The document summarizes a study on regulation and investment in European high-speed broadband infrastructure. It presents a theoretical model to analyze how regulation of access prices affects investment incentives for incumbent telecom firms and cable operators. The model considers infrastructure competition between these firms. The results indicate that access regulation can have ambiguous effects and depends on factors like the relative strength of various economic effects. Empirically, the study uses firm-level data across EU countries to estimate the impact of access regulation on next generation network investment by different types of firms.
1. Regulation and Investment in
European High-Speed
Broadband Infrastructure
Wolfgang Briglauer, Carlo Cambini, Michał Grajek
ZEW, Politecnico de Torino & EUI-FSR, ESMT
2. Motivation/1
The discussion about regulation and infrastructure
investments in network industries is hardly new
A hot topic today in telecoms:
Next Generation Networks (NGNs) expected to be a
significant contributor to economic growth; e.g. Czernich
et al. (EJ, 2011): 10% increase in the broadband
penetration rate results in 1-1.5% increase in annual
GDP per-capita
Which regulatory framework?
Providing enough incentives to invest in NGNs
Avoiding market monopoly/duopoly by incumbent(s)
The transition from copper to fiber is going slowly
Coexistence between legacy networks and NGNs
3. Our contribution
Current literature:
Existing theory helps to understand the problem; however
it does not consider the presence of alternative
infrastructured operators (i.e. cable firms)
Models with two firms (an incumbent and an entrant)
Scant empirical evidences
Our paper
Extends current models to account from another source
of infrastructure competition: presence of cable operators
with own network and no need for access;
Simultaneous moves
Empirical analysis at EU27 level
4. Economics of Migration to broadband
Bourreau, Cambini and Dogan (2012 IJIO; 2014 JRE): The
effect of a lower price of local loop unbundling on investment
in fiber coverage is ambiguous, due to 3 conflicting economic
effects:
Replacement effect:
Entrant’s profits with copper = opportunity cost to invest in fiber
A lower access price to copper stronger replacement effect (less
investment)
Wholesale revenue effect:
Incumbent face an opportunity cost of investing in fiber = loss of access
revenues from copper (to the extent that the incumbent’s investment
stimulates the entrant’s investment)
Lower access price to copper lower wholesale revenue effect (more
investment)
Retail-level migration effect:
When the access charge on the copper network is low, the prices for
standard broadband services are low. So, more difficult to encourage
customers to switch from the copper network to the fiber network.
High access price to copper stronger migration effect (more
adoption)
5. Cambini/Jiang (2009); Grajek and Röller (2012, JLE)
stricter regulation tends to deprive incentives to invest
Empirical evidence on NGN investment is scant
stricter regulation deprives incentives to invest in NGN
infrastructure (Wallsten and Hausladen, 2009, early NGN
deployment)
Briglauer et al. (2013, IEP) use real NGN investment
data, but only focus on service-based competition (no
unbundling price)
Only one study explicitly assesses role of the unbundling
price with respect to NGN investment (Briglauer, 2015,
JRE) showing a positive relationship
Empirical literature
6. Build theoretical model to guide empirical analysis
Based on Bourreau et al. (BCD, 2012)
Accommodate major actors in the investment game: telecom
incumbents, telecom entrants and cable companies
Consider cable’s partial coverage and investment costs
asymmetry between cable and incumbents
Incorporate both the regulated access to copper and fiber
Guides empirical model: Estimated equations = log-linearized
best response functions
Provide empirical evidence of the effect regulation on
NGN investment
Use company-level data from 27 EU nations over 2004-2014
Use the # of fiber access lines as the investment measure
Our contribution
7. A benchmark model
Three firms: (telecom) incumbent, a cable operator
and a serviced-based telecom entrant.
Incumbent and cable firms invest in NGN; the
telecom entrant is passive (relaxed in Appendix)
The three firms compete in the retail broadband
market
The market is fully covered with “basic” broadband
The incumbent owns the copper network
The cable owns its own network and does not seek
access
The entrant can access the incumbent’s copper network
(through local loop unbundling), at regulated price a
The incumbent and cable firms simultaneously
8. Investment Cost
Each local market is composed of different areas,
ordered according to the cost of deploying the NGN
area z
cost of NGN
in area
High density Low density
9. Profits
In each area, the incumbent and the entrant earn
profits according to the networks they use.
