Allocating universal service subsidies using
electronic multi-round reverse auctions:
Telecommunications in Tanzania
World Bank Group
29 June 2023
Final Report
© 2023 The World Bank Group
1818 H Street NW, Washington, DC 20433
Telephone: 202-473-1000; Internet: www.worldbankgroup.org
Some rights reserved
1 2 3 4 21 20 19 18
The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the World Bank Group, its Board of Executive
Directors, or the governments they represent. The World Bank Group does not guarantee the accuracy of the data included in this work. The boundaries,
colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the
legal status of any territory or the endorsement or acceptance of such boundaries.
Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of the World Bank Group, all of which
are specifically reserved.
Rights and Permissions
This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.
org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for
commercial purposes, under the following conditions:
Attribution—Please cite the work as follows: World Bank Group. 2023. Allocating universal service subsidies using multi-round reverse auctions:
Telecommunications in Tanzania. Washington, DC: World Bank.
Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by
the World Bank Group and should not be considered an official World Bank Group translation. The World Bank Group shall not be liable for any content
or error in this translation.
Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original
work by the World Bank Group. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and
are not endorsed by the World Bank Group.
Third-party content—The World Bank Group does not necessarily own each component of the content contained within the work. The World Bank Group
therefore does not warrant that the use of any third party-owned individual component or part contained in the work will not infringe on the rights of
those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to reuse a component of the work, it is your
responsibility to determine whether permission is needed for that reuse and to obtain permission from the copyright owner. Examples of components can
include, but are not limited to, tables, figures, or images.
All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433,
USA; e-mail: pubrights@worldbank.org
Allocating universal service subsidies using
electronic multi-round reverse auctions:
Telecommunications in Tanzania
World Bank Group
29 June 2023
Final Report
Preface
This report has been prepared by Dr. Chris Doyle, The completion of this study was conducted under
Head of Telecoms, CEPA1 for the World Bank’s Digital the supervision of the World Bank (WB) co-Task-Team
Development Global Practice. The funding has been Leaders for the DTP, Tim Kelly and Sara Ballan. It has
provided by the Digital Development Partnership (DDP) also benefitted from discussions held with stakeholders
Trust Fund2 and is given in support of the implementation during the DTP Implementation Support Mission held over
of the Digital Tanzania Project (DTP – P160766)3. 1-5 August 2022. In particular, meetings were held with
representatives of telecommunications operators and
The report explores the use of more innovative and government officials at UCSAF. Meetings took place in Dar
transparent procurement techniques, such as interactive es Salaam and Dodoma.
electronic auction platforms, for the allocation of universal
service awards. Although the specific case study is for An earlier version of the report benefitted from comments
mobile broadband awards in Tanzania, it could easily be provided by Khalid Bin Anjum, Senior Procurement
applied to other universal service infrastructure (such as Specialist, World Bank, Tanzania; Siddhartha Raja,
fiber or satellite), and in other countries. Senior Digital Development Specialist, World Bank;
Andy Dymond, Universal Service Consultant, and Knut J.
The project has been undertaken in close collaboration Leipold, Lead Procurement Specialist, World Bank. Isabel
with the Universal Communication Services Access Fund Neto, Practice Manager IDD04, World Bank, chaired the
(UCSAF), based in Dodoma, Tanzania, and the Ministry review process in December 2022.
of Information, Communications and Information
Technology (MICIT).
1 “CEPA” is the trading name of: Cambridge Economic Policy Associates Ltd (Registered: England & Wales, 04077684); CEPA LLP (A Limited Liability
Partnership Registered: England & Wales, OC326074); and Cambridge Economic Policy Associates Pty Ltd (ABN 16 606 266 602).
2 For more information on the DDP, see: https://www.worldbank.org/en/programs/digital-development-partnership
3 The Digital Tanzania Project (DTP) is funded by The World Bank and has the following project development objectives: to increase access to high
quality internet services for government and citizens, and to improve the government's capacity to deliver digital public services. The DTP is an
investment project financing implemented by the Government of the United Republic of Tanzania through the Ministry of Information, Commu-
nication and Information Technology (MICIT) and the President’s Office for Public Service Management and Good Governance (POPSMGG) with
financing support from the International Development Association (IDA) – see Digital Tanzania Project P-160766
Contents
Preface i
Figures, tables and boxes iv
abbreviations and acronyms vii
Executive summary x
1. Introduction 1
1.1. Background 1
1.2. Telecommunications: access and services 2
1.3. The Digital Tanzania Project (DTP) 3
1.4. Rural broadband policy and DTP 5
1.5. Procurement methods for disbursing funds to promote 6
universal service and universal access objectives
1.6. Current UCSAF and MICIT policy: SRA format 8
1.7. MRRA and price discovery 11
1.8. Report structure 13
2. The Tanzania telecommunications market 14
2.1. Fixed and Mobile network operators 14
2.2. Mobile money 15
2.3. Tower companies 16
2.4. Regulatory landscape 17
3. Universal service policy in tanzania 18
3.1. Universal Communications Service Access Fund (UCSAF) 18
3.2. UCSAF objectives and responsibilities 19
3.3. Internet penetration policy 19
3.4. Disbursing public funds and UCSAF 20
4. Auctions and universal service funds 21
4.1. The MRRA format 22
4.2. The default MRRA format: the clock auction 29
5. International MRRA case study 25
5.1. FCC Auction 904 – MRRA Clock Auction 25
5.2. Auction 904 – outcome 26
6. Lessons for Tanzania 27
6.1. Legal framework 27
6.2. Procurement experience 27
6.3. Capacity building and education 27
v
7. Steps towards a pilot: multi-round reverse clock 28
auction– recommendations
7.1. Step 1: Identification of sites and number of sites 28
7.2. Step 2: Estimate maximum allowable subsidy for each site 28
7.3. Step 3: The budget 28
7.4. Step 4: Application form and pre-qualification 29
7.5. Step 5: Qualified bidders 29
7.6. Step 6: Electronic auction platform 29
7.7. Step 7: Auction education 29
7.8. Step 8: Auction 30
7.9. Step 9: Auction results and public announcement 30
7.10. Step 10: Auction impact evaluation 30
8. Next steps 31
8.1. Wider lessons 31
Appendix
Appendix A telecommunication operators in tanzania 34
Appendix B Regulators in the communications space 39
Appendix C Auction format: multi-round reverse clock auction detailed rules 41
C.1. Before the auction starts
C.2. Bidding rules
C.3. Activity rules
C.4. Processing the clearing round
C.5. Information policy
Appendix D: Numerical illustration of multi-round reverse auction 47
Appendix E Electronic auction platforms 53
E.1. Recommended EAP for pilot auction
E.2. Independent module alongside taneps
Appendix F Slides presented at mission in dodoma 55
Bibliography 60
vi
Figures, tables and boxes
Figure 1 Map of Tanzania 4
Figure 2 Tanzania population density map 4
Figure 3 The bid success rate of UCSAF reverse auctions 6
Figure 4 MNO subscriber shares and subscriber volumes 9
Figure 6 Auction 904 with over 5 million areas eligible for support 26
Figure 7 NICTBB Network Map 37
Table 1 Key Recommendations xi
Table 2 Recommendations for auction pilot xi
Table 3 Next steps xii
Table 4 Bidder estimates of funding required to earn a normal return 2
Table 5 Next steps in the multi-round reverse auction process 3
Table 6 Evaluating aggregate cost 3
Table 7 Within round information policy 8
Table 8 End of round information policy 10
Table 9 Clearing round and subsequent rounds information policy 12
Table 10 Five areas and maximum allowable subsidy 12
Table 11 Illustrative auction round-by-round 16
Table 12 Round 1 bidder values and aggregate cost 21
Table 13 Round 2 bidder values and aggregate cost 31
Table 14 Round 3 bidder values and aggregate cost 44
Table 15 Implied price support 45
Table 16 Price determination 46
Table 17 Scenario 1 post clearing round bidding 46
Table 18 Scenario 2 post clearing round bidding 47
Table 19 Illustrative auction round-by-round 47
Table 20 Round 1 bidder values and aggregate cost 48
Table 21 Round 2 bidder values and aggregate cost 48
Table 22 Round 3 bidder values and aggregate cost 49
Table 23 Implied price support 50
Table 24 Price determination 50
Table 25 Scenario 1 post clearing round bidding 51
Table 26 Scenario 2 post clearing round bidding 51
Table 27 Scenario 3 post clearing round bidding 52
Box 1 Static Reverse Auction Format: Illustrative Example 10
Box 2 Price Discovery and Multi-Round Auctions: Illustrative Example 12
vii
Abbreviations and Acronyms
2G, 3G, 4G, 5G Second, Third, Fourth, Fifth Generation mobile telephony service
ARPU Average Revenue Per User
BoT Bank of Tanzania
BAFO Best and Final Offer
CDMA Code-Division Multiple Access
DCA Development Credit Agreement
DDP Digital Development Partnership
DTP Digital Tanzania Project
EAP Electronic Auction Platform
EASSy Eastern Africa Submarine Cable System
E-Commerce Electronic Commerce
FCC Federal Communications Commission (United States)
FCCT Fair Competition Commission Tanzania
GNI Gross National Income
GSMA Global System for Mobile Communications Association
HTT Helios Towers Tanzania
ICT Information and Communication Technology
ICTC ICT Commission
IDA International Development Association
IDI ICT Development Index
IP Internet Protocol
ITT Invitation to Tender
ITU International Telecommunication Union
KYC Know Your Customer (due diligence)
LCC Least Connected Countries
LGAs Local Government Authorities
MAB Most Advantageous Bidder
MAS Maximum Allowable Subsidy
MBB Mobile Broadband
Mbit/s Megabits per second
MDAs Ministries, Departments and Agencies
MFP Ministry of Finance and Planning
MIC Millicom International Cellular
MICIT Ministry of Information, Communication and Information Technology
viii
MNO Mobile Network Operator
MOCN Multi Operator Core Network
MORAN Multi Operator Radio Access Network
MRRA Multi-Round Reverse Auction
MPLS Multi Protocol Label Switching
MSP Minimum Scale Percentage
NICTBB National ICT Broadband Backbone
NPS National Payments Systems
PO-PSM President’s Office of Public Service Management
POPSMGG President’s Office, Public Service Management and Good Governance
PPA Public Procurement Act, 2011
PPRA Public Procurement Regulatory Authority
PS Permanent Secretary
RAN Radio Access Network
RBD Regional Broadband for Development
RCIP Regional Communications Infrastructure Programme
RFP Request for Proposal
ROI Return on Investment (a measure of profit over cost of investment)
SIM Subscriber Identity Module
SMRA Simultaneous Multi-Round Ascending
SRA Static Reverse Auction
TANePS Tanzania National e-Procurement System
TANESCO Tanzania Electricity Supply Company
TBC Tanzania Broadcasting Commission
TCRA Tanzania Communications Regulatory Authority
TMAS Total Maximum Allowable Subsidy
TTCL Tanzania Telecommunications Corporation Limited
TZS Tanzania Shilling
UAT User Acceptance Testing
UCSAF Universal Communications Service Access Fund
UNCITRAL United Nations Commission on International Trade Law
USD United States Dollar
VSAT Very Small Aperture Terminal
WB World Bank
ix
Executive summary
This report, commissioned by the World Bank using funds Feedback from meetings with key stakeholders, especially
from the Digital Development Partnership (DDP) looks feedback from mobile network operators (MNOs), has
at the application of a multiple round reverse auction been positive in respect of a pilot MRRA. The pilot will
(MRRA) hosted on an electronic auction platform (EAP) likely be used for additional awards of cell tower coverage
for allocating universal service funds in Tanzania. subsidies, but could equally be used for other universal
Although the specific case study is for mobile broadband service infrastructure goals, for instance providing last
subsidy awards in Tanzania, the learnings apply to other mile fiber connectivity for schools.
universal service infrastructure awards (such as fiber The report covers among, other things:
or satellite) and to other countries disbursing funds to • A summary and assessment of the application of a
promote infrastructure development. MRRA hosted on an EAP.
• A brief assessment of reverse auctions conducted by
The project was undertaken in close collaboration with the other Universal Service agencies around the world,
Universal Communication Services Access Fund (UCSAF) in particular focussing on best practice in the United
and the Ministry of Information, Communications and States which has the most experience in this field.
Information Technology (MICIT). The report has also • An assessment of the ‘readiness’ of the Tanzanian
benefitted from inputs from the private sector, notably the market to adopt such an approach
mobile network operators currently active in the Tanzania
market, and from peer review by World Bank specialists The next steps in the process should be for UCSAF, MICIT
and expert external consultants. and the WB to form a working party tasked with designing,
procuring and implementing the pilot auction. Extensive
The report is part of a programme designed to implement consultation should take place with other stakeholders,
sub-component 2.2 of the DTP on extending rural especially the MNOs, before the pilot takes place, most
broadband. That sub-component, with an allocation of likely starting in 2023/24. This report sets out how the
funds of around USD 29.5 million, aims to complete the pilot should be framed.
country’s universal service infrastructure, by providing
new cell tower infrastructure for mobile broadband The report recommends the application of a particular
in areas that are currently underserved or unserved MRRA format for allocating universal service subsidies.
(greenfield sites), and to bring more Tanzanians online by Chapter 2 reviews the telecommunications market and
upgrading narrowband (2G) cell sites to provide 4G or looks at operators, services and regulatory agencies.
higher mobile broadband service. Universal service policy in telecommunications in Tanzania
is reviewed in chapter 3.
The report recommends the deployment of a pilot MRRA
for disbursing funds extending rural broadband coverage. In chapter 4 auction principles and how in practice
This would supplement, and perhaps eventually replace auctions can be designed to allocate universal service
the current method of allocating subsidies which relies funds is examined. Chapter 5 looks at international best
on a static reverse auction (SRA), blind bid approach practice, as undertaken by the Federal Communications
administered by UCSAF. Commission in the US. In chapter 6 draws lessons for
Tanzania.
The report provides recommendations on how best to
conduct a pilot. It is recommended that a descending clock Chapter 7 presents recommendations on a pilot MRRA
reverse auction is piloted in accordance with international and this is followed in chapter 8 with next steps.
best practice. There are five appendices. Appendix A describes the main
licensed operators in the Tanzania market. Appendix B
describes the main regulatory agencies of relevance to
universal service in telecommunications.
x
Appendix C sets out in detail the rules of the auction format proposed. Appendix D is a numerical example of the
auction proposed and Appendix E is a discussion on electronic auction platforms. A bibliography is at the end.
Key Recommendations
A To pilot a MRRA to allocate universal service funds in 2023/24
B To host the pilot on an EAP
C Prepare and scope a detailed specification for an EAP in accordance with the auction rules
set out in this report
Table 1 Key Recommendations
Table 2 below provides a simplified overview of the recommendations in Chapter 7 of the report addressing
specifically the implementation of the MRRA.
Recommendations
1 Select up to 100 greenfield sites
2 Select sites from at least three regions and include Zanzibar
3 Apply and develop a financial model to calculate the Maximum Allowable Subsidy (MAS)
for each selected site
4 The budget in each region of the pilot should not exceed 90% of the Total MAS (TMAS) in
each region
5 Interested parties can be invited to submit bid applications in response to a call for
expressions of interest (EOI)
6 List of Qualified Bidders would selected and made public on UCSAF auction website
7 UCSAF shall use a secure EAP hosted on the internet (by purchase or rental)
8 UCSAF shall explore feasibility of using an EAP as an independent module accessible via
TANePS
9 UCSAF to undertake bidder information sessions, mock auctions, and engage in bidder
and stakeholder education
10 UCSAF to run auction on the EAP, with individual cell sites as separate lots
11 UCSAF to announce winners and publish auction results
12 UCSAF to conduct an impact evaluation of the pilot and draw lesson for future policy
Table 2 Recommendations for auction pilot
xi
Table 3 below shows the next steps discussed in Chapter 8.
Item Task Owner
1 Undertake to complete a MRRA using an UCSAF with support from EAP contractors
electronic auction platform during 2023
2 Scope out the feasibility of using an EAP as UCSAF with manager of TANePS
an independent module accessible on the
TANePS platform
3 Develop high-level financial model to inform UCSAF in consultation with MNOs
MAS values
4 Capacity building in UCSAF UCSAF and donor agencies
5 Stakeholder engagement UCSAF with other relevant agencies (e.g. TCRA)
6 Detailed workplan for implementing the UCSAF and WB
MRRA
7 Wider assessment of the feasibility of multi- WB
round procurement auctions in Tanzania and
in other countries
Table 3 Next steps
The report emphasizes that the application of a MRRA delivered on an interactive EAP has the potential to result
in substantial benefits for Tanzania.
Investment in the process would also build capacity that could be leveraged in other sectors of the economy.
xii
1 Introduction
The World Bank has retained Dr. Chris Doyle of CEPA as • Market analysis of the main interactive auction
a consultant for the purpose of undertaking a study of the platforms available on the commercial market, to
use of interactive auction platforms for universal service develop a list of potential suppliers
awards in Tanzania. The funding is provided by the Digital • An assessment of the level of interest from cellular
Development Partnership (DDP) Trust Fund and given operators, and other stakeholders (such as tower
in support of the implementation of the Digital Tanzania companies) to take part in the bidding process
Project (DTP – P160766).
The report also recommends the potential use of
The report is part of sub-component 2.2 of the DTP on interactive auction platforms in other WB lending
extending rural broadband. That sub-component, with programs which have a rural broadband component,
an allocation of funds of around US$29.5 million, aims to such as the proposed Digital Malawi Phase II (P160533),
complete the country’s universal service infrastructure, where the main focus is on fixed broadband extension in
by providing new cell tower infrastructure for mobile secondary cities and rural areas, and to schools.
broadband (MBB) in areas that are currently underserved
or unserved (greenfield sites), and to bring more
Tanzanians online by upgrading narrowband (2G) cell sites 1.1 Background
to provide 4G or higher mobile broadband service. Tanzania is on the coast of East Africa and in 2021 its
The project has involved close collaboration with the population was estimated at 63.6 million (see Figures
Universal Communication Services Access Fund (UCSAF), 1 and 2).4 Following two decades of sustained growth,
based in Dodoma, Tanzania, and with officials working Tanzania reached an important milestone in July 2020
with the Ministry of Information, Communications and when it formally graduated from low-income country
Information Technology (MICIT). to lower-middle-income country status. Tanzania’s
achievement reflects sustained macroeconomic stability
The study has been conducted under the supervision of that has supported growth, in addition to the country’s
the World Bank (WB) co-Task-Team Leaders for the DTP, rich natural endowments and strategic geographic
Tim Kelly and Sara Ballan. The report presents a list of position.5
policy recommendations with regard to the deployment of
a pilot multi-round auction. The WB estimates GDP per capita at USD1,099 and
The content of the report covers among other things: reported a real GDP growth rate of 4.3 percent and a GDP
• An overview and summary of procurement methods per capita growth rate of 1.3 percent in 2021, following
for disbursing funds to promote universal service a 1.0 percent per capita GDP contraction in 2020 (due
objectives mainly to the effects of the COVID-19 pandemic). The
• A summary and assessment of the use of interactive national poverty rate6 is estimated to have declined
auction platforms used by other Universal Service marginally in recent years and stood at 44.9 percent in
agencies around the world 2018, driven by the recovery of employment and nonfarm
• An assessment of the “readiness” of the Tanzanian business revenue. Growth is expected to strengthen over
market to adopt such an approach the next two years. The real GDP growth rate was forecast
at 4.6 percent in 2022.7
4 Source: World Bank Development Indicators, https://data.worldbank.org/country/TZ Nearly 42 million (or around two-thirds) of the population
are defined as living in rural areas, see Table 1, TCRA (2021) “2021 Country Data” at https://www.tcra.go.tz/uploads/text-editor/files/Country%20
ICTs%20Indicators%202021_1645555763.pdf
5 See https://www.worldbank.org/en/country/tanzania/overview
6 The World Bank measures its poverty headcount ratio as the percentage of the population living on less than $2.15 a day at 2017 purchasing power
adjusted prices.
