1

The economic perspective of the mobile networks in EuropeDimitris Katsianis (University of Athens∗ , GR), Ilari Welling, Maria Ylönen1 (NOKIA, FI), Dimitris Varoutas,

Thomas Sphicopoulos (University of Athens, GR), Nils Kristian Elnegaard, Borgar T. Olsen (Telenor, NO),Lucien Budry2 (Swisscom, CH)

∗ National and Kapodistrian University of AthensDepartment of Informatics and Telecommunications, Panepistimiolopis, Athens, Greece

d.katsianis@di.uoa.gr

1 Introduction

Over the next years, about 70 licenses for UMTS (Universal Mobile Telecommunication System)spectrum will be distributed in Europe, either through auctions or comparative hearings. Now, morethan ever, mobile operators are making critical decisions, which will shape their business over the nextten years. The main challenge is to maximize investment efficiency to provide next-generation, mobiledata services, at announced speeds of up to 2 Mbps. Such systems, generally referred to as IMT-2000or UMTS, are expected to emerge in 2001 in Japan and in Europe, hence the intense preparations inprogress at the manufacturing, regulatory, and operator levels.

The European ACTS project TERA (Techno-Economic Results from ACTS, AC364 [1]) is among thefirst groups in the world to publish figures [2][3][4][5] on the economic feasibility of next-generationmobile systems for mobile operators in Europe. This article presents the techno-economic perspectiveof these technologies aiming to provide guidelines for strategies in the move towards broadbandmobile services. The analysis has been carried out during 2000 and is based on 3G roll-out scenariosin two “typical” European countries with contrasting profiles. The first is a large (population: 60 million)country with moderate mobile penetration of about 40% in 2000. The second is a “fast-track” smallercountry (population: 4 million) in which mobile penetration is around 60% today and slated to grow to80% in upcoming years. The objectives are to compare both incumbents’ and newcomers’ (greenfield)business cases in these two country types.

The techno-economic evaluation is followed by a sensitivity analysis in order to identify the impact ofthe important parameters as the operator’s market share and overall level of demand for next-generation mobile systems.

Following an overview of the current context and the different paths, which may be taken toward 3rd

generation mobile systems, the general approach and main assumptions adopted for the businesscase have been described. The deployment area, the selected network architecture evolution, theservices offered, the demand, and the related tariffs have been included. Final results are presentedand discussed. A complete glossary of acronyms is provided at the end of this article.

2 Context and evolution paths

Today, Europe’s prevalent digital cellular GSM standard provides mobile telephony and low rate dataservices. Current trends are toward the emergence of new operators, lower tariffs for both traffic andterminals, as well as new services (location specific) and new types of terminals including dual bandand dual mode, one-touch handsets for children, communicators, and palm pilots [6].

HSCSD, or High Speed Circuit-Switched Data, a GSM upgrade introduced, supports data rates of upto 57.6 kbps by grouping 4 GSM time slots. HSCSD was first commercially launched in August 1999 inFinland. A new terminal is required for the customer.

GPRS, or General Packet Radio Service, another enhanced GSM technology introduced in 2000,supports up to 115.2 kbps packet switched mobile data alongside circuit-switched telephony.

1 No longer with Nokia2 No longer with Swisscom

2

Hardware upgrading is required for the present GSM core network, as well as a new terminal for theuser.

EDGE, or Enhanced Data for GSM Evolution, which includes advanced versions of HSCSD(EHSCSD) and GPRS (EGPRS), may be introduced as of 2001. Since EDGE uses a different codingscheme providing 48 kbps per time slot for an overall rate of up to 384 kbps, extra hardware andsoftware must be added to the HSCSD or GPRS systems. Again, a new terminal for the user isrequired.

Commercial UMTS is expected to go into operation for mobile telephony and data services at up to 2Mbps for a stationary customer, 384 kbps for customers with high mobility (120 km/h) and 128 kbps forcustomers at full speed (500 km/h). Early on, UMTS will serve only hot spots at rates of up to 384kbps, while GPRS or EDGE systems should be used for full coverage. Once again, the user will berequired to change terminals, a fact, which must not be neglected by the operator. Figure 1 illustratesthese different evolution paths which are the main alternatives options for a GSM operator.

GSM

HSCSD

GPRS

UMTS

EHSCSD

EGPRS

1999 2000 2001 2002

9.6 kbps /14.4 kbps

115.2 kbps

57.6 kbps

144 kbps

384 kbps

2 Mbps (stationary)384 kbps (120 km/h)144 kbps (500 km/h)

circuit

packet

4 GSM time slotssoftware upgrade

+ new terminal

hard- & soft-ware upgrade+ new terminal

(14.4 kbps per slot)

new network+ new terminal

new coding(48 kbps/slots)

hard- & soft-ware upgrade( new coding: 48 kbps/slots)

EGPRS

384 kbps

new network+ new terminal

3

22; 3

1; 2 ;3

4; 5; 6

5

5

1; 66

4

Figure 1 - Mobile evolution steps

It is pointed out that that, while the intermediate steps are overlaid onto a GSM network, the UMTStechnology requires full buildout from scratch. Incumbent operators may, however, re-use existingGSM sites. This is the major advantage of an incumbent operator in order to provide advancemultimedia mobile services.

3 The business case

3.1 General Assumptions

As initial estimations [5][7] have ruled out short-term profitability for UMTS, a 10-year time frame,spanning the period 2000 to 2009 has been chosen. In order to calculate discounted cash flows, whichtake into account the time value of money, a discount rate of 10% has been selected. This value is amean value among the major European Telecommunication Operators.

The first step was the gathering of information on tariffs and market penetration of both fixedbroadband and mobile services in various European countries. Existing forecasts for UMTS serviceswere gathered, and an analytical market model was developed [8]. Parallel to service informationgathering, the proposed radio and core network architectures were derived from public sources likeETSI UTRA documentation (drafted on the basis of work conducted within the ACTS/FRAMES project[9]) and relevant ACTS projects such as RAINBOW [10].

3

For the UMTS case study, two kinds of operators have been studied. In the first case, it is assumedthat an existing GSM operator implements GPRS as an evolutionary path towards UMTS. In 2002, itstarts the UMTS service, where both voice and data services are offered. In the second case, a newoperator (possibly a fixed network operator) invests directly in UMTS in a competitive environment.The options will be described in more detail in the architecture section.

3.1.1 Area studied.

The profitability of advanced mobile services is tested in two different kinds of European countries.One is a typical Nordic country (later referred to as the small country) with relatively low populationdensity and high mobile penetration. The other area represents a central or southern Europeancountry with higher population density and lower mobile penetration (later called the large country). Inboth cases, the areas are composed of a dense downtown zone surrounded by a less denseperipheral zone, and a larger suburban area. Note that the zones are not necessarily contiguous; thedense urban zone is taken to represent the most dense areas of the country, the five or six largestcities, for example. The urban areas are assumed to surround these most dense zones, and thesuburban area covers the rest of the country, to the exception of rural areas. As mentioned previously,since UMTS is expected to be deployed in hot spots, it is assumed that rural areas will be coveredