(Interconnection Module Lectures #2 & #3)
Challenges for Convergence:
Interconnection
William Lehr
ESD 68 :
Communications and Information Policy
Massachusetts Institute of Technology
April 4 & 6, 2006
©Lehr, 2006
Interconnection Lecture Outline
q
q
q
q
q
Examples of interconnection in telecoms
Why regulate interconnection?
Basic economics of interconnection
Goals of interconnection regulation
Current models for interconnection
Ÿ Cost-based pricing
Ÿ Negotiated pricing (reciprocal compensation)
Ÿ Bill and Keep
2
©Lehr, 2006
Readings
q
q
q
Sicker, Douglas (2002), "Further Defining a
Layered Model for Telecommunications Policy,"
draft mimeo, October 2002.
DeGraba, Patrick, “Bill and Keep at the Central
Office As the Efficient Interconnection Regime,”
OPP Working Paper Series No. 33, Federal
Communications Commission, December 2000.
Kende, Michael “The Digital Handshake:
Connecting Internet Backbones,” OPP Working
Paper Series No. 32, Federal Communications
Commission, 2000.
3
©Lehr, 2006
What is interconnection issue?
q
q
Two (or more) networks exchange traffic, they need to
be interconnected.
Ÿ Physical point(s) of interconnection
Ÿ Technical/operational issues
Ÿ Commercial relationship: who pays what?
Why problem for convergence?
Ÿ From silos è platforms
Ÿ Regulation still based on silos
4
©Lehr, 2006
Interconnection & Access Pricing (Theory)
q
q
One way access:
Ÿ Incumbent sells essential input to entrant
Ÿ Incumbent could be vertically integrated or not (does
incumbent compete in retail market with entrant?)
Ÿ e.g., Local loop unbundling
Two way access:
Ÿ Network interconnection problem
Ÿ Reciprocal needs to terminate traffic
Ÿ One or both could have market power
Ÿ e.g., Internet peering or transit, mobile/wireline network
interconnection charges, international settlements
5
©Lehr, 2006
Telephone Network: a network of networks
Distribution & Feeder
A
A-B Local call
A-C LD call
LD POP
B
Local switch
Long Distance
Network
Local switch
Local
Transport
Local switch
Local switch
LD Access
LD POP
Single carrier network or multiple networks?
Which party pays: Calling party or both?
©Lehr, 2006
C
6
The Communications Landscape
T-1 Line
DLC
Class 5
Switch
LAN
DLC
Analog Modem
Users
WAN
DSL
Modem
Users
Cable Head
End
DSLAM
Cellular Base
Station
Cable Modem
User
ACCESS
Customer Premises
Figure by MIT OCW.
7
©Lehr, 2006
A more realistic picture
User
User
User
User
User
User
Little
ISP
Little
ISP
Corp
Campus
Backbone
(big ISP)
User
Backbone
(big ISP)
Backbone
(big ISP)
User
from Dave Clark’s Lecture…
©Lehr, 2006
User
User
Little
ISP
The ISP lives here..
The ISP does not live
at the end-points.
User
8
Interconnection Models
Network A
retail
Network B
retail
wholesale
e.g., International LD, Local/LD, Mobile/wireline
Network C
wholesale
Network D
retail
e.g., LD/LD, Local/Local, Mobile/mobile, Internet backbone
9
©Lehr, 2006
Interconnection Models
Network A
retail
Network B
wholesale
Network C
wholesale
retail
e.g., Multihop routing. B is transit network
Network B
Network A
Network C
retail
retail
Network D
e.g., Multilateral peering point
10
©Lehr, 2006
Current models for interconnection
q
Examples:
Ÿ
Ÿ
Ÿ
Ÿ
q
International settlements: negotiated rates for terminating
calls. May not be symmetric, generally well above costs.
Long distance pay per minute access charges for local
termination.
VoIP calls avoid charges
Internet peering using “bill and keep”
Different prices for similar situations: inefficient
pricing
11
©Lehr, 2006
Interconnection Models
q
q
q
q
q
©Lehr, 2006
Technology of networks: same or different?
Ÿ Wireline/wireline, wireline/wireless,
packet/circuit, etc.
Type of traffic? (e.g., Web browsing vs. telephone call)
Ÿ Balanced or asymmetric flows?
Ÿ QoS needs: delay sensitivity? BER sensitivity?
Size of networks: same or different?
National or international?
Regulated or negotiated?
Different costs, business relationships, and regulatory
treatment. Not a problem when telcos were regulated
monopolies…
12
Elements of Interconnection Agreement
q
q
q
q
Scope and Purpose of Interconnection
Ÿ Who are parties?
Ÿ Types of traffic? Networks? Architecture?