7 See https://www.worldbank.org/en/country/tanzania/overview
1
1.2 Telecommunications: access and services
The telecommunications market in Tanzania is largely comprised of mobile communication services. The market is
competitive with six licensed mobile network operators (MNOs) and a number of smaller specialised service providers
offering a range of services. The MNOs provide 99.9% of active lines in operation and in March 2023 mobile penetration
of the population stood at 97.2%. By contrast there were only 84,517 fixed telephone lines in operation.8 The largest
mobile operator, Vodacom, has a 4G population coverage of more than 42% and a 3G population coverage of more than
63% (see Appendix A for details about licensed communications operators in Tanzania).9
MNO Registered SIMs Share Launch 5G
17,054,716 27.6% Yes
26%
8,146,036 13.2%
30% 15,171 0.0% Yes Number of telecom
27.6% Subscriptions and
16,599,038 26.8% Market Shares
13.6% 0%
2.4% 1,510,985 2.4% Yes
18,553,779 30.0%
61,879,725 100%
Table 4 MNOs, subscriber volumes and shares at March 202310
In September 2022, Vodacom switched on Tanzania’s first 5G network, a stand-alone 5G network largely covering Dar
es Salaam with a plan to reach out to more than 230 other locations in the country, including Arusha, Dodoma, Mwanza,
Iringa, Kagera, Zanzibar, and Mbeya.11 Tigo launched a 5G network in Dar Es Salaam, Dodoma, and Zanzibar in February
2023 through upgrading its 4G facilities and in collaboration with Ericsson is aiming to expand its service over the next
few years.12 In June 2023, Airtel announced it had launched 5G services in Tanzania.13
In respect of internet access penetration, which for most users is via MBB devices, according to the Tanzania
Communications Regulatory Authority (TCRA) this stood at 52% in March 2023.14 This represents a significant
improvement over recent years. The International Telecommunications Union (ITU) (2015) noted Tanzania was positioned
as 152 out of 166 on the global ICT Development Index (IDI), one of 42 of the Least Connected Countries (LCC) in the
world.15 In July 2022, ITU estimated internet use at 20% of the population.16
Although the population coverage statistics present a healthy picture in regard of access to communications, geographic
coverage varies significantly across the country and many areas have only a limited 2G or 3G mobile service. In Table 5 it
can be seen that access to MBB services providing 4G or higher service extends over 40% of the country and covers 58%
of the population.
8 Source: Communication Statistics, Quarter ending March 2023, TCRA (2023)
9 See Vodacom Tanzania Public Limited Company Annual Integrated Report for the year ended 31 March 2021 at https://vodacom.co.tz/annualRe-
sults
10 Source: TCRA (2023) op cit.
11 See Vodacom (2022) at https://www.vodafone.com/new/technology/vodacom-switches-on-tanzania-first-5g-network
12 Ericsson Press Release (2023) “Tigo Tanzania and Ericsson launch 5G and enhance existing network” 15 February.
13 Business Post (2023) “Airtel Rolls Out 5G Services in Nigeria, Tanzania, Zambia”, 21 June.
14 Source: TCRA (2023) op cit.
15 ITU (2015).
16 See ITU Digital Development Dashboard at https://www.itu.int/en/ITU-D/Statistics/Dashboards/Pages/Digital-Development.aspx
2
Description Mobile Standard March 2023
Percentage of the population covered 3G 74%
by a MBB network signal (3G, 4G or
higher) 4G 58%
Percentage of geographical coverage 3G 59%
by mobile network signal (3G, 4G or
higher) 4G 40%
Table 5 MNOs, subscriber volumes and shares17
The data in Table 5 indicate that competition over the last few years has resulted in Tanzania achieving a respectable
coverage in communication services. However, compared against world averages, Tanzania, like other least developed
countries (LDCs), has a substantial access gap. The ITU estimated that in 2022 for LDCs there was on average “an access
gap of 17 per cent of the population that cannot access the Internet”.18 This gap is due in large part to the low affordability
or relatively high cost of services, as well as due to the relatively high cost of network rollout.
Table 6 below shows Tanzania in comparison to other LDCs, the World and neighbouring countries.
Coverage: % Affordability: MBB
Access: % individuals
Region/Category/Country population covered by basket as % of GNI per
using the Internet
4G network capita
World (2022) 66.3 88 1.5
LDCs (2022) 36.1 49 5.8
Angola 32.6 33 2.7
Democratic Republic of Congo 22.9 40 10.3
Kenya 65.0 97 3.1
Malawi 24.4 69 9.4
Mozambique 17.4 50 9.4
Tanzania 50.6 52 4.6
Uganda 10.3 31 5.4
Zambia 21.2 91 2.4
Table 6 International comparison of MBB coverage and affordability in 2021 (unless otherwise stated) 19
1.3 The Digital Tanzania Project (DTP)
Like most other countries, Tanzania is seeking to promote wider access to broadband services and the internet, especially
in rural areas. The DTP has set an objective to increase access to high quality broadband internet services for government,
businesses and citizens, and to improve the government’s capacity to deliver digital public services. The DTP20 is geared at
accelerating digital economy towards industrialization by supporting implementation of the Government’s National Five-
Year Development Plan. This is intended to enhance Government efforts towards achieving a knowledge-based society
for socio-economic development.
17 TCRA (2023) op cit.
18 ITU (2023), page 6.
19 Sources: ITU (2023) op cit., Table A2.3; A4AI “Data-Only Mobile-Broadband Basket” for 2021; allAfrica (2021) “Tanzania: Vodacom Network Cover-
age Provides Vital Rural Connectivity” 24 September 2021.
20 See The United Republic of Tanzania Ministry of Communication and Information Technology (MCIT) Digital Tanzania Project (DTP) Stakeholders
3
Figure 1 Map of Tanzania21
Figure 2 Tanzania population density map22
Engagement Plan (SEP) March 2021 at https://www.mawasiliano.go.tz/uploads/documents/sw-1616059071-Digital_Tanzania_Program_%20
SEP_March_2021.pdf
21 Source: World Bank at https://maps.worldbank.org/
22 Ibid.
4
The DTP programme has the following three components: extend data-enabled (3G or higher) network coverage
to three million people currently living in areas of
1 Digital Ecosystem – (i) Digital Enabling Environment the country that are currently unserved by a mobile
(Establishment of National ICT Professional and cellular signal and will upgrade existing 2G networks
Innovation Centre, Establishment of ICT equipment to 3G and above.
refurbishment, Scanning of ICT Regulatory
Environment) (ii) Infrastructure to Support National 5. The RBD will be conducted in collaboration with
Development and E-Commerce (Enhancing UCSAF. The mechanism proposed to be used in this
the National Addressing and postcode system, intervention is a “reverse auction subsidy”23, similar
Development of the National Statistical Information to the one used under RCIP-3, and as used in multiple
Management System, National e-Commerce WB programs, such as Niger Smart Villages (P167543)
initiatives) and Digital Malawi (P160533).24
2. Digital Connectivity – (i) Connected Government 26. This report is a study on reverse subsidy auctions
(connecting all unconnected Ministries, Departments intended to inform how best to achieve the goals of
and Agencies (MDAs) and Local Government RBD and Digital Connectivity.
Authorities (LGAs) to high-speed broadband through
extension of the Government Communication
Network and increasing capacity of the Government 1.4 Rural broadband policy and DTP
Bandwidth) (ii) Rural Broadband for Development According to the DTP Implementation Status & Results
(expanding rural mobile coverage and upgrading Report (2021), the proportion of the population covered
service from 2G to enable broadband connectivity and by a 3G or higher speed mobile network was 68% at
utilization of spectrum white space) the end of 2021 (citing data from the Global System for
Mobile Communications Association (GSMA)). The target
3. Digital Platforms and Services – (i) Digital Services under the DTP is to see that increase to 75% or higher
and Productivity platforms (OneStop Service by end December 2026.25 Most of this increase will need
Centres - Huduma Pamoja Centres) in every investment in rural areas of Tanzania.
region (offering access to public online services
and in person transactions, transformation of the In addition to widening and extending MBB coverage,
traditional economy into digital economy) (ii) Data the DTP also sets out targets in a number of other digital
Centre Infrastructure (Enhancement of data centre ecosystems, including improvements in the government’s
infrastructure) (iii) Digital Literacy and Capacity capacity to deliver digital public services.
Building (Government ICT Cadre Training Programme
& Citizen Digital Literacy and Awareness Program) To achieve the targets set out, the DTP envisages UCSAF
will play a key role. The Fund is working with the MICIT to
4. This report focuses on one aspect of relevance to implement sub-component 2.2 of the DTP on extending
component 2: Digital Connectivity. In particular, this rural broadband. Subsidy programmes are targeted at
report looks at a specific issue of relevance to sub- site locations for the installation of new or upgraded
component 2.2 ‘Rural Broadband for Development’ telecommunications infrastructure. The site locations are
(RBD) which builds on a previous program, the third defined according to administrative divisions: ‘wards’.26
phase of the Regional Communications Infrastructure Under DTP, some 763 wards have been identified to
Programme (RCIP-3). The RBD activity plans to receive cell coverage for the first time, while a further
23 An auction is viewed by economists as a mechanism (a set of rules) that elicits information from bidders to enable the auctioneer to achieve an
objective. A simple objective might be identifying the person who values the object highest and therefore should be awarded the object. This person
would be prepared to bid the highest amount for the object and a simple ascending bid auction would identify them and hence achieve the objective.
When the auctioneer is a government agency, the objective is usually to ensure the best allocation (set of winners) for society. If an auctioneer knew
all the relevant information about bidders, an auction would be redundant as the auctioneer’s objective could be achieved by directly allocating the
objects. A reverse auction is an auction where multiple sellers compete to sell (provide) a good or service to a single buyer. A common example is
B2B procurement auctions. By contrast, a forward auction is where multiple buyers compete to buy a good or service from a single seller. Common
examples of forward auctions are art auctions, treasury auctions, real estate auctions, etc. A Double Auction involves multiple buyers and multiple
sellers, e.g. a stock exchange, certain electricity markets, etc. A reverse auction subsidy works by soliciting bids from applicants (sellers or providers
of service) for support payments (subsidies) and where the buyer (in this case government) is seeking to disburse limited funds at the lowest possible
cost (to the most advantageous bidder). See Klemperer (2004) and Ausubel (2008).
24 Reverse auctions are one of many universal service instruments available. Other instruments include licence obligations, end-user subsidies, desig-
nation of universal service providers, access deficit charges, etc. See Muente-Kunigami and Navas-Sabater (2010).
25 See table on page 2 in the Implementation Status & Results Report, World Bank, DTP at https://documents1.worldbank.org/curated/
en/099702004262273872/pdf/P160766004a6f90080ac4f070bafef9eaef.pdf
26 The administrative divisions of Tanzania are controlled by Part I, Article 2.2 of the Constitution of Tanzania. Tanzania is divided into thirty-one re-
gions (mkoa). Each region is subdivided into districts (wilaya). The districts are sub-divided into divisions (tarafa) and further into local wards (kata).
Wards are further subdivided for management purposes: for urban wards into streets (mtaa) and for rural wards into villages (kijiji). The villages may
be further subdivided into hamlets (kitongoji).
5
488 cell sites have been earmarked for upgrade from 2G to 3G or higher. Other innovations include opening up bidding
to companies other than cellular communications operators (e.g. cell tower companies, VSAT service providers, etc.) and
the use of national roaming. Cellular operators who have not previously worked with UCSAF, such as Halotel, will also be
invited to bid.
The DTP uses a combination of development loans and funds accumulated from licensed operators to provide subsidies
aimed at extending rural mobile and rural broadband coverage. The subsidies awarded combine with private capital to
support investments. For example, under RCIP-3, USD30m of IDA subsidy leveraged around USD70m in investment from
the private sector (i.e. a ratio of slightly better than 1:2).
In the following section we describe at a high-level procurement methods for disbursing funds to promote universal
service objectives. This is followed by a section looking at the procurement SRA format27 that has been used to date in
Tanzania to disburse subsidies.
1.5 Procurement methods for disbursing funds to promote universal service and
universal access objectives
Universal service and access objectives are set by policy makers and in the telecommunications sector aim to make
services accessible to the widest number of people (and communities) at affordable prices.28 The purpose of such policies
is to promote social benefits29, greater economic efficiency and equity by reducing gaps30 (see Figure 3 below) in access
and service across households and areas.
100% of Communities
High cost areas
Universal
Access Access
Gap
Sustainability Frontier (4s)
Sustainability Frontier (4d)
100% of Population
Market Efficiency Frontier (3s)
Supply
Existing level of Access (2d)
Existing level of Access (2s) Market
Market
Gap Gap
Low cost areas
Current Penetration (1) Affordability Efficiency Frontier (3d) Universal
Access
High income households Demand Low Income Household
Figure 3 The Gaps Model31
27 An auction format defines the rules for the announcement of prices, the placement of bids, the updating of prices, the auction close, and the way
winners are selected. We set out and describe the rules of a well-designed MRRA format in chapter 4
28 This definition is set out in an influential ITU document, ITU (2013). For an early overview on universal service in telecommunications see also
Dymond and Rogati (1997). On funding gaps and universal service, see Navas-Sabater, Dymond and Juntunen (2002) and Emiliani (2008). For a good
overview of universal service fund policies around the world up to around 2010, see GSMA (2013).
29 An excellent example of social good is presented in World Bank (2023) where evidence shown demonstrates that internet availability increases jobs
and reduces poverty in African countries.
30 Perceiving universal service and universal access as filling ‘gaps’ was set out conceptually in Navas-Sabater et al. (2002) op cit. and Stern and
Townsend (2006). There are two gaps: a market gap and an access gap, both affecting the supply and demand sides of the market, see Figure 3.
31 Reproduced from Stern and Townsend (2006) p. 21, Summary Report.
6
Achieving universal service and access can be done in a Procuring universal service competitively usually involves
number of different ways.32 For example, on the supply both administrative and market elements. Administrators
side policy it can be affected by direct regulation through scope the services and infrastructure needed and also
the setting of specific licence obligations in regard of determine the criteria for selecting suppliers. Market
coverage and pricing. In this instance the cost of universal elements involve respondents supplying cost information
access and service is borne directly by consumers. In low through a competitive process.37
income countries this option is unlikely to yield success, as
there will not be enough consumers with sufficient income The method of procuring providers to supply universal
willing to pay the relatively high prices required to fund service obligations using a reverse auction within a
service in poorer areas. competitive context falls broadly into one of three broad
categories:
Alternatively, or in addition, universal service funds
managed by government or regulatory agencies can be 1 . Static Reverse Auction (SRA): Administrators
used to subsidise infrastructure investments in areas or scope services required, a request for proposals
market segments where gaps have been identified. The (RFP) is issued (either to a limited group or publicly),
source of funds is usually from within the industry33, respondents submit responses to the RFP which are
general taxation34 and credit facilities from multilateral evaluated according to a scoring method set out in
development banks35. the RFP. The scoring criteria should be objective and
include a mix of metrics regarding investment and
UCSAF in Tanzania is an example of a universal service cost. In a World Bank context, most advantageous
fund. In some countries universal service funds are bidder(s) (MAB) are those respondents submitting
channelled by government or regulatory agencies to the highest ranked bid(s). In most cases, in a SRA the
operators with a special designation.36 This approach selected suppliers receive funds as bid (first-price
tends to be appropriate in markets where there is at auction). A variant of the SRA includes the possibility
least one operator that has considerable geographic and for post-bid negotiation with MABs.38
population coverage.
Instead of designating particular operators as having 2. Multi-Stage Best and Final Offer (BAFO): As with a
responsibilities for meeting universal access and service SRA, administrators scope services required, a request
objectives, depending on market circumstances, the for proposals (RFP) is issued (either to a limited group
disbursement of universal service funds may be better or publicly), respondents submit responses to the RFP
undertaken via competitive procurement. This approach which are evaluated according to a scoring method
tends to work in environments where sufficient operators set out in the RFP. Those submitting Substantially
are relatively symmetrically positioned in the market, Responsive Bids39 are invited to respond in a second
where there are substantial gaps in access and service stage with a BAFO.
and where there is healthy competition among suppliers.
All these conditions hold in Tanzania, as we shall discuss
further in this report.
32 Stern and Townsend (2006) op cit. provides an early excellent overview of approaches towards funding universal service. GSMA (2006) gives an
early assessment of how mobile communications can help universal access.
33 For example, in the United States universal service is paid for by contributions from providers of telecommunications based of an assessment on
their end-user revenues, see https://www.fcc.gov/general/universal-service.
34 For example, in the United Kingdom the government has provided funding to stimulate the rollout of advanced services and fibre networks, see
DCMS (2018) Future Telecoms Infrastructure Review.
35 In 2021 the World Bank (IBRD and IDA) lent in the region of $750 million for information, communications and technology projects in Eastern and
Southern Africa, see slide 20 in Annual Report 2021 World Bank Lending (Fiscal 2021).
36 For example, in the European Union a national regulatory authority may designate after market analysis certain operators as universal service
providers. Any financial transfers to such designated operators are required to be undertaken on an objective basis, and should be transparent,
non-discriminatory and proportionate. This is to ensure that universal service cost transfers result in the least distortion to competition and to user
demand. See Directive (EU) 2018/1972 of the European Parliament and of the Council of 11 December 2018 Establishing the European Electronic
Communications Code, OJ L 321/36.
37 An excellent early account of the application of market elements in procuring universal service is Milgrom (1996)
38 Most universal access auctions that have occurred have typically been a version of the SRA. For example, see Sepulveda (2004) pages 85-94 for
an early discussion. Sepulveda describes how until the early 2000s, universal service auctions featured a single round, with Chile being among the
first countries to use a competitive SRA process in 1995. Wu (2014) is a useful updated summary of approaches and she also highlights how India
pioneered the application of MRRAs for universal service funds in the early 2000s. On the Indian experience see also Jain and Raghuram (2010).
39 According to World Bank procurement methods, a Substantially Responsive Bid is assessed to be complete and without major deviation from the
eligibility, technical and commercial requirements in the RFP procurement documents, see World Bank (2018) Procurement Guidance. On procure-
ment policy and the World Bank, see Procurement Framework and Regulations for Projects After July 1, 2016.
7
3 . Multi-Round Reverse Auction (MRRA): As with (1) and (2) above, administrators issue a RFP and determine potential
suppliers in a pre-qualification process. Respondents who meet pre-qualification criteria are invited to participate
in a multi-round bid process (auction) that ends with successful bidders offering the best terms. The MABs receive
funding determined by their bids and may receive amounts determined by marginal losing bidder(s) (equivalent to a
second-price auction).
Each of the above can be deployed to achieve the goals of universal service and universal access. In Table 7 below we set
out the respective suitability and pros and cons of each category.
Multi-stage: Best and Final
Static Reverse Auction Multi-Round Reverse Auction
Offer (BAFO)
Suitability One lot or a few lots, easy Complex lot or lots, common Multiple/many lots, lots
to define lot(s), value value relevant, potentially definable by a few dimensions
largely private40 with several suppliers, BAFO – e.g., price and quality,
known common value41, allows auctioneer to adapt common value relevant,
limited number of suppliers, procurement requirements several or many suppliers
negotiation allows for around information revealed
welfare gains (win-win) during bid phase
Pros Process fairly Allows some price and value Process facilitates price and
straightforward, largely discovery value discovery, second-price
administrative, works well in formats incentivize bidders
less competitive settings to bid truthfully, better value
for money
Cons No price or value discovery42, Complex, requires probity Lengthier process, relatively
bidders may overstate costs, assurance provider for complex, onerous for smaller
less good value for money negotiations, bidders may bidders, needs more bidder
overstate costs education
Table 7 Procuring providers within a competitive context
1.6 Current UCSAF and MICIT policy: SRA format
UCSAF43 and MICIT currently use a SRA format for allocating subsidies under the DTP. The format involves the issuing
of an Invitation to Tender (ITT).44 The ITT is advertised publicly and sets out technical and financial criteria to guide the
award of subsidies to help promote further communications infrastructure investment in underserved areas.
The process involves the announcement of a list of wards eligible for financial support. Bidders, usually MNOs, are
required to comply with technical specifications in order to be eligible to bid for financial support. UCSAF and MICIT
present a maximum allowable subsidy (MAS) available for each ward. In a recent tender process to upgrade 2G networks
to 2G/3G and/or 4G networks, UCSAF announced a list of 488 wards where each was allocated the same MAS of TZS 20
million.45
40 An auction where bidders have private values for the object(s) is one where each bidder estimates the value of the object(s) and this is independent
of other bidders’ valuations. For example, an MNO estimates the return to build and operate a cell tower in a particular location and its estimate
would not change if it knew about other MNOs estimates.
41 A common value auction is one in which the object(s) has the same true value for each bidder. In practice bidders form estimates of this often hidden
value based on research and information. For example, if an MNO estimates a return for a cell tower at X, it would likely change this estimate if it
knew about the estimates of other MNOs. Other examples of common value objects often seen in an auction context are financial assets, oilfields
and wind farm locations
42 Price and value discovery arise in dynamic auction settings for common value objects as participants learn about the hidden values from observ-
ing the bids placed by other bidders. By discovering more about the hidden values, bidders can make better informed decisions that lead to more
efficient outcomes. For example, a bidder might learn from other bidders’ submissions that the return associated with serving a particular ward with
an upgraded telecoms infrastructure is higher than initially anticipated. For more information on price discovery and auction formats, see Milgrom
(2017).
