Deriving this sum with respect to ki yields:
d P_j πj,-j
∂ki
= Ci(ki ) s2 χ
-j
s-j
sibi
4 --j s--j b--j
a- bjcj
c-j
bj—-
j2
-bj
__ _ s--j (a - b--j c--j )
2 P—j s—j b—j
+ ci(ki)7j X s-j aα - cj--2j
-j
Σ2—j s--j(α b—jc—j)
2 P—j s—j b—j
b_________Sibi_________
j 4 P__j s--j b--j
(29)
=ci(ki) 2 X s-j a- - bj cj
-j
c-j
bj—-
j 2
-bj
Σ-j s--j (a - b--j c--j )
2 P—j s—j b—j
sibi
4 __j s__j b__j
<0
Profits from outgoing calls are reduced as retail prices for outgoing calls to competitors’
networks increase with the competitors’ termination rates. Additionally, also the per-
unit price-cost margin is reduced as the absolute increase in retail prices is lower than
the absolute increase in termination rates. Therefore, we find a negative indirect effect
of i’s investments on j’s profits from outgoing calls.
A.4 Data Selection and Data Adjustments
We employ data from multiple sources including information about the fully vertically
integrated MNPs in the EU 15 as well as Norway and Switzerland, data about termi-
nation regulation and the termination rates, information about individual transmission
technology per provider and data about population concentration.
Financial information and customer information per MNP are taken from the Merrill
Lynch European Wireless Matrix 2006 and 2007. Merrill Lynch offers detailed informa-
tion for each provider including traffic and financial information on a national level on
a quarterly basis. From this data set we have selected information on calling minutes,
market shares in terms of customers, the total market size and information about invest-
ments. As a proxy for investments we use capital expenditures (capex) having in mind
that capex also includes investment in assets which are not directly linked to network
infrastructure, like buildings. Nevertheless, investments in assets which are not related
to physical network infrastructure are expected to be only a minor share of the total
capex.
Termination rates are taken from the Ovum data base. Ovum provides termination rate
information for each MNP. Note that at least for 2001 we have to adjust termination
rates due to the introduction of the Euro in most countries of our sample. For most
of the MNPs Ovum provides on-peak and off-peak prices and traffic-dependent average
36