for the audience you want there is no wastage. As he put it in his
witness statement:
“I believe that the industry is generally aware of the applicable ratios of value
as between the various media types and that prices are affected by this
awareness, as negotiations occur in this context.” 26
[41] Although the Commission’s preference was for the narrower market it did appear to
24 See Tribunal record, witness statements file, report of Simon Roberts, pages 219 and 207. He refers to
reports which suggest that the age profile for YFM is 1624 and for 5FM 2030. See page 207. He suggests that
for Kaya and Highveld the age profiles are similar with 50% of their listeners being over 35. Kaya however in
internal documents suggests listeners between 2534 comprise 30% of their core market and that 68% are
between 2549. See Exhibit 11 page 3
25 We use the term demographically here loosely to include factors such as race, language, age and content.
26 See Tribunal record, Witness statement file, page 118.
10
concede that the market could be capable of being more widely populated than its preferred
trio, and hence it conceived as well of the wider market that might include 5FM, YFM and
Metro FM. 27
[42] In his original report the Commission’s expert, Dr Roberts, provided for the market
shares in his wider market (i.e. including Metro, 5FM and YFM) based on their total
advertising revenues, during the course of hearings, he altered his methodology and
provided for different market shares by taking into account ‘weighting up’ for Gauteng
listeners. As we understood this term, advertisers would, even if they were national, ‘weight
up’ for Gauteng audiences as they are considered more valuable. The conclusion reached
after conducting the ‘weighting up exercise’, revealed in a new revised table , exhibit 32, was
that the market shares as originally reflected in Table 8 of his witness report, of Metro, 5FM
and Jacaranda declined, while those of the other three, particularly Highveld, Kaya and YFM
significantly increased. 28
[43] Besides the weighting up exercise in Exhibit 32, during the course of his testimony
Roberts introduced several further permutations to the market share calculations in his
witness report. Some were based on the number of hours listened to ( Exhibit 31) another
(Exhibit 33), was a series of permutations to deal with age profiles, seemingly performed as
a response to Primedia’s expert’s criticism. The interminable tables and graphs handed in by
experts during the course of proceedings has thrown little more light on the matter and
considerably more mud. It suggests the degree of uncertainty inherent in this debate over
market definition, who is in that market once defined and then still, how to slice up what
proportion of those participants’ listeners based on some demographic ( race, age, LSM
profile, content of the station talk or music and, if music, the genre ), one should have regard
to.
to.
[44] Even when the respective economists reworked examples to concede some ground
to their opponents case, conclusions to be derived from the data varied, as the data,
predictably, was very sensitive to the exclusion and inclusion of certain participants, but less
predictably, once modified HHI’s are taken into account also proved to be highly sensitive to
assumptions about whether one is dealing with a full merger or a passive financial
investment. This is because the formula for calculating the change in concentration (delta)
varies, depending on which assumption is being made. Crudely expressed, the mathematics
of modified HHI’s leads to lower deltas for passive financial investments than for a full
merger assuming all else is equal. 29
[45] We do not wish to burden this decision with all the permutations that were introduced
in the course of proceedings, but we consider below two tables introduced at a late stage in
the enquiry which serve to illustrate our point.
27 See Appeal Court record, page 881.
28 Kaya went from 6 to 11, YFM from 12 to 18, and Highveld went from 23 to 28.Jacaranda also broadcasts into
Limpopo and Mpumalanga hence its share decreased when the up weighting exercise was performed.
29 See footnote 33 below.
11
[46] We first consider the Commission’s Exhibit 32 that we alluded to earlier. This is
selected as its represents the Commission’s refinement of its earlier work and one on which
it relied during Dr Roberts’s testimony for its concentration thesis. 30
Exhibit 32
Table 8 reworked: 2005 advertising spend, promotion in wide and narrow market,
weighted for Gauteng listeners.
Ad revenue weighted according to listeners in Gauteng
Wide Narrow
Jacaranda 69202575 13% 21%
Highveld Stereo 204369175 38% 61%
Kaya 59978611 11% 18%
YFM 96875541 18%
Metro 61148018 11%
5FM 43937725 8%
Source: AC Nielsen Media Research, Discovery file 5, p110: Proportion of listeners in
Gauteng.
2006/3 Radio Diaries Mar/JuneAdult
Examination of 2005 advertising revenue based on the stations in the wide and narrow
definitions, weighted to take into account listeners in Gauteng, reveals that Highveld
commands 61% of the narrow market and 38% of the widest market. With the inclusion of
Kaya, this would increase to 79% in the narrow market.
[47] If we observe the figures in Exhibit 32 we note its sensitivity to changing market
definition, from the socalled narrow market to the wider market. On the Commission’s
narrow market Highveld has 61%, which reduces to 38% assuming the wider market.
30 See Appeal Court record page 853, where after the exhibit is put to Roberts he states that “.. we see from
the advertising side that this is a very concentrated market and the merger increases concentration .
12
Likewise Kaya moves from 18% to 11% respectively. Note as well that the Commission
excludes Classic, which, as we see later, AME would include as a participant in this market.
[48] In this example, the premerger HHI for the wider market is 2243 and for the narrow
market 4486. A market with an HHI above 1800 is considered to be highly concentrated. In a
concentrated market an increase in concentration from pre to post merger of over 100 is a
matter for concern. On this example, in the wider market if there was full control, the
increase in concentration or delta, would be 836. 31 This delta is alarmingly high. But on the
assumption that this is not a full merger, but instead the acquisition of a silent financial
interest, by Primedia in Kaya FM, and that financial interest is 19%, the delta would be
79,4.32 This delta is below the normally accepted range of concern. 33
[49] Let us try the same exercise with a table proposed by the merging parties, which
represents their closest compromise to the position of the Commission. This we find in
Exhibit 25 – again a table created only during the course of the hearing. Here the shares
are:
Exhibit 25
Advertising market shares with Metro included in Gauteng market
OWNER ADVERTISING MARKET SHARES
KAGISO (JACARANDA) 12.26%
PRIMEDIA (702/94.2) 42.84%
KAYA 10.55%
YFM 17.06%
CLASSIC 6.45%
SABC (METRO) 10.80%
TOTAL 100.00%
HHI 2547
CHANGE IN HHI 904
31 (2x 38x11).
32 (38x11x19%).
33 For how the formulas are derived see O’Brien and Salop supra, page 5956. For a silent financial interest,
one multiplies the respective market shares of the acquiring and target firm by the percentage interest the
acquiring firm has in the target firm; in a full merger, one multiples the two market shares by two. On the
Salop/O’Brien formula if there was socalled Coasian joint control the result would be the same as for a full
merger.
13
(Note that here, unlike in the Commission’s table, Primedia is bulked up by the inclusion of
702, a Primedia station. This of course is adverse to Primedia’s interest, but is properly
made given their conception of the market as being less stratified than the one that the
Commission asserts. SA FM is not included.)
[50] In the market set out in table in Exhibit 25, the premerger HHI is 2547. For a full
merger the delta is 904. 34 thus a matter of great concern. For a modified HHI assuming the
partial Nail interest at 25%, this would be a delta of 113 35, and for a modified HHI, assuming
the actual financial interest of only Primedia at 19%, this would be – 85.9. 36 Given that the
literature suggests that it is the actual financial interest of Primedia that should be calculated,
not the full Nail interest which includes the approximately 6% stake owned by the unrelated
Capricorn, this delta falls below the level of concern.
[51] What this consideration of the two examples illustrates is the wide gulf between
assumptions; (i) whether the market is narrow or wide; and (ii) whether this is a full merger or
the acquisition of a passive financial interest
[52] The Commission’s concentration conclusions have been based on an assumption by
the Commission that there will be what O’Brien and Salop would term a ‘full merger’, as
emerges from the cross examination of the Commission’s economist Dr Roberts by counsel
for Primedia:
ADV GAUNTLETT: Your exhibit 33 deals with the matter as if this was a complete
merger, not so?
Dr ROBERTS: That’s the basis on which we proceed for the majority of the analysis,
yes.37
[53] The same concession is made later in respect of exhibit 31. 38 One can assume the
same holds for Exhibit 32 whose Table 8 we reflect above. Thus the Commission has built a
case on concentration based on the assumption of a full merger and not a passive non
controlling interest – an assumption that we will come to later. Dr Roberts statement that this
was the way they proceeded with the “for the majority of the analysis” is also noteworthy,
because not having come to grips with the absence of control , or having assumed it
34 ( 2x42.8x10.55)
35 (25%x42.8x10.55)
36 (19%x 42.8x10.55). Note that the actual figure is 18, 17% we have rounded it off upwards.
37 See Appeal Court record, page 879. The assumptions around control also appear from Dr Roberts witness
statement where he states, “The merging parties’ position is that they will not have control over Kaya FM given
the size of the stake they are purchasing. We have not assessed these issues in any detail and draw from the
Commissions’ analysis. …we understand the rationale to be that the merging parties seek to ‘turn Kaya around’
which can only be possible if they are effectively in control.” See Tribunal record witness statements page 2345.
38 See Appeal Court record, page 879.
14
uncritically, the Commission has had difficulties now reframing its case, after our February
decision, based on one assumption ( a full merger) to fit a case it never seriously pursued at
the time of the hearing ( passive financial investment), whose premise depends on an
analysis of diversion ratios away from the acquiring firm to the acquired firm, an analysis that
has not been performed, but only theorised. We comment on this more fully below.
[54] AMEs’ market definition is “the Gauteng regional market for the supply of high
income listeners to advertisers on radio stations.” AME agrees with the Commission that
Kaya FM, Highveld and Jacaranda compete in that market, but they add in Classic FM, 702
and YFM, as well. They exclude Metro FM because they exclude all national stations. Like
the Commission they exclude the African language stations, on the basis that they do not
attract any high income advertising.
