Securing the infrastructure
 -international reflections
   Professor Howard Williams
Main themes
• Distributional issues
• BB eco system
• Infrastructure
Distributional Issues
• We need ask the fundamental question about the specific
  values and returns we expect from BB infrastructures?
• Trade perspectives and call patterns
• This is not a straight forward question !
• There are profound distributional issues – Sprint and Apple
  in the USA; $15bn transfer
• Colombia has reached a key point; over 4 million
  connections, high growth rates in the recent past.
• Who are those without access?
• Who is extracting value from those who have access?
• Theoretical issues – greatest distortion flow from
  distorted/subsidised input prices (Diamond/Mireless)
Overselling BB – Charles Kenny
                                                 Year     Predicted Values at Given GDP/Capita Average Values   Average Values
Income level                                                 1,000 5,000       10,000   30,000                  Poor    Rich

Secure Internet servers (per 1 million people)     2001     neg     21        38        64                        5          66
                                                   2008     neg     109       205       357                       15         331

Fixed broadband subscribers (per 100 people)       2001     neg      0.6      1        1.7                         0         1.8
                                                   2008     neg      5.7     10.1      16.9                       1.3        16

Internet users (per 100 people)                    2001     neg       9       15        24                        2              25
                                                   2008     neg      24       36        54                        12             50

Mobile cellular subscriptions (per 100 people)     2001     neg      21       32        51                        6          53
                                                   2008     24       71       91        124                       52         115

Fixed line subscriptions (per 100 people)          2001     neg      19       28        42                        8.6        41.8
                                                   2008     neg      17       24        37                        8.3        35.1

% of Firms Using Email                             2009     49       69      77        91                         59         81
% of Firms using Own Website                       2009     16       39      49        66                         27         55
Literacy rate, adult total (% )                    2008     66       83      91        103                        75         95
School enrollment, tertiary (% gross)              2008      6       32      43        61                         20         60
Value Lost Due to Power Outages (% of Sales)       2009      6        4       3         1                          6          2
ICT Exenditure ($/capita)                          2008     59      290      576      1714                        206       1429
GDP Density ('000/km)                              2008     neg     8518    17690     32229                       417       26926
Rural population (% of total population)           2008     68       48      39        25                         57         28

(Average GDP/Capita)                               2001                                                          3,526      25,278
                                                   2008                                                          3,553      24,926
Two sided markets: Low spend customers
                              benefit most from receiving calls
 Through CPP, revenue from received calls allows low spending users to be
 connected, even though the scale of subsidy is small.
                               Contribution to total ARPU, by value
                                                                                                           > Although high spending
                                                                                                             customers receive a lot of
Making calls




                                                                                                             calls, the revenue from
                                                                                                             this is greatly exceeded by
                                                                                                             what they pay for making
                                                                                                             calls.
                                                                                                           > The majority of revenue
                                                                                                             for the lowest spending
Receiving calls




                                                                                                             group comes from
                                                                                                             receiving calls.
                                                                                                           > Low spending users are
                                                                                                             able to maintain a pre-pay
                   11%      17%        17%        13%        9%      7%     5%       4%       7%     10%
                  Percentage of the total mobile users in each ARPU band
                                                                                                             account without an
                  Source: Vodafone customers in Delhi                                                        ongoing subscription.
                                                                                                           > The lowest spending group
                                                 7             Making Broadband Accessible for All
                                                                                                             represent 2012 of users but
                                                                                                                   03 July
                                                                                                                           11%
                                                                                                             only 1% of revenue.
Bringing broadband to the majority of citizens
Fibre offers the fastest speeds at the greatest cost (suitable for high demand users);
wireless is the cheapest and fastest way to reach universal broadband coverage.
 Fibre                                        Wireless
 > Optical fibre offers broadband connections up                   > Wireless broadband can offer speeds from <1Mb/s
   to 100Mb/s                                                        (3G and its developments HSPA+) to over 40Mb/s
 > Fibre is very costly to install; as access network                (LTE)
   it is only commercially viable in densely-                      > Bringing wireless broadband to rural areas will
   populated, affluent areas                                         require significant investment but remains the
 > 70% of the cost of a next generation fibre                        cheapest access technology
   network is in the last 100m of the access                       > Advanced services, such as e-health, can be reliably
   network                                                           provided using wireless broadband



                                                                                     Fibre
                                                                       100
                                                                                                                                Streamed
                                                               Speed (Mbit/s)                Cable                        LTE   HD video


                                                                                10                                              Advanced
                                                                                                     ADSL         HSPA+         e-health

                                                                                                             3G                 YouTube
                                                                                1                                               Internet,
                                                                                                                                Email
                                                                                              03 July 2012

                          8           Making Broadband Accessible for All
Relative viability of fibre and wireless broadband in India
       Only the dense urban areas have sufficient aggregate monthly income per km2 to
       support the cost of a fibre access network, but almost all districts support wireless.

       Network cost as a % of district income                                                      Maharashtra illustrates the urban / rural split
                 Wireless                                                                          > The equivalent monthly cost of fibre per km2 can be
                                                                                                     determined by the revenue per customer that WIK
                                                                                                     calculated as being necessary to support a viable
Rural geotype




                                                                                                     network, for each geotype.
                                                                                                   > Placing each district into its geotype, we can
                                                                                                     compare the aggregate monthly income per district
                                                                                                     to the calculated monthly income (Net District
                                                                                                     Domestic Product/NDDP).
                                                                                                   > ITU data shows the demand for telecom services is
                                                                                                     commonly 2-5% of state GDP.
Dense rural geotype




                                                                                                   > The cost of fibre would represent 1% of NDDP in
                                                                                                     Mumbai and 4% in Thane. These are ‘dense urban’
                                        Only in Mumbai and Thane                                     and ‘less suburban geotypes’. In all other districts,
                                        districts does the cost of fibre                             the equivalent cost of fibre would be a much greater
                                        approach wireless, as a % of                                 % of district income, rising to 61% of NDDP in
                                        monthly NDDP                                                 Washim and 177% in Gadchiroli.
                                                                                                   > The equivalent cost of wireless broadband is 3% or
                                                                                                     less of district income.
        Source: State Economic Census for Maharashtra, WIK study of fibre network cost, Vodafone
        analysis



03 July 2012                                                                                                                                              9
Relative viability of fibre and wireless broadband in Jo’burg
 Repeating exercise for suburbs of Johannesburg shows that fibre roll-out likely to be
 challenging for the majority of the population
                                             Joburg highlights economics of bbd investment
 Fibre network cost as a % of income

                                             > Repeat exercise for districts of Jo’burg.
                                             > The richest parts of Jo’burg are in high cost
                                               deployment areas; whereas low income areas are in
                                               dense areas which are lower cost to deploy.
                                             > The cost of fibre would represent over 10% of
                                               income Diepkloof. Whereas, high income levels in
                                               Parkview means it is profitable to deploy fibre.
                                             > The equivalent cost of wireless broadband for
                                               Diepkloof is 4% or less of income.


                                             Wireless is less than half the cost of
                                             fibre for low income areas of
                                             Johannesburg




03 July 2012                                                                                      10
Relative viability of fibre and wireless broadband in India
 Charting the equivalent monthly cost of fibre and wireless networks as a % of
 monthly district income demonstrates that fibre only affordable in dense urban
 districts, but that wireless is less than 4% of monthly NDDP in all except the most
 rural districts.
 Maharashtra                     Karnataka                 Rajasthan
     Gadchiroli wireless 8% / fibre 177%                                                       Churu    wireless 3% / fibre 131%
        Washim                                                                               Bikaner    wireless 3% / fibre 127%
   Osmanabad                                                                                 Barmer     wireless 3% / fibre 170%
         Hingoli                                                                           Jaisalmer    wireless 12% / fibre 573%
      Yavatmal                                       U.Kannada
          Beed                                            Koppal                               Jalore
          Jalna                                  Chamarajanagar                              Nagaur
      Buldhana                                            Bijapur                               Tonk
          Dhule                                     Chikmagalur                             Jodhpur
       Amravati                                      Chitradurga                               Baran
    Sindhudurg                                           Raichur                                 Pali
     Nandurbar                                              Bidar                        Cittoragarh
    Chandrapur                                            Gadag                            Jhalawar
        Wardha                                      Chikballapur                     Hanumangarh                                    Rural
      Parbhani                                           Hassan                                Bundi
        Gondia                                           Tumkur                                Karoli
      Ratnagiri                                        Shimoga                                 Sirohi
   Ahmednagar                                            Kodagu                      Sriganganagar
        Solapur                                        Bagalkot                              Udaipur
         Satara                                           Bellary                           Bhilwara

       Nanded                                          Gulbarga                         Dungarpur
          Latur                                           Haveri                     Swaimadhopur
         Akola                                          Mandya                             Dholpur
     Bhandara                                       Ramanagara                                Sikar
       Jalgaon                                         Belgaum                           Jhunjhunu
         Sangli                                     Davanagere                         Rajsamand                                    Dense rural
   Aurangabad                                              Kolar                            Dausa
        Raigad                                            Udupi                              Ajmer
        Nashik                                           Mysore                          Bharatpur
      Kolhapur                                         Dharwad                                Kota
       Nagpur                                     Bangalore rural                            Alwar
          Pune                                       D.Kannada                              Jaipur