43 Further information about UCSAF is in chapter 3.
44 In this report we use ITT and RFP interchangeably.
45 The MAS in USD was in the region of $8,350 for upgraded sites.
8
Those eligible to bid are invited to submit one set of bids for funding.46 Each bidder can express no more than one bid
value for each ward and is not permitted to request a subsidy exceeding the MAS. Box 1 provides an explanation of the
mechanics of the SRA format used in Tanzania and that was applied in the recent tender for cell sites in underserved
areas.47
UCSAF has completed thirteen funding rounds using a SRA and the ‘Bid Success Rate’, defined by UCSAF as the number of
wards receiving bids divided by total the number of wards in the award, is shown in Figure 4 for the first eleven rounds.48
Out of these eleven funding rounds shown in Figure 34, no more than 90% of wards received a subsidy in any of the SRAs
and on average 61% of wards received subsidy funding. Over the last four funding rounds shown in Figure 4, the bid
success rate has been around 50%.
Bid Succes Rate (%)
90%
83% 81%
69% 66%
65%
49%
46% 44% 45%
34%
PHASE1 PHASE 1APHASE 1BPHASE 2A PHASE 2B BSZPH1 BSZPH2 PHASE 3 PHASE 4 PHASE 5 BSZPH6
Figure 4 The bid success rate of UCSAF reverse auctions49
Factors influencing the bid success rate include: the expected returns of bidders, the format of the auction, and the MAS
for each ward (the latter form part of the auction format). In some instances the MAS may not be sufficient to allow for a
reasonable commercial return. In the last award shown in Figure 1.4, 55% of wards did not receive a successful bid. This
implies more than half of all wards did not receive any funding and investment in those areas was less likely to occur.
It was claimed in the previous section that other procurement formats, such as BAFO or MRRA, could perform better than
the SRA because of price discovery. As the telecommunications market is competitive in Tanzania and with at least three
or four operators in a strong position to bid for subsidies, the MRRA format would seem most appropriate and could lead
to superior outcomes. In an MRRA in which bidders may express bids over multiple rounds, price discovery is more likely
to lead to a higher bid success rate and better value for money.
46 The MAS is equivalent to the role played by a reserve price in an auction.
47 Source: Evaluation Report and Recommendation for Award of Subsidy for New Cell Sites in Underserved Areas DTP – Phase 1, Tender number:
ME.006/WB-DTP/P.160766/306158/2022-2023/HQ/NC/13, February 2023.
48 The bid success rate used by UCSAF differs from the measure usually applied in analyses of reverse auctions for subsidies. Typically, the preferred
measure of success looks at the subsidy awarded as a proportion of the total maximum allowable subsidy (TMAS). In early auctions in Chile, for
funds to support the rollout of public payphones, subsidy requests were as low as 25% of the TMAS, see Sepulveda (2003). In the recent Tanzania
brownfield subsidy award tender in November 2022, (Tender ME.006/WB-DTP/P.160766/285652/2022-2023/HQ/NC/14) there were 488 lots
available with a MAS of TZS 20m per ward. The UCSAF measure of bid success rate was 62.3%, whereas the subsidy request was 84.7% of the TMAS
for the 304 awarded sites. In the greenfield tender, described in Box 1, the subsidy request was around 87% of TMAS for the 437 awarded sites (a
UCSAF bid success rate of 57% out of the 764 in the award).
49 Source: UCSAF presentation, World Bank Mission Workshop, Dodoma, 1-5 August 2022.
9
Box 1 SRA Format used by UCSAF
UCSAF’s reverse auction in 2022/23 for greenfield sites “Award of Subsidy for New Cell Sites in Underserved Areas
DTP Phase 1” typifies how UCSAF has operated an SRA for awarding subsidies. In this tender there were 764 sites
(wards) available for funding. Each ward had a MAS of TZS 145,000,000 (about USD $60,190) or TZS 150,000,000
(about US $ 62,266). In Table 8 below we show the data for the Kigoma region which had 9 wards.
Five bidders qualified to bid, though only four bidders in the auction submitted bids at or below the MAS. The results
for the Kigoma region are shown below, the figures in bold being winning amounts.
Ward Bidder 1 Bidder 2 Bidder 3 Bidder 4
1 145,979
2
3
4
5
6 133,400
7
8 142,000
9 131,950 138,000
Table 8 SRA bids submitted in ‘000s TZS
Only one ward was competitive in Kigoma (ward 9), with the allocated subsidy of 131,950,000 representing 91% of
the MAS. In ward 1, the allocated subsidy was 97% of the MAS, in ward 6 the allocated subsidy was 92% of the MAS
and in ward 8 the allocated subsidy was 98% of the MAS. UCSAF’s measure of bid success for the Kigoma region was
44%.
The process involves bidders submitting one subsidy bid request and the bidder submitting the lowest bid in a ward
is selected as the winner. Bidders do not have an opportunity to submit new revised bids, though in some instances
there is scope for post-bid negotiation with MICIT that oversees the tender process.
10
In the SRA used by UCSAF there is no opportunity for 1.7 MRRA and price discovery
price discovery and this limits the effectiveness of the The Bank team has advised the Government to pilot
format. The format is also prone to bidders overstating the use of an EAP that will allow for multiple rounds of
costs, due to common value uncertainty and what is bidding, conducted in an interactive manner, to achieve
known as the winner’s curse.50 a better allocation of sites to bidders, at the lowest price,
and with an improved geographic split. The deployment of
In the BAFO there is a limited possibility for price an EAP is considered further in Chapter 7 and Appendix E.
and value discovery, as bidders would only learn from The attraction of conducting a MRRA format is price
information implicit in one round of bids. As we discuss discovery – which enables bidders to improve their
further in chapter 4 below, there are strong reasons to estimates of required subsidies.
suspect revealing bid values over multiple rounds in a
MRRA format would allow bidders to discover more and As in any multi-round auction, bidding continues until
this is beneficial. These benefits arise from price discovery demand is brought into line with supply. In the case of a
and modern EAPs enable these benefits to materialise MRRA for subsidy, this occurs when demand for financial
relatively quickly. support does not exceed the total subsidy made available.
Box 2 below speculates what a MRRA format might look
It is well understood by auction theorists that multi-round like based on the Kigoma greenfield auction presented
auction formats promote price discovery and have the Box 1.
potential to be more effective at achieving policy goals.51
For example, in the case of UCSAF, the MRRA format Later in this report we provide a detailed example of
could lead to higher bid success rates and, for any given a MRRA format that could be deployed by MICIT and
bid success rate, it is likely less funds would be disbursed UCSAF.52 As seen speculatively in Box 2, multiple rounds
and there would be higher value for money as price by facilitating price discovery can yield improvements
discovery promotes competition. Benefits of this kind are in the bid success rate and value for money. At worst a
shown in chapter 5 for the case study of a reverse auction MRRA format is always as good as a SRA.
for broadband subsidies in the United States.
50 In a SRA the winner’s curse has the effect of bidders submitting subsidy requests above their estimates – because the thought process of a bidder
supposes that winning implies the other bidders have higher estimates. Consequently bidders manage this risk by submitting a higher demand for
subsidy support, or no bid at all. The winner’s curse occurs in auctions for objects that have common value. In a SRA for a common value object, a
bidder knows that winning the object most means that insufficient subsidy has been requested. This is because the best estimate of the required
subsidy is the average of all the bidders’ estimates. A winning bid is based on only the bidder’s estimate in a SRA and will therefore be expected to be
above the best estimate – which is an average of all the bidders’ estimates. Knowing ahead of the auction that winning means asking too little, (the
winner’s curse), leads every bidder to increase their requests to avoid asking for too little. This has the effect of causing bidders to choose requests
that do not reflect their estimates of the value of the object, compromising price discovery and likely resulting in inefficiency. It may also lead to poor
value for money. The effect of the winner’s curse can be mitigated by having multiple rounds in an auction. The format enables bidders to update
estimates of common value based on information disclosed about other bidders’ estimates during the bidding rounds. An example of this format is
the open-outcry English auction deployed by auction houses. This format features a reserve price and an ascending price whenever demand exceeds
available supply. For more on the winner’s curse, see Kagel and Levin (2002).
51 The benefits of multi-round auctions have been demonstrated in many spectrum auctions held across the world over the last three decades. Rather
than bidding blind, multi-round auctions provide bidders opportunities to react to information discovered during the auction. This view is echoed in
GSMA (2021).
52 Box 2 is a simplified version of the format we recommend for the disbursement of public funds for universal service in Tanzania. The format we
recommend is explained more fully in Chapter 7 and Appendices C and D.
11
Box 2 Price Discovery and the MRRA format: Speculative illustration
In Box 1 bidders submitted one set of bids in the SRA. Here we speculate on what might have happened had UCSAF
and MICIT held a MRRA. We suggest two further rounds of bidding might have occurred. The speculative figures are
shown below in Tables 9 and 10 (we assume the MAS in each ward is 150,000).
Ward Bidder 1 Bidder 2 Bidder 3 Bidder 4
1 145,979
2
3
4
5
6 133,400 130,000
7
8 142,000
9 131,950
Table 9 Round 2 bids in MRRA submitted in ‘000s TZS
In round 2, after round 1 requests are revealed, price discovery and bid flexibility leads Bidder 4 to shift from Ward
9 to Ward 6.
Ward Bidder 1 Bidder 2 Bidder 3 Bidder 4
1 145,979
2
3
4 133,400
5
6 130,000
7
8 142,000
9 131,950
Table 10 Final round 3 bids in MRRA submitted in ‘000s TZS
In round 3, the final round, Bidder 1 shifts from Ward 6 to a new bid on Ward 4. The other bids are not altered. The
auction process ends.
In this speculative world, the UCSAF measure of bid success in the region is higher at 55.5% (five ward covered
versus four in the SRA), and the subsidy requested as a proportion of the TMAS is 91% - which is below the SRA case.
12
1.8 Report structure Appendix C sets out in detail the rules of the auction
The remainder of this report explores the possibility of format we propose. This includes charts outlining the
applying MRRAs for allocating universal service funds main steps in the recommended process. This is followed
and recommends the application of a particular format. In in Appendix D with a numerical example and Appendix E
chapter 2 we start by reviewing the telecommunications with a discussion on electronic auction platforms.
market of Tanzania, and looking at operators, services
and the regulatory agencies. Universal service policy in Appendix F reproduces the slide deck presented at the
telecommunications in Tanzania is reviewed in chapter 3. Mission held in Dodoma on 3 August 2022, attended by
This assessment leads us to conclude that a MRRA should key representatives from UCSAF (including Ms Justina T.
be considered in Tanzania. Mashiba, CEO) and the Ministry of Finance and Planning
(MFP) (including Mr Eliakim C. Maswi, CEO). A version of
In chapter 4 we briefly discuss auction principles and how the slide content shown in Appendix F was also discussed
auctions can be designed in practice to allocate universal with representatives from the MNOs AirTel, Halotel,
service funds. Chapter 5 looks at international best Tigo and Vodacom on 1 and 2 August in Dar es Salaam.
practice, as undertaken by the Federal Communications Feedback from the stakeholder engagement was positive.
Commission in the US. In chapter 6 we draw lessons for A bibliography is presented at the end.
Tanzania.
In chapter 7 we set out recommended steps towards a
pilot MRRA and follow this in chapter 8 with suggested
next steps.
Five appendices accompany the main report.
Appendix A describes the main licensed operators in
the Tanzania market. Appendix B describes the main
regulatory agencies of relevance to universal service in
telecommunications.
13
The Tanzania
2 telecommunications market
The telecommunications market in Tanzania 2.1 Fixed and Mobile network operators
is overseen by policy set by Government, with The majority of communication services in
current policies defined in the 2016 “National ICT the country are supplied through mobile
Policy”.53 The most significant policy occurred back telecommunications providers in a highly competitive
in 2005 when Tanzania extensively liberalized the market, as shown in summary in Table 4 in Chapter
telecommunications sector and established a new 1 above. We have already discussed in the
regulator the TCRA to oversee licensing and market introduction above the current status of the mobile
enforcement. The liberalization policy was aimed at telecommunications market, see Chapter 1, Section
accelerating the take-up of communications services, 2.
which were at very low levels by international
standards.54 The incumbent operators in Tanzania are Tanzania
Telecommunications Corporation Limited (TTCL),
In this chapter we set out the telecommunications which until liberalization in 2005, operated as a
market landscape in Tanzania today. This provides monopoly on the mainland, and Zantel, as a monopoly
a helpful context for the subsequent discussion on provider in Zanzibar. These companies have
universal service and the policy instruments used to experienced much restructuring since and have, in
promote improved universal service. the case of Zantel, resulted in privatization.55 These
two operators provide much of the high capacity
We start by looking at fixed and mobile network fixed network infrastructure, as well as offer fixed
operators and follow this with a look at mobile line services to end-users and MNOs.
money services, which have been an important driver
stimulating the growth of both telecommunications
and financial services. The emergence more recently
of cellular phone tower companies is also discussed.
The chapter ends with a brief overview of the
regulatory landscape.
53 For example see the National ICT Policy 2016 which updated an earlier policy framework set out in 2003 at https://www.ega.go.tz/uploads/publica-
tions/sw-1574848612-SERA%202016.pdf
54 Early initiatives promoting greater coverage involving the African Development Bank started in the 1990s when tele-density in Tanzania was well
below 1 per cent. See “Tanzania - Telecommunications II Project 1992-2000” at https://projectsportal.afdb.org/dataportal/VProject/show/P-TZ-
GB0-002
55 TTCL was privatised in 2001 but by 2016 was again wholly owned by the Government of Tanzania. Zantel is 85% owned by the private company
Millicom with the other 15% held by the Government of Zanzibar.
14
TTCL was given management of the government’s The BoT is required to ensure that payment
national fibre backbone, the National Information systems are stable and secure. This duty led to a
and Communication Technology Broadband key regulatory development providing consumer
Backbone (NICTBB), which services the whole protection in Tanzania in the shape of various
country; and under the current arrangements private regulations that ensure due diligence on customers
operators must contract for service with TTCL on all (Know Your Customer (KYC)) and in conjunction with
the routes it serves.56 the TCRA a requirement for the registration of all
Appendix A provides more details about the SIM cards.58
operators in Tanzania.
In early 2008, the fintech company E-Fulusi launched
the first mobile wallet in Tanzania, MobiPawa, shortly
2.2 Mobile money57 followed by the launch of M-Pesa by Vodacom
The absence of an extensive retail banking in April 2008. Over the following two years, the
system in the country has both promoted mobile other major MNOs launched their own mobile
money services and caused such services to be financial services, Z-Pesa (Zantel), AirTel Money
a key driver in promoting the uptake of mobile (AirTel) and Tigo Pesa (Tigo). Mobile money services
telecommunication services in Tanzania (and more grew substantially between 2012 and 2018, with
generally across the region). The innovation driving subscribers almost tripling from about 8 million
mobile money started in Tanzania in 2007 (shortly active subscribers to 23 million active subscribers.
after its appearance in nearby Kenya) when MNOs The figures published by the regulator TCRA in Table
asked the authorities to allow users to convert 11 below shows data on mobile money subscriptions
purchased airtime into cash and transfer this to other by at the end of March 2023.59
users across the mobile network.
In 2006, the Bank of Tanzania (BoT) amended the
Bank of Tanzania Act to give themselves, as the
Central Bank, powers to oversee and regulate non-
bank entities offering payment services. In 2007,
following the successful launch of M-Pesa in Kenya
and extensive discussion with fintech innovators and
MNOs, the BoT adopted a “test and learn” approach
to test, and monitor, the deployment of digital
financial services. MNOs were required to partner
with banks under a BoT ‘letter of no objection’ to
ensure the protection of consumer funds.
The government of Tanzania is extending the NICTBB with a view to achieving its ICT vision. NICTBB is being implemented in five (5) phases. Phases
I and II and sub-phase I of phase III have been completed. Phases I and II involved the construction of 7,560 km of NICTBB to cover all regional head-
quarters, cross-border connectivity to the neighbouring countries of Kenya, Uganda, Rwanda, Burundi, Zambia and Malawi, and international con-
nectivity to the submarine cables of the Eastern Africa Submarine Cable System (EASSy) and Seacom. Phase III sub-phase I involved the construction
of an Internet Protocol (IP)/Multiprotocol Label Switching (MPLS) layer on top of the core NICTBB, a connected optical fibre cable between Main-
land Tanzania and Zanzibar, and the first national data centre in Dar es Salaam, which is operational as a tier 3 data centre. The implementation of
Phase III subphase II will involve the expansion of NICTBB networks to all district headquarters as well as the construction of national data centres
in Dodoma and Zanzibar. The Government is engaging the private sector under a public-private partnership arrangement to construct ICT infra-
structure that includes metro fibre networks in urban areas, and the construction of missing links through a consortium of operators. Through this
initiative, a total of 2,595 km (missing links: 2,020km and metro: 575km) have been deployed as phase IV of the NICTBB. Source: page 26 in “Report
on the Implementation of the Istanbul Programme of action for LDCs for the decade 2011-2020” March 2021 available at https://www.un.org/ldc5/
sites/www.un.org.ldc5/files/tanzania.pdf The NICTBB is managed and operated by TTCL on behalf of the government, through the MICIT. Recently
the government has promoted further fibre investment by encouraging TTCL to partner with the electricity company TANESCO, see https://itweb.
africa/content/5yONP7EgONeqXWrb 13 September 2021. TTCL has also recently applied to the International Financial Corporation (IFC) of the
World Bank Group (WBG) to seek additional financing.
57 This section has benefited from the content in Ephraim and Mhina (2021). See also di Castri and Gidvani (2014).
58 To register a SIM card in Tanzania it is a requirement to provide details of your ID, a fingerprint and have a passport style photo taken.
59 See TCRA (2023) Table 2.1, op cit.
15
MNO Mobile Money Service Subscribers % Market Share
Provider
Airtel 9,507,726 21%
Halotel 3,821,098 9%
Tigo 13,608,203 31%
TTCL 1,221,065 3%
Vodacom 16,195,476 36%
TOTAL 44,353,568 100%
Table 11 Mobile money subscriptions at March 2023
2.3 Tower companies
Tanzania’s MNOs have, as in the case of many countries, divested management and ownership of their telecom
towers to independent tower companies. In 2013, Vodacom sold 100 per cent of its 1,149 towers to Helios
Towers Tanzania (HTT)60 on the basis that it would lease back the infrastructure subject to a long-term contract
(Wakama 2013).61 In 2019, HTT owned 3,650 sites in Tanzania and, according to its website, it had a market
share of 68 per cent in 2018 (see Helios Towers 2019).
Helios Towers reported that it had an average 2.18 MNOs on each of its towers in Tanzania in 2018 (see Helios
Towers 201962), indicating that significant physical infrastructure sharing takes place.63 HTT have completed
four tower transactions since 2011 and recently reported it manages 4,068 sites with 9,121 Tenants, giving an
average of 2.24 tenants per site.64
AirTel Africa continues to offload its telecom infrastructure in an effort to reduce group debts, selling 1,400
towers throughout Tanzania for USD 175 million. AirTel Africa has announced the sale of 1,400 cell towers
to a joint venture between a subsidiary of the SBA Communications Corporation and UK-based Paradigm
Infrastructure Solutions. The towers represent the entire portfolio belonging to AirTel subsidiary AirTel
Tanzania. Paradigm holds a minority stake of the joint venture and maintains responsibility for the JV’s day-to-
day operations.65 The JV is operating as Minara Tanzania.66
60 HTT is part of the Helios Towers group, a UK-based telecommunications company listed on the London Stock Exchange and a constituent of the FTSE
250 Index. The company was established in Mauritius with financial backing from George Soros, Millicom and Bharti Airtel in 2009. It was the subject
of an initial public offering on the London Stock Exchange in 2019.
61 Wakama, A.S. (2013). ‘Vodacom Tanzania Acquires Helios Towers Tanzania Network’. IT News Africa, 22 July. Available at: https://www.itnewsafrica.
com/2013/07/vodacom-tanzaniaacquires-helios-towers-tanzania-network/
62 HTT(2019) Annual report and financial statements at https://www.heliostowers.com/media/1786/ht-ar-2019.pdf
63 Regulation in Tanzania requires MNOs to build larger towers capable of supporting no less than three sets of different Radio Access Network (RAN)
equipment.