Conclusion on relevant market
[55] As we observed earlier coming to a conclusion on the relevant market has proved
inordinately difficult and the more data we received from the duelling teams of economists
each manufacturing a new a table of data to refute some assumption of the other, the more
the boundaries of the market receded into fog as opposed to clear lines.
[56] Radio markets are complex to analyse and although we had the benefit of the
testimony and research of three economists in the course of hearing this matter, none of
them emerged with a more probable version of the market than the others. At best the
economists knocked holes into the more tenuous assumptions of their opponents, and so we
can discount certain of the more border line theories of the relevant market, but we were still
left with a middle ground, where consensus could not be found and on whose assumptions,
wildly conflicting notions of the extent of the concentration could be made. This is not a
wildly conflicting notions of the extent of the concentration could be made. This is not a
criticism of the efforts of these economists – indeed they all demonstrated great diligence in
examining the available data – but the nature of the industry.
[57] The problem for all the economists was that evidence of cross elasticities of demand
was unavailable. 39 Although all stations or most of them at least have a publicly available
rate card, there is seldom a relationship between what they charge in reality and what is on
the rate card. Mr Wilkins the advertising executive who testified for Primedia suggested,
39 The Commission’s experts asked advertisers if they would switch to Metro, if Kaya or Highveld increased
prices by a small but significant amount. They indicated they would not. The Commission observes that this is
consistent with the posted rates data it looked at. (See Report of Johannesburg Economics, Tribunal record,
witness statement file, page 210). When cross examined on this point by Primedia’s counsel, Roberts admitted
the same exercise had not been performed as between Highveld and Kaya, because Kaya had not been included
in the data set that they had access to. (See Appeal Court record, pages 908911).See also Theron report,
Tribunal record, witness statement file, page 17 where it is stated “No product market in this industry defined by
the traditional SSNIP test will contain fully and perfectly substitutable products in a way that is sensitive to the
real degree of attenuation of such substitutability that attends these markets ” They go on to state that it is far
more meaningful to state propositions about relative relations between potential competitors than to make
conclusions about absolute boundaries to markets. See Tribunal record, witness statement file, page 17.
15
fairly dismissively, that he never uses rate cards. 40 Thus although research agencies furnish
plenty of data on trends, this is premised on rate cards being reflective of the prices at which
sales are concluded which the evidence suggests it never is.
[58] Added to this is that radio stations change format and audience during the day. The
socalled ‘drive time’ slots, when commuters listen to radio to and from work, are considered
the ones for which there is most demand. In the course of a single day Highveld’s rates go
from being more than three times those of Kaya’s for an equivalent slot during morning drive
time, to in the late evenings and early morning slots, dropping below those of Kaya’s. 41
[59] Absent reliable econometric evidence and inconclusive evidence from the
demographics of audience and advertiser profiles, the next best evidence we have of
who competes in the market are the opinions of the stations themselves. While
Primedia witnesses testified to this in the hearing, some caution must be expressed
about this evidence, given their interest in the outcome. One piece of evidence that
cannot be regarded as having been drawn up in contemplation of this hearing are
the internal documents of Kaya FM which were obtained under subpoena prior to the
hearing.
[60] In the most recent document available in this set, at the time of the hearing, an
overview is presented as part of a strategy document dated January 2006.
[61] There we find reference to the fact that Metro FM is considered a ‘direct’ competitor
of Kaya’s, and various observations about Metro’s strategy. 42 On the other hand
Jacaranda and Highveld are described as ‘indirect’ competitors.
[62] Let us consider some further quotes from the same document:
“Kaya’s cost per thousand is considerably lower than our indirect competitors
like Highveld and Jacaranda.”
“However since Metro FM’s transmitter splits and audience size in Gauteng their
CPT is now marginally lower.”
CPT is now marginally lower.”
“Metro FM are still direct competition”
“The only audience growth for Kaya FM is by targeting Metro, Highveld and
Jacaranda’s 2549, LSM 610 “
“Highveld’s positioning from a white content driven music format to a more urban
crossover format has backfired. 43
40 See Appeal Court record, page 43 where he testifies that he does not have rate cards in his office.
41 See Theron report, Tribunal record, witness statement file, pages 578
42 Exhibit 11 page36.
43 Exhibit 11 pages 3436 – these items are quoted randomly from the document and do not appear in any order.
16
[63] In another internal Kaya document, dated 25 August 2005, a report is being made by
a member of Kaya’s management to the board in an assessment of its competitors. The
person reports on the competitive analysis of a show by a Kaya presenter against SA FM,
702 and Metro. Later in the same document follows a demographic analysis across, as it
describes it, “ Kaya’s major competitors .” Mentioned in this category are Highveld,
Jacaranda, Metro FM, SA FM, 5 FM, Y FM and Jozi FM.
[64] Although each side to the debate found some phrase in the Kaya documentation to
bolster its version of the relevant market, it is not as equivocal as this outcome might
suggest. While the Kaya documentation does not help to decide definitively who is in and
who is out, it does help to evaluate the competitive proximity of stations.
[65] In our view, while we cannot determine definitively who the participants in this market
are we can evaluate the extent of competition between the stations that are relevant to
consider in this merger.
[66] Does this failure to corral the market behind a fence we can call the relevant market
constitute a crisis in appreciating the application of section 12 A to the facts of this case? We
would suggest not. As Bishop and Walker write:
“ ..although the definition of the relevant market plays an important role in the
competitive assessment of mergers, it is important that this role is not
overstated.”44
[67] The authors discuss the difficulties in examining the effects of mergers in
differentiated markets – a problem that we have here. If no two radio stations are exactly the
same we would be more worried about a merger between two stations that can be
considered closer substitutes. 45 The intuition here is that if for a given price rise certain
marginal consumers of product A would no longer choose to buy A, but would instead go to
marginal consumers of product A would no longer choose to buy A, but would instead go to
buy B, because they consider that to be A’s closest substitute; thus we might consider that
the presence of an independent B, constrains A; but that given a merger between the two,
that constraint would be removed, as the merged firm would retain the lost sales to A, by its
acquisition of B.
[68] The authors suggest that the degree of closeness between products can be
44 See Simon Bishop and Mike Walker, “The Economics of European Law’, Second Edition (Sweet and
Maxwell), page 262 read with remark on 260.
45 This approach finds support in Theron's report where she remarks “ … in other words, it is easier to ask
whether A and B are more or less meaningful competitors in a market than say B and C, than to ask which
competitors must be regarded as in the market and which as outside it.” See Tribunal record, witness file, page
17.
17
calculated by using what is termed a diversion ratio. 46 While several methods for doing this
are suggested, none have been performed here. The only evidence that we have on
closeness is demographic; the Commission suggests that listener profiles between Kaya FM
and Highveld are increasingly overlapping. The Commission relies for this proposition on the
fact that Kaya FM has a large black audience and that Highveld’ s black audience figures
are now at 30%, having gone up considerably from when the station was state owned. The
significance of even this is refuted by Primedia, who suggest that after its rapid rise, the
black audience figure has remained stable. 47 The Kaya internal documentation quoted
above, suggests that Highveld’s crossover attempts failed. Primedia does not regard Kaya
FM as Highveld’s closest substitute – it suggests that Jacaranda and 5FM are and that
Kaya’s closest rival is Metro. 48
[69] Absent data on this point and given the stabilisation of Highveld’s black audience, the
next best evidence on closeness is the internal marketing material discovered. On this
material as previously discussed the probabilities point to the stations not being their
respective closest substitutes. Rather, the Kaya documents, with their repeated and detailed
reference to Metro FM, suggest that it is their most direct competitor.
[70] Although we are obliged in terms of the Act, to examine the strength of competition in
the relevant market we are not obliged to determine, when this is not feasible, a closed list of
who may be considered to participate in that market. This is such a case.
[71] In this case what we are able to do is reject some of the more tenuous claims for
market definition made by the relevant parties to this litigation. Into that bracket we would put
the merging parties claim that each station is a monopoly on its own. No evidence seems to
support this and even the merging parties advanced this more tentatively than they did other
support this and even the merging parties advanced this more tentatively than they did other
aspects of their case. So too, would be the Commission’s suggestion that the market
comprises its narrow market of Kaya, Highveld and Jacaranda. There is no econometric
evidence to support this conclusion, and against it, as we have noted, the next best
evidence is the internal documents of the stations, and those of all three players, Jacaranda
as manifested by Radmark and Kagiso, Primedia and Kaya, point to a wider pool of
competitors. Again too, the Commission’s subsequent research was aimed at proving its
case still held for what they referred to as the wider market, notwithstanding its reservations
about whether the market was indeed this wide.
[72] We would be with the Commission and AME in suggesting that the market is one for
radio advertising in Gauteng for listeners roughly in the LSM 610 category, who listen to
predominantly English language stations. We would however hesitate in segmenting this
market any further on the base of age, race or content – there is not enough convincing
evidence that these factors are important in excluding stations from the market.
