The Future of Mobile Communications

1 An Open Access Wireless Market Supporting Public Safety, Universal Service, and Competition

Peter Cramton Linda Doyle

2 Summary

• The cellular market is extremely inefficient, with a small number of carriers dictating supply and pricing. • Implementing an Open Access Wireless Market (OAWM) will allow supply and demand to determine prices by time and location, with the following consequences: – It will Increase innovation, generate new use cases, and diversify traffic patterns – It will eliminate the barriers to entry • This will transform the industry by increasing network utilization, reducing the price of data per GB and increasing data consumption dramatically. • Rivada is actively working on implementation opportunities globally. From monopoly to vibrant competition

• Original “Ma Bell” telecommunications Monopoly

• Spectrum auctions • Mobile communications Oligopoly

• Open Access Wireless Market (OAWM) • Wholesale-only provider eliminates the conflict of interest in the carrier retail / wholesale market • Neutral network with transparent, non-discriminatory pricing to MVNOs and wholesale consumer Competition • Time and location pricing enhances network efficiency and promotes low-cost entry

4 Mobile networks Wholesale market Open access network (ISO) Service providers

Mobile virtual Proprietary network operator network 1 (MVNO) Retail (MNO1) Users/devices market MNO1

A B C D E F G Proprietary network 2 MNO2

(MNO2)

Same market model as electricity

successfully operating for two decades 5 The Mobile market today is highly concentrated & retail centric

Spectrum Network Operators Retail Market Vertical Integration

Licensed Spectrum Spectrum Network

Wholesale $189 B $50 B $25 B

Retail $213 B

Wholesale Market

$56 B Mobile market tomorrow will be more diversified Spectrum Network Operators Retail Market Vertical Integration

Licensed Spectrum Spectrum Network

Wholesale $189 B $50 B $25 B

Retail $213 B

Additional Capacity in Constraine d Markets Shared Spectrum Wholesale Market Excess Capacity

$324 B OAWM $56 B Wholesale $3.0 B $873 B $542 B Spectrum Holders Frictionless Access Excess Capacity $519 B Transparent Prices $168 B Net Neutrality Repeal $370 B $636 B $23 B Neutral Wholesale $16 B $95 B Spectrum Network $711 B $52 B $1.5 B Wholesale $58 B $84 B $98 B MVNO & wholesale evolving

Wholesale • Wholesale-only provider eliminates the conflict of interest in the carrier retail / wholesale market

Neutral • Neutral network with transparent, non- discriminatory pricing to MVNOs and wholesale consumers

Time and Location pricing • Time and location pricing enhances network efficiency and promotes entry Range of potential customers is wide & varied

— Virgin Mobile — ACN Traditional MVNOs — Credo — Ready Mobile — Ting — TracFone There are many

Cable Operators/Rural — Comcast — SouthernLinc interested parties — Cox — C-Spire Carriers — Charter — Cellular One — iWireless — US Cellular and more to come

— Google — Amazon — Apple — Netflix Wholesale Tech Companies / — FreedomPop — IBM Innovators — Republic — Hewlett-Packard • Diversifying the wireless — Dell — Microsoft ecosystem — TracFone — iWireless — Q Link — Safelink LifeLine — Total Call — TruConnect — Boomerang — Tag Mobile — Budget PrePay — Ready Wireless • Innovating business models

— Deutsche Telekom — International MNOs — Vodafone — Orange — Telefonica — KDDI • Diversifying the traffic — T-Mobile — AT&T Existing MNO’s Domestic MNOs — Sprint — Regional Carriers — Verizon — Rural Carriers patterns

— Aeris — Fitbit — Kore/Wyless — UBI — Numerex — Zubie M2M M2M Adopters — Spireon — Auto OEMs — Geotab — Gemalto — UPS — Trimble

9 Neutral wholesale network reduces the tension between competition and efficiency

Competition

+ OAWM National Needs Right Conditions

Neutral Universal Service Global Standards Wholesale Competition Technological Flexibility Public Safety Demand Heterogeneity Data explosion Scarcity More Value Chain Agents

Network Sharing More Value Policy Innovations Business Innovations Chain Agents e.g. MORAN or MOCN Spectrum Access Neutral Wholesale Infrastructure Access No Influence Passive Infrastructure Sharing Time and Location Pricing ISO e.g. Tower sharing 4 MNO 3 MNO 2 MNO

1 MNO Independent operation The FUTURE Efficiency Demand aggregation allows neutral host to enjoy economies of scale previously available only to dominant operators The Current Model

• Market size is a determinant of profitability

• Smaller network operators cannot compete due to economies of scale Wholesale market represents less than 10% of entire global market The Current Model MVNO Share of Market by Country

Netherlands Germany • Inherent conflict of interest between Denmark seller (carrier) and buyer (MVNO) Austria • Buyers are competing with their own France Korea suppliers Spain Japan Italy • This hinders competition and keeps Chile MVNOs small Colombia Poland Mexico Argentina

0.00% 10.00% 20.00% 30.00% 40.00% Market Share - % Operators are traditionally poor at maximizing value of assets Network Utilization in 2016 The Current Model

• Networks are built to meet peak demand

• Network utilization is low

Spectrum Usage in 2016 • Spectrum utilization is low

• Investment diminished

Source: http://dfmonitor.eu/networkeconomics/ Operators charge same price despite huge swings in demand

Manhattan – early morning Manhattan - evening The Current Model

• There is a big mismatch between data

15 needs and pricing

10 Hourly Price • Data needs and value vary by time and Average Price 5 location Price $ / GB

0

80% • Data prices are constant

60%

40% Average Utilization: 35% 20%

Network Utilization 0% 1 3 5 7 9 11 13 15 17 19 21 23 Applying open access has dramatic impact on asset usage and revenue Manhattan – early morning Manhattan - Evening The open access model

• A neutral time and location based

15 market will drive increased utilization

10 Hourly Price • Prices reflect the value of data at every 5 time and location

Price $ / GB Average Price

0

80% • Average prices decrease significantly

60% Average Utilization: 65%

40% • Average utilization increases due to low Average Utilization: 35% prices 20%

Network Utilization 0% 1 3 5 7 9 11 13 15 17 Hour19 of day 21 23 Location, location, location

Washington DC 6am Manhattan 6pm The open access model 14

12 • Prices reflect the value of data at every time and location

10 • Local Prices reflect local data needs 8 Price $ / GB 6 • Utilization increases everywhere

4

2

0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Anonymized data consumption Operator traffic patterns on a national carrier network

7.00% 6.00% 5.00% • Data consumption by hour of day 4.00% • The network is better utilized in the evening 3.00% (Netflix Effect) but a lot of capacity is wasted 2.00% in the early morning. 1.00% % of data consumption 0.00% 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Hour of day

6.00%

5.00% • The network is broken down into 100 buckets, each with an equal number of sites 4.00% ordered by usage. The usage in the busiest 3.00% bucket is 1,700 times greater than the least 2.00% used bucket. % of data consumption 1.00% • Data consumption is highly concentrated in a 0.00% 1% 10% 19% 28% 37% 46% 55% 64% 73% 82% 91% 100% few buckets, producing low utilization in many areas without much traffic Location Percentile Open access will have a dramatic The open access model effect on date use and price 7.1 $/ GB – Current Model GB/Month/Sub $/GB 30.00 4.50 • Coverage Design for a National Network on the 700 MHz band 4.00 25.00 22 GB/Month/Sub – Rivada Model 3.50 • Current Model:

20.00 3.00 • Price per GB will drop from $7.1 to $2.3 • Consumption will rise from 4 GB to 22 GB 2.50 2.3 $/ GB – Rivada Model 15.00 2.00 • Key economic assumptions: 10.00 1.50 • Price elasticity: 1.3 GB per Month 1.00 • Elasticity of substitution: 0.3 5.00 4 GB/Month/Sub – Current Model Price per GB 0.50

0.00 0.00 35% 45% 55% 65% 75%

Target Utilization An increase in network utilization has a The Rivada Model material impact on revenue & enterprise value

$ B

9.00 Revenue • Better utilization means more revenue 31% 37% 8.00 25% 18% 7.00 • Existing assets are utilized better 6.00 5.00 • New investments have better 4.00 returns 3.00 2.00 1.00 0.00 • When compared with a traditional 35% 45% 55% 65% 75% Network Utilization retail-centric network, the Neutral $ B Wholesale Model with the OAWM can 25.00 Enterprise Value 58% 45% increase revenue and enterprise value 20.00 31% 15% 15.00 Increasing utilization from 35% to 65%, 10.00 • increases: 5.00 • Revenue by 31% 0.00 35% 45% 55% 65% 75% Network Utilization • Enterprise value by 45% Characteristics of open access Spectrum Sharing • Market for wireless capacity, not spectrum • Feasible only for marginal quantities of • Capacity (throughput) is generated from both spectrum spectrum and network assets • Based on best practices from existing time and • Vertical sharing between drastically locational markets. Core tenets are efficiency, different uses transparency, simplicity and fairness • Military vs Commercial • All users have access to the same technology • Mobile communications vs Point-to- • Feasible for large quantities of spectrum across Point frequency bands • Horizontal sharing between similar users • Compatible with traditional bilateral wholesale • Redundant and idle investments contracts • Extremely difficult to manage • Simple to manage: • Interference • Interference is managed at the network level • Incentives • Transparent, market-driven pricing aligns incentives • See whitepaper at: OpenAccessWireless.com Open access facilitates a more Current license model efficient way of allocating spectrum Licenses are auctioned to the highest bidder • Bids reflect the value of: • Spectrum is paid recurrently: • Using the spectrum • Price reflects the fair market value value of using the spectrum • Rents from monopolistic competition • Rents from blocking entry • Innovations get accounted in the price • Consumer prices ($/GB) are high • Inferior services • Nominal barriers to entry: • Slow deployment • Local and fast deployment is possible • Constrained coverage • Innovation flourishes – new uses of data • Barriers to entry are high • Competitive prices • Few service providers • Few users • Incentivizes spectrum hoarding Open access is a perfect merger remedy Based on experience: • Similar to recent 4-to-3 merger • Set aside a portion of capacity to the open access remedies in Germany and Ireland, but market much more responsive to changing • Assignment and prices determined to maximize gains market from trade in open competitive process

• Assignment and prices respond to changing market • Similar to electricity merger remedy: auction portion of generation as “Virtual • Market evolves with environment Power Plants” e.g. EDF 2001-2012 • Merged entity receives competitive market value for network resource, and ongoing interest in success of service providers Product design

• Product should be directly valued by consumer • Network throughput at particular location and time interval and latency • A market for throughput and latency not spectrum • Energy (MWh) in an electricity market • Aggregation in both time and location to simplify and improve liquidity • Example: Particular cell over one hour (GB/h)

23 All markets use single-price auction Price ($/GB)

Winning buyers Supply

Clearing price P*

Demand Winning sellers

Q* Quantity traded Quantity (GB/h) 24 Multiple opportunities to trade: Yearly, monthly, hourly On peak Off peak

Hour 7 Hour 15 Hour 3

Hourly Hourly … area A area B

Hourly area J Demand Monthly area M Low Medium High

Yearly aggregates monthly aggregates hourly in time and location 25 Service provider estimates demand and stages purchase in three markets Yearly auction = buy 165 GB per hour, for every peak hour in the year, in the yearly area

5 GB 5 GB 5 GB 5 GB 5 GB 5 GB

5 GB 10 GB 15 GB 15 GB 5 GB 5 GB

5 GB 10 GB 10 GB 10 GB 10 GB 5 GB

5 GB 5GB 5 GB 5 GB 5 GB 5 GB

26 Refine estimate and make adjustment in monthly market Yearly auction = buy 165 GB per hour, for every peak hour in the year, in the yearly area