64 https://www.heliostowers.com/where-we-work/tanzania/
65 See https://paradigm-infra.com/news-story/first-closing-of-tower-acquisition-transaction-in-tanzania/
66 See https://minara.co.tz/
16
2.4 Regulatory landscape
The liberalized telecommunications market in 3. Application service the reselling of electronic
Tanzania operates under a myriad of regulators. communication services to end users
Direct regulation of operators is overseen by
4. Content service, a service offered for sound, data,
the TCRA who apply a licensing regime based on
text or images whether still or moving except where
an approach pioneered by Malaysia known as a
transmitted on private communication
Converged Licensing Framework. There are four
categories of licence available:
Most of the network operators described above
1. Network facility the provision of any element hold licences in each category. The TCRA maintains a
or combination of physical infrastructure used database of licence holders on its website.
principally for, or in connection with, the provision of Appendix B provides a summary of the principal
Content services and other Application services, but actors in the regulatory space.
not including customer premises equipment67
2. Network service a service for carrying information
in the form of speech or other sound, data, text
or images, by means of guided or unguided
electromagnetic energy, but not including services
provided solely on the customer side of the network
boundary
67 The template for a network facilities licence is available at https://www.tcra.go.tz/uploads/documents/sw-1619014620-Network%20Facility%20
(NF).pdf In this document it can be seen that the term length is defaulted at 25 years for a network facilities licence and in addition to an initial fee,
the licensee shall pay a royalty based on turnover.
17
Universal service
3 policy in Tanzania
Over the years various Governments have emphasized 3.1 Universal Communications Service
the need for greater coverage of telecommunications and Access Fund (UCSAF)
universal service. The road to universal service was set out UCSAF has been operational since 2010 and is responsible
before liberalization in the National Telecommunications for both collecting funds and disbursing monies for the
Policy of 1997.68 Ambitions for promoting ICT were also purpose of promoting communications development
laid out in The Tanzania Development Vision 202569 and in overseen by the Universal Communications Service
the Rural Development Strategy 2001.70 Access Fund Regulations.74
In 2003 the Government defined a new National ICT It is funded by a universal service levy applied on the
policy in which it recognized the importance of funding gross operating revenues of licensed service providers, by
rural coverage: “While there is provision for a Rural Government and by loans provided through international
Telecommunication Development Fund, this has not yet agencies.75 Currently the UCSAF Regulations allow a
been implemented, leaving matters of universal access to levy charge of 1%, which was gradually increased from an
the liberalised environment.”71 Following on from this, the initial 0.3% in 2015-16 to 1% in 2019-20.76 Other sources
Universal Communication Service Access Act was passed of funds include: Government, the TCRA, Parliamentary
in 200672 which established the guidelines for the creation Allocation, and Development Partner grants (including
of UCSAF. WB loans).77
Universal service remains a high priority as demonstrated Funding raised was combined with project funds from the
in the 2016 National ICT Policy: “One of the major initiatives RCIP which initially received USD 100 million financed
that the government has pursued is the improvement of the by the WB through an International Development
ICT infrastructure to bridge the digital divide and lower the Association (IDA) credit facility. The Development Credit
cost of communications”73 Agreement (DCA) between the Tanzania government
and the WB was signed on 22 July 2009 and the credit
became effective on 20 January 2010. The project was
implemented jointly by the Ministry of Works, Transport
and Communication (MWTC) and the President’s Office-
Public Service Management (PO-PSM).78
UCSAF has conducted thirteen SRAs since 2010 which
have resulted in the granting of universal service funds to
68 See section 2.1 in National Telecommunications Policy, October 1997 at https://www.tcra.go.tz/uploads/documents/sw-1619081634-National%20
Telecommunications%20Policy%20of%201997.pdf
69 For example, see page 20 in “The Tanzania Development Vision 2025”, published 2000 available at http://www.tzonline.org/pdf/theTanzaniadevelop-
mentvision.pdf
70 “Rural Development Strategy”, Prime Minister’s Office, December 2001 at http://www.tzonline.org/pdf/ruraldevelopmentstrategy1.pdf
71 See page 8 in the United Republic of Tanzania, Ministry of Communications and Transport “National Information and Communications Technologies
Policy” 2003 at http://www.tzonline.org/pdf/ictpolicy2003.pdf
72 https://www.tcra.go.tz/uploads/documents/sw-1619084421-The%20Universal%20Communications%20Service%20Access%20Act%20of%20
2006.pdf
73 National ICT Policy 2016 p.1 see note 26.
74 See https://www.ucsaf.go.tz/uploads/documents/sw-1635428722-Kanuni%20za%20MFUKO.pdf
75 Sources of funds for UCSAF are set out in general terms in the UCSAF Act 2006, section 23, op ci
76 There is the perspective of reviewing the universal service levy upwards up to a ceiling of 1.5% as provided for by the UCSAF Act (section 18(4)).
Source EU (2021) Action Document for Digital4Tanzania – e-Governance Support Programme available at https://www.gtai.de/resource/
blob/788132/e262205f40141c2ea2c13c84f61c4b27/PRO20220127788124%20-%20Annex3.PDF
77 For a summary of early interventions, see pp.41-42 in GSMA (2013).
78 The original availability period of 40 years concessionary loan covers the timeframe of five (5) years with effect from January 2010 through Febru-
ary 2015. The principal amount plus interest is to be repaid in equal semi-annual instalments over a period of 30 years effective from 15 Septem-
ber 2019, upon expiry of the grace period of 10 years. The interest component of the loan facility is calculated at the margin rates of 1% and 2%,
between 15 September 2019 and 15 March 2029; and between 15 September 2029 and 15 March 2049, respectively.
18
licensed communications operators. As noted in Ministry
of Finance and Planning (MFP, 2021), this has helped in 6. Advising the TCRA on matters relating to
“Increasing the number of fixed and mobile subscribers from communications for all;
21.2 million (2010) to 48.9 million in (2020). This increase 7. Looking for appropriate ways to provide
is attributed to the continued investment in communication subsidies (including auctions) for the delivery of
services, including in rural areas via the UCSAF”.79 telecommunications services in rural areas and in
areas that are not commercially viable;
8. Legalize contracts for the implementation of
3.2 UCSAF objectives and communications service delivery responsibilities;
9. Operating and developing the Fund as established;
responsibilities 10. Establish good procedures for managing, calculating
UCSAF was established in 2006 with the aim of delivering
and collecting communication levies for all from
and facilitating communication services to citizens living in
service providers;
areas with no commercial appeal to telecommunications
service providers.80 UCSAF is responsible for enabling 1. Calculate, plan and distribute communications
accessibility and participation by communication grants to all;
operators in the provision of communication services, 2. Recommend effective Fund policies to the
with a view to promoting social-economic development Minister of Communications Sector;
of the rural and urban underserved areas through 3. Managing the Fund’s financial resources and
ICT interventions as well as to ensure availability of ensuring that the Fund’s expenditure benefits the
communication services in rural and urban underserved community;
areas and to bridge the digital divide between rural 4. Consulting and collaborating with Government
populations and urban populations. A key objective Ministries, Departments or Independent
of establishment of UCSAF is to bridge the existence Authorities regarding public service matters; and
5. Developing guidelines for the operation of the
communication gap between urban and rural population.
Fund.
The ‘Fund’ is a body corporate with a board whose
members are appointed by the relevant government
Minister.
UCSAF has a number of responsibilities: 2.3 Internet penetration policy
The Government announced in early 2021 plans to
1. To identify areas of communication projects that may increase internet penetration in the country from the level
receive grants from the Fund – such a rural broadband of 43% in December 2020 to 80% by 2025, and to extend
development; access to telecommunications to a further 173 unserved
2. Set criteria for identifying areas of villages that need villages and to switch off 2G.81 By March 2023 about 33.1
and cannot access telecommunications services million had access to internet services, a penetration rate
because they have no commercial appeal; of over 50%.82
3. Establish an effective financial management system;
4. Set appropriate criteria for granting grants to
telecommunications service providers for deploying
communications in designated areas;
5. Conduct various surveys and monitor the
development of communications in rural and non-
commercial areas;
79 See page 25 in MFP (2021) “Report on the Implementation of the Istanbul Programme of Action for LDCs for the Decade 2011-20: Tanzania Coun-
try Report”, March, available at https://www.un.org/ldc5/sites/www.un.org.ldc5/files/tanzania.pdf
80 See https://www.ucsaf.go.tz/
81 Source https://itweb.africa/content/KA3WwMdD99eMrydZ
82 See Table 3.1 in TCRA(2023) op cit.
19
3.4 Disbursing public funds and UCSAF
Since its establishment in 2009 UCSAF has worked in UCSAF disburses funds via contract to bidders whose bids
periodic phases. In June 2020 Rural Telecom Project phase have been determined as Most Advantageous Bids. These
5 started. UCSAF provides subsidies to private operators are the bids of the bidders that meet the qualification
and in the first four phases of activity from 2013 to 2019 criteria and whose bids have been determined to be:
it mobilized TZS 120 billion (USD 50 million). The level
of subsidies is steadily increasing, and it is expected to (a) substantially responsive to the particular bidding
increase in the coming years.83 document; and
(b) the lowest evaluated cost.
Funds have been disbursed in thirteen award programmes
since 2010. The process for disbursing funds follows a As already discussed in chapter 1, participating bidders
path where UCSAF initially identifies areas in need of have one opportunity only to submit a request for public
support. Once areas have been identified, UCSAF then support in each area listed in an SRA award process. As we
arranges a process to allocate the funds. argued above, if the bidding process had multiple rounds
of bidding, there would be a greater chance of securing
The process that UCSAF has applied to date is based on a higher bid success rate. This theme is picked up in the
a mechanism called Request for Bids. UCSAF publishes a following chapters.
list of wards (administrative areas) available for possible
receipt of public support and sets a MAS for each area
under consideration. For example, 200 areas (wards) may
be listed as part of an award process, and each will have a
value attached designating the MAS. Typically UCSAF has
set a MAS across wards such that total funds available are
divided almost equally. Although the MAS is usually the
same for most lots (wards), the costs of service provision
and levels of demand may be very different across wards.
83 See p.7 in EU(2021) op cit
20
Auctions and universal
4 service funds
In choosing how to disburse universal service funds,
Principle of Public SRA Multistage MRRA
government agencies have a number of policy
Procurement with BAFO
options available (as shown in Chapter 1, Table 7
above). Whichever option is chosen, the goal of public Competition
procurement should adhere to the following key
principles: Transparency
Accountability
• Competition, Table 12 Qualitative assessment of procurement options against
• Transparency, and principles of public procurement
• Accountability.84
Competition ensures that suppliers are forced to compete The MRRA scores highest in regard of competition
for the limited funds available, which drives down costs because with several or many potential suppliers it offers
and improves quality. Transparency ensures that bidders the ability for robust, real-time price competitions with
know what they are bidding on and why, so they can make multiple rounds of bidding for continued reductions in
informed decisions and resources are allocated efficiently. subsidy requests. A multistage auction with BAFO is also
Accountability ensures that government officials who more competitive than a SRA, but it is limited in terms of
award contracts are held accountable for their decisions, rounds and hence subsidy reductions.
minimizing corrupt influence and bias.
In the previous chapter we noted UCSAF seeks to ensure The MRRA scores best for transparency, as multiple round
that the allocation of universal service funds delivers bidding makes it clearer to all parties what subsidies are
value for money and its processes adhere to the principles available in the market and price discovery over multiple
above. More generally, the Public Procurement Regulatory rounds increases the precision of cost estimates among
Authority (PPRA), which oversees public procurement in potential suppliers. The multistage with BAFO is also able
Tanzania, sets a vision that public procurement systems to offer greater transparency than a SRA, but as it usually
should have “integrity, offering best value for money”.85 occurs with two rounds, it is more limited in terms of price
discovery.
In this chapter we argue that auctions and the MRRA
format executed on an EAP are best suited to deliver Finally, a well-designed MRRA and multi-stage auction
public procurement policy objectives in a competitive with BAFO both perform best in terms of accountability,
setting involving the promotion of mobile broadband as the outcome of the process is a result of fair and open
coverage. competition among suppliers. The SRA is prone to more
bureaucratic judgment and very limited price rivalry with
In our view the MRRA format is at least as good as no price discovery, and may as a consequence be less
alternative options, as highlighted in Table 10 below, accountable.
which scores each option qualitatively against the above
key public procurement principles.
84 See Armstrong (editor) (2012) especially chapters 1 and 7.
85 See https://www.ppra.go.tz/
21
In our view and in the view of many others, a well- The design of auction pays careful attention to the
designed MRRA is an effective way to discover prices, detailed structure governing bidding and participation.
achieve value for money and promote efficiency in the The main design issues are described in the following ten
presence of competition.86 Over the course of the last 20 elements, which are considered in the context of UCSAF’s
years or so there have been a number of reverse auctions responsibilities:90
in the telecommunications sector.87 It is the MRRA we
focus on in this report, as competition and the scale of the 1. Lot category
universal service programme in the telecommunications A key component in designing an auction is the
sector in Tanzania facilitates its application. description of lots for award – here the area in which
the subsidy is offered and technical specifications
are relevant. For UCSAF, this is achieved by choosing
4.1 The MRRA format areas that coincide as closely as possible with areas
In a reverse auction with multiple objects (such as a served or potentially served by MNOs cell sites. To
reverse auction involving hundreds of lots), a single round ease the administrative process, UCSAF chooses areas
of bidding does not provide bidders the opportunity to that align service areas with political administration
change bids in response to information revealed by other areas known as wards. Note that more than one ward
bidders in their bids. This means bidders are not able to may be served by a single, well-sited cell tower.
exploit variations in value for substitute or complement
objects. Consequently, under a single round of bidding a 2. Lot size
bidder is more likely to be exposed to the risk of paying How lots are dimensioned can influence the way
too much for an inferior combination of objects if some bidders approach bid strategy and impact on auction
objects that are desired and have synergies are won by outcomes. For example, lots can be presented
other bidders (known as exposure risk88). individually, as is typically the case by UCSAF, or they
may be organized in categories such as regions, etc.,
MRRA formats are designed to mitigate exposure risk, to ease auction logistics. The more packaged are lots,
promote price discovery and as a result are more likely to the reduced scope bidders may have to assemble their
yield an efficient allocation of public subsidy funds. preferred packages. On the other hand, pre-arranging
In a multi-round auction, as in all auctions, the design and lots into convenient batches (say regional blocks), may
clarity of the auction rules are critical to success: in this simplify auction logistics.91
instance ensuring an efficient allocation of funds. This was
amplified by Wallsten (2009):89 3. Maximum Allowable Subsidy (MAS)
The MASs in UCSAF auctions are calculated with
“In addition, other countries have used reverse auctions reference to available credit facilities, estimates of
to provide universal service with some success. Their cost of service and estimates of demand. The MAS
experiences demonstrate convincingly that reverse auctions should be sufficiently high enough to attract credible
can bring down subsidies substantially. Their experiences also interest, but not too high to avoid poor value for
demonstrate that, as in any auction, the rules matter a great money, and not too low to deter participation.
deal.”
86 As Nobel Prize winning economist Paul Milgrom has emphasised, multi-round auction formats promote efficiency and price discovery, see Milgrom
(2004). Wallsten (2009) asserts “that reverse auctions can be an effective tool for revealing information about the true cost of providing universal
coverage and for reducing expenditures on subsidies.” The benefits of electronic reverse auctions have been noted more widely in government pro-
curement, for example see Hawkins (2010), and Shalev and Asbjornsen (2010). According to the Office of Federal Procurement Policy in the United
States, electronic reverse auctions have brought savings of around 12-14% for several Federal agencies in the United States, see Memorandum for
Chief Acquisition Officers, Anne E. Rung “Effective Use of Reverse Auctions”, 1 June 2015 at https://www.whitehouse.gov/wp-content/uploads/lega-
cy_drupal_files/omb/procurement/memo/effective-use-of-reverse-auctions.pdf
87 An early contribution on reverse auctions in the telecommunications sector is Wallsten (2009). In 2009 seventy one eminent economists in the
United States wrote an article favouring the use of reverse auctions (procurement auctions) in promoting broadband stimulus programmes, see
Milgrom et al. (2009). For a more recent empirical assessment see Oh (2021). The empirical assessment by Wu (2014) is also a very good overview of
applications of reverse auctions in the telecommunications sector.
88 Exposure risk, also known as aggregation risk, occurs in multi-unit auctions in which there are synergies across lots, see Cramton (2002).
89 Wallsten (2009) op cit. page 397.
90 See also Paul Milgrom, Lawrence Ausubel, et al. (2012) “Incentive Auction Rules Option and Discussion”, submitted by Auctionomics and Power
Auctions to the FCC available at https://apps.fcc.gov/edocs_public/attachmatch/FCC-12-118A2_Rcd.pdf
91 Increasing the number of lots in an auction offers more discretion for bidders but comes at a risk of additional computational complexity, especially
in regard of bid strategy, winner determination (depending on how bids are expressed) and valuation. In some circumstances this can result in inef-
ficiency. In an econometric analysis of the US ‘C Block’ spectrum auctions of 1995-96, in which the FCC offered 480 distinct licence areas, Fox and
Bajari (2013) argued that had the FCC chosen fewer licensed areas the outcome would have been more efficient.
22
4. Bidder Categorization 7. Assignment (which winner receives which lots)
UCSAF may choose to categorize interested parties In a MRRA and other universal service reverse
(bidders) depending on industry background, size, auction formats, the assignment of areas to winners is
location, experience, etc. In the subsidy awards to straightforward because the lots are defined as set out
date, UCSAF has granted funds to communications in (1) above. In other auctions in telecommunications,
bidders may initially bid for generic lots (such as the
providers such as MNOs, though the technical
quantum of radio spectrum) and once demand is
specifications do not always preclude others (such as equated with supply, a second stage assignment of
tower companies) from participating. UCSAF might winning lots to specific frequencies may take place. In
choose to designate certain parties as entrants if there principle generic ‘geographies’ could be established in
is concern about market competition. a reverse auction setting, although this would unlikely
yield material benefits.
5. Bid collection (how bid information is gathered
from bidders)
The way UCSAF collects information from bidders 8. Pricing rule (how bids determine the subsidies
depends on the type of auction. As discussed, UCSAF received by winners)
collects information using a SRA. Alternatively, The pricing rule determines what the winning bidders
bid information may be gathered over multiple receive in regard of subsidy. Rules may require bidders
rounds in a MRRA. A single round auction could be receive-as-bid (first-price), pay-as-clear (all successful
executed quickly and where the bidders are well- bidders receive the same market clearing price) or
receive an amount equal to the nearest unsuccessful
informed about the value of the lots and common
bidder (second-price). In the assignment stage it is
value uncertainty is not significant, this may be usual to apply second-price solutions.
optimal. However, we feel the conditions in the
subsidy award auctions do feature sufficient common 9. Activity rules (constraints on bid activity)
value uncertainty to warrant consideration of the Activity rules are crucially important and govern
MRRA format. However, a MRRA, depending on the number and type of lots a bidder can bid for
design, can facilitate strategies that lead to strategic subsidy in an auction. In general these may cover
overstatement of subsidy requests (equivalent to three dimensions: (i) the minimum quantity of lots
demand reduction in a forward auction). This gaming a bidder is required to bid for; (ii) the maximum
behaviour can be mitigated through other rules quantity of lots a bidder can bid for, and (iii) an overall
constraining the nature of bidders’ bidding activity bid activity constraint. Their purpose is to promote
and rules affecting information disclosure, see (9) and straightforward bidding (i.e. bids based on a bidder’s
(10) below. valuation rather than influenced by other bidders’
valuations), competition and progression.
6. Winner determination (how winners are
identified) Bid activity rules can help progress an auction and
A common winner determination procedure is to encourage bidders to bid in accordance with their
award lots to bidders submitting the lowest bids. If valuations.92 Typically, bid activity rules are applied by
there is scope to present packaged bids, it may not be the auctioneer using a points measure of bid activity.
straightforward to assign each lot to a winning bidder.
In this case, the auctioneer would usually apply an For example, each ward is assigned points by the
algorithm to search for the best assignment of bids auctioneer before the start of an auction. The points
that win (the usual winner determination criterion structure could reflect a pre-auction estimate of the
relative value of lot categories. A bidder’s maximum
would be to look for the winning combination of bids
activity in an auction is usually determined by its initial
that yields the lowest subsidy request, where each eligibility, which is the total number of activity points
bidder submits only one bid in that combination, and
associated with the bidder’s deposit.
the allocation is feasible).
92 Activity rules may also allow for waivers, where bidders can sit out a round and remain active in the auction without submitting a bid. The motivation
for waivers is to provide bidders additional time to consider ongoing participation. The downside of waivers is bidders may use them strategically.
Another activity rule may permit bid withdrawals, to protect bidders against the risk of winning uneconomic lots. The downside of withdrawals is
strategic bidding aimed at price driving (which in the context of a reverse auction means bidding purposely to lower the amount of subsidy a rival
may receive on a lot).
23
As part of bid activity rules, the auctioneer chooses an 4.2 The default MRRA format: the clock
activity requirement for each round. This usually works auction
by determining the bidder’s eligibility in each round.
The modern default auction format for multi-round
For example, the activity requirement might be such
that a bidder can preserve its eligibility as long as its auctions is known as a clock auction.94 A clock auction Is
processed activity (the points it uses when bidding in a when the auctioneer announces prices and for the case of
round) is 90% or more of its eligibility.93 If a bidder’s a reverse auction these decline over time (i.e. successive
processed activity were less than 90%, its eligibility rounds).
going into the next round is lowered.