[73] We believe that the proper approach in a case such as this is to consider the market
46 For more on this see Bishop and Walker, supra, pages 2678.
47 See Moyane’s evidence, Appeal Court transcript, page 159
48 See Moyane’s evidence that Highveld regards Jacaranda and 5 Fm as its main competitors, Appeal Court
transcript,155
18
as composed of a number of firms which constitute better or worse substitutes for one
another. Based on the material we have from the stations: for Kaya FM, the target firm, the
first circle of close substitutes would comprise firstly, Metro FM its closest substitute, then
Highveld, Jacaranda, YFM and 5 FM. In the next circle would be 702, SA FM, Classic FM
and Radio Jozi – these are stations that are possible substitutes, but on the evidence seem
less close than those in the first circle. For Highveld the closest substitute would be
Jacaranda, and then 5 FM, and then Kaya. Its next circle may be the same as it is for Kaya,
but some in the second circle, may be closer to it than to Kaya. We are not certain that the
market extends only to the first circle and there may be circumstances for some advertisers,
as they are not a homogenous group either, that stations in the second circle are an
adequate substitute, and this group of advertisers may be sufficiently large to deter a
successful post merger price increase, assuming Primedia was in a position to control
Kaya’s pricing post merger.
[74] Note that our first circle corresponds with what the Commission term their wider
market. The difference between our approaches is that we consider the first circle as a
possible universe of the market and, that if it errs, it does so by narrowing the market more
than might be justified, whilst the Commission comes to the opposite conclusion.
[75] The evidence of who is in the market, albeit it cannot be determined definitively, is
relevant for several purposes. Firstly, the question of whether Kaya FM and Highveld can be
considered rivals. On this point our answer is yes. Secondly, whether the post merger
market becomes highly concentrated, and if so, whether that might lead to concerns about
anticompetitive effects .Thirdly, to evidence on possible patterns of diversion, which as we
anticompetitive effects .Thirdly, to evidence on possible patterns of diversion, which as we
will see later is relevant to the anticompetitive effects of passive acquisitions. Fourthly, in
analysing incentives to coordinate, we must know something of who is in the market,
whether they have similar incentives to coordinate, and if they do, how much of the market
might be party to the coordination.
[76] Let us now consider these aspects. On the first point we have no hesitation in finding
that Kaya FM and Highveld can be considered rivals, although neither firm is the best
substitute for the other, both are in the first circle.
[77] We now consider concentration. If the merger is going to raise serious concerns in
respect of levels of concentration, a number of assumptions have to hold. Firstly, even if we
cannot definitively determine whether the market extends beyond the first circle to include
firms in the second as well, we can test for concerns over concentration, by initially
considering the narrower first circle. If concentration concerns do not arise in this smaller
universe, then it follows they would not arise if the market is construed more widely, to
comprise firms in the second circle as well. Secondly, there would need to be a full merger
i.e. Primedia would need to be able to take control over Kaya FM. As the MHHI analysis we
19
considered in the two examples above showed – when there was a full merger the change in
concentration was at the level of concern – when it was a passive financial interest it fell
below that level of concern.
[78] We have already found that post merger Primedia will not be able to control Kaya
Fm. Our findings on fact on the control issue are set out in our February decision and need
not be repeated again here. If it cannot control Kaya FM then the merger will not lead to the
acquiring firm being able to control the prices of the target firm. Thus at best for the
Commission and AME we can say that the merger leads to Primedia acquiring a passive
financial interest in the business of a rival firm. Having found that the passive financial
interest is the correct interest to calculate for the purpose of making conclusions on
concentrations, we find that even on a narrow market assumption, by this we mean that the
market is limited to those firms in the first circle as opposed to including those in the second,
the change in concentration does not exceed the level of concern. 49
[79] (Note that we have assumed that the passive financial interest is
approximately 19% and not 24,9% as the remaining interest is beneficially owned by
Capricorn and thus for passive investment analysis purposes is not relevant. Since
the stake is not a controlling one at this level, it is not relevant either that Primedia,
as between it and Capricorn, votes this stake qua shareholder and director.)
[80] It then leaves as the remaining questions for us to decide whether the passive
financial interest would create unilateral incentives for the acquiring firm to raise prices or
adverse coordinated effects. We go on to examine these in turn.
Passive shareholding – unilateral effects
[81] Because as a matter of fact we have found the merger will not enable Primedia to
acquire control over Kaya, the anticompetitive effects associated with the acquiring firm
being able to control the pricing decisions of the target firm are not present here. That leaves
for us to consider, as the only remaining theory of harm based on a unilateral effects theory,
an enquiry into whether an acquisition of a passive financial interest by Primedia, could have
anticompetitive effects.
[82] AME argue that even if Primedia is unable to affect the behaviour of Kaya FM, it
would still have an incentive to raise prices at Highveld, because, post merger, if Highveld
lost any advertising revenue as a result, that revenue would be diverted to Jacaranda and or
Kaya or a combination of the two. To the extent that some of that revenue was diverted to
49 See Exhibit 25 as well and see also the figures provided by Theron in her supplementary report where she
does this calculation in respect of the Commission’s initial version of the narrow market and comes to the
conclusion that the change in concentration would be 93.7.Exhibit 10 page 14 Table 11. When the Commission
recalculated the figures for the narrow market, as found in Exhibit 32 set out above, Kaya FM and Highveld’s
market shares increase, and on this scenario, the delta is around 200, However, as stated above, we reject the
view that the market is composed only of Highveld, Jacaranda and Kaya.
20
Kaya, Primedia would benefit albeit its economic interest in Kaya was only 18, 17%.
[83] This theory is expressed in its economic expert James Hodges’ report:
‘ Currently, when Primedia considers increasing the price it charges for
advertising at Highveld or 702 it weighs (a) the higher price it receives on its
advertising against (b) the reduced demand for its advertising slots. The
reduction in demand occurs as advertisers go to other radio stations (e.g.
Kaya) or choose not to advertise on radio at all. Presently, Primedia considers
these lost sales to other radio stations as a cost of raising its price and so this
places downward pressure on the prices it charges. If Primedia has a stake in
Kaya it benefits from the increased demand and profit that Kaya experiences
from Primedia raising its prices. This reduces the cost to Primedia of raising
its prices which gives Primedia a greater incentive to increase prices and so it
is more likely to do so. Thus we can expect Primedia stake in Kaya to result in
upward pressure on the prices that Primedia charges advertisers” 50
[84] AME has not presented any data to show that this theory has any practical
application.51 Recall that in terms of the theoretical model of O’Brien and Salop, if this
strategy is to work the loss of revenue to the acquiring firm needs to be compensated by the
diversion to the target firm. 52 Nor has it presented any evidence that Primedia has
considered such a strategy. Absent these two features we are asked to accept an entirely
speculative model. There are a number of reasons why this theory is unlikely.
1) Highveld rates, as per rate card so exceed those of Kaya presently that Kaya does
not seem to have constituted any form of restraint to Highveld in the past. In the
lucrative morning drive time slot Highveld charges a rate 3, 8 times higher than that
lucrative morning drive time slot Highveld charges a rate 3, 8 times higher than that
50 Tribunal record, witness statement file, page 190.
51 See Hodge’s evidence, Appeal Court record, page 829 where he states, after being asked about whether
diversion is not more likely to go to firms closer to the acquiring firm than to those that are far away, Hodge
explains that one would have to know if the closest substitute had capacity e.g. during a drive time slot if it did not
it was a factor to take into account but he then concludes his remarks by saying “ so the diversion ratio is not in
any way going to be a very simple aspect to examine.”
52 The diversion ratio has been applied as tool for measuring the unilateral effects of mergers on firms selling
differentiated products. (See O’Brien supra page 598). The authors explain that the diversion ratio “measures the
amount of a firm’s sales that are diverted to the firm’s merger partner in response to a price increase, relative to
the total substitution away from the firm. As a technical economic matter the diversion ratio is the ratio of the
cross elasticity of demand to the own elasticity of demand, multiplied by the ratio of the quantity of the firm to
that of its merger partner. See O’Brien supra page 563 footnote 13.
21
of Kaya. 53 Granted we must be cautious about rate cards, as suggested by Wilkins,
but even so the discrepancy is so vast, it is difficult to conceive that if the rates at
Highveld rose further, and thus the discrepancy widened even more, that advertisers
at the margin who chose to desert Highveld, would not have a wide range of
substitutes to choose from, not only Kaya FM. The danger to Highveld if this
diversion was a factor that they considered in a post merger unilateral price increase
that would, premerger not prove profitable, is that it could not fall with sufficient
reliability into the pockets of Kaya, and, even if it did, that given the rate discrepancy,
this could ever make up for the losses in Highveld. This is even assuming that it is
not diverted to some other advertising medium other than radio as even Genesis,
AME’s experts, contemplate as a possibility. 54 We know as well that Highveld does
not consider Kaya a close substitute and that Kaya from its internal documents
considers itself an ‘indirect’ competitor of Highveld, whereas it considers Metro FM a
direct competitor. If these are the views of the stations respective managements, it
suggests that Highveld is unlikely to consider the benefits of a compensatory
diversion of profits to Kaya FM as a likely prospect, if it unilaterally increases its own
rates post merger.
2) In Highveld, Primedia enjoys 100% of the profits; for every advertising customer
diverted to Kaya FM it would only enjoy the prospect of approximately 19%. Thus in
order to make diversion profitable the aggregate of the increase in profits at Highveld
and the profits on its share of the diverted revenue to Kaya, would need to exceed
the loss of profit from advertisers who switched away from Highveld, as a result of
the loss of profit from advertisers who switched away from Highveld, as a result of
the price increase. A price increase decision at Highveld that has no regard to the
stake in Kaya, is one that is not merger specific, and hence, of no concern to us now.
It is only the price increase that is facilitated by the safe knowledge that ‘what I lose
on the swings I gain on the roundabouts’ that we are interested in now.
3) We would need to know the relative profitability’s of the stations. Even if revenue is
diverted to Kaya in sufficient amounts, if its cost base is higher than Highveld’s, the
return on that diverted revenue in Kaya may not be as high as it is in Highveld. Note
53 See Theron’s report Tribunal record, witness statement file pages 578.
54 Tribunal record, witness statement file, page 190, where in discussing what Primedia factors into account in
raising advertising prices premerger it weighs up the increase in revenue from the higher prices as against lost
revenue from a reduction in demand. “The reduction in demand occurs as advertisers choose to go to other radio
stations (e.g. Kaya) or choose not to advertise on radio at all.”