Monthly auction = sell 8 GB per hour, for every peak Monthly auction = buy 15 GB per hour, for every peak hour in the month, in the red area hour in the month, in the green area

4 GB 4 GB 4GB 7 GB 8 GB 8 GB

4 GB 9 GB 17 GB 17 GB 8 GB 8 GB

4 GB 9 GB 12 GB 12 GB 12 GB 8 GB

4 GB 7GB 7 GB 7GB 7 GB 8 GB

Monthly auction = buy 20 GB per hour, for every peak hour 27 in the month, for the blue area Finalize estimate one hour ahead and make final adjustment to demand

Hourly area H Peak hour product Hour Change Net 07 Buy 3 GB 11 GB 08 Sell 2 GB 6 GB … … … 17 Buy 1 GB 9 GB

8 GB 28 Sequence of auctions Hour Yearly Monthly Hourly ahead auction auctions realization auction Example April: April 5, hour 15: April 5, hour 15: Buy 8 GB Sell 1 GB Buy 3 GB -1 GB deviation in each hour 7 GB net 10 GB net 9 GB demand • Three opportunities to trade • Reduces risk of service provider • Facilitates planning of service provider • Provides price transparency • Mitigates market power 29 Prototype auction platform

• To illustrate market • Demonstrate proof of concept

30 Auction design

• Uniform-price auction for each product • Preferences expressed as piecewise-linear strictly-decreasing demand curves • Consistent with underlying preferences • Unique clearing prices and quantities • Yearly and monthly auctions: simultaneous ascending clock • Hourly auction: sealed bid

31 32 Sample demand for bidder Manhattan (peak) monthly

33 Addressing hourly deviations between actual and purchased demand • Neither system operator nor service provider can control demand perfectly to assure quantity consumed = quantity won • Deviations are inevitable • Final settlement should motivate service providers to limit deviations

34 Hourly settlement for deviations • = price in hour h in area k that balances as-bid demand with estimated supply • = total quantity bought by bidder i in hour h in hourly area k 𝑝𝑝ℎ𝑘𝑘 (includes yearly, monthly, and hourly net purchases) • 𝑞𝑞𝑖𝑖𝑖𝑖� = actual quantity consumed by bidder i in hour h in hourly area k • = = deviation between actual quantity consumed and quantity bought𝑄𝑄𝑖𝑖𝑖𝑖� • 𝐷𝐷Tolerance𝑖𝑖𝑖𝑖� 𝑄𝑄𝑖𝑖𝑖𝑖�= percentage− 𝑞𝑞𝑖𝑖𝑖𝑖� tolerance band (e.g., 10%); no penalty if deviation is within Tolerance • Penalty Factor = a factor that is applied to square deviations above Tolerance • Adjustment for deviations in the real-time market is = × + where

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖� 𝑝𝑝ℎ𝑘𝑘 𝐷𝐷𝑖𝑖𝑖𝑖� 0 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖� • = 𝐷𝐷𝑖𝑖푖𝑖𝑖 × × 𝑖𝑖𝑖𝑖 𝑞𝑞𝑖𝑖푖𝑖𝑖 ≤> 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑖𝑖𝑖𝑖� 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 � 2 𝐷𝐷𝑖𝑖푖𝑖𝑖 ℎ𝑘𝑘 𝑖𝑖𝑖𝑖� 𝑞𝑞𝑖𝑖푖𝑖𝑖 Standard efficient𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 settlement𝑃𝑃𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑝𝑝if deviation𝐷𝐷 𝑖𝑖𝑖𝑖 is within Tolerance;𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 penalty based on squared deviation outside of Tolerance to induce best estimate and control of demand to match winnings 35 Likely implementations • Mexico (competition) • 90 MHz of 700 MHz set aside for open access • RFP to select implementer in 28 Sep 2016 • United States (public safety) • 20 MHz of 700 MHz set aside + $7 billion • RFP to select implementer in November 2016 • European Union (merger remedy) e.g. UK: Three & O2 • Proposed mergers leading from 4 to 3 carriers • Merged entity allocates portion of network to open access

36