The dynamics of merger and acquisition waves in the UK: an

Download Report

Transcript The dynamics of merger and acquisition waves in the UK: an

Does EC balances efficiency gains against
anti-competitive effects?
A Preliminarily Empirical Evaluation
Zafeira Kastrinaki
EC Competition Enforcement Data
ACLE, 10-11 April, 2008
Amsterdam
Motivation
In the past EC has received criticism for not making proper economic analysis,
for example in 2002, the EC lost three merger cases in court: Airtours/
First Choice, Schneider/Legrand and TetraLaval/Sidel. In all three cases,
the CFI strongly criticised the Commission’s economic analyses.
The Commission has now formally stated that it will take account of
substantiated efficiency claims that are merger-specific, verifiable, and
beneficial to consumers
In theory, this shift in attitude provides merging parties with an additional
layer of arguments to overcome a presumption of adverse competitive
effects created by high combined shares.
Can such arguments also make a real difference in practice?
Warwick Business School
2
Research Objectives
Against this background, it seems worthwhile to evaluate
whether the Commission have historically place an emphasis
on economic factors such as anticompetitive effects or
efficiency gains when investigating and prohibiting mergers
We also examine whether EC is forward looking is a sense of
considering future merger when investigating a given merger
case . The efficiency offence argument does not find any
justification under a forward looking AA (Motta and
Vasconcelos, 2005)
Warwick Business School
3
Earlier Literature
Lindsay et al. (2003) using a sample of 245 mergers for the period 1990–
2002, the authors find that high market shares and barriers to entry are
the main causes of prohibitions, while dummies indicating that the parties
were incorporated in the USA or in a Nordic country have no significant
effects
Bergman et al. (2005) using a random sample of 96 mergers for the period
1990-2002 find that the probability of a phase-2 investigation and of a
prohibition of the merger increases with the parties’ market shares. The
probabilities increase also when the Commission finds high entry barriers
or that the post-merger market structure is conducive to collusion
Some additional empirical studies of EU’s merger regime
Roller and Neven (2002), Aktas et al. (2003),and Duso et al. (2003) analyze
the relation between the Commission’s decisions and the movements of the
share prices on the stock market
Warwick Business School
4
The Empirical Model (1)
Notif.
Date
Phase I (z)
Or
Phase II (k)
Prohibition (p)
Cleared (d)
P  t z  Tz  t z  t z / T  t 
hz  t   lim
t 0
t
hkp  t   lim
P t p  Tp  t p  t p / Tp  t p 
rate of prohibitions (2)
t
t 0
hkd  t   lim
rate of phase I (1)
P  td  Td  td  td / Td  td 
t 0
Warwick Business School
rate of clearance in phase II (3)
t
5
The Empirical Model (2): a proportional hazard model
hn  t / X   hon exp  X n n   where n=z, p, d(4)
Eq. 4 gives the hazard function or rate (or probability) of transition
hon
Xn
is a baseline hazard measuring the effect of time past
It is assumed constant within pre-specified n groups
However, it may differ across them
is a vector of covariates
n
is a parameter vector

n
is a random variable (>0) which summarises the impact of
unobservable firm specific effect that scales the no-frailty
component with unit mean, finite variance
Warwick Business School
6
Definition of Explanatory Variables
Cumulative Abnormal Return of merging firm (CARm) which equals 1 if
CARm>0 and 0 otherwise
Cumulative Abnormal Return of competitors (CARc) which equals 1 if
CARc>0 and 0 otherwise
Efficiency gains of merging firms which equals to 1 if CARm>0 , CARc<0
and 0 otherwise ( eg. Banerjee and Eckard, 1998)
Expected change in the cumulative number of mergers in sector j in the
interval [t, t+1], measured by {S(t+1)-S(t)}
Where,
S(t): Cumulative number of acquisitions in sector j up to and including time t
However, there is a possibility that the merger announcement signals that a rival
is more likely to become a merger target in which case the implied sign
pattern would be the same as for the collusion or efficiency hypotheses,
(McGuckin et al. 1992)
Thus,
Likely targets among rivals are not considered when above variables are
calculated
Warwick Business School
7
Sampling and Data
The sample consists of mergers examined by EC:
All phase II mergers over the period 1990-2007
A randomly matched sample of phase I mergers over the
period 1990-2007
Identity of merging firms and competitors is obtained from EC
merger decisions
However, due to difficulties in identifying necessary data the
final sample consists of 880 firms:
102 phase II cases (of which 15 prohibitions)
123 phase I cases
655 competitors
Data are sourced from the Official Journal publications of the EC as well as
Datastream database.
Warwick Business School
8
Abnormal Returns Estimation
(an event study methodology)
1. Estimation of the market model: Ri,t =α+βRm,t+εi,t
(5)
where,
Ri,t : firm i’s stock price at time t
Rm,t: market index for the sector and country that firm i belongs to
Over 180 trading days, starting from 30 days prior to the merger
announcement day (Scholes-Williams (1977) method)
2.Firm i’s abnormal return around the announcement day t is calculated:
ARi ,t  Ri ,t  Rˆi ,t  Ri ,t  ˆ  ˆ Rm,t
(6)
Under the null hypothesis of efficient markets, abnormal returns have zero
mean and finite variance
3. Firm i’s cumulative average abnormal is then calculated:
 5
(7)
CARi   AR
 5
Warwick Business School
9
Results (1): Competing risks hazard estimation
Dependent Variable
Risk of phase I
Risk of clearance in
phase II
Risk of prohibition
Independent Variables
CARm
-0.0172 (0.01)
CARc
-0.2305 (3.6***)
Efficiency
0.1553 (1.03)
Expected mergers
0.1553 (0.99)
Theta
1.2680
Likelihood-ratio test of theta=0
Warwick Business School
-0.0100 (0.02)
0.3608 (3.79***)
0.1319 (1.28*)
-0.1319 (1.03)
1.3920
Χ2(1)=11.1
-0.0037 (0.14)
0.4781 (4.24***)
-0.0770 (0.71)
-0.0770 (0.71)
0.9823
Χ2(1)=22.57
Χ2(1)=8.91
10
Results (2): Competing risks hazard estimation
Dependent Variable
Risk of phase I
Risk of clearance in
phase II
Risk of prohibition
Independent Variables
Time dummies
D7
0.3386 (1.16)
0.1192 (1.55*)
0.3618 (1.21)
D6
0.2930 (1.00)
0.0907 (1.32*)
0.3097 (1.03)
D5
0.7021 (2.20***)
0.1034 (2.48***)
0.6415 (2.24)
D4
0.3805 (1. 78**)
0.0655 (1.28*)
0.3385 (1.28*)
D3
0.7021 (1.02)
0.0648 (1.85**)
0.5715 (1.85**)
D2
0.0107 (0.72)
0.0243 (0.72)
0.2395 (0.72)
D1
0.0806 (1.26)
0.0407 (2.60***)
0.7872 (2.60***)
D0
0.0181 (1.66*)
0.0202 (1.71**)
0.5467 (1.71**)
D99
0.0873 (1.94**)
-0.0527 (1.99**)
0.5959 (1.99**)
D98
-0.0327 (1.73**)
-0.0305 (2.48***)
D97
Warwick Business School
0.5497 (1.20)
0.0890 (1.16)
-0.0055 (0.10)
0.6262 (1.16)
11
Concluding Remarks
EC seems to consider efficiency gains claimed by merging
firms. However, there is no strong evidence in favour of EC
balancing efficiency gains against anticompetitive effects
It seems that EC has a myopic behaviour as it does not
consider expected mergers when it judges a given merger case
EC does not consider merging firms interests . However, rivals
gains influence the decision process
These preliminarily results trigger a more detailed examination
as regards the role of efficiency issues (for example, using
more direct measures of efficiency gains) in merger control,
especially after the 2004 reform
Warwick Business School
12