A very simple clock auction format is as follows. Assume
We set out in Appendix C.2 the activity rules for an a policymaker has $100 universal service funds for an
MRRA subsidy auction. area in which one operator is invited to supply service.
Suppose three bidders qualify to bid for subsidy. The
10. Information disclosure (what information is auctioneer starts by announcing support at $100 and all
made available to all bidders when) three bidders’ express interest. The three bidders are tied
Information of relevance in an auction relates to at $100. The auctioneer lowers the price in a subsequent
bids submitted, who is bidding, who bid what, etc. round and announces $95 support. Assume one bidder
Information disclosure procedures determine what
drops out, leaving two contesting the subsidy. After a
information is made public before, during and after
an auction. It also relates to information disclosed to further round at which the auctioneer announces $90,
bidders before a round, during a round, and after each only one bidder is left in the auction.
round.
In Appendix D we provide a detailed illustrative example
Transparency and disclosure are viewed as favourable of a clock auction involving multiple areas. We propose
for price discovery and accountability. On the this format for application in a pilot to be deployed by
other hand, too much disclosure may foster anti- MICIT and UCSAF. The clock auction format pilot we
competitive bidder collusion and gaming. Examples propose is a variation of one used by the FCC in the
of information disclosure rules include making United States, which is the focus of the next chapter.
bidders’ identities anonymous during the auction and
revealing excess demand information in summary
form. Such rules are intended to deter bidders from
gaming.
We set out in Appendix C.5 information disclosure
rules for an MRRA subsidy auction.
93 In this case 90% represents the activity requirement which is set by the auctioneer. The FCC in the United States often starts with an activity
requirement of 85% or 90% in spectrum auctions and after a certain number of rounds it increases the activity requirement to 95% or higher. The
ramping up of the activity requirement trades-off allowing bidders flexibility during the early rounds, when price and package discovery are likely to
be of greater value and promoting auction progression, and strategic bidding issues.
94 A clock auction is one in which the auctioneer moves the price (up for ascending, down for descending) for an object in accordance with a clock. For
example, the price might decline by $1 every minute. The most famous clock auctions are held in the Netherlands every day for flowers, see https://
www.royalfloraholland.com/en/about-us/who-we-are/royal-floraholland/visit-the-auction In short, a clock auction describes an auction where the
auctioneer moves the price.
24
International MRRA
5 case study
In chapter 1 section 5 we noted that reverse auctions The Phase I auction, which began on October 29, 2020,
for minimum subsidy and universal service funds have and ended on November 25, 2020, awarded support to
occurred in several countries, including Chile, Peru, bring broadband to over five million homes and businesses
and India. Many of these auctions used the SRA format. in census blocks that were entirely unserved by voice and
However, the largest allocation of universal service funds broadband with download speeds of at least 25 Mbits/s.
has taken the form of MRRAs, with the largest occurring in Auction 904 involved the disbursement of funds of up to
India and the United States.95 In this chapter we focus on USD 16 billion to subsidise broadband rollout. Auction
the MRRA format developed and applied by the FCC in the 904, and its smaller precursor Auction 903, was a multiple
United States. round reverse clock auction featuring a descending price
in which bidders submitted demands for financial support
The FCC has pioneered several large-scale reverse over multiple rounds.
auctions aimed at promoting universal service in rural
broadband services. The experience of the FCC represents Auction 904 invited participants on a technological neutral
international best practice. The structure of the MRRA basis and involved important financial and quality of
model used by the FCC is flexible, scalable and highly service dimensions, where the latter had assigned weights
suited for deployment on an EAP. In the next chapter we reflecting a judgment about performance “Tier” (speed of
draw lessons from the FCC’s methods for UCSAF and service) and “Latency” (T+L weights). The motivation for
MICIT in Tanzania. assigning weights was to incentivise higher-speed (T) and
lower latency (L) services. As we shall discuss, the weights
We focus on the MRRA format as applied the FCC Auction serve to favour bidders offering higher speed and lower
904 which took place in 2020. latency.
The scale of the auction was immense, with up to USD 16
5.1 FCC Auction 904 – MRRA Clock billion of universal service funding available to support
Auction service in over 5 million areas, as shown in Figure 5.
One of the largest reverse auctions (by value) for universal
service fund allocations in telecommunications has
occurred in the United States. The reverse auction was
held in relation to the FCC’s management of the Rural
Digital Opportunity Fund designed to help bridge the
digital divide. On August 1, 2019, the FCC adopted a
Notice of Proposed Rulemaking proposing to establish the
USD 20.4 billion Rural Digital Opportunity Fund to bring
high speed fixed broadband service to rural homes and
small businesses that lack it.96
On January 30, 2020, the FCC adopted the Rural Digital
Opportunity Fund Report and Order, which established
the framework for the Rural Digital Opportunity Fund,
building on the success of the Connect America Fund
Phase II auction by using reverse auctions in two phases.
95 A good summary of the approach taken to disbursing universal service funds in India using a MRRA format is in Annex 2 in Malik and de Silva (2005).
96 See https://www.fcc.gov/auction/904
25
Figure 5 FCC Auction 904 with over 5 million areas eligible for support
The FCC used a descending clock auction format. In each area (or group of areas) a maximum subsidy allowance (reserve
price) was set, and the auction started with a ‘percentage price’ multiplied into the reserve prices. The percentage price
was common to all areas, and it was the percentage that descended round-to-round whenever the total demand for funds
exceeded the available funds.
The auction cleared (ended) when demand for support in aggregate was equal to USD 16 billion or less. The auction closed
after:
• the budget cleared, and
• there were no areas remaining with competing bids.97
5.2 Auction 904 – outcome
The outcome of the auction resulted in winning bidders receiving funding to deploy high-speed broadband to over 5.2
million unserved homes and businesses. This was a bid success rate of almost 99%, higher than that achieved in any of the
UCSAF reverse auctions. A total of 180 bidders won auction support from the 386 qualified bidders that participated.
The auction unleashed robust price competition that resulted in more locations being awarded at less cost to Americans
who pay into the Universal Service Fund. Some 5,220,833 locations were assigned support in the auction had an initial
reserve price of over $26 billion. Competition among bidders over multiple rounds meant that the final price tag to cover
the locations was just over $9 billion. Only 56% of the universal service funds USD 16 billion was needed to achieve a near
100% bid success rate.
The Chair of the FCC, Ajit Pai, commented about the auction:98
“I’m thrilled with the incredible success of this auction, which brings welcome news to millions of unconnected rural Americans who
for too long have been on the wrong side of the digital divide. They now stand to gain access to high-speed, high-quality broadband
service. We structured this innovative and ground breaking auction to be technologically neutral and to prioritize bids for high-
speed, low-latency offerings. We aimed for maximum leverage of taxpayer dollars and for networks that would meet consumers’
increasing broadband needs, and the results show that our strategy worked. This auction was the single largest step ever taken to
bridge the digital divide and is another key success for the Commission in its ongoing commitment to universal service.”
97 Only one bidder could win in each area.
98 Source: https://www.fcc.gov/document/fcc-auction-bring-broadband-over-10-million-rural-americans
26
Lessons for
6 Tanzania
There are lessons for Tanzania and other countries from In our view, Tanzania is fully capable of deploying a
the success of the FCC’s multi-round universal service MRRA for universal service funding and this could yield
auctions. In this brief chapter we highlight the lessons for substantial savings resulting in a more efficient process. In
Tanzania. We start by showing that the legal environment the next chapter and in Appendices C and D, we describe
in Tanzania is compatible with holding a MRRA. in full the proposed MRRA.
We next highlight the experience Tanzania has with
conducting online interactive procurement processes and 6.3 Capacity building and education
note it has much experience. An extension into a MRRA The application of a MRRA using an EAP will present
for universal service funding allocations using an EAP institutional challenges, require capacity building
should be possible with sufficient capacity building and and education. For example, the MRRA format used
education. by the FCC, as described in the previous chapter, is a
comprehensive format that can be tailored and simplified
6.1 Legal framework for application in a pilot in Tanzania.
We are confident a MRRA could be conducted in Tanzania.
The relevant law governing tender processes for the The agencies that would manage the process, MICIT and
disbursement of public subsidies is set out in the Public UCSAF, would need to work closely with stakeholders and
Procurement Act 2011 (PPA).99 Section 67 of the PPA others to develop an EAP to host a MRRA. An important
deals with competitive tendering asserts: element of the process will be education of participants
“all eligible prospective tenderers [will have] an equal and the wider public, as well as briefing key management
opportunity to tender for the required works services” in the respective ministries.
An online MRRA can be made compliant with this
requirement, as all prospective tenderers, or bidders,
would be subject to the same auction rules.
Note section 6 of the UCSAF (2006) Act allows for the “set
the bidding conditions for the awarding of funds”.
6.2 Procurement experience
Tanzania has much experience with online interactive
e-procurement systems through the Tanzania National
e-Procurement System (TANePS). However, we
understand this has not been organised to date to
host a MRRA across potentially hundreds of lots
simultaneously.100 In regard of the pilot that we
recommend, we suggest that this could possibly be
organised as an independent module linked to TANePS.
99 The Public Procurement Act, 2011 at https://www.tanroads.go.tz/common/uploads/acts/p168205f13169a8d90f5662e0f8fe6ba.pdf
100 TANePS is a web-based, collaborative system, developed in accordance with the requirement of public procurement laws, to facilitate public
procurement processes in Tanzania. It offers a secure, interactive, dynamic environment for carrying out procurement of all categories, complex-
ity or value. TANePS supports processes of procuring Goods, Works, Consultancy, Non-Consultancy and Disposal of assets. The system supports
various public procurement procedures including user registration, tender notification, tender preparation and submission, online tender evalua-
tion, contract awarding, creation and management of catalogue, creation and management of framework agreements and auctions and payments.
TANePS is managed and developed by European Dynamics. We recommend that the TANePS tool is considered as the front-end to manage a multi-
round reverse auction on behalf of UCSAF. See https://www.taneps.go.tz/epps/home.do
27
Steps towards a pilot: multi-round
7 reverse auction – recommendations
We propose that MICIT and UCSAF pilot a MRRA Step 2: Estimate maximum allowable
delivered over an EAP for the purpose of allocating subsidy for each site
universal service funds.101 The recommended steps in MICIT and UCSAF should develop a high-level financial
the auction pilot are set out below (a summary of the model populated with data on population density, existing
recommendations is shown in Tables 1 and 2 in the mobile network coverage (if any), powerline access,
Executive Summary). Appendices C and D set out the fibre access, road access, inbound phone volumes and
detailed rules of the auction format to be used in the pilot income levels. The purpose of the financial model is to
and Appendix E discusses available EAPs on the market provide a rough estimate of the MAS for each selected
that could potentially be deployed in the pilot. site. Setting different values for the MAS across sites may
not be possible if UCSAF does not have discretionary
powers.103 If UCSAF were to develop a financial model
Step 1: Identification of sites and for the purposes of modelling MSA values, it should be
number of sites made public and shared with stakeholders to ensure
The first step is to identify the sites (based on wards) transparency of the process. An initial model could be
suitable for inclusion in the pilot auction. The purpose developed under a consultant contract. Thereafter,
of the pilot is to assess the effectiveness and viability of UCSAF should work with other agencies and stakeholders
applying a MRRA. As this will be the first time it has been to maintain the financial model.
applied in Tanzania for the purpose of allocating universal
service funds, limiting the number of sites (wards) to no The financial model is intended as a guide for MICIT and
more than 100 would ease the administration of the pilot UCSAF and should not be viewed as a replacement for the
for all parties involved (MICIT, UCSAF and bidders). The process of price discovery during the MRRA. In practice,
pilot should be for greenfield sites (sites with no or very MICIT and UCSAF should apply conservative values and
limited existing mobile coverage).102 err towards overstating the MAS at each site. It is the
Recommendation 1 Select up to 100 greenfield sites job of competition and subsidy request reductions in the
MRRA to reveal the hidden true costs of service at each
Sites should reflect geographic diversity and we site.
recommend that they are selected from at least three Recommendation 3 Apply and develop a financial model
regions of Tanzania, and ideally from as many regions as to calculate the MAS for each selected site104
possible. Some sites should be selected from Zanzibar.
Recommendation 2 Select sites from at least three
regions and include Zanzibar
101 A pilot was also used to determine the benefits of extending subsidies to promote mobile phone coverage in Tanzania, see GSMA (2018) “Tanzania
rural coverage pilots: Performance report” available at https://www.gsma.com/mobilefordevelopment/wp-content/uploads/2018/02/GSMA_Tan-
zania_Report_Jan.pdf
102 The auction format and process could be applied to brownfield sites (sites with limited coverage or only 2G coverage), but the attraction of green-
field sites is a higher likelihood of competition.
103 If UCSAF can set different MAS for different wards, these should be based on objective measurable criteria. For example, the MAS may be calculat-
ed using a financial methodology that allows each efficient operator at a selected site to enjoy an expected normal rate of return when in receipt of
the MAS. The return on investment would be a discounted stream of net revenues over capital employed. Sites in receipt of a UCSAF subsidy MAS
would enjoy a higher rate of return. The financial model for each site would seek to identify a MAS that results in an expected ‘normal’ return on
capital employed.
104 It is important to note that the MAS in the dynamic multi-round auction case is more akin to a reserve price guide rather than an absolute ceiling.
In Appendices C and D the full description of the auction rules and their illustration indicate the possibility that winning bidders may receive more
than the MAS value at some sites.
28
Step 3: The budget Step 5: Qualified bidders
After completing steps 1 and 2, UCSAF should determine Following the end of the application form process,
a budget “B” for the pilot. This should be based on the UCSAF shall announce publicly which applicants have
total maximum allowable subsidy (TMAS). The TMAS is attained qualified bidder status. UCSAF shall confirm
equal to summation of MAS values across the sites in the in confidential communication with each bidder the list
pilot. We recommend that B is set at no more than 95% of of areas the bidder is eligible to bid for. Each notified
the estimated TMAS. The expectation is price discovery qualified bidder will only be aware of the sites it has
and competition will yield at savings of at least 5% of the expressed an interest in bidding for and not those of other
estimated TMAS. qualified bidders. Communication among applicants and
Recommendation 4 The budget “B” in each region of the qualified bidders about such matters is prohibited under
pilot should be B ≤ 0.95 TMAS the rules of the procurement process.
Recommendation 6 Qualified bidder status public
If a budget implied by Recommendation 4 is not possible announcement
for the 100 sites proposed, then MICIT and UCSAF should
reduce lower the total number of sites below 100 by Step 6: Electronic auction platform
removing the sites with the highest MAS values. It should UCSAF shall make use of a EAP for the pilot.106 The EAP
lower the number of sites until the condition in Step 3 is shall have clear bidder and auctioneer interfaces. UCSAF
satisfied. shall provide guidance to qualified bidders in the form of
a manual or set of instructions for using the EAP. The EAP
The lower the value set for B, the more likely that fewer will be available via a secure server and compatible with
sites will be successfully covered in the MRRA. In practice standard Internet browsers. Access by UCSAF officials as
MICIT and UCSAF should aim to set B close to and around the auctioneer and by qualified bidders will be password
95% of TMAS. protected and subject to two-factor authentication (or
multi-factor authentication), or through the issuance of
digital certificates.107 UCSAF should explore the possibility
Step 4: Application form and of managing the EAP as an independent module that can
pre-qualification be accessed through TANePS.
UCSAF shall issue an invitation for interested parties to Recommendation 7 UCSAF shall use a secure EAP hosted
apply for qualified bidder status. Qualification shall take on the internet, through acquisition or rental
account of technical, financial and competition law criteria. Recommendation 8 UCSAF shall explore the feasibility of
Participation need not be restricted to licensed MNOs. using the EAP as an independent module accessible via the
For example, tower companies, financial intermediaries TANePS system
and others could participate. Where an interested party is
not a licensed MNO, it should be required to offer access Step 7: Auction education
on a non-discriminatory basis to licensed MNOs to the UCSAF shall supply all qualified bidders access to the
UCSAF supported facilities. Interested parties shall inform EAP for training purposes. Following an open bidder
UCSAF in their application form which regions they intend information session, one or more mock auctions should
to submit bids in and how many sites in each region they be held using hypothetical data for areas and bidders
are interested in bidding for – only regions indicated should be assigned scripts to follow in the mock auction.108
in the application form may be bid for by the applicant. The mock auction should be held close to the start of the
Applications will remain confidential. For each region and date of the actual auction. Participation by bidders will be
the number of sites indicated in the application form, a remote or virtual, and bidders would be anonymised.
charge equal to 2% of the highest MAS values in the region Recommendation 9 UCSAF to undertake a bidder
could be levied, as a bid security.105 information session, one or more mock auctions, and
Recommendation 5 Interested parties are invited to should produce a step-by-step video uploaded to its
complete a bid application in response to a call for website explaining the EAP processes
expressions of interest
105 The charge amount could differ from 2% and could be subject to a prior public consultation. The levy applied is intended to serve two purposes.
First, it incentivises would be participants to list only those sites it has a serious intent on potentially serving. Second, the levy contributes towards
UCSAF administrative cost.
106 We set out in Appendix E further details about the EAP.
107 Standard practice for operating EAPs is such that client connections are encrypted, and authentication of individual users is managed through
digital certificates and passwords. Logging of each and every interaction between auction bidders and the auction software provides for a complete
digital audit trail. Real-time diagnostics allows the auctioneer to detect Internet problems that might affect the running of an auction. Back-up is
automated. Communication via messaging should be possible within the EAP.
108 Bidders in the mock auctions will follow instructions in accordance with a script provided by UCSAF. The intention is to allow participants to famil-
iarise themselves with the process.
29
Step 8: Auction Step 10: Auction impact evaluation
UCSAF shall run the reverse auction on the EAP allowing After step 9, UCSAF shall undertake a public consultation
multiple rounds where bidders submit bids remotely. conducting an impact evaluation of the multi-round
Appendix C describes the rules of the auction that should reverse auction pilot. UCSAF should engage with
be reflected in full in the EAP. UCSAF should investigate stakeholders and seek to draw lessons for future auction
whether auction regulations are required processes.
Recommendation 10 UCSAF to run auction on EAP Recommendation 12 UCSAF to conduct an impact
evaluation of the pilot multi-round reverse auction
Step 9: Auction results and public
announcement
After completion of the auction process, UCSAF shall
announce publicly by notice on their website the names
of bidders who have won subsidies, the sites covered,
and the amount universal service funds allocated in the
process. UCSAF shall also make public by publishing on its
website the full auction round data, subject to commercial
confidentiality conditions.
Recommendation 11 UCSAF to announce winners and
publish auction results
30
8 Next steps
We are recommending that Tanzania pilots a MRRA for the purpose of allocating universal service funds. This process
should start in 2023/24. In the previous chapter we set out our recommendations that should enable the relevant
agencies, and UCSAF in particular, to achieve this goal. In Table 13 below we bring together the main tasks required to
facilitate and conclude the process.
Item Task Owner
Undertake to complete a multi-round reverse
1 auction using an electronic auction platform during UCSAF with support from EAP vendor
2023
Scope out the feasibility of using an EAP as an
2 independent module accessible on the TANePS UCSAF with manager of TANePS
platform
Develop high-level financial model to inform MAS
3 UCSAF in consultation with MNOs
values
4 Capacity building in UCSAF UCSAF and donor agencies
5 Stakeholder engagement UCSAF with other relevant agencies (e.g. TCRA)
Detailed workplan for implementing the multi-
6 UCSAF and WB
round reverse auction
Wider assessment of the feasibility of multi-round
7 procurement auctions in Tanzania and in other MFP and WB
countries
Table 13 Next steps in the multi-round reverse auction process
8.1 Wider lessons
This study and report have looked specifically at the application of a MRRA for the allocation of universal service funds
in the communications sector in Tanzania. It specifically focuses on supporting the development of mobile services and
extending their coverage.
We believe that the auction pilot provides an important foundation for application in other sectors and in other WB
lending programs. Tanzania should consider the feasibility of using a MRRA format for other programmes involving the
allocation of subsidies. This is addressed as Task 7 in Table 13. An example would be the installation of access to the
Internet via fixed telecommunications providers.
The WB should explore the feasibility of applying the auction format considered in this report in other projects, such as
the proposed Digital Malawi Phase II (P160533), where the main focus is on fixed broadband extension in secondary cities
and rural areas, and connecting schools.
31
Although we have emphasised throughout the report that the MRRA offers scope for greater efficiency by promoting
price discovery and more competition, it should be borne in mind that no procurement format is a perfect solution and
disappointments can occur. For example, for commercial or political reasons, participants may fail to show for whatever
reason. The proposed MRRA promises more engagement and transparency and sets out a successful new order at best,
though there are still plenty of risks related to administrative capacity, or to creating and maintaining full bidder attention
and support.