22
that in their hypothetical on the profitability of unilateral passive investment O’Brien
and Salop consider relative profitability’s of the acquiring and the target firm. 55
Neither the Commission nor AME has done this exercise, and, given the
discrepancies in the size of the holdings and the rates enjoyed by the respective
stations during the most profitable advertising periods, it seems, absent data
modelling a diversion ratio, unlikely. 56
4) Even if Kaya has some compensating revenue diverted to it, given that the Board of
Kaya is not controlled by Primedia, there is no certainty that this benefit will be
passed on as a dividend to shareholders and not invested further in the station. While
increased investment might mean Kaya attacks the market shares of Metro FM –
there is no guarantee that it might not also consider the more lucrative audiences of
Highveld. The quotation from the strategy documents above suggests that Kaya FM
sees audience growth coming from any of Highveld, Metro or Jacaranda.
[85] The idea that Highveld would now contemplate a price increase that they would not
have contemplated without the merger, because they are now comfortable with the notion
that what they lose to rivals they will gain back in their share of Kaya’s earnings, sounds so
devoid of commercial reality as to not require further contemplation.
[86] Of course if Primedia increased its price and Kaya followed suit, hence making it a
less attractive destination for Primedia advertisers contemplating switching, this might make
the increase profitable. If this is a valid theory, it is one that presents itself as a coordinated
effects theory, and for this reason is better analysed under this rubric as we do below.
Coordinated effects
[87] In Main Street 333(Pty) Ltd / Kumba Resources Ltd we observed that:
“a merger may give rise to coordinated effects concerns in two instances. In
“a merger may give rise to coordinated effects concerns in two instances. In
the first instance it may strengthen an existing coordination. In this instance
there would need to be evidence of an existing coordination, and secondly,
that the merger is likely to strengthen that coordination. The second instance
55 See O’Brien supra, 576
56 We do not have evidence comparing the profits of the stations, but in a letter to Icasa dated, 6 July 2004,
Pheladi Gangwa the regulatory affairs manager for Highveld, gives some interesting comparisons. Highveld had
at the time 1,3million listeners and generated revenue of R205, 2m. Kaya Fm had an audience of 1,1million and a
revenue of R51, 7 million. (See Tribunal record merging parties bundle page 255.)Thus on these figures although
Highvelds’ audience is only 18% higher than Kaya’s its revenues are nearly 400 % higher.
23
is that the merger increases the likelihood that the firms will coordinate. Here
there may be no evidence of an existing coordination, but evidence that post
merger, it will be probable.” 57
[88] We will examine the coordinated effects theories advanced by the Commission and
AME in this way.
a) Evidence of an existing coordination
[89] The Commission argues that there is evidence of coordination premerger between
Highveld and Jacaranda and that the merger will offer an opportunity to consolidate this co
ordination as once Primedia is represented on the board of Kaya FM, there will be an
opportunity for further and enhanced coordination between the controllers of Highveld,
Jacaranda and Kaya – Primedia and Kagiso – to meet and improve coordination, and both
will have an interest in adding Kaya to the coordination. The evidence that the Commission
relies upon for the premerger coordination are suggestions that for certain periods
Highveld and Jacaranda priced similarly.
[90] The Commission relies on price similarities between Jacaranda and Highveld as
evidence of a preexisting coordination. Theron points out that these rely only on the rate
cards, and as has been demonstrated by Wilkins, the Primedia expert on advertising, rate
cards bear little relationship to actual prices obtained in the market. Indeed he went so far as
to assert that he does not keep rate cards in his office. Wilkins view is bolstered by an
internal report from the managing director of Radmark the company that was then
responsible for obtaining advertising for a number of non–Primedia radio stations, including
Kaya. In this report the author Coen Gous writing in May 2005 observes: “There appears to
be an unofficial price war amongst virtually all media at the moment. Discounts of 80% plus
are common.” 58
[91] However whether the prices do in fact track one another is also contradicted by
[91] However whether the prices do in fact track one another is also contradicted by
another media specialist quoted by Genesis in its report, who states that price differences
between Jacaranda and Highveld were as much as 30% different. Thus we have no basis
for concluding that the prices between these stations do track one another. Nor were these
assertions put to Primedia’s management witnesses Moyane and Kirsh for their comment.
When Kirsh is asked more generally about prospects for coordination he states that he has
never heard it taking place in the radio industry anywhere and asks for any example – the
case of Highveld coordinating with Jacaranda is not put to him at this moment as one might
57 2006(2) CPLR 601 CT.
58 Extracts from Managing Director’s report – Radmark Board meeting held on 5 May 2005, Tribunal record,
merging parties bundle, page 203.
24
have expected it to, if it had any reliable basis.
[92] Nor is there any evidence of past coordination between Kagiso and Primedia. The
only evidence promised during the case of coordination between Primedia and Kagiso was
that tendered in an unsworn witness statement by Mr Rivak Bunce the managing director of
AME’s two radio subsidiaries. In his statement Bunce alleges:
“ Primedia will also be able to be in direct contact with Kagiso through Kaya
since Kagiso owns a stake in Shanike … There is a strong incentive for these
two stations to collude in decision making concerning Kaya in order to make
sure that Kaya does not compete with Jacaranda and Highveld. Kagiso
adopted a similar strategy with P4 in Durban when it acquired a minority
holding in it. “ 59
[93] As we noted in our February decision Bunce was not called to testify – this despite
the fact that of the factual witnesses who were to be called, he was the only one who in a
witness statement, indicated that he could testify to the possibility of coordination. Given
that both Kirsh and Theron refuted the coordination thesis when it was put to them in cross
examination by the Commission, AME knowing this, had an opportunity to put its evidence
alleging past coordination through Bunce. 60 Its failure to do so suggests that it was not
confident in sustaining this version.
b) an increase in the likelihood that the firms will coordinate
[94] Various scenarios were proffered to suggest that the merger will increase the
likelihood of coordination.
[95] The first is that if Primedia acquired control over Kaya FM via Nail, this would
facilitate the possibility of coordination between Primedia and Kaya, and this would have
potential anticompetitive effects, as it would allow Primedia to raise prices for advertising and
if there was a concern that this advertising might be diverted to rivals, Primedia would be
if there was a concern that this advertising might be diverted to rivals, Primedia would be
able to ensure that Kaya followed suit in raising its prices, hence making the price rise
profitable.
[96] We have already found that Highveld will not be in a position to control Kaya FM and
hence this theory is not a plausible scenario without the acquiescence of Kaya’s other
shareholders or management.
59 See witness statement of Rivak Bunce, paragraphs 223. See Tribunal record witness statement bundle page
144.
60 For Kirsh, see Appeal Court transcript, page 389, and for Theron pages 6634
25
[97] This leads into the second scenario which suggests just this. This presupposes an
examination of what incentives exist for them to acquiesce and, if they did, who on behalf of
the station would facilitate the coordination. The agent to facilitate the coordination could
be either its managers (socalled Coasian joint control referred to in the earlier discussion
over the O’BrienSalop article who have coined this term) or the other shareholders, if not all,
at least a sufficient number, who with Primedia, could control Kaya FM.
[98] In our discussion in our February decision on the possibility of the transaction leading
to joint control we had already rejected this possibility. However to the extent that we are
being asked to reconsider that conclusion in the light of a fullblown section 12A analysis, we
will revisit this issue.
[99] The thesis advanced by AME, which we understand the Commission shares, is that it
is in the interests of Primedia’s Highveld and Radio 702, Jacaranda and Kaya FM to raise
their prices because “ then there would be no loss of customers from one station to another
because the stations would remain as attractive to advertisers relative to one another.” 61
[100] The shareholders lurking behind these radio stations, Primedia, Kagiso and possibly
Caxton, would find this equally attractive as they would enjoy increased revenue without any
loss of market share. Kagiso at the time we heard the merger had a 65% interest in
Jacaranda, and this has, according to its most recent annual report increased to 80%. 62
[101] Theron, for Primedia, had critiqued this theory in her evidence, making the point that
what need to be demonstrated were merger specific reasons for the coordination. The
theory advanced by AME she argued, did not explain why the transaction made co
ordination any more likely than it would be without it. AME in response contended that the
presence of a Primedia appointee on the Kaya board, served to increase opportunities for
presence of a Primedia appointee on the Kaya board, served to increase opportunities for
coordination, through opportunities for information exchange, and, provided a means to
monitor cheating. 63
[102] It does not follow, even if there were such agreement, that faced with prices rises
from all these stations, advertisers would continue to support them to the same extent as
previously. It is equally plausible that faced with coordinated increases, some stations might
benefit more than the others. Given the differentiation between stations they would face
different demand curves and in the face of general rate rises, advertisers might see greater
value in continuing to support some at the expense of others. Fear of this would
disincentivise coordination over pricing. 64 More importantly it would need to be shown that
the stations would see coordination as more profitable than taking market share from one
61 See AME heads of argument in further written submissions paragraph 50.
62 See Roberts witness statement Tribunal record, witness file page 199. See also Kagiso Media Annual report
2007 page 26.
63 See Genesis report record page 191.
64 There is some suggestion in the internal documents that Kaya has been adversely affected when rates had
risen as advertisers did not support stations considered marginal and so it seems when budgets were tight,
favoured confining their expenditure to mainstream stations. In a Radmark document the managing director
observes, “CG advised that the market was currently extremely difficult. Marginal stations like Kaya, were
expected to be omitted from media schedules….Concern was expressed that Classic FM and Kaya would be
26
another. Highveld’s’ lucrative market share is an obvious source of attack for Jacaranda and
Kaya, who may have no interest in coordination’s benefits, unless they outweigh the
benefits associated with increased market share. There is no evidence in the discovered
documents that such strategies have been contemplated in the past.