Ÿ Points of Interconnection
Quality of Service and technical specifications
Ÿ Quality of service and performance standards
Ÿ Technical interconnection specs and capacity
Ÿ Infrastructure sharing, collocation
Ÿ Traffic measurement and routing
Billing and payment terms
Ÿ Pricing
Enforcement/Dispute Resolution
13
©Lehr, 2006
Why regulate interconnection?
q
Promote interconnection: larger networks more valuable
Ÿ Positive network externalities
• Scale & Scope economies à lower costs
• Complementary goods à more choice
• More people to call (subscriber externality)
q
q
Coordinate interoperability à standards
Control market power
Ÿ Promote competition à facilitate entry
Ÿ Protect consumers from monopoly power
14
©Lehr, 2006
Challenge of Regulating Interconnection
q
Promoting “interconnection” à easy when regulated end-to-end monopoly
Ÿ International is negotiated bilateral/multilateral treaty (trade issue)
Ÿ Interconnection rates include implicit subsidies, but lots of other
regulatory levers to address distortions
• Control of “rate base” monitors investment
• Retail rate regulation protects consumers
Ÿ
q
q
Silos minimize challenge of cross-platform interconnection
But, Convergence à Telecom becomes a “network of networks”
Ÿ Traffic passes between networks owned/operated by different carriers, or
across regulatory boundaries.
Ÿ Need physical point(s) of interconnection and business rules (pricing,
QoS) to exchange traffic.
And, Competition à Transition to wholesale regulation
Ÿ Interconnection is a “wholesale market”
Ÿ Between carriers, services are ingredient to a retail service
15
©Lehr, 2006
Interconnection and Market power
q
Interconnection rates set to exploit/leverage market power
Ÿ Originating monopoly problem
• Is their competition for subscribers? If so, then competition assures
originating carrier cannot extract surplus rents.
– Switching costs (e.g., incomplete information re: alternatives – pay phones;
lack of address portability – email addresses, etc.)
– International mobile roaming
– MCI “Friends & Family”: discriminate between on-net and off-net calls
• No? Then access a bottleneck.
Ÿ
Terminating monopoly problem
• Only one path to terminate
• Subscribers care more about what they pay than what those who call
them pay
• Incentive for terminating network to set high fees
16
©Lehr, 2006
Interconnection and Market Power
q
q
q
q
Incentives to interconnect?
Ÿ Network externalities: larger network more valuable
Ÿ No market power, providers interconnect to increase value of both networks
Ÿ Competition for subscribers (which network to join?)
If market power, then may seek to abuse interconnection
Ÿ Natural monopoly, scarce resource, or first-mover advantage
Ÿ Incumbent w/ large network has market power relative to smaller (newer) networks
Ÿ Collusion: bilateral setting of high rates (international settlements, mobile roaming)
Modes of abuse
Ÿ Denial of access: foreclose competition
Ÿ Discriminatory access: inferior access to 3rd parties relative to affiliated subsidiary
Ÿ Monopoly pricing: price access significantly above cost
Regulatory response
Ÿ Common Carriage à non-discriminatory access and interconnection obligation
Ÿ Mandatory unbundling and interconnection
Ÿ Price and terms of interconnection regulated
Ÿ Line of business restrictions (preclude retail entry)
17
©Lehr, 2006
Regulating Carrier Interconnection
q
q
Regulating both retail and wholesale rates problematic
What price to set for interconnection?
Ÿ Efficiency: P=Incremental cost of termination
• Economic (forward-looking), not accounting costs.
• Costs of network “access” recovered on originating end (unbundling)
Ÿ
Wholsale rate > cost à arbitrage, inefficient bypass (distort investment)
• Historically, interconnection prices include subsidies (for universal service, for
non-traffic sensitive “access” costs, etc.)
q
Who sets rate?
Ÿ Regulators: Expensive proceedings to set cost-based rates
• Contribution to shared/common costs? Implicit subsidies?
Ÿ
Markets: Arbitrage enforces “Law of One Price”
• International Bypass, Voice-over-IP
Ÿ
Negotiated: mandate “reciprocal compensation”
• OK if costs symmetric, but what if not? Mobile v. Wired. Traffic asymmetric.
q
Which party pays?
Ÿ Calling (Sending) party pays: problem of mobile termination
Ÿ “Bill and Keep”
18
©Lehr, 2006
Unified Carrier Compensation Scheme
q
q
q
q
Drivers:
Ÿ Convergence: symmetric regulation
Ÿ Liberalization: markets not regulation
Ÿ Globalization: promote free trade (e.g., WTO)
FCC Unified Intercarrier comp regime (2001): Bill & Keep?
European Commission: Interconnection directive
Ÿ Competitive markets: allow flexible negotiation
Ÿ When competition lacking, regulators may enforce
interconnection, which includes rate setting
Ÿ Symmetric rules
One size fits all??
19
©Lehr, 2006
Calling party pays
q
Calling party pays incremental cost of termination
Ÿ
Ÿ
Ÿ
Ÿ
q
Reciprocal compensation
Ÿ
Ÿ
Ÿ
q
q
Doesn’t address call externality (value called party)
Good incentive for quality of service when terminating
Vulnerable to terminating monopoly problem
Vulnerable to monopoly leveraging if market power
Technology same
Negotiated termination fees, but requirement for
reciprocal rates reduces bargaining power of incumbent
e.g., debate over ISP Reciprocal Compensation in U.S.