The MRRA we propose, if organized well and if the various stakeholders support it, promises to have a chance of bringing
the key participants to the table and to improve on the past hit rate. Evidence of its application suggests that considerable
benefits are possible from applying MRRAs on an EAP for allocating universal service funds.
32
Appendix A Telecommunication Operators In Tanzania
Operator, service
Description
launched
AirTel – 2001 AirTel Tanzania Limited is the third-largest telecommunications company in Tanzania by
number of customers.
The company is part of AirTel Africa, a pan-African mobile network operator and the largest
mobile service provider in Africa outside of South Africa, active in 14 countries on the
continent.
AirTel Africa plc is majority owned by the Indian communication services company Bharti
AirTel. In 2019 an agreement between Bharti AirTel and the Government of Tanzania
resulted in a shareholding of 51% Bharti AirTel and 49% Government of Tanzania. Previously
the split in ownership was 60% in favour of Bharti AirTel and 40% Government of Tanzania.
As of March 2023, AirTel Tanzania is the second largest MNO with 17.05 million subscribers,
some 27.6% share of mobile subscribers.
Predecessor brands are Zain and Celtel
Azam – 2018 Azam Telecom (T) Limited acquired 2x10 MHz for a total price of US$10 million in June
2018. The company is part of the Bakhresa Group of Companies (BGC) owned by Said Salim
Bakhresa. Under the terms of the auction, the spectrum frequency licence issued stipulates
coverage obligations that by end 2021, population coverage for broadband services meeting
the specified quality of service level should be equal to 60%; and by end of year 2024,
population coverage should be equal to 90%.
Service is not yet available. Company operates in digital terrestrial television space.
Halotel – 2015 Trading name of Viettel Tanzania Public Limited Company, providing voice, messaging,
data and communication services. Owned by Viettel Global JSC, a state-owned investment
company from Vietnam. The company has placed over 18,000 km of optical fiber cabling.109
Viettel started in Tanzania in 2011 and planned $736 million in investment after the then
president Jakaya Kikwete made a state visit to Vietnam.110 The government wanted to
encourage rural mobile connectivity and by allowing Viettel to operate in the country it
facilitated more competition. According to (Anyanzwa 2019)111 Halotel operates in a specific
niche, targeting rural areas with low-cost products and services.
The licence granted to Halotel required the company to invest in fibre-cable rollout across
the country before it could build in cities.
As of March 2023, Halotel had 8.15 million subscribers amounting to a 13.2% share of all
subscribers.
109 See https://halotel.co.tz/
110 Source: http://viettelglobal.vn/halotel-becomes-driver-of-tanzanias-socio-economic-development.html
111 Anyanzwa, J. (2019). ‘Party Call: Telecoms in East Africa Set for Mergers, Acquisitions’. The East African, 29 May. Available at: https://www.theeas-
tafrican.co.ke/business/Telecoms-in-EastAfrica-set-for-mergers-acquisitions/2560-5136200-10k196iz/index.html
34
Operator, service
Description
launched
Smart – 2014-19 Smart Telecom launched in 2014 and shut down in 2019 after submitting a business closure
request to the TCRA. The company officially closed its business in June 2019. ‘Smart’ started
its operations in Tanzania after it acquired what used to be known as Benson Informatics.
‘Smart’ was a partnership between Industrial Promotion Services (IPS) Kenya (a division
of the Aga Khan Fund for Economic Development) and Cyprus-based, Russian-owned
Timeturns Holdings (49%).
It launched its network in March 2014, utilizing the existing nationwide technology-neutral
mobile licence awarded in 2009 to CDMA operator Benson Informatics and switched on a
4G LTE service in August 2015.
‘Smart’ managed to raise its customer numbers from only 528 subscribers it had inherited
from Benson Informatics as of June 2015. TCRA figures show that in October 2015, ‘Smart’
reached its peak, with a total of 1,800,169 subscribers, accounting for 4% of the market
share of 38 million subscribers at the time.
Subscriber volumes declined after 2015, reaching 1,385,390 subscribers in January 2016
before going further down to 853,390 in October 2016 and further dropping to 132,404 in
March 2019.
Smile – 2013 Smile Communication Tanzania Ltd (Smile TZ) was registered in July 2009 and began
commercial operations in May 2013. The company is a subsidiary of Smile Telecoms Holdings.
Smile’s first country of operation in Africa was in Tanzania and it was the first company to
launch commercial 4G LTE services, starting in Dar es Salaam. Today its coverage extends to
regions covering the largest cities.
Smile Telecoms Holdings Limited is a private company limited by shares incorporated under
the laws of Mauritius with head office in London, England.
As of March 2023, Smile was the smallest mobile operator with only 15,171 subscribers
(mainly business), amounting to a 0.01% share of all subscribers.
Tigo – 1994 Tigo started as a joint-venture between Millicom International Cellular (MIC) based in
Luxembourg, Ultimate Communications Limited of Tanzania, a private company, and Tanzania
Posts and Telecommunications Corporation owned by the government. In February 2004,
MIC’s parent company Millicom International Cellular took over the 26% stake held by
the government to then control 84% of the company’s equity. In 2006 MIC bought out the
remaining 15.9% held by Ultimate to have 100% ownership.112
In April 2022 MIC announced it was divesting its entire holding of MIC Tanzania which had
also merged with Zantel (see below) to pan-African group Axian and Tanzanian entrepreneur
Rostam Azizi in a deal worth around $100 million.113
As of March 2023 Tigo is the third largest operator with 16.6 million subscribers, amounting
to a 26.8% share of all subscribers.
Predecessor brands were Mobitel, Simu Poa, Buzz
112 Source: https://www.commsupdate.com/articles/2012/04/24/government-to-launch-probe-into-tigo-tanzanias-ownership-paper-says/
113 Mr. Rostam Azizi is a wealthy individual who is said also to own around 18% of Tanzania Vodacom (see https://en.wikipedia.org/wiki/Rostam_Aziz)
See also: https://www.barrons.com/articles/millicom-tigo-completes-multi-year-africa-exit-with-tanzania-sale-01649163970
35
Operator, service
Description
launched
TTCL – 1994 (mobile) Tanzania Telecommunication Corporations formerly known as Tanzania Telecommunications
Company Limited (TTCL) was established by an Act of Parliament ‘The Tanzania
Telecommunications Company Incorporation Act of 1993’.
The Company came into operation following a split of the former parastatal organisation
known as Tanzania Posts and Telecommunications Company (TP&TC) and began operations
on January 1, 1994.
Subsequent to continued economic liberalization policy of the Government of Tanzania, TTCL
was privatized in February 2001, where a Consortium of MSI of the Netherlands (Celtel) and
Detecon of Germany acquired 35% shares of the company from the Government of Tanzania.
In September 2013 Celtel’s Parent Company MTC Group was acquired by Zain International
BV of Kuwait. Along with the Celtel brand, Zain also acquired the 35% share in TTCL. From
the 2005 agreement Celtel and TTCL were considered separate legal entities and Zain had
no direct control over the companies’ operations. In June 2010, Bharti AirTel purchased
mobile operations in 15 African countries from Zain and inherited the stake.
In 2016, Bharti AirTel agreed to sell its 35% shareholding back to the Government of
Tanzania and TTCL is wholly owned by the Government.
As of December 2021 TTCL had 1.8 million subscribers, amounting to a 3.4% share of all
mobile subscribers.
The main focus of TTCL is overseeing the government’s national fibre network NICTBB. The
high capacity fibre ring is part of the National ICT policy. TTCL also provides a fixed network
service serving around 75,000 lines.
As of March 2023 TTCL had 1.5 million mobile subscribers, amounting to a 2.4% share of all
subscribers.
Figure 6 NICTBB Network Map114
Vodacom – 2000 The company was created in late 1999 as a subsidiary company of Vodacom based in
South Africa. Vodacom Tanzania was a joint venture between the Vodacom Group, with
approximately 65 percent, with the remaining 35 percent owned by Tanzanian shareholders.
Vodacom Tanzania announced officially its commercial operations in August 2000 and
became the largest mobile operator in the country within one year of launch. It remains the
largest mobile communications network operator in Tanzania.
Vodacom Tanzania was licensed in Tanzania with its local partners Planetel Communications
holding 36% and Caspian Construction holding 16%. Later Planetel decreased its stake to
16%, while Caspian’s increased to 19%.
Vodacom Tanzania was listed on the Dar es Salaam Stock Exchange on 15 August 2017.
Vodacom Tanzania and its subsidiaries (together ‘the Group’) are majority owned by
Vodacom Group Limited (75% holding), which in turn is majority owned (about 60%) by the
Vodafone Group based in the United Kingdom.
As of March 2023 Vodacom was the largest mobile operator with 18.6 million subscribers,
amounting to a 30.0% share of all subscribers.
114 Source: http://www.nictbb.co.tz/map.php
36
37
Operator, service
Description
launched
Zantel – 1999 (2005 Zantel is the brand of Zanzibar Telecom Limited which started in 1999 with a minority
on the mainland). Since shareholding held by Etisalat (Emirates Telecommunications Corporation) headquartered in
2021 merged with Tigo Abu Dhabi.
and re-branded as Tigo
In 2006 Etisalat became the majority owner with a 51% share and by 2014 had a 65%
share. Of the remaining 35%, Meeco International of Tanzania owned 17% of Zantel and the
government of Zanzibar, a semi-autonomous nation in a political union with Tanzania, held
18% of Zantel.
In 2015 Millicom International Cellular (MIC) announced that, having received the necessary
regulatory approval, it had acquired an 85% stake in Zanzibar Telecom (Zantel) with the
government of Zanzibar retaining 15%.
In April 2022 MIC announced it was divesting its entire holding of MIC Tanzania including
Tigo (see above) and Zantel to pan-African group Axian and Tanzanian entrepreneur Rostam
Azizi.
As of March 2022 Zantel’s subscriptions have been combined with Tigo. As of December
2021 Zantel had 1.1 million subscribers, amounting to a 2.0% share of all subscribers at that
date.115
115 TCRA (2021) “COMMUNICATIONS STATISTICS Quarter 2 - 2021/2022 December 2021” at https://www.tcra.go.tz/uploads/text-editor/files/
Communications%20Statistics%20December%202021%20(1)_1645554800.pdf
38
Appendix B Government agencies in the communications
space
Organization Governance and policy objectives
UCSAF – 2006 The Universal Communications Service Access Fund
The UCSAF was established under the Universal Communications Service Access Act
2006.116 Under the Act, the Minister responsible for communications published Regulations
which govern the operations of the Fund came into force in April 2009. The Fund has a Ten
(10) Member Board which was inaugurated in October 2010.
Towards achieving a vision of: “Achieve fully accessible and equitable communication
services in Tanzania” the Universal Communications Service Access Fund (UCSAF) aims to
strengthen the capacity to partner with telecommunications service providers in the country
to ensure availability of communication services in rural and urban under-served areas.
UCSAF is focused on Identifying areas that may be subsidized with funds of the Fund or
that qualify for allocation of subsidy by the Fund; with the aim of ensuring that all citizens
have access to communication services that can contribute significantly to the economy of
individuals and the country.
Section 6 of the UCSAF (2006) Act allows for the “set the bidding conditions for the awarding
of funds”.
The UCSAF is funded by a universal service levy on communications licensees based on gross
turnover, not exceeding 1.5%. Additionally, funds may derive from the government and other
entities, such as World Bank lending programmes.
Since inception 1,235 wards have received UCSAF public support to extend mobile
communication services. As of September 2021, 829 were completed and the construction
of towers in 406 wards was underway. Some 12.9 million people in 3,292 villages have
benefitted from the projects. Since 2012, 11 award processes for public support have taken
place.117
116 See https://www.ucsaf.go.tz/uploads/documents/en-1632143155-SHERIA%20ZA%20MFUKO.pdf
117 Reported in the East African 18-24 September 2021 at https://www.ucsaf.go.tz/uploads/press_releases/en1654515343-EA%20-%20Tele-
comm%20Day%20(3)%20final%20copy.pdf
39
TCRA – 2003 Tanzania Communications Regulatory Authority
The Tanzania Communications Regulatory Authority (TCRA) is a governmental body
responsible for managing the telecommunications, broadcasting and postal sectors in
Tanzania. It was established by the Tanzania Communications Authority Act No. 12 of 2003
to manage electronic and postal services and communication frequencies in the United
Republic of Tanzania. The Authority came into operation on November 1, 2003 combining
the former Tanzania Communications Commission (TCC) and the Tanzania Broadcasting
Commission (TBC).
The main strategic objectives of the TCRA are:
• To modernize TCRA activities using appropriate technology; strengthen quality research
into controlled services, staff capacity and competence.
• Promoting efficiency; reliability and ensuring the security of the communication
infrastructure and its use.
• To promote the efficiency and affordability of telecommunications services and increase
access to postal and ICT services in areas that are not adequately serviced and those
that are without services.
• Protect stakeholders’ interests and promote awareness of their rights and
responsibilities.
• Monitor the performance of managed services and ensure compliance with the Laws and
Regulations to the appropriate standards.
• Coordinate the implementation of national, regional and international sector
commitments.
Fair Competition Established under the Fair Competition Act (FCA) of 2003
Commission (FCCT) – Through the powers conferred upon it by the FCA, the FCC deals with all issues of anti-
2003 competitive conduct, abuse of dominance and has provision for curtailing mergers
and acquisitions if the outcome is likely to create dominance in the market or lead to
uncompetitive behavior. It also carries the consumer protection regime administered by a
department at arms-length with the Commission. The adjudications of consumer related
cases are done in courts.
The FCC is given powers to gather information, to conduct investigations and to impose
sanctions for violations of the law. Its power in regard of communications is concurrent with
the TCRA.
Bank of Tanzania (BoT) Bank of Tanzania (BoT)
– 1966
The Bank of Tanzania is the Central Bank of the United Republic of Tanzania established by
the Bank of Tanzania Act of 1965. It became operational on 14th June 1966.
The legal and regulatory framework that governs mobile payments in Tanzania includes the
Bank of Tanzania Act 2006 (S. 6) (“the BoT Act”), section 6.
Since 2012 the BoT has taken a progressive approach to designing a regulatory framework
that has considerably contributed to the growth of a competitive market where MNOs are
contributing to the progression of electronic finance systems.
National Payment Systems (NPS) Act 2015 to regulate the Payment Systems (NPS) and
promote sound banking and financial system that includes the payments, clearing and
settlement systems conducive to the development of the national economy. The Act is
further supported by the powers under the Bank of Tanzania Act 2006 which broadly allows
the Bank to supervise the payment systems services and products offered by both banks and
non-bank institutions.
The Bank of Tanzania issues circulars and guidelines to ensure that the payment systems
operations in the country are running smoothly. Various stakeholders are also involved in the
formulation of payment systems Acts and Regulations.
40
Appendix C Auction format: multi-round reverse clock
auction detailed rules
In this appendix we set out in detail the rules of the MRRA delivered via an EAP to be piloted in Tanzania. The
model described here is comprehensive and should probably be simplified for application in the Tanzanian
context. For example, MRRA could be applied without weights (see below) and could also be applied without
package bids (also described below). These simplifications are easy to implement and would result in a more
straightforward auction, require less capacity building and less bidder education. On the other hand, such
simplifications are likely to reduce the potential for price discovery to be fully effective.
To keep things simple, we recommend that the format is applied separately across each of the regions
identified in the pilot. It is presumed at the beginning of the auction that demand for funds exceeds the budget
in each region.
The reverse auction format proposed is a multi-round auction taking place in each region. The auctioneer
controls the price level in the auction by announcing a clock percentage price point that decreases round-by-
round.
The clock percentage price point multiplies into the MAS or reserve prices of areas in a region and depending
on bids expressed, this is used to elicit whether demand for funds exceeds budget available. As the clock
percentage price point declines, the reverse auction in a region will reach a round when the budget clears for a
region, i.e. when demand for funds is no greater than the budget available. This occurs in the round in which the
estimated total cost of the bids (demand for funds) at the current round’s clock percentage is no greater than
the budget assigned to the region.
In the round which clears a region, the areas in the region will be assigned to winning bidders. Any unassigned
areas for which bidders have expressed bids will, after the clearing round, be subject to further rounds of
bidding until these have also been assigned.118 Once all bids have been processed in a region and the budget
exhausted, the auction ends.
C.1. Before the auction starts
1. To participate in the auction applicants must qualify by meeting criteria set out by policy maker e.g. licensed telco,
tower company, years of experience, etc. This is achieved by completing a detailed pre-qualification form. The purpose
of pre-qualification is to assess the technical and managerial competency and financial soundness of the interested
bidders. This feature is typical in all universal service fund award schemes.
2. To qualify for the auction, applicants must specify which regions they are interested in bidding to supply service and
the total number of areas119 in each region they are interested in bidding to supply service. If an applicant qualifies
as a bidder, it is allowed only to bid in regions specified on their application form. The information about regions and
areas supplied by each applicant will not be made public until after the conclusion of the auction.
3. Applicants must specify the service category for each region it intends to submit bids for. It is possible to specify
multiple categories of service.120 A weighting system is used based on the category of mobile generation (G) and
the category of infrastructure sharing (S): the total weight of a bidder in a region will comprise each G+S on their
application form. The purpose of the weight is to favour 4G over 3G and to favour infrastructure sharing that
minimises competition related issues.
118 As described below, such further rounds will involve carry-forward bids.
119 The applicant only needs to indicate how many areas in a region and does not need to name areas. For example, suppose region A has 10 areas. The
applicant could indicate it is interested in bidding on up to 5 areas in region A. If the applicant qualifies to bid, it can bid on any areas in region A
but may not present more than five areas in its bids in any round during the auction. In accordance with recommendation 5 in chapter 7, the bidder
would need to pay an application fee of 2% multiplied into the five highest MAS values in the region. For example, suppose the MAS values were 10
for eight wards and 20 for two wards. The application fee for indicating five sites would be 2% × [20+20+10+10+10]=1.4.
120 A simplified version of the auction would set out uniform service and not require the application of weights as set out here.
41
• 4G has a G=0 weight
• 3G has a G=20 weight
• Infrastructure sharing (Multi Operator Radio Access Network (MORAN)) has a S=0 weight
• Multi Operator Core Network (MOCN) sharing has a S=10 weight
• Non-infrastructure sharing has a S=30 weight
4. Each area in a region has a reserve price R (i.e. a maximum allowable subsidy (MAS) in an area) that is pre-
determined by consultation with stakeholders and through financial modelling and/or benchmarking. The level of
R per area in each region will be made clear on the pre-qualification form.
C.2. Bidding rules
5. The auctioneer announces at the start of each round in each region a clock price point percentage P.
6. The opening clock price point percentage P in the auction is equal to the largest 100+G+S indicated in the qualified
applicants’ bidding forms at pre-qualification.
7. A bidder may submit a bid during a round from round 2 onwards at the clock price point percentage announced
or at another price point percentage between the clock percentage in the round and the previous round’s clock
percentage (up to two decimal points). If the previous round clock price percentage were 150 and the current
round is clock price percentage is 140, a bidder could, for example, submit a bid at a price percentage 145.97 for
an area or group of areas in a package bid (see below). The bidder is allowed to attach different price percentages
to multiple bids submitted in a region in the same round. For example, a bidder might bid at 145.97 for area A and
140 for area B.
8. The auctioneer sets the clock price point percentage P in round 1 at a decrement no lower than 5 than the
opening P. If the opening P were 130, the auctioneer might set the round 1 price point percentage at 120.
9. The implied support for a bid in an area is given by the formula: , here P is for a given price point
percentage expressed by the bidder that may equal the clock price point percentage announced by the auctioneer
or be at a higher price point percentage up to the previous round’s clock price point percentage. A bidder with
positive weights (G+S>0) against a bidder with zero weighting (G+S=0) will receive less support. This provides an
incentive to compete with a lower weight and may also reflect lower costs.
10. The bidder’s total implied support for its bids at the round’s clock percentage cannot exceed the budget for the
region.
11. If the auction does not clear in the most recent round, the auction will carry forward all the bids that were
submitted at the clock percentage price into the current round from the previous round. Bidders may express
new bids for an area that they bid for in the previous round at a lower percentage price point than the clock
percentage price in the previous round; these new bids override carry-forward bids.121
12. The auctioneer determines a decrement in P in the current round in each region after round 1 whenever the
maximum implied support bid for each area in a region exceeds the budget B available in the region at the price
point percentage in the previous round. (This calculation is shown further in a numerical example in Appendix D.)
Typically the decrement would lie be between 5 and 20.
13. A bidder can submit a bid for an individual area in a region in a round but cannot submit more individual bids than
the total number of areas expressed for the region on the bidder’s application form.
14. A bidder can submit a package bid for a group of areas in a region in a round, but the areas included in a package
bid cannot feature in any other package bid or individual bid made by the bidder in the same round. All areas in a
package bid must be expressed at the same G+S.122
15. In a package bid, the bidder must apply a minimum scale percentage (MSP). The MSP states that any subset of areas
in the package can only be assigned to the bidder if the sum of the reserve prices for the subset as a percentage of
the total of reserve prices for all areas in the package exceeds the MSP. The auctioneer may impose a maximum
figure for the MSP at or below 100% before the start of the auction, e.g. 75%. Subject to the maximum MSP, a
bidder may vary its MSP attached to a package bid across different rounds. The idea of the MSP is to provide
a bidder with protection against being exposed to the risk of winning a small subset of its package bid – the
presumption being synergies (cost savings and/or revenue yield) across the package may be considerable.
16. If a bidder submits a bid for an area in an earlier round, the bidder must attach the same G+S in the current round
to the same area as in the earlier round.
17. The price point in a bid must not be less than G+S+1. Thus, the implied support amount must be at least 1% of the
area’s R to be acceptable.
121 If bidders were to stop bidding in round n, the auction would clear, and assignment would be based on a percentage price lying between the rounds
n and n+1.
122 A simplified version of the auction could be undertaken without package bids. However, packaging allows bidders to better tailor their demands.
42
C.3. Activity rules
18. A bidder’s activity is the total of the implied support for all of the bidder’s bids in a round.
19. If the clearing round has not occurred, a bidder’s activity cannot exceed the activity in the previous round.123 If
a bidder’s activity in round 5 were 250, the bidder cannot exceed this activity in round 6 and any other future
round.
20. If the clearing round has not occurred, a bidder’s activity for areas not included in the previous round cannot
exceed a maximum percentage M of the bidder’s total implied support at the previous round’s clock percentage.
An auction for a region with 10 areas has a clock percentage 70% in round 3, decreasing from the clock
percentage 80% in the previous round. If a bidder submits a package bid with zero weight at the clock percentage
70% for areas {1,2,3,4} and submits a bid for area 10 with zero weight at percentage price 75%, the total implied
support at the clock percentage is defined by the sum of the reserve prices for areas 1,2,3 and 4 multiplied by the
clock percentage, say 400×0.7=280. (The bidder’s activity is 280+0.75×100=355, where 100 is the reserve price
for area 10.) In round 4 the bidder’s activity cannot exceed 355, and the bidder cannot bid on any area or areas in
{5,6,7,8,9} if the bidder’s activity for these bids is greater than M×280. If M=10%, this implies the bidder cannot
switch bids where the activity exceeds 28.
21. The value of M in round 2 will be determined by the auctioneer and announced to qualified bidders at the start of
the auction. The auctioneer will start with a value for M in round 2 no higher than 30 and will not set a value for M
below 10 in any subsequent round. The value of M in the current round for any round from 3 onwards will not be
above the value of M in the prior round.
22. After the clearing round no switching is allowed. A bidder can only bid for an area that it bid for in the clearing
round at the percentage clock price and the area is unassigned.
23. After the clearing round a bidder can only bid for a package if it is the same or a subset of a package for which it
bid at the previous round’s clock percentage and the area does not contain any areas that have been assigned.
24. After the clearing round partially assigned packages having one unassigned area are such that bidders can only
bid for the area individually.124
C.4. Processing the clearing round
25. The clearing round occurs when the maximum demand for support across all areas measured at the clock
percentage is no greater than the budget available.125 The auction should follow a process in the clearing round in
the following order:
• Determine which areas are to be assigned to winning bidders with any remaining any areas being unassigned
and open to further bidding in subsequent rounds.126
• Calculate the clearing price point as the highest price point not exceeding the clock percentage in the
previous round, taking account of second prices, such that the total support amount is no greater than the
budget.
• Allocate the support payment for each assigned area using a second-price rule.
• If the clearing price point equals the previous round’s clock percentage, any carried-forward bids from the
previous round are assessed for possible assignment.
26. Determining the assignment of areas: bids should be assessed in order of ascending price. Bids at the same
price point are assessed in order of ascending G+S weight. Ties should be broken by applying pseudo random
numbers.127 An area can be assigned if, at a given G+S weight, it has not previously been assigned and if no other
bidder bid for it at the round’s clock percentage at the same or lower G+S weight. An area can be assigned if, given
other areas already assigned and the price point of the bid, the cost of the area assignment does not result in
total aggregate cost (see (a)-(c) below) exceeding the budget for the region. If the bid is part of a package bid, the
areas that are available to be assigned meet the MSP. All bids are processed, even if the budget does not allow an
assignment.
123 Delaying bids by a bidder to discover something about the demand of other bidders is therefore costly for the bidder as it necessarily lowers activi-
ty and restricts future bidding options.
124 Suppose bidder 1 bid for a package including areas {1,2,4} and bidder 2 bid for area 4 in round n. Clearing occurs in round n and areas 1 and 2 are
assigned to bidder 1 and areas 3 and 4 are unassigned. Area 4 can be bid for individually and no bidder can submit a package bid including area 3
with area 4.
125 It is measured at the clock percentage as this reflects the demand for funds at the lowest level of support announced. If all the bidders in a region
were to bid for support above the clock percentage, then the auction would clear as demand for funds at the clock percentage would be zero.
126 Carried-forward bids ensure unassigned areas are assigned should no further bids be received.
127 Pseudo random numbers are generated by algorithms that provide near random number generation.
43
The system calculates the aggregate cost at a given price point P as:128
• For areas not yet assigned that were bid at the clock percentage, the implied support amounts at the clock
percentage counting each area once at the costliest scenario, i.e. at the lowest weighted bid.
• For all areas already assigned and received no other bids less than P, the support amounts implied by P.
• For all areas already assigned which received other bids below P, the support amounts implied by the greater
of the price point bid by the bidder receiving the assignment and the lowest price point at which any other
bidder bid for the area.
Example Here we illustrate how the aggregate cost is estimated. Assume the budget is $220 and there are six
areas 1,2,3,4,5 and 6. There are two bidders who submit bids of weight G+S=0. In round 9 the auction clock
percentage is 75% and in round 8 it was 80%. Table 14 below shows the bids submitted in round 9.
Bidder 1 at G+S=0 Bidder 2 at G+S=0
1 at 75% Package bid {4,5,6} at 76% with MSP 50%
2 at 75%
3 at 77%
Table 14 Evaluating aggregate cost
Round 9 clears the auction as the aggregate cost of bids at the clock percentage 75% is $150<$240 budget.
The bids submitted at the round clock percentage 75% are considered first. Areas 1 and 2 are assigned to bidder 1
as there are no other bids at the clock percentage for these areas.
Bidder 2’s bid is considered next as the percentage price point increases to 76%. At this price percentage, the
estimated aggregate cost of the package bid is $228, plus the cost of the areas assigned to bidder 1 at 76%
is $154. The total cost is therefore $382>$240. Hence the package bid areas cannot be assigned together.
Allocating any two areas in the package bid also leads to excess demand for funds. Allocating any one area in the
package bid fails the MSP constraint.
Bidder 1’s bid for area 3 is considered next at 77%.The aggregate cost of areas 1 and 2 that have already been
assigned at 77% is $154. The cost of the bid for area 3 is $77. The total cost is therefore $231 which is below the
budget of $240. Thus area 3 is assigned to bidder 1.
All bids have been processed and the given percentage price point P=77% is below the clock percentage 80% of
round 8. Hence, there are no carried-forward bids to consider and the auction proceeds to calculating P*. Each
assigned of the three areas has a cost [(P-0)/100]×100, so aggregate cost is 3×P.
P* must be no greater than 80% and no less than 77%. It is the highest value within the range permitted that
allows funds to be disbursed without exceeding the budget $240. Therefore 3P≤240 implies P*=80%.
27. Calculating the clearing price point: Once the system has computed which bids in the clearing round have been
assigned it calculates what is called the clearing price point P*. This is the highest P (rounded to 0.01%) that is no
greater than the previous round’s clock percentage such that the aggregate cost is no greater than the budget.
28. Support payment determination for bids submitted in the round: For areas assigned to a bidder in the clearing
round that received no other bids below P*, the support payment is the amount implied by P*. In our numerical
example in Appendix D, areas 2, 4 and 5 are all assigned with support payments determined at P* – as each of
these areas did not receive a bid below P* from a non-winning bidder. For areas which did receive bids from other
bidders below P*, the support payment is based on the G+S of the assigned bid and then the greater of (1) that
implied by the price point bid by the winning bidder and (2) the support implied by the lowest price point at which
any other bidder bid. In our example, area 3 is assigned to bidder 1 in the clearing round, having bid at 90% at
weight G+S=0. However, in the clearing round bidder 3 with weight 20 bid at P=95<P*=95.56.
128 The key here is the assumption that the value of P is given. Thus, for any given P the system calculates according to the instructions shown in (a)-(c).
In a numerical example in Appendix D we expand on this point.
44
Therefore the determined support payment for bidder 1 is 95%×20=19.129
29. If the clearing price point P* is less than the previous round’s clock percentage, bid processing ends and no
carried-forward bids are considered. Otherwise, when P* equals the clock percentage of the previous round, the
carried-forward bids of the previous round would be processed.
30. After the clearing round: if there are areas that have not been assigned there will be further rounds if funds
permit this. For example, areas unassigned include any that received multiple bids at the clock percentage in the
clearing round. This latter point means that bids carried-forward or new bids made will be processed in ascending
order of their G+S weight first, then percentage price.130
C.5. Information policy
31. The information made available to bidders is critical to the success of the auction. As is conventional in well-
designed modern multi-round auctions, the identity of eligible bidders is kept anonymous until the auction
process has finally ended. Thus eligible bidders will be assigned anonymous labels. Eligible bidders will only know
that should an auction take place in a region, there is more demand for subsidies than funds available.
32. Limiting information about participants is intended to minimise gaming. In other words, bidders should focus on
submitting bids that reflect their estimates of required support amounts, rather than place bids based on that
they perceive other bidders are doing.131
33. It is in the interest of the auctioneer to reveal as much information about demand for support payments as
possible without compromising the competitive integrity of the auction. The purpose is to facilitate price and
package discovery which is desirable for efficiency. Tables 15-17 below summarise the information disclosed to
each bidder during every round up to and including the clearing round.
Within round information policy Description
A. Clock percentage The opening clock percentage in each region is made public before the auction
starts. Before the start of each round, the clock percentage for the round is shown
to all bidders participating in the bidding in a region
B. Implied support for bids The total sum of the implied support amounts across all the bidder’s submitted
in a region bids in each region in a round
C. Implied support for bids The total sum of the implied support amounts across all the bidder’s submitted
at the clock percentage in bids made at the round clock percentage in each region – where C≤B for each
a region region
D. Areas bid in each region The number of areas bid in each region by the bidder
E. Areas at the clock The number of areas bid in each region by the bidder at the round clock
percentage in each region percentage for the region
F. Maximum region activity The total amount of implied support the bidder may submit in the round for each
region based on activity in each region in the previous round. In each region, the
number of areas in the bidder’s bids cannot exceed the number of areas listed on
their application form
G. Switching areas activity For each region the maximum activity allowed for bids on areas not bid for in the
constraint previous round based as a percentage of bids submitted in the previous round’s
clock percentage
Table 15 Within round information policy
129 This requirement ensures that the payment is always equal to the second price bid if the budget allows. Although bidder 3 in our example below bid
overall less when account is taken of its weight 20, the second price is calculated by expressing any lower bid at an equivalent weight to the winning
bidder. Had bidder 3 submitted a bid at P below P* but above that of bidder 1 (e.g., bidder 3 say bid at 92% with a weight of 20, the round would still
clear and the implied support payment for bidder 1 would have been determined as 97%×20=18.4 and the clearing price point P*=96.57). Note
that the higher value for P* arises because bidder 3 demands less support at 92% and hence less budget is provided to bidder 1 for area 3, leaving a
residual that can be allocated across other areas.
130 Note this is the reverse of the sequencing order in the clearing round.
131 It is well understood in auction theory that truthful bidding is more likely to result in efficient outcomes in a well-designed auction process.
45
34. After the completion of a round before the clearing round. Table 16 below indicates the information policy in
regard of each bidder.
End of round information policy Description
A. Area demand bids Number of bids for every area in each region in which the bidder is eligible to
bid, expressed as 0,1,2, etc. The bidder may receive information about areas for
which it is ineligible to submit bids.
B. Lowest weight In each region in which the bidder is eligible to bid, for the areas which received
1 or more bids the lowest G+S weight amongst those bids
C. Aggregate cost In each region in which the bidder is eligible to bid, the difference between
the aggregate cost at the clock round percentage and the region’s budget (this
provides a useful indication of the likelihood of the next round being a clearing
round for a region)
Table 16 End of round information policy
35. At some round during the auction the auctioneer will announce in a region the clearing round. The information
policy in regard of each region for each bidder in the clearing round and thereafter is shown in Table 17 below.
Clearing round and subsequent Description
rounds information policy
A. Areas unassigned At the end of the clearing round in a region those areas which have not yet
been assigned e.g. as a result of multiple bids at the same G+S lowest weight
(taking account of any carried-forward bids)
B. Areas assigned At the end of the clearing round the areas in a region that have been assigned
(at this stage each bidder would know areas it has been assigned but will not
know who has been assigned other areas that have been assigned)
C. Minimum assigned support At the end of the clearing round, for each area assigned, a bidder will know the
minimum implied support which will be no less than the bidder’s demand for
support
D. Maximum assigned support At the end of the clearing round, for each area not yet assigned, bidders will
know the maximum implied support available
E. The clearing point After processing all the bids at the end of the clearing round, announce the
percentage clearing point percentage
F. The clock percentage For areas not yet assigned, the clock percentage will be announced for the
round after the clearing round and in each round thereafter until such areas
have been assigned
Table 17 Clearing round and subsequent rounds information policy
36. Once all areas have been assigned and subsidies calculated, the auctioneer shall make public the information
about the winning bids.
Appendix D provides a detailed numerical example illustrating how these rules could work in practice.
46
Appendix D: Numerical illustration of multi-round reverse
auction
1. The way the auction format works is illustrated in a simple numerical example below.
2. There are five areas available in a region which has been allocated a total funding budget of B=85. The support
available in each area cannot exceed the maximum allowable subsidy amounts shown in Table 18 below. The
budget of B=85 is 85% of the TMAS=100 and is compliant with Recommendation 4 in Chapter 7.
Area Maximum allowable subsidy
1 10
2 20
3 20
4 25
5 25
Table 18 Five areas and maximum allowable subsidy
3. There are three bidders B1, B2 and B3. Bidders B1 and B2 applied to supply 4G service and offer MORAN
infrastructure sharing in the region, resulting in weights G+S=0. Bidder 3 applied to supply 3G service and offer
MORAN infrastructure sharing, resulting in a weight G+S=20.
4. The opening clock percentage is announced before the auction as 120%. This is based on the highest G+S weight
which is associated with B3’s application. The switching percentages are set at maximum 20% in round 2 and
maximum 10% up to the clearing round.
5. Table 19 shows the round-by-round auction outcome. The auctioneer applies a maximum MSP of 75% for
package bids. The bidders submit package bids at 50% MSP. The highlighted row is the clearing round.
Aggregate
Round Clock B1 Bids B2 Bids B3 Bids Assignments and
Cost at Clock %
# % G+S=0 G+S=0 G+S=20 Support Amounts
minus Budget
2 at 110% &
1 110 {1,2,3,4} at 110% 1 at 110% 12.5
{3,4,5} at 110%
2 100 {1,2,3,4} at 100% 1 at 100% {1,3,4,5} at 100% 10
B1 assigned 2 for
{1,2,3} at 90% & 4 {1,4,5} at 90% & 3 19.31 & 3 for 18.4
3 90 1 at 90% -5
at 98% at 92% B3 assigned 4 for
19.14 & 5 for 19.14
4 80 1 at 85% 1 at 82% 1 at 80% B2 assigned 1 for 8.5
Total support
Budget
85 amount 84.5
Table 19 Illustrative auction round-by-round
1. Round 1 All bidders submit bids at the clock percentage 110%. B1 submitted a package bid for four
areas {1,2,3,4}. B2 bid for area 1 at the clock percentage 110%. B3 submitted a bid for area 2 at 110%
and a package bid at 110% for areas {3,4,5}.
47
Table 20 below shows how the bidders’ activities and aggregate cost are calculated from these bids.
Implied support for Implied support Implied support
Area Maximum
B1 at 110% for B2 at 110% for B3 at 110%
1 10 10 10
2 20 18 20
3 20 18 20
4 25 22.5 25
5 22.5 22.5
Total bids by each
75 10 81
bidder (activity)
Aggregate cost at 110% 97.5
Table 20 Round 1 bidder values and aggregate cost
2. Round 2 All bidders submit bids at the clock percentage 100%. B1 submitted a package bid for four areas {1,2,3,4}.
B2 bid for area 1 at the clock percentage 100%. B3 submitted one package bid for areas {1,3,4,5}.
Table 21 below shows how the bidders’ activities and aggregate cost are calculated from these bids.
Area Implied support for Implied support for Implied support for Maximum
B1 at 100% B2 at 100% B3 at 100%
1 10 10 8 10
2 20 20
3 20 16 20
4 25 20 25
5 20 20
Total bids by each
75 10 64
bidder (activity)
Aggregate cost at 100% 95
Table 21 Round 2 bidder values and aggregate cost
The bid activity of each bidder cannot exceed their activity in round 1. Bidders B1 and B2 submit activities equal
to that in round 1, B3 submitted a reduced bid activity at 64. Further, B3 presented a switch in demand, shifting
demand from area 2 to area 1. Under the rules, this switch is allowed only if the activity associated with area 1
(the demand switch) is no greater than 20% of B3’s activity of 81 in round 1. As the activity of B3 for area 1 is 8
and 20% of its activity in round 1 is 16.2, the demand switch is allowable. If the demand switch constraint were
10%, as it is in round 3 onwards, it would still be allowed.
Round 2 is not a clearing round as the aggregate cost 95 at the clock percentage100% exceeds the budget 85 by
10; hence the round does not clear. The auctioneer therefore lowers the clock percentage in the next round to
90%. B3 submits the highest bid activity. As round 2 is not a clearing round, a bidder’s activity in round 3 cannot
exceed its activity in round 1.
3. Round 3 All bidders submit bids at the clock percentage 90% but B1 and B3 also include bids at an intra-round
percentage rate above 90%. B1 submitted a package bid for three areas {1,2,3} and a bid at 98% for area 4. B2 bid
for area 1 at the clock percentage 90%. B3 submitted package bid for areas {1,4,5} at 90% and a bid for area 3 at
92%.
48
Table 22 below shows how the bidders’ activities and aggregate cost are calculated from these bids.
Area Implied support for Implied support for Implied support for Maximum (only bids
B1 at 90% [activity B2 at 90% B3 at 90% [activity at 90% considered)
for bids above 90% for bids above 90%
shown in parenthesis] shown in parenthesis]
1 9 9 7 9
2 18 18
3 18 (14.4) 18
4 (24.5) 17.5 17.5
5 17.5 17.5
Total bids by each
69.5 9 56.4
bidder (activity)
Aggregate cost at 100% 80
Table 22 Round 3 bidder values and aggregate cost
The bid activity of each bidder cannot exceed their activity in round 2. Every bidder lowers its bid
activity in round 3.
Round 3 is the clearing round as the aggregate cost 80 at the clock percentage 90% is less than the
budget 85.
4. Assignment determination. The bids are considered in ascending order of percentage point. Bids at the
same percentage point are processed in ascending order of the S+G weight. The bids at 90% by B1 and
B2 are considered first. Assume that B1 is chosen by a pseudo-random number generator.
• B1 submitted a package bid with a MSP 50%. Area 1 cannot be assigned because B2 also submitted a bid at
zero weight at 90%. Areas 2 and 3 and are available to be assigned. (B3 did bid for Area 3 but at a percentage
point above the clock percentage.) Areas 2 and 3 of B1’s bid exceed the MSP and therefore these areas can be
assigned to B1.
• B2’s bid is considered next. This was for Area 1. This cannot be assigned because B1 also bid for Area 1 at
90% with a zero weight.
• B3 submitted a 90% package bid {1,4,5} with weight S+G=20 and MSP of 50%. Area 1 is not yet available as
the other bidders also bid for it at zero weight. Areas 4 and 5 are available and meet the MSP. Thus Areas 4
and 5 are assigned to B3.
• B3’s bid for Area 3 at 92% is considered next. This area has already been assigned to B1.
• B1’s bid for Area 4 at 98% is considered next. This area has already been assigned to B3.
5. Price point determination. This is calculated as the highest price point (say in multiples of 0.01%)
between 90% and 100% at which the aggregate cost is no greater than the budget. In our terminology
we defined this as a number P* between 90 and up to 100. The idea is to exhaust as much of the budget
as possible, reflecting bids for the areas made and the assignments made in the clearing round. Table 23
shows the implied price support for a price point P.
49
Area Assignment status Implied support at P, Comment
where 90≤P≤100
1 Not yet assigned 9 This is the implied support at clock percentage 90% at
zero weight which is the most expensive scenario
2 B1 (P/100)×20 This is the implied support at P at zero weight
3 B1 (min[P,92]/100)×20 If P≤92 the implied support is calculated at P for a zero
weight, if P>92 the implied support is at 92% with zero
weight, reflecting the bid made by the second placed
bidder B3 but at the winner’s weight
4 B3 (min[P,98]-20)/100)×25 If P≤98 the implied support is calculated at P for a weight
of 20, if P>98 the implied support is at 98% with a weight
20, reflecting the bid made by the second placed bidder
B1 but at the winner’s weight
5 B3 ((P-20)/100)×25 This is the implied support at P at weight 20
Table 23 Implied price support
6. The highest price point P* in the above table which applies to the assigned areas 2-5 is 96.57%. Table
24 illustrates the maximum support amounts at P*=96.57%. The clearing price point P* is below the
previous round’s clock percentage (100%) and this ends the processing of bids in the clearing round. If
P*=100%, the processing would continue by examining carried-forward bids.132
Area Assignment status Maximum support amount at P*
1 Not yet assigned 9
2 B1 19.31
3 B1 18.40
4 B3 19.14
5 B3 19.14
Total support amount 84.99 rounded to 85.00
Table 24 Price determination
7. Round 4 The clock percentage is set at 80% by the auctioneer. Each bidder can only bid for the
unassigned Area 1. Bidders’ bids from round 3 are carried forward at 90%.
• Scenario 1. B1 submits a bid for Area 1 at 85%. B2 submits a bid for Area 1 at 82% and B3 submits a bid
for Area 1 at 80%. In this instance the bids are processed in ascending order of the S+G weight; after this
they are then processed in ascending price order (this is the reverse of the sorting that takes place in the
clearing round). B1 and B2 have submitted bids at S+G=0 and B3 submitted a bid at S+G=20. B1 and B2 bids
are therefore considered first. B2’s bid is considered first, as it is at a lower price point. As no other bidder
has submitted a lower price bid for Area 1 at S+G=0, B2 is assigned Area 2. The price point determining the
support payment is determined by the second smallest bid, which in this case is B1’s bid at 85%. Therefore B2
is assigned Area 2 and receives support payment 8.5. The final outcome under this scenario is shown in Table
25.
132 Recall, carried-forward bids are the bids made in the previous round at the clock percentage. Processing would consider carried-forward bids for
areas that have not yet been assigned. In the numerical example here, all areas have been assigned or feature competition and will result in further
rounds of bidding. Hence, there are no carried-forward bids to process in the clearing round. In the round after the clearing round (round 4), a
bidder can choose to bid for an area yet to be assigned (Area 1), notwithstanding any carried-forward bid at the previous round’s clock percentage
90%.
50
Area Assignment status Maximum support amount at P*
1 B2 8.5
2 B1 19.31
3 B1 18.40
4 B3 19.14
5 B3 19.14
Total support amount allocated 84.49
Table 25 Scenario 1 post clearing round bidding
• Scenario 2. Only B3 submits a bid at 85%. In this case the bids of B1 and B2 made in round 3
carry-forward at 90%. As there is a tie between B1 and B2, the system would separate the bids
of B1 and B2 by a pseudo-random number.133 Suppose B1 is chosen as the winning bid. Area 1 is
assigned to B1 at 90%. The final outcome under this scenario is shown in Table 26.
Area Assignment status Maximum support amount at P*
1 B1 9
2 B1 19.31
3 B1 18.40
4 B3 19.14
5 B3 19.14
Total support amount allocated 84.99
Table 26 Scenario 2 post clearing round bidding
• Scenario 3. B2 submits a bid of 87% and B3 submits a bid at 85%. It follows B1 submits the
carried-forward bid at 90%. In this case the bid of B2 is processed first as it has a zero weight
and a lower price point than B1. B2 would be assigned Area 1 at 90%, the next-smallest bid at
zero weight. The final outcome under this scenario is shown in Table 27.
133 The system separates ties of carried-forward bids after the clearing round rather than proceeding to a new round. This is because the tied bidders
have not presented new bids and the role played by lowering the percentage price point would be redundant. Bidders who have presented bids
for an unassigned area would know in advance that not submitting a new bid carries a risk that another eligible bidder could outbid them. The
second-price rule provides an incentive to bid truthfully (i.e., bid for support that honestly reflects the cost of providing service).
51
Area Assignment status Maximum support amount at P*
1 B2 9
2 B1 19.31
3 B1 18.40
4 B3 19.14
5 B3 19.14
Total support amount allocated 84.99
Table 27 Scenario 3 post clearing round bidding
8. Round 4 is the final round, as all areas are assigned and all bids have been processed.
Appendix E Electronic auction platforms
52
1. Many government agencies around the world have made use of multi-round auctions to sell and procure goods
and services. The usual approach is for agencies to use an EAP, in which an auction is hosted online over the
Internet (public or private) allowing the auctioneer and bidders to participate remotely. By removing the need to
be physically present at an auction address and for bids and other auction information to be posted electronically,
transaction costs are lowered. Further, remotely hosting auctions enables a complete digital audit trail to be
maintained.
2. An EAP is an application run on a server hosted by an agency or usually a third party supplier. In the specialised
field of EAPs involving multiple goods and sometimes complicated winner determination algorithms, there are a
number of specialized vendors. The following are among the prominent in the field:
Auctiometrix (www.auctiometrix.com)
Auctionomics (www.auctionomics.com)
DotEcon (www.dotecon.com)
Innovative Auctions (www.innovativeauctions.com)
Power Auctions (www.powerauctions.com)
3. The services provided by an EAP specialist include:
• Assessing which auction format best achieves the goals of the client
• Planning the entire award or procurement process
• Drafting the auction rules and ensuring these are compliant with relevant legislation
• Advice on the consultative process with stakehAolders
• Advice and assistance during the pre-qualification process and evaluation
• Production of auction guidelines and manuals
• Ensuring auction rules are fully captured in the software
• Auction training for administrators and bidders
• Deploying and managing auction
• Advice on publication of winners
• Post-auction evaluation
4. A typical contract with an EAP provider enables the client as auctioneer to use an EAP under licence for the
duration of a specific award or procurement process. The licence will usually be written such that the intellectual
property of the EAP is retained by the vendor. The vendor offers its generic platform (e.g. WebBidder is the
generic platform of DotEcon134 and Power Auctions135 offers a SaaS136 model) which it then modifies to meet the
clients’ needs.
5. The end-to-end process for a full-scale EAP can be lengthy. From choosing the EAP supplier through to auction
deployment can take many months. The software that is produced will usually need to undergo both UAT (user
acceptance testing) and pen testing (penetration testing).
6. UAT is where the auction software is tested by the client or a third-party on behalf of the client, to determine
whether it can be accepted or not by the users of the software. In effect UAT is an approval process and is
performed once the functional, system and regression testing137 are completed. UAT validates the software
against the auction requirements and is carried out by entities familiar with the auction requirements.
7. UAT is the last testing carried out before the software goes live and is the last chance for the client to test the
software and measure if it is fit for purpose. It involves testing all electronic forms respond as they should, that
the front-end of the auction and any back-end algorithms are entirely consistent with the detailed auction rules.
This usually requires scripting scenarios and testing for edge cases. In addition to the companies listed above,
134 See https://www.dotecon.com/expertise/auction-software/
135 Power Auctions offers its platform as ‘Software as a Service’ (SaaS), see https://www.powerauctions.com/software.
136 Software as a service is a software licensing and delivery model in which software is licensed on a subscription basis and is centrally hosted. SaaS is
also known as "on-demand software" and Web-based/Web-hosted software. This is how most EAP vendors offer services to their clients.
137 Regression testing is defined as a type of software testing to confirm that a recent program or code change has not adversely affected existing fea-
tures. Regression testing is nothing but a full or partial selection of already executed test cases which are re-executed to ensure existing function-
alities work fine. It ensures that new code changes do not have side effects on the existing functionalities. In other words, the old code should still
work once new code changes are completed.
53
specialized companies such as the Smith Institute in the UK undertake UAT of sophisticated EAPs.138
8. Penetration testing is a security exercise where a cybersecurity expert attempts to find and exploit vulnerabilities
in a computer system. The purpose of this simulated attack is to identify any weak spots in a system’s defences
which attackers could take advantage of. Pen testing is important in high-stakes auctions, especially to guard
against malicious attacks that undermine the system. Specialist companies provide these services and often
deliver their outcomes remotely.
E.1. Recommended EAP for pilot auction
1. In Chapter 7 of this report we have recommended an auction format for deployment as a pilot in Tanzania by
UCSAF. To proceed it is necessary to make use of an EAP. The options available are:
• Develop, build and maintain an EAP
• Procure an EAP under licence, manage and maintain in Tanzania
• Procure an EAP under licence, outsource its management and maintenance
2. Observations of other universal service agencies, such as the FCC in the United States, indicate that option (c)
is the preferred choice. This reflects that the highly specialized skills required are typically unavailable, that the
vendor has greater experience, and it costs less. In this setting UCSAF would manage the auction but interact
with the chosen EAP vendor during the entire process.
E.2. Independent module alongside taneps
3. UCSAF should explore with the chosen vendor the implementation of the EAP alongside the established TANePS
platform.
4. UCSAF should prepare and scope a detailed specification for an EAP in accordance with the auction rules set
out in Appendix C.
138 See https://www.smithinst.co.uk/spectrum/
54
Appendix F Slides presented at mission in dodoma
On 3 August 2022 the slides shown below were presented to government stakeholders at Dodoma. Attendees included
key representatives from UCSAF (including Mrs Justina T. Mashiba, CEO) and MFP (including Mr Eliakim C. Maswi, CEO).
A version of the slide content was also discussed with representatives from the MNOs AirTel, Halotel, Tigo and Vodacom
on 1 and 2 August in Dar es Salaam. Feedback from the stakeholder engagement was positive.
Smart Reverse Auctions
How to achieve the
best possible outcome
for disbursing
universal service
funds
Dr. Chris Doyle
August 3, 2022
55
Purpose
• Policy maker wants to promote regional broadband uptake
• Sets policy – affects licence terms, regulation and may result in
transferring support payments to operators
• Assist service providers by providing financial support
• Demand for financial support may exceed funds available
• How do we resolve this?
Solution
• Administer grants by application or hold a reverse auction?
• A solution should ensure that support payments are targeted effectively
• An auction also lessens some of the bureaucratic burden of disbursement
• Private sector stakeholders see it as a fairer and more transparent
process
Policy recommendation for UCSAF
• Consider deploying a pilot to see whether a multiple round clock auction
can achieve more effective use of public funds
• Could try the auction process in several wards in one region
• Start by estimating realistic supports needed to ensure break-even –
reserve payments
• Auctioneer announces clock percentage
• Bidders can vary bids through rounds and react to information disclosed
during the process
• Can be applied to many other settings where public funds are available
and competition for funds exceeds amounts available
56
Grants: Pros
• Allows bureaucratic oversight of applicants
• Ensures the right entities obtain support
Grants: Cons
• Scoring methods may be challenged
• Susceptible to influence
• Applicants may overstate investment intentions
• Funds may not be fully disbursed
Reverse Auctions: Pros
• Can be designed to ensure available financial support is targeted
effectively
• Can lessen bureaucratic burden of disbursement by making use of auction
‘discovery’
• Private sector stakeholders often see it as a fairer and more transparent
process
Reverse Auctions: Cons
• If badly designed can result in poor outcomes
• May not be needed if demand for support funds below available support
57
International Best Practice
• Auctions have been used by some government agencies around the world
to:
• Assigning radio frequencies
• Allocate financial support for broadband and fibre
• Example: United States, the approach is to allocate support payments by
auction
• August 1, 2019, the FCC adopted a Notice of Proposed Rulemaking (NPRM)
proposing to establish the $20.4 billion Rural Digital Opportunity Fund to bring
high speed fixed broadband service to rural homes and small businesses that
lack it
• Multiple round (clock) auctions used
• FCC Auction #904 disbursed nearly $16 billion
Auction 904
• Bidders won funding to deploy high-speed broadband to over 5.2 million
unserved homes and businesses, almost 99% of the locations available in
the auction
• A total of 180 bidders won auction support, to be distributed over the next
10 years
• Bidding procedures prioritized bids for higher speeds and lower latency
• The auction unleashed robust price competition that resulted in more
locations being awarded at less cost to Americans who pay into the
Universal Service Fund
• 5,220,833 locations assigned support in the auction had an initial reserve
price of over $26 billion over the next decade; through vigorous
competition among bidders, the final price tag to cover these locations is
now just over $9 billion
58
Auction #904 FCC
59
Bibliography
Armstrong, Sue (editor) (2012) “Public Procurement Regulation: An Introduction”, University of Nottingham,
report on principles and practice of regulations on the United Nations Commission on International Trade Law
(UNCITRAL) Model Law of the procurement of goods, construction and services (see https://uncitral.un.org/
Ausubel, Lawrence M. (2008). “Auctions (Theory)”. In: The New Palgrave Dictionary of Economics. Palgrave
Macmillan, London. https://doi.org/10.1057/978-1-349-95121-5_2746-1
di Castri, Simone and Lara Gidvani (2014) “Enabling mobile money policies in Tanzania: A “test and
learn” approach to enabling market-led digital financial services” February, at https://www.gsma.com/
mobilefordevelopment/wp-content/uploads/2014/03/Tanzania-Enabling-Mobile-Money-Policies.pdf
Cramton, Peter (2002), “Spectrum Auctions,” in Martin Cave, Sumit Majumdar, and Ingo Vogelsang, eds.,
Handbook of Telecommunications Economics, Amsterdam: Elsevier Science B.V., Chapter 14, 605‐639
Dymond, Andy and Sonja Oestmann (2004) “Chapter 3: The role of sector reform in achieving universal access”
in ITU (2004) Trends in Telecommunication Reform 2003 – Promoting universal access to ICT: Practical Tools for
Regulators, ITU: Geneva, Switzerland pages 51-63. http://handle.itu.int/11.1002/pub/807b3d5b-en
Dymond, Andy and Kayani Rogati (1997) “Options for promoting rural telecommunications development”
World Bank Technical Report number 359 at https://elibrary.worldbank.org/doi/abs/10.1596/0-8213-3948-6
Emiliani, Luis D. (2008) “Universal service and universal access to telecommunications: a review” The 36th
Research Conference on Communication, Information and Internet Policy, September 26 28. Arlington, VA
Ephraim, Innocent and Daniel Mhina (2021) “FSDT Case Study: The Story of Digital Finance in Tanzania”,
Financial Sector Deepening Trust at https://www.fsdt.or.tz/wp-content/uploads/2021/05/FSDT_DFS_
CaseStudy.pdf
Fox, Jeremy T. and Patrick Bajari (2013) “Measuring the Efficiency of an FCC Spectrum Auction” American
Economic Journal: Microeconomics vol. 5(1): pp.100–146, http://dx.doi.org/10.1257/mic.5.1.100.
GSMA (2006) “Universal Access: How Mobile can Bring Communications to All”
GSMA (2013) “Universal Service Fund Study” available at https://www.gsma.com/publicpolicy/wp-content/
uploads/2016/09/GSMA2013_Report_SurveyOfUniversalServiceFunds.pdf
GSMA (2018) “Tanzania rural coverage pilots: Performance report” available at https://www.gsma.com/
mobilefordevelopment/wp-content/uploads/2018/02/GSMA_Tanzania_Report_Jan.pdf
GSMA (2021) “Auctions Best Practice” GSMA Public Policy Position, September at https://www.gsma.com/
spectrum/wp-content/uploads/2021/09/Auction-Best-Practice.pdf
Hawkins, Timothy (2010) “Enhancing reverse auction use theory: an exploratory study” Supply Chain
Management, v.15, no.1, 2010, pp. 21-42
ITU (2013) “Universal service funds and digital inclusion for all” available at http://handle.itu.int/11.1002/
pub/80c867af-en
60
ITU (2015) “Measuring the Information Society” available at https://www.itu.int/en/ITUD/Statistics/
Documents/publications/mis2014/MIS2014_without_Annex_4.pdf
ITU (2023) “Measuring digital development – Facts and Figures: Focus on Least Developed Countries March
2023”
Jain, Rekha and G. Raghuram (2010) “Role of Universal Service Obligation Fund in Rural Telecom Services:
Lessons from the Indian Experience” Indian Institute of Management W.P. No. 2009-06-03
Kagel, John H. and Dan Levin (2002) Common Value Auctions and the Winner’s Curse, Princeton University Press
Klemperer, Paul (2004) Auctions: Theory and Practice, Princeton University Press
Malik, Payal and Harsha de Silva (2005) “Diversifying Network Participation: Study of India’s Universal Service
Instruments” Discussion Paper WDR0504, World Dialogue on regulation for Network Economies, September
at https://lirneasia.net/2005/11/study-of-indian-universal-service-instruments/
Milgrom, Paul. (1996) “Procuring Universal Service: Putting Auction Theory to Work” Lecture at the Royal
Swedish Academy of Sciences, 9 December
Milgrom, Paul (2004) Putting Auction Theory to Work (Churchill Lectures in Economics). Cambridge: Cambridge
University Press
Milgrom, Paul et al. (2009) “Comments of 71 Concerned Economists: Using Procurement Auctions to Allocate
Broadband Stimulus Grants” available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1377523
Milgrom, Paul, Lawrence Ausubel, et al. (2012) “Incentive Auction Rules Option and Discussion”, submitted by
Auctionomics and Power Auctions to the FCC available at https://apps.fcc.gov/edocs_public/attachmatch/FCC-
12-118A2_Rcd.pdf
Milgrom, Paul (2017) Discovering Prices: Auction Design in Markets with Complex Constraints Columbia University
Press
Muente-Kunigami, Arturo and Juan Navas-Sabater (2010) “Options to increase access to telecommunications
services in rural and low-income areas”, World Bank Working Paper No. 178, Washington DC
Navas-Sabater, Juan, Andrew Dymond, and Niina Juntunen (2002) “Telecommunications and Information
Services for the Poor” World Bank Discussion Paper at https://doi.org/10.1596/0-8213-5121-4
OECD (2005) “Leveraging Telecommunications Policies for Pro-Poor Growth: Universal Access Funds with
Minimum-Subsidy Auctions,” OECD Papers, OECD Publishing, vol. 5(1), pages 1-58. https://doi.org/10.1787/
oecd_papers-v5-art1-en
Oh, Sarah (2021) “Using Reverse Auctions to Stretch Broadband Subsidy Dollars: Lessons from the
Recovery Act of 2009”, working paper Technology Policy Institute, Washington DC, available at https://
techpolicyinstitute.org/wp-content/uploads/2021/03/Oh-Reverse-Auctions-Lessons-from-BTOP-Jan-2021.
pdf
Sepulveda, Edgardo (2004) “Chapter 5: Minimum-subsidy auctions for public telecommunications access in
rural areas” in ITU (2004) Trends in Telecommunication Reform 2003 – Promoting universal access to ICT: Practical
Tools for Regulators, ITU: Geneva, Switzerland pages 85-94. http://handle.itu.int/11.1002/pub/807b3d5b-en
61
Shalev, Moshe Eitan and Stee Asbjornsen (2010) “Electronic Reverse Auctions and the Public Sector – Factors
of Success” Journal of Public Procurement, vol. 10, no. 3, pp. 428-452, Fall 2010, Available at SSRN: https://ssrn.
com/abstract=1727409
Stern, Peter, and David Townsend (2006) “New Models for Universal Access in Latin America” Final Report,
Regulatel, World Bank & ECLAC Project on Universal Access for Telecommunications in Latin America
Wallsten, Scott (2009) “Reverse Auctions and Universal Telecommunications Service: Lessons from Global
Experience”, Federal Communications Law Review, vol.61, issue 2, pp. 373-394 available at https://www.
repository.law.indiana.edu/cgi/viewcontent.cgi?article=1531&context=fclj
Wellenius, Bjorn (2002) Closing the Gap in Access in Rural Communications: Chile 1995-2002, World Bank,
Washington DC https://doi.org/10.1596/0-8213-5092-7
World Bank (2023) “Digital Africa: Technological Transformation for Jobs”
World Wide Web Foundation (2018) “Universal Service and Access Funds: An untapped resource to close the
gender digital divide” WWW Foundation, A4AI and UN Women, March
Wu, Irene S. (2014) “Maximum impact for minimum subsidy: reverse auctions for universal access in Chile and
India”, info, vol. 16 no. 5, pp. 46-58, https://doi.org/10.1108/info-06-2013-0037
62