[103] Indeed what evidence we can glean from past documents is that Kagiso’s board was
being advised to view its stake in Kaya as an opportunity to leverage that station and
Jacaranda as a:
“Powerful combined force against competitors, especially Highveld.” 65
[104] The most telling evidence for AME and the Commission is the presence of a
Primedianominated director on the Kaya FM board. They point out that this is a right that
Primedia refused to waive as a means of getting the deal through, and insists on asserting.
This of course is true. The sole reason we have what is an intermediate merger before us, is
Primedia’s reluctance to accept the Commission’s original stipulations on how directors were
to be appointed by it, as a condition for approval. Even when on appeal to us, Primedia
made proposals for conditions it would find acceptable if the deal was found to be
anticompetitive, it still insisted on enjoying this right, albeit that the director on the new
formulation would be someone ‘independent’, as understood in the company law sense of
that term. Primedia asserts that this right is there to protect its investment and not to create a
listening post for a director who would report back to those who control Highveld and 702. 66
[105] But even if we assume in favour of the objectors that the director will report back
strategic information to Primedia; the question is whether in respect of pricing information
that is not information known to them already. Here the record is contradictory.
Paradoxically, the merging parties who assert that the informational opportunities of having a
director on the Kaya board are overrated in the context of the benefits of coordination, are
the very ones who in a different context when dealing with market definition, through Wilkins,
assert that advertising rate cards (pricing information in the public domain) are not worth the
paper they are printed on, as every deal is the subject of negotiations outside of rate card
rates. This would suggest that pricing information is opaque and that having a man inside
the rival’s walls would facilitate coordination and help monitor cheating. The Commission’s
advertising expert Muller by contrast, suggests that he would have a pretty good idea as to
what rates others are achieving, because agencies such as his own and Wilkins,’ are placing
advertising for many clients with every station, and so they have developed a feel for what’s
on offer out there. On Wilkins evidence and Muller does not dispute this, most advertising is
adversely affected by recent rate adjustments.” See Extract from Managing Director’s report – Radmark board
meeting on 4 December 2003. See Tribunal record, merging parties bundle page163.
65 Presentation to Kagiso Media board, dated 4 August 2004, entitled “Kaya audience growth and benefit to
Kagiso Media.” See Tribunal record, Merging parties bundle page 248.
66 Primedia of course have to take care in their governance of their investment in Kaya. In terms of section 4(2)
of the Act an agreement between firms in a horizontal relationship to engage in a horizontal restrictive practice
may be presumed if any one of those firms has a substantial interest in the other, or they have at least one
director or substantial shareholder in common, and any combination of those firms engages in a restrictive
horizontal practice.
27
placed through agencies which do the price negotiations on behalf of clients, and only a
handful of agencies are responsible for placing most of the advertising placed though
agencies. If this is the case pricing whilst opaque to the public, is not opaque to agencies. If
it is not opaque to agencies which interact with stations all the time, some communication of
what rates are being achieved in the market place is likely to filter through.
[106] If Muller is correct, the information gleaned by a nonexecutive director, who
Primedia may appoint to the Kaya board, in the aggregated way these matters are
presented by managements to boards, may not be that valuable in relation to what Primedia
may already know from market interactions with intermediaries. However even if it is, Kaya
FM, absent an incentive to coordinate, does not have the incentive to supply a rival with
information that may monitor its pricing, even if that rival offers it a conduit. One of the
problems with the coordination theory is its lack of reciprocity. Even if Kaya’s board was
predisposed to coordinate pricing with Primedia it has no way of monitoring Highveld’s
adherence – or no more than it does presently, since it has no person on the boards of
Primedia’s stations. In summary, then the presence of a board member only creates
asymmetries in respect of access to information and monitoring and does not wholly resolve
the cartel cheating problem.
[107] The other problem is who has the incentive to coordinate at Kaya FM.
[108] We are told that Caxton and Kagiso lurk behind the remaining shareholders, but
apart from the diagram annexed to our February decision we don’t know much more.
Significantly neither of these two firms are direct shareholders and for all we know the
appointees to the board are not their people and don’t report back to them. As appears from
Appendix 3 to our February decision, which sets out the Kaya shareholder structure, it is
Appendix 3 to our February decision, which sets out the Kaya shareholder structure, it is
Shanike investments, TCT and Mokgosi who appoint the directors, Kagiso’s interest is one
higher up in the chain of holdings, and we have not been provided with evidence of what
rights this investment up the chain entitles it to. In terms of this structure Kagiso has 24, 9%
of the shares in Shanike investment with the balance being owned by Tiso and Makana
Investment Corporation. Whilst Shanike in turn holds a direct investment in Kaya FM of 24,
9% its indirect investment in TCT is 50%, with the other shareholder being Caxton.
[109] One of the problems for the incentive to coordinate theory is to establish whether the
coordination benefits all three stations better than the counterfactual, the competition
scenario – or whether the coordination benefits only Highveld and Jacaranda, and not
Kaya, or, at best, only marginally Kaya. Given that both Primedia and Kagiso are more fully
invested in Highveld and Jacaranda respectively, and have small financial interests in Kaya,
which is also a small station, they may prefer a coordination that favours these stations, at
the expense of Kaya.
[110] But this is not an interest shared by the other shareholder or stakeholders in Kaya
FM. For instance Caxton has no interest in stations that compete with those in the relevant
market, what interest does it then have in the coordination. Surprisingly the party best
placed to answer some of these issues AME is the least forthcoming – its controller Caxton
is well placed to understand Kaya FM, its funding control and incentives. Yet neither Caxton,
28
nor indeed any executive from AME, Kaya FM or Caxton was called. Granted noone
testifies before the Tribunal that he is potential recruitment material for a cartel, but some of
the building blocks of how Kaya, Caxton and Kagiso interrelate might have been adduced;
e.g. whether the relationship has changed since Kagiso acquired indirect interests in Kaya
through Shanike, whether Caxton and Kagiso have any agreements between themselves as
to the management of Kaya FM, whom they consider their competitors, how management
of the station relates to the shareholders, have Kagiso and or Caxton been able to influence
the appointment of any directors to the Kaya FM board, and if so what influence have they
historically had on determining station policy; this type of evidence might have got them
some way up the ladder of proving their assumption. They did not begin to climb the first
rung.
[111] From this convoluted structure it would appear that to the extent that Kagiso wants to
exercise influence, this influence is mediated by obtaining the consent of three other firms,
whose financial interests may be, but are not necessarily identical, to those of Kagiso. 67
This makes the case for why Kaya’s shareholders other than Kagiso might have an interest
in coordination post merger that they do not already have now, tenuous indeed.
[112] If the coordination is going to prove difficult if shareholders are to be the agents of
the coordination, what of Kaya’s management’s potential to perform this function?
[113] This is where what O’Brien and Salop refer to as ‘Coasian joint control’ becomes
relevant. Here the instrument of the coordination is not the shareholder, but the
management of the target firm. This coordination, as theorised, is premised on
management of the target firm attempting to align the interests of the target and acquiring
firms, so that they are akin to a full merger. But as the authors point out the dilemma for the
managers of the target firm is that one of their shareholders, the owner of the acquiring firm
(Primedia) wants them to cooperate, but the others would prefer that the target firm cheat
by cutting prices. This would prevent the Coasian outcome from being achieved. 68
[114] But apart from that difficulty there is no evidence in this case that Kaya management
would operate as a management in some widely held firm, with a degree of autonomy from
its shareholders. Kaya FM is a tightly held firm with only four shareholders. The idea of a
management autonomy ‘. to negotiate a joint wealth maximizing outcome’ that is not
supported by shareholders of Kaya FM is ludicrous.
[115] Other writers who write on the theoretical nature of the effect of passive investment
on the potential for coordination, suggest that we would be more concerned if the acquiring
firm was the industry maverick.
Ezrachi and Gilo write that:
67 In our February decision we refer to documents that indicate that Kagiso’s financial interest in Kaya may be
12, 5%. See footnote 65 of that decision supra.
68 See our earlier discussion on Coasian joint control where we cite this passage from O’Brien and Salop. Note
that the authors wryly remark in discussing this concept “Like most Coasian theoretical outcomes, however, this
outcome faces the problem of realworld transaction costs.” See supra page 582.
29
“Accordingly, if a firm other than the industry maverick invests in a competitor,
passive investment is likely to have no coordinated anticompetitive effects,
but rather only unilateral anticompetitive effects..On the other hand if the
industry maverick invests in a rival, coordinated anticompetitive effects may
occur, in addition to other competitive concerns.” 69
[116] That is because if the maverick was responsible for aggressive price competition
premerger that enthusiasm might be dampened because of the interest it now has in the
target. In this case Primedia can hardly be described as the industry maverick.
[117] The last theory of coordinated effects to consider is a form of market division by
maintaining the existing differentiation that exists between the stations. In terms of this
theory Primedia takes up the acquisition in order to ensure that Kaya FM does not target
itself more at a Highveld audience and thus by winning more audience make itself more
attractive to advertisers. While Primedia may well have an incentive to protect its more
valuable Highveld investment from encroachment, it would need the support of the other
shareholders to do so. We have not been given any reason why they should choose to do
so. It seems unlikely that maintaining differentiation is a strategy that benefits all three
stations equally. If stealing Highveld market share is attractive to the other Kaya
shareholders who do not have a stake in Highveld, why would they be interested in the co
ordination. Presumably the answer from AME would be that otherwise Highveld would attack
Kaya’s audience. If this is correct this is a form of coordination that is not merger specific
and could occur without it. As the Kaya documents suggest, stations monitor each others’
contents closely – you only need to turn a dial to do so having someone on a rivals board
does not materially improve information exchange in order to do so.
does not materially improve information exchange in order to do so.