                                                                                          Banswara                                  Less suburban
         Thane
                                                                                                                                    Dense urban
        Mumbai                                         Bangalore

                0%      20% 40% 60% 80%                         0% 20% 40% 60% 80%                     0%     20% 40% 60% 80%

03 July 2012                               All                                                                                                11
Professor Howard Williams

  howard.williams@strath.ac.uk
Additional highlights from the presentation include:
- The average internet user in Colombia spent 20.4 hours online during September,
consuming 1,606 pages of content and averaging 42 online visits during the month.
- 86% of Colombians visited a social networking destination in September, with
Facebook leading the category. Visitors averaged 4.6 hours on the site during the
month.
- Nearly 7 out of 10 Colombians visited a photo sharing site in September led by
Facebook.com Photos.
- An average searcher in Colombia conducted 184 searches in September, resulting
in a total of 2 billion queries conducted in Colombia during the month.
(comScore, December 2010)

Experiencia Internacional en Políticas Públicas para cerrar la brecha digital

  • 1.
    Securing the infrastructure -international reflections Professor Howard Williams
  • 2.
    Main themes • Distributionalissues • BB eco system • Infrastructure
  • 3.
    Distributional Issues • Weneed ask the fundamental question about the specific values and returns we expect from BB infrastructures? • Trade perspectives and call patterns • This is not a straight forward question ! • There are profound distributional issues – Sprint and Apple in the USA; $15bn transfer • Colombia has reached a key point; over 4 million connections, high growth rates in the recent past. • Who are those without access? • Who is extracting value from those who have access? • Theoretical issues – greatest distortion flow from distorted/subsidised input prices (Diamond/Mireless)
  • 4.
    Overselling BB –Charles Kenny Year Predicted Values at Given GDP/Capita Average Values Average Values Income level 1,000 5,000 10,000 30,000 Poor Rich Secure Internet servers (per 1 million people) 2001 neg 21 38 64 5 66 2008 neg 109 205 357 15 331 Fixed broadband subscribers (per 100 people) 2001 neg 0.6 1 1.7 0 1.8 2008 neg 5.7 10.1 16.9 1.3 16 Internet users (per 100 people) 2001 neg 9 15 24 2 25 2008 neg 24 36 54 12 50 Mobile cellular subscriptions (per 100 people) 2001 neg 21 32 51 6 53 2008 24 71 91 124 52 115 Fixed line subscriptions (per 100 people) 2001 neg 19 28 42 8.6 41.8 2008 neg 17 24 37 8.3 35.1 % of Firms Using Email 2009 49 69 77 91 59 81 % of Firms using Own Website 2009 16 39 49 66 27 55 Literacy rate, adult total (% ) 2008 66 83 91 103 75 95 School enrollment, tertiary (% gross) 2008 6 32 43 61 20 60 Value Lost Due to Power Outages (% of Sales) 2009 6 4 3 1 6 2 ICT Exenditure ($/capita) 2008 59 290 576 1714 206 1429 GDP Density ('000/km) 2008 neg 8518 17690 32229 417 26926 Rural population (% of total population) 2008 68 48 39 25 57 28 (Average GDP/Capita) 2001 3,526 25,278 2008 3,553 24,926
  • 7.
    Two sided markets:Low spend customers benefit most from receiving calls Through CPP, revenue from received calls allows low spending users to be connected, even though the scale of subsidy is small. Contribution to total ARPU, by value > Although high spending customers receive a lot of Making calls calls, the revenue from this is greatly exceeded by what they pay for making calls. > The majority of revenue for the lowest spending Receiving calls group comes from receiving calls. > Low spending users are able to maintain a pre-pay 11% 17% 17% 13% 9% 7% 5% 4% 7% 10% Percentage of the total mobile users in each ARPU band account without an Source: Vodafone customers in Delhi ongoing subscription. > The lowest spending group 7 Making Broadband Accessible for All represent 2012 of users but 03 July 11% only 1% of revenue.
  • 8.
    Bringing broadband tothe majority of citizens Fibre offers the fastest speeds at the greatest cost (suitable for high demand users); wireless is the cheapest and fastest way to reach universal broadband coverage. Fibre Wireless > Optical fibre offers broadband connections up > Wireless broadband can offer speeds from <1Mb/s to 100Mb/s (3G and its developments HSPA+) to over 40Mb/s > Fibre is very costly to install; as access network (LTE) it is only commercially viable in densely- > Bringing wireless broadband to rural areas will populated, affluent areas require significant investment but remains the > 70% of the cost of a next generation fibre cheapest access technology network is in the last 100m of the access > Advanced services, such as e-health, can be reliably network provided using wireless broadband Fibre 100 Streamed Speed (Mbit/s) Cable LTE HD video 10 Advanced ADSL HSPA+ e-health 3G YouTube 1 Internet, Email 03 July 2012 8 Making Broadband Accessible for All