Incentives to collude? (mobile roaming)
Implications for retail rate regulation?
20
©Lehr, 2006
Bill and Keep
q
q
q
Recover all costs from network’s own subscribers
Ÿ Wholesale rate for interconnection = 0
Ÿ Carriers each pay own costs for interconnection
Used in Internet backbone. Could be used more generally.
Benefits?
Ÿ Simple to implement. No inter-carrier fees paid.
Ÿ Deregulatory: no longer need to set prices for termination.
Ÿ Efficient if:
• Costs of termination symmetric & traffic balanced à net payment~0 anyway.
• Costs termination close to zero
q
Issues:
Ÿ Hot potato routing
Ÿ Asymmetric costs/values (e.g., mobile/wireline)
Ÿ Asymmetric traffic (Web browsing, streaming media)
Ÿ Incentive to terminate with high quality? (Free riding)
21
©Lehr, 2006
Interconnection Tussle
q
q
Issues/Perspectives
Ÿ Efficient pricing: usage v. flat rate charges, elimination of
implicit subsidies
Ÿ Market power? (Terminating or originating monopoly)
Ÿ VoIP?
Ÿ Usage v. Flat rate charges?
Ÿ Jurisdiction?
Stakeholders:
Ÿ Rural Telcos à high rates, retain subsidies, regulate VoIP
Ÿ ILECs à move usage subsidies into SLC, move to BnK
Ÿ CLECs à competitive neutrality (cost-based), reciprocal comp
Ÿ States à retain state autonomy to set local/intrastate rates
Ÿ FCC à BnK to simplify and increase cross-platform competition
22
©Lehr, 2006
Additional Slides
©Lehr, 2006
Costs of terminating traffic
q
Economic not accounting (historic)
Ÿ
Ÿ
Ÿ
Resources priced at opportunity cost
Forward-looking: ignore sunk/history irrelevant
Incremental: short-run or long-run?
• Short-run: take capacity as fixed. Exclude fixed/sunk.
– Marginal costs = dTC/dq
• Long-run: investment in capacity.
– Long-run Incremental Cost (LRIC)
Ÿ
q
How to estimate?
Ÿ
Ÿ
Ÿ
q
• Exclude costs already recovered in access (origination)
Per minute (switching), per call (set-up), per month (capacity)?
Market data (comparables?)
Engineering cost models
Accounting data, adjusted to reflect productivity gains
Costs variable? e.g., Hot potato routing.
24
©Lehr, 2006
Externalities
q
q
q
Externality: benefits (or costs) imposed on others as result
of individuals actions.
Ÿ Prices which do not reflect all benefits (costs) result in
too little (too much) usage
Ÿ Examples: pollution, traffic jams, spectrum interference
Solution: internalize the externality so individual
cost/benefit reflects all impacts
Ÿ Example: pollution fines, road tolls, spectrum fees
Relevant examples for interconnection
Ÿ Network (“subscriber”) externality (positive)
Ÿ Calling externality (positive)
Ÿ Congestion externality (negative)
25
©Lehr, 2006
Network Externality
q
q
q
q
Bigger network more valuable. Impact positive.
Ÿ Direct: expanded connectivity. More options for
calling.
Ÿ Indirect: more complementary goods, lower costs
Subscriber externality
Ÿ Early adopters convey benefit on later (justify
penetration pricing?)
Ÿ Diminishing marginal returns
Examples: Universal service, Microsoft Windows, Internet
Should small network pay more when connecting to big
network?
26
©Lehr, 2006
Calling Party Externality
q
q
q
q
q
Both called and calling party benefit from call
Typically only calling party pays: makes fewer calls than
optimal
Costs of terminating calls may not be symmetric
Ÿ e.g., Mobile to wireline, Web browsing
Ÿ Origination vs. termination (e.g., switch usage)
Ÿ Not always positive: SPAM
Solutions:
Ÿ Both parties pay (in US, mobile caller and called party pay)
Ÿ Inter-temporal alternating direction of origination
Ÿ Flat rate billing
Should called and calling party pay? Metering/privacy?
27
©Lehr, 2006
Congestion externality
q
q
Caller’s traffic slows down everyone else’s traffic
when network congested. Delay imposed on
other’s is ignored by sender.
Solutions:
Ÿ Congestion pricing: internalize externality
Ÿ Peak-load pricing: time varying prices (e.g.,
time of day tariffs)
28
©Lehr, 2006
Arbitrage
q
q
q
“Law of One Price”
Ÿ Close substitutes ought to have similar prices.
Ÿ Buy one and sell other.
Examples:
Ÿ Call-back in International Telephone
Ÿ VoIP to avoid telephone charges
Ÿ Reciprocal Comp: ISPs and CLECs in US
Is it efficient?
Ÿ Forces prices in line with costs (e.g., financial markets)
Ÿ Makes difficult to sustain regulatory subsidies
29
©Lehr, 2006