[118] Again, apart from the suggestion, there is no evidence to show that this would prove
profitable or that it has ever been contemplated or attempted. What evidence we do have
about changing format comes from Kaya’s internal documents. There we see from the
portions quoted above that in Kaya’s estimate Highveld’ s attempts to differentiate its
programming have proved unsuccessful and so it has no reason to fear that Highveld will
differentiate itself sufficiently to attack the Kaya audience base. Without that concern it has
little incentive to agree to this form of coordination when it sees a benefit in attacking the
Highveld audience base, and feels vulnerable to an attack from Metro. The fact that none of
the SABC stations become party to this coordination makes all the parties vulnerable to
attacks from them when, absent the coordination, one of the strategies open to them was to
attack one another. This type of coordination seems unlikely and inherently unstable.
[119] Also this is an entirely speculative theory without any evidence to support it, nor
does it fully account for the interests of the other shareholders in Kaya, and whether those
69 See Ariel Ezrachi and David Gilo “ EC Competition Law and the Regulation of Passive investments among
competitors”. Oxford Journal of Legal Studies, Summer 2006 at 327, page 332.
30
shareholders have an identity of interest. We have made the point before that Kagiso has an
interest in a competitor, Jacaranda, but there is no evidence that any other shareholder or
stakeholder in Kaya FM has a stake in radio stations in Gauteng.
[120] In summary then the possibility of coordination to exist post transaction exists as a
theoretical possibility – but there are also a number of scenarios that equally plausibly point
to coordination, being either unlikely or failing. In order to make the case for coordinated
effects the evidence needed to be stronger than the mere holding of an interest in a rival and
the right to appoint a director to its board. Available information could have been led to build
these foundations, but neither the Commission nor AME did so. As a result the theory is
dependant on making a number of assumptions all of which are premised on shaky
foundations.70
[121] A theory of harm based on the passive investment in a rival leading to coordinated
effects between rivals post merger, has not been established.
Control “material influence”
[122] As we indicated earlier in this decision we have no reason to depart from the factual
findings on control that informed our February decision.
[123] A recent case in the United Kingdom based on the question of what constitutes
material influence has been decided by their Competition Commission and there control was
found to exist when the acquirer acquired only a 17% stake in the target firm. We examine
that decision now to see why it does not alter our conclusions.
[124] First we consider our section 12(2) (g) which is linguistically similar to the English
provision: In terms of section 12(2) (g) “A person controls a firm if that person has the ability
to materially influence policy of the firm in a manner comparable to a person who, in ordinary
commercial practice, can exercise an element of control referred to in paragraphs (a) to (f).”
commercial practice, can exercise an element of control referred to in paragraphs (a) to (f).”
[125] In the UK case, the Competition Commission had had referred to it by the Secretary
of State for Trade and Industry, a merger in which British Sky Broadcasting Group (BSkyB)
took a stake of 17,9 % in rival ITV Plc. It is thus a stake analogous to the 18, 17% financial
stake Primedia will indirectly own in Kaya.
[126] A threshold issue for the Commission to decide was whether a relevant merger
situation had been created. Under section 23 of the Enterprise Act, a merger situation is
created when two or more enterprises cease to be distinct. Under section 26(1) they cease
to be distinct when they are brought under common ownership and common control. Under
70 Although both the Commission and AME raised coordinated effects in their original submissions
during our first hearing, they did not loom large in their analyses, which were largely based on a
unilateral effects theory, assuming control over Kaya FM by Primedia. Since our February decision
and the findings of fact on control, the emphasis has understandably shifted in the arguments
presently before us, and so coordinated effects as a theory, has found its way to centre stage. But it
became as a result a theory in search of a factual substructure to support. Since the focus of the
hearing was elsewhere, the theory has been left dangling in the air.
31
section 26(3):
“A person or group of persons, able directly or indirectly, to control or
materially to influence the policy of a body corporate, or the policy of any person in
carrying on an enterprise but without having a controlling interest in that body
corporate or in that enterprise, may, for the purpose of subsections (1) and (2) be
treated as having control over it.”
[127] The English test of material influence is similar to our own. On what basis did they
find control when we had not? BSky B is interesting because the UK Commission found that
BSkyB had acquired “material influence” over ITV, despite its very low stake. But in its
analysis the Commission relied on two factors not present in the matter before us;
1) They found that after examining voting records of ITV, a publicly held company, that
a shareholder with 17% could have a sufficient majority to constitute at least 25%
plus one of the voting quorum present, and thus could block a special resolution
2) That it was the largest shareholder in the company and was very large relative to the
size of the next largest shareholder at just more than 5 %, and the next, three with
just over 3%. 71
[128] Neither of these factors is present in this transaction. Nail is not the largest
shareholder; it is joint second behind Thebe which has 45, 2%. Its stake at 24, 9% is
insufficient on its own to block a special resolution (presumably why it was set where it was)
and given that this is a private company with four shareholders, shareholder apathy is
unlikely to be a factor at general meetings.
[129] Note that the conclusion in BSkyB is in any event reported to be on appeal. 72 But
even if the decision is upheld, and we agree in any event with the reasoning, that case is
distinguishable on the facts from the one in casu, and hence we have no reason to alter our
distinguishable on the facts from the one in casu, and hence we have no reason to alter our
previous decision on control.
Conclusion
[130] For the reasons given in our February decision, the merger between Primedia and
Nail does not lead to Primedia acquiring control of Kaya FM either solely or jointly.
[131] The merger therefore needs to be evaluated as to whether Primedia's acquisition of a
71 See paragraph 3.39
72 See report in Business Times on line dated February 22, 2008.( business.timesonline.co.uk )
32
passive investment in Kaya FM would lead to a substantial prevention or lessening of
competition in the market for radio advertising in the Gauteng region.
[132] We find that even if that market is restricted to the narrow number of players that we
have defined as constituting the first circle of stations that compete in that market, the
merger will not lead to either anticompetitive unilateral or coordinated effects.
[133] It is common cause that the merger raises no public interest issues which would alter
our conclusion on the competition aspects. The merger is approved without conditions.
Postscript
[134] This merger has raised a paradox about the application of merger control to socalled
indirect mergers in consequence of the Competition Appeal Court decision.
[135] Its seems clear that in order for the competition authorities to have jurisdiction over a
transaction it must constitute a merger – i.e. there must have been an acquisition of control
by an acquiring firm over the business or part of the business of a target firm. Thus if there is
a transaction in which A acquires control over T, that will constitute a merger. It seems clear
as well that if as a result of a direct transaction that may be described as a merger, A by
virtue of acquiring control over T (we will refer to this as the triggering transaction) also
acquires control over firms controlled by T – let us say firms T1, T2 and T3, then a merger
between A and those firms occurs and as part of the same triggering transaction, the
competition authorities are entitled to examine the consequences of A’s acquisition of T1, T2
and T3, and to do so without the need for this series of indirect mergers to be notified and
evaluated separately. They can be treated as part of the same enquiry.
[136] What is less clear, and arose in this case, is if T does not enjoy control over T1, but
has some passive investment in it, does the fact that the triggering transaction constituted a
merger in respect of A’s acquisition of T, make the indirect acquisition ( we will refer to this
from now as the secondary acquisition) of a non controlling interest by A in T1, a merger as
well?
[137] In our February decision our answer to this question was no. Unless you could show
that the secondary acquisition constituted a merger itself i.e. lead to an acquisition of control,
you did not need to proceed to examine it under section 12A. What is good for the triggering
transaction must be good for the secondary acquisition. It would be wholly anomalous if the
triggering transaction required control for the competition authorities to have jurisdiction over
it, while a secondary acquisition could be reviewed even if control was absent.
[138] As Primedia pointed out in argument before us, but not it appears before the Court, it
would mean that if it had acquired the Nail stake directly, instead of indirectly, this would not
have been a transaction that required notification. It would be an extraordinary anomaly if
our Act was interpreted as creating two merger regimes – a triggering transaction having
control as a sine qua non and a secondary acquisition where it was not.
33
[139] For that reason we take the view that for the competition authorities to exercise
jurisdiction over a secondary acquisition, it must be shown that the acquiring firm will directly
or indirectly acquire control over the subject matter of the secondary acquisition. Absent
such a showing one need not proceed to examine the secondary acquisition in terms of
section 12 A.
[140] This is not a question of conflating jurisdiction with the section 12A enquiry. Clearly
the intention of the Act is if a firm acquires control by virtue of the triggering transaction it
need not renotify separately each distinct secondary acquisition if they constitute control.
Indeed, the Act may be read as requiring their disclosure so that the Competition
Commission can take a view about whether the secondary acquisition leads to an acquisition
of control. It is however another matter to dispense with control as a requirement for
deciding whether the secondary acquisition should be considered a merger.
[141] It is as it were a residual issue of jurisdiction which if not expressly provided for in the
Act seems obviously there by implication. We say on an ordinary reading of section 12 this is
there by implication. In terms of section 12(1) (a) “…a merger occurs when one or more
firms directly or indirectly acquire or establish direct or indirect control over the whole or part
of business of another firm” ( Our emphasis)
[142] A merger is about control, direct or indirect. It follows that if you have, as a
consequence of a transaction that triggers a merger notification, a secondary acquisition in
which the acquiring firm acquires a noncontrolling investment in another firm, that
secondary acquisition does not meet the definition of a merger, and need not, once this has
been determined, be proceeded with under 12A. We have jurisdiction, as part of the
jurisdiction assumed over the triggering transaction, to enquire as to whether the secondary
jurisdiction assumed over the triggering transaction, to enquire as to whether the secondary
acquisition constitutes a merger, and then, if it does, to examine it under section 12A.
Sometimes, as in this case, those enquiries may be simultaneous – but if, as a matter of
fact, control is not established, there is no need to consider section 12A. It is on this latter
point that our views appear to diverge with those of the Court. We see control as a sine qua
non of all mergers direct and indirect – they appear to see control as a sine qua non of the
triggering merger, because this is a question of jurisdiction, but not the indirect merger,
because jurisdiction has already been established by the triggering transaction. It seems that
for the Court in the secondary merger, control becomes relevant as a factor in the analysis,
but is not a prerequisite to that analysis being made. It could be entirely absent and there
still might be an anticompetitive effect as a result of a passive investment.
[143] Does our approach mean that passive investments by way of a secondary acquisition
would not be capable of adjudication even though economic theory suggests that there may
be instances of anticompetitive effects? Yes, that is what it means. However, it appears to
be the only way of addressing the paradox that if the same effects on competition were part
of a direct acquisition then the competition authorities would not have jurisdiction in any
event. But this is a policy problem of applying a control based merger system – it is not a
problem capable of being solved by reading in a separate jurisprudence for secondary
34
acquisitions that does not apply to the triggering acquisition. 73
[144] Nor does the literature of those who advocate concerns about the anticompetitive
effects of passive acquisitions suggest that the problem is necessary curable in terms of
current merger control systems.
[145] Salop and O’Brien it must be noted are writing in the context of US law.
[146] Unlike our law the triggering mechanism of US antitrust – the Clayton Act and the
Hart Scott Rodino Act (HSR) do not impose a requirement of control:
“Section 7 of the Clayton Act covers the acquisition of ‘any part’ of the stock of
another company. The statute does not require the acquisition of stock
sufficient to confer control; nor does it contain a threshold or minimum stock
purchase amount. It simply acquires acquisition of any part of a company’s
stock where the effect may be to substantially lessen competition. For
example, acquisitions of less than 25% of a company’s stock have been found
to violate section 7.” 74
[147] The HSR Act, which provides for premerger notification, exempts from reporting,
acquisitions solely for the purpose of investment when the securities acquired do not exceed
10% of the outstanding voting securities of the issuer. 75
[148] The closest these provisions get to notions of control are the regulations made under
the HSR Act which have to define when an acquisition falls under the “solely for investment’
exemption. These regulations state that this applies when the acquirer has no intention of
participating in the formulation, determination or direction of the basis business decisions of
the issuer.
[149] Yet even so the authors suggest legal certainty is lacking in respect of partial
acquisitions:
“Thus, the HSR Act and regulations add some clarity in distinguishing partial
ownership acquisitions that are merely passive investments from those that
ownership acquisitions that are merely passive investments from those that
73 In our economy the problem is less compelling than it might be in the US or European Union. The types of
cases we have referred to in the comparative literature involve widely held companies where a rival has taken a
small stake in terms of the size of the overall share capital but a stake that is typically larger than that of any other
shareholder of the firm. In our economy widely held companies without any controlling ( in the Company Act
sense) or materially influencing shareholder( in the Competition Act sense) are less frequent because our
economy is smaller and institutional shareholders are more pervasive as controllers rather than as passive
investors. The problem is nonexistent in privately held companies where shareholders typically enter into
agreements over control.
74 O’Brien and Salop, supra page 565.
75 O’Brien and Salop, supra page 5667
35
confer control over the acquired firm. But the law remains highly uncertain and
provides very little guidance for antitrust practitioners trying to assess the antitrust
risk of partial stock acquisitions . The economic paradigm set forth below provides a
more systematic, principled approach to evaluating partial ownership acquisitions.”
[150] What Salop and O’Brien are trying to achieve in their article is to suggest that, as a
matter of economics, the pessimism expressed by Areeda and Turner regarding the
quantification of the effects of partial ownership was no longer valid. 76 Their article both
suggests that partial acquisitions may be susceptible to anticompetitive effects and that
these effects can be quantified. Whilst US legislation as we noted is much more susceptible
to adjudicating these issues as part of conventional merger regulation even then the authors
do not suggest this is happening, because in their conclusion they state:
“Unlike merger analysis, where the acquiring firm automatically controls the
acquired entity after the merger, the analysis of partial ownership involves a
careful assessment of the degree of corporate control conferred by the
ownership interest.” 77
[151] By contrast in the European Union, merger notification, as with our law, is triggered
by a change of control. Section 3 of Article 3 of the EC merger regulations establishes sole
or joint control as “rights, contracts or any other means which either, separately or in
combination ... confer the possibility of exercising decisive influence on an undertaking”
76 Areeda and Turner would recognise that the absence of control is not a jurisdictional bar to the jurisdiction of
the Clayton Act, owing to its language, but nevertheless, they warn of the practical problems when they caution
that “Testing for prohibited effect is much more subtle, however, when control is neither attained or
contemplated.” ( 1203 a) They go on later to observe that in contradistinction to a full merger, where independent
competition is replaced by common control: “A noncontrolling acquisition has no intrinsic threat to competition at
all.”(1203(d)) Note that Areeda and Turner do not say that partial acquisitions do not have anticompetitive
consequences. They would agree that they do, but they are sceptical about their susceptibility to adjudication.
Partial acquisitions create for them adjudicative problems in two areas. One, we have noted in the earlier
discussion on the O’Brien and Salop critique, is about the ability to quantify anticompetitive effects. The second is
about material influence and how one can measure what is enough to constitute enough to influence either firms’
behaviour.
Their proposed solution is to have a per se rule as to what size of acquisition would constitute a substantial
partial interest, ( they suggest at least more than 5%) and, once that has been determined, to state that the
merger even though a partial merger be then analysed as if it were a full merger . Thus if a full merger between
the firms would be condemned so should the substantial partial merger. Thus their solution eliminates
adjudication over what constitutes material influence and the anticompetitive effects of possibly passive financial
interests by creating presumptions. We do not have the luxury of this solution which would require an amendment
to our legislation. What it does illustrate that even in jurisdictions where there is no statutory bar to assuming
jurisdiction over transactions in the absence of proof of control, experienced commentators are reluctant about
whether they should be adjudicated except in a small class of transaction for whom they would offer a per se rule,
followed by a presumption, as opposed to an effects based test.
77 See O’Brien and Salop supra, 602
36
[152] Ezrachi and Gilo who in their article are examining the question of whether
European merger regulation can be used to monitor the acquisition of passive investments
conclude that it cannot:
“Consequently a passive investment falling short of establishing joint or sole
control would not fall under the realm of the ECMR” 78
[153] They go on to note the reasons why in a discussion of the European Commission’s
Green paper of 2001:
“As part of the consultation ... the Commission explored the possibility that
‘minority shareholding (potentially coupled with interlocking directorships) may
alter the linked companies’ incentive to compete and may thus have an
impact upon market conditions.’ Nevertheless the Commission, backed by
Member States, concluded that the regulation of passive minority
shareholding under the new Merger Regulation should be outside the scope
of the merger regime. Most member states agreed with the Commission’s
position, saying it would be disproportionate to extend the Merger Regulations
application to passive investment. The Commission and Member states thus
vowed to retain control as the criterion for assessment under the Merger
regulations, and favoured it over alternative tests based on shareholdings.” 79
[154] The authors consider the problems associated with dispensing with the control
standard and suggest by way of reform an ex post jurisdiction over this species of
transaction.80
[155] Thus what we have in sum is an acknowledgement by some writers that passive
financial investment can give rise to anticompetitive effects and the belief that these effects
can be quantified. But, there is also an acknowledgement that these issues are not subject
to present merger analysis premised as it is on notions of control. There is recognition that at
present under EC law, there is no jurisdiction to consider such transactions as mergers,
because control is a jurisdictional requirement for the ECMR to operate. Under the US law,
which is section 7 of the Clayton Act, authorities acquire jurisdiction over a transaction
78 See Ezrachi and Gilo supra, 335.
79 See Ezrachi and Gilo supra, 337.
80 See Ezrachi and Gilo supra, 348. “Therefore rather than the ex ante monitoring involved in prior notification
that is applied to transactions involving decisive influence, passive investments could be assessed by an ex post
analysis. Moreover, the Commission’s power to conduct such an ex post analysis would only be available in the
markets in which the level of concentration and the level of passive investment exceed a predefined threshold.”
37
because it is an acquisition that has ‘effects’ that substantially lessen competition. The
provision makes no reference to control and indeed courts have held that a company need
not control another in order to violate the Clayton Act. 81
[156] As Areeda and Turner have explained:
“Nor should the courts hesitate to find that section 7 confers jurisdiction to
consider the anticompetitive effects of partial acquisitions, even where control
is neither attained nor contemplated.” 82
[157] The requirements for what transactions must be notified to the authorities as a
merger, again do not require control as an element for determining whether the transaction
must be reported. Because of the way US law operates, the fact that a transaction is not
notifiable does not mean it falls outside the scope of the Clayton Act. Thus, unlike in our
system, what constitutes a merger, and what constitutes a transaction that must be notified,
are not coterminous. Control is relevant under the reporting rules in relation to certain
subsidiary issues such as the determination of what entities are to be included the acquiring
or target firm when calculating whether the transaction meets ‘size of person’ thresholds. 83
Indirectly notions of what constitutes control are animated by an exemption for passive
investments.
[158] But significantly the US agencies can take enforcement action against a transaction
even if they cleared it initially. Private parties may take action as well. This means that
mergers can be evaluated ex post as well, and this is significant, as Ezrachi and Gilo
suggest that if passive investments are to be evaluated as part of merger control, ex post is
the time to do it. Thus conceptually notions of control play a different role in US merger law,
the context in which O’Brien and Salop write, than they do in our own. But even those
the context in which O’Brien and Salop write, than they do in our own. But even those
authors note that to the extent that notions of control have been considered in the context of
whether a transaction is exempt from notification as a passive investment (note, not exempt
from the jurisdiction of merger control in terms of section 7 of the Clayton Act), notions of
what transactions confer control remain highly uncertain. There is therefore no consensus
that passive investments should be a subject of merger law. Indeed, as noted, the
Europeans have considered and rejected this as casting the net too wide on business. Even
writers like Gilo and Ezrachi, who recognise the anticompetitive effects of these transactions,
do not see the solution coming from ex ante analysis, as our system requires – but
81 See Antitrust Law Developments (third) American Bar Association, page 280, citing the Supreme Court
Decision in Denver and Rio Grande Western Railroad Company v United States , 387 US. 485, 501, (1967) where
the Court held, “a company need not acquire control over another company in order to violate the Clayton Act.”
82 See Areeda and Turnner Antitrust Law, paragraph 1203(b) (1980).
83 “ The Act (15 U.S.C. 18a (i) (1)) provides that any action taken by the agencies, or failure to take action, shall
not bar any future action by the agencies with the respect to the acquisition.”(See United States merger
notification and procedures template, website of the International Competition Network.)
38
recommend a post facto analysis. Areeda and Turner as we noted earlier, whilst recognising
that control is not a jurisdictional prerequisite for merger review under the Clayton Act,
nevertheless recognise the problems this may cause for the sound administration of a
merger system, and suggest the adoption of per se rules and presumptions – the effect of
which are to eliminate some transactions that fall below the threshold from consideration and
treat others which exceed the threshold as if they were full mergers – a fiction, since they
are not. Notably their proposed solution does not advocate applying special formulas to
analyse this type of merger. Whatever the merits of the debate O’Brien and Salop have with
them over the reliability of such formulas, the underlying caution of Areeda and Turner
remains undiminished – testing for the effects of such transactions is ‘ much more subtle’
when control is neither attained nor contemplated.
[159] What concerns us is a likely interpretation of the Court’s decision that may result in a
confusing dichotomy in merger analysis. Direct or triggering acquisitions are judged by the
control standard – no control, no merger, no need to notify, and, short of a prohibited
practice, no subsequent remedy even if it was a transaction that gave rise to the
anticompetitive effects that have been associated with passive investments. By contrast,
indirect acquisitions of a passive financial investment that come to us in the jet stream of a
triggering transaction are subject to scrutiny for their anticompetitive effects.
[160] Regrettably this anomalous outcome was not argued before the Court – indeed
Primedia on appeal appears to have made concessions that did not require the Court to
consider this issue – and so the Court accepted as common cause that it did not need to
look at this merger as creating a secondary acquisition anomaly. Critically as well, it is
look at this merger as creating a secondary acquisition anomaly. Critically as well, it is
unclear what role control still plays in section 12A enquiries. The Court in one remark
suggests that control remains a consideration, but what is meant by this if control is absent,
is unclear. The Court on the other hand endorses the O’Brien and Salop approach which as
we noted would find that passive, i.e. non controlling investments, may in the instances
cited by them have anticompetitive effects. AME reads into this that we therefore have
jurisdiction to use merger control to remedy this problem, whilst Primedia does not. As
Primedia stated in this heads of argument submitted after the matter had been referred back
to us by the Court:
“It is critical to acknowledge that the CAC did not hold that it was illegitimate
for the Tribunal to have considered the issue of control to have decisive
cogency of the theories of harm presented in the case. Indeed the CAC made
it clear that ‘the question and nature of the scope of control which [Primedia]
could exercise over Kaya is an important consideration in this part of the
enquiry’ (i.e. the part considering the impact of competition –par 51). What it
held was that the Tribunal wrongly assumed that, without such control, there
was no need for it even to assess the theories of competition harm put
forward by those attacking the merger. The CAC therefore did not suggest
39
that the question of control was to feature only in the first of the enquiries it
identified. Its reference to an “holistic enquiry’ (para48) clearly shows that. It
merely held that the role played by control ( whether as part of the
jurisdictional enquiry or as part of the enquiry into harm) should be clearly
separated as between these two enquiries and that appreciation of the
distinction between these roles and the enquiries did not appear from the
Tribunals’ reasons.” 84
[161] Contrast that with the reading of AME who state in their heads of argument in the
light of the CAC’s decision:
“The acquisition, by a firm, of a noncontrolling interest in a competitor firm,
even of an entirely passive investment, is likely to give rise to anticompetitive
effects.”85
[162] It is more likely that when the Court referred to the “nature and scope of control being
an important part of the enquiry”, it had in mind the approach of O’Brien and Salop not that
it is a prerequisite to the finding of an anticompetitive effect, but that if it is present, it alters
the manner of analysis in terms of section 12A. Thus if control was found to be absent then
the acquisition would amount to a passive financial interest, which, the theory suggests, only
changes the incentives of the acquiring firm and not the target. Conversely, if control was
found to be present, both the acquiring and target firms’ incentives might change post
merger, and as a result we would analyse the competitive harms associated with control
differently from the way we would a passive financial interest. Also as we have shown above
the formulas used to calculate the concentration index would vary, depending on whether
the acquisition gives rise to control and, if so, the extent of control. 86 What we understand
the Court to mean is that control is not a prerequisite to conducting substantive analysis of
the Court to mean is that control is not a prerequisite to conducting substantive analysis of
secondary transactions. However it is important as a “factor” in the analysis, because where
it is present, it influences how you conduct the analysis. On this reading, the relevance of
control is that it answers ‘how’, not ‘whether’ we conduct an enquiry under section 12A.
[163] This however creates a separate problem. The standard for the evaluation of direct
84 See Merging parties heads of argument dated 25 January 2008, paragraphs 1415.
85 See AME’s heads of argument dated 25 January 2008, paragraph 11.
86 As we noted above in terms of the MHHI model the calculation of a passive financial interest has a different
formula to the one for a full merger when one calculates the change in concentration.
40
and indirect mergers must be the same. We respectfully request that if this issue comes
before the Court again that this aspect be considered and clarified. It obviously has far
reaching implications, not only for the law governing merger review, but for the entire
administration of Chapter 3 of the Act. The Commission receives countless merger
notifications in which passive, noncontrolling investments are embedded. Are these now all
to be subject to a full Section 12A review because of the possibility that such an acquisition
may lead to anticompetitive effects of the variety considered here, despite the fact that the
triggering event for consideration of a merger, a change in control, has manifestly not
occurred? If so, at what level of shareholding should such a consideration be triggered?
Given that as Areeda and Turner observed, the effects of such mergers are ‘subtle’, can our
merger control system, constantly under pressure to clear mergers expeditiously, afford the
luxury of the lengthened enquiries this would entail, given that the possible anticompetitive
effects of such transaction are rare?
[164] We should clearly have considered the issue more expansively in our February
decision, because by not doing so, we have made ourselves vulnerable to the suggestion
that we had conflated the jurisdictional and substantive enquiries. Rather what we were
recognising was that the triggering transaction and the secondary acquisition must be
adjudicated by the same standard that requires control as an antecedent. That
anticompetitive effects may arise out of transactions that fall short of control we recognise as
theoretically sound, but it does not form part of our merger review regime which has control
as its centre piece. Not to recognise this creates a dual system, where jurisdiction over
passive financial investments is assumed by chance, not by legal requirement, an outcome
that undermines the integrity of law enforcement and the legal value that the law should
that undermines the integrity of law enforcement and the legal value that the law should
apply consistently to like situations.
[165] There is a further problem which is also severe. We have to ask ourselves how a
change from a passive shareholding to control is to be treated for the purposes of merger
review. On our approach a finding that there is no control means that the secondary
transaction is not considered by us as a matter of substantive law, and hence, if the passive
investment later becomes a controlling one, the firms would have to notify it, as we indicated
Primedia would need to do in this case . 87
[166] If we follow the Court’s approach the situation becomes more blurred. If we approve
a merger where a secondary acquisition as here, is only a passive investment, would the
firm be obliged to notify that again if it acquired control? The merged firm might argue that
the competitive assessment has been adjudicated and that we have no jurisdiction to hear
the merger again. We have previously decided that a change from joint control to sole
control would trigger a new notification, even if the new sole controller was part of the joint
control previously approved or deemed to have been approved. We would it seems have to
treat a move from a passive investment to control, be it solely or jointly, in the same way.
Thus were Primedia’s stake in Kaya FM to become a controlling one, they would have to
notify the transaction to the Commission. 88
87 See our February decision paragraph 80.
88 In argument before us counsel for Primedia appears to concede this fact. See page 95 of the Tribunal
41
______________ 09 May 2008
N. Manoim Date
Concurring: Y Carrim and D Lewis
Tribunal Researcher: J Ngobeni
For the merging parties / appellants : Adv. JJ Gauntlett SC and Adv. F Snyckers,
instructed by Edward Nathan Sonnenbergs
For the Commission : Adv. DN Unterhalter SC and Adv. NH
Maenetje, instructed by AMMM Inc.
For the Intervenor : Adv. JWG Campbell SC and Adv. MA
Wesley, instructed by Zenwill Lacob Attorneys.
transcript, in the remittal proceedings, dated 30 January 2008.
42