  • 9.
    Relative viability offibre and wireless broadband in India Only the dense urban areas have sufficient aggregate monthly income per km2 to support the cost of a fibre access network, but almost all districts support wireless. Network cost as a % of district income Maharashtra illustrates the urban / rural split Wireless > The equivalent monthly cost of fibre per km2 can be determined by the revenue per customer that WIK calculated as being necessary to support a viable Rural geotype network, for each geotype. > Placing each district into its geotype, we can compare the aggregate monthly income per district to the calculated monthly income (Net District Domestic Product/NDDP). > ITU data shows the demand for telecom services is commonly 2-5% of state GDP. Dense rural geotype > The cost of fibre would represent 1% of NDDP in Mumbai and 4% in Thane. These are ‘dense urban’ Only in Mumbai and Thane and ‘less suburban geotypes’. In all other districts, districts does the cost of fibre the equivalent cost of fibre would be a much greater approach wireless, as a % of % of district income, rising to 61% of NDDP in monthly NDDP Washim and 177% in Gadchiroli. > The equivalent cost of wireless broadband is 3% or less of district income. Source: State Economic Census for Maharashtra, WIK study of fibre network cost, Vodafone analysis 03 July 2012 9
  • 10.
    Relative viability offibre and wireless broadband in Jo’burg Repeating exercise for suburbs of Johannesburg shows that fibre roll-out likely to be challenging for the majority of the population Joburg highlights economics of bbd investment Fibre network cost as a % of income > Repeat exercise for districts of Jo’burg. > The richest parts of Jo’burg are in high cost deployment areas; whereas low income areas are in dense areas which are lower cost to deploy. > The cost of fibre would represent over 10% of income Diepkloof. Whereas, high income levels in Parkview means it is profitable to deploy fibre. > The equivalent cost of wireless broadband for Diepkloof is 4% or less of income. Wireless is less than half the cost of fibre for low income areas of Johannesburg 03 July 2012 10
  • 11.
    Relative viability offibre and wireless broadband in India Charting the equivalent monthly cost of fibre and wireless networks as a % of monthly district income demonstrates that fibre only affordable in dense urban districts, but that wireless is less than 4% of monthly NDDP in all except the most rural districts. Maharashtra Karnataka Rajasthan Gadchiroli wireless 8% / fibre 177% Churu wireless 3% / fibre 131% Washim Bikaner wireless 3% / fibre 127% Osmanabad Barmer wireless 3% / fibre 170% Hingoli Jaisalmer wireless 12% / fibre 573% Yavatmal U.Kannada Beed Koppal Jalore Jalna Chamarajanagar Nagaur Buldhana Bijapur Tonk Dhule Chikmagalur Jodhpur Amravati Chitradurga Baran Sindhudurg Raichur Pali Nandurbar Bidar Cittoragarh Chandrapur Gadag Jhalawar Wardha Chikballapur Hanumangarh Rural Parbhani Hassan Bundi Gondia Tumkur Karoli Ratnagiri Shimoga Sirohi Ahmednagar Kodagu Sriganganagar Solapur Bagalkot Udaipur Satara Bellary Bhilwara Nanded Gulbarga Dungarpur Latur Haveri Swaimadhopur Akola Mandya Dholpur Bhandara Ramanagara Sikar Jalgaon Belgaum Jhunjhunu Sangli Davanagere Rajsamand Dense rural Aurangabad Kolar Dausa Raigad Udupi Ajmer Nashik Mysore Bharatpur Kolhapur Dharwad Kota Nagpur Bangalore rural Alwar Pune D.Kannada Jaipur Banswara Less suburban Thane Dense urban Mumbai Bangalore 0% 20% 40% 60% 80% 0% 20% 40% 60% 80% 0% 20% 40% 60% 80% 03 July 2012 All 11
  • 12.
    Professor Howard Williams howard.williams@strath.ac.uk
  • 13.
    Additional highlights fromthe presentation include: - The average internet user in Colombia spent 20.4 hours online during September, consuming 1,606 pages of content and averaging 42 online visits during the month. - 86% of Colombians visited a social networking destination in September, with Facebook leading the category. Visitors averaged 4.6 hours on the site during the month. - Nearly 7 out of 10 Colombians visited a photo sharing site in September led by Facebook.com Photos. - An average searcher in Colombia conducted 184 searches in September, resulting in a total of 2 billion queries conducted in Colombia during the month. (comScore, December 2010)