Service base entrant is passive, i.e. does not invest
Four possible cases/areas
Incumbent with Old Network, Cable with Old Network, Entrant
rents the old network (O,O,O)
Incumbent with NGN , Cable with Old Network, Entrant rents the
old network (N,O,O)
Incumbent with Old Network, Entrant with NGN, Entrant rents old
network (O,N,O)
Incumbent and Entrant with NGN, Entrant rents old network
(N,N,O),
10. Profits
General profit functions (net of investment cost)
Incumbent: 𝜋1
𝑁,𝑂,𝑂
𝑎 𝜋1
𝑂,𝑁,𝑂
(𝑎)
Cable: 𝜋2
𝑂,𝑁,𝑂
𝑎 𝜋2
𝑁,𝑂,𝑁
(𝑎)
Co-existence of technologies:
For k, l = O, N, 𝜋𝑖
𝑘,𝑙,𝑂
(𝑎) 0 for i=1,2,3,
Effects of a:
For k, l = O,N, (a) d𝜋1
𝑘,𝑙,𝑂
(𝑎)/da ≥ 0, for all a ≤ amax
(b) d𝜋3
𝑘,𝑙,𝑂
(𝑎)/da ≤ 0, for all a ≤ amax;
(c) d𝜋2
𝑘,𝑙,𝑂
(𝑎)/ da≥ 0, for all a ≤ amax.
12. Best response and equilibria
z1
z2
m
z1
c
z1
c
z2
m
z2
)()( ,,
1
,,
1 aa ONOONN
)()( ,,
1
,,
1 aa OOOOON
)()( ,,
2
,,
2 aa OOOONO
)()( ,,
2
,,
2 aa OONONN
Weak strategic substitutes
13. Benchmark case: Main results
Asymmetric equilibria
Either the incumbent or the cable invests more and
become a fiber leader, i.e. a monopolist in high-speed
broadband provision in some (intermediate) areas
Regulated access to copper network a relevant
Cable operator never seeks access to the copper
network, because it has its own cable network. However,
a inflates basic broadband prices and influences
competition
For the incumbent, same retail-migration and wholesale
revenues effects
For cable, a new effect comes out, the business-stealing
effect: the higher is a, the higher the cable’s profit on old
technology, hence the lower the incentives to migrate to
𝑘,𝑂,𝑂
14. Extension 1: cable’s partial coverage
Different cases emerge
Data: in 68% of EU countries cable operators are the
leader in NGN investment. Cable’s leadership in the
model.
Assume that cable’s existing coverage is zc < 𝑧 zm
Profits are:
15. Extension 1: equilibrium
Asymmetric equilibrum
Results:
Incumbent’s decision affected by two countervailing effects
(retail migration and whoelsale revenues effects).
Cable’s decision is positively affected by a.
However, because of substantially higher investment costs,
cable companies typically do not roll out NGA infrastructure in
the areas previously uncovered by the cable TV network
discontinuity in investment cost i.e. c(𝑧 ) = +.
New equilibrium: z2 = 𝑧 , not a function of a.
16. Extension 2: regulated fiber access
Main difference to the benchmark case
The cable and the entrant can access incumbent’s fiber
network at a regulated price 𝑎 (but not the other way
around!)
Profits:
17. Extension 2: regulated fiber access
Asymmetric equilibria:
A) Incumbent is the leader:
B) Cable is the leader:
Main results
The regulated access price 𝑎 matters and positive impact
on incumbent’s investment. More stringent regulation
reduces the incentives to invest
For cable, when it is the leader, no effect of the access
charge (due to asymmetric regulation). Instead,
ambiguous effect if it is the follower
18. Extension 3: cost asymmetry
z1
z2
m
z1
c
z1
c
z2
m
z2
)( ,
1
,
1
NONN
)( ,
1
,
1
OOON
)( ,
2
,
2
OONO
)( ,
2
,
2
ONNN
1
~z
D
D
c
z2
m
z2
2 firms’ scenario
More likely for cable
to become the leader
19. Sum up of the results/1
The impact of mandated access to the legacy
network on incumbent telecom firm’s
investment.
The impact of access price a is ambiguous and depends
on the relative strength of the retail migration and the
wholesale revenue effects.
The impact of mandated access to the legacy
network on cable firm’s investment.
The impact of access price a on the investment incentives
of the cable firm is also ambiguous and depends on the
relative strength of the retail-migration and the business
stealing effects. If existing coverage is only partial and
extremely high cost to add areas, then coverage is
unaffected by a
20. Sum up of the results/2
The impact of mandated access to the NGA
network on incumbent telecom firm’s
investment.
The impact of access regulation to NGA on the
investment incentives of the incumbent firm is
unambiguously negative.
The impact of mandated access to the NGA
network on cable firm’s investment.
The impact of access price to NGA on the investment
incentives of the cable firm is positive only in case the
cable firm is the investment follower, but not in case of
leadership.
21. Regulatory
variables
Access charge:
Unbundling price
llu_price
Infrastructure
Competition
Intermodal
competition:
Share of total mobile
lines to the total
number of fixed lines
fms
Intramodal
competition:
Legacy: Total
number of active
fixed landlines per
100 inhabitants
New entrants’ retail
market share in
fixed broadband
lines
bb_ne
Control
variables
Demand side:
GDP_pc; education
Cost side:
Mdwell_perm
Dependent
variables
Incumbent lines:
Total number of
NGN lines
(ln_inc_nga)
Cable lines:
cable
entrant NGA
Lines (ln_cable_nga)
Other
entrants NGA
lines
(ln_other_nga)
Sources: FTTH Council/IDATE, EIU, EUROSTAT/COCOM, EUROMONITOR, EU Progress Report, ITU, IMF, ECB, WB
Yearly panel data on EU27 MS from 2003-2014
Data
Instrument. var:
EU price level
llu_euprice
Fixed effects:
Relevant time-
invar. cost, demand
and institutional
factors
NGA regulation:
Active remedies
dummy
Nga_reg
External instrument:
PCs number; nga_inst
22. Relevance of local loop unbundling in EU
16%
28%
37%
46%
54%
62%
66%
72%
75%
76%
78%
21%
18% 18% 19%
17%
13%
11%
9%
7% 6%
4%
0%
10%
20%
30%
40%
50%
60%
70%
80%
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Own network
Full ULL
Shared access
Bitstream
Resale
23. Empirical model and identification
Incumbent’s investment equation (best response)
Cable entrant investment equation (best response)
Identification
Simultaneous equations: Exclusion restrictions: legacy
business
Dynamic panel model: Arellano-Bond instruments
Other endogenous variables: Anderson-Hsiao instruments
Additional “geographic” instruments for regulation – both
price and NGA regulation
I
ti
I
i
I
titi
I
ti
I
ti
I
ti
I
ti
I
ti
ελXngaentδlegacyincδ
regngaγpricelluγngaincβngainc
,,,2,1
,2,11,,
Θ)_ln(_
__)_ln()_ln(
E
ti
E
i
E
titi
E
ti
E
ti
E
ti
E
ti
E
ti
ελXngaincδlegacycableδ
regngaγpricelluγngaentβngaent
,,,2,1
,2,11,,
Θ)_ln(_
__)_ln()_ln(
26. Conclusion
Access regulation on the old copper network affects
incumbent’s NGN investment incentives
Stricter access regulation (both the extension to
NGN and lower LLU price) discourages incumbent’s
incentives to invest in NGN
This effect is due to competition with the telecom entrants
Retail migration dominates wholesale revenue effect
Cable entrants’ incentives to invest in NGN are
unaffected by the access regulation
Cable entrants are leaders in the NGN deployment in
most markets (i.e. don’t seek regulated access)
Cable entrants are less affected by the retail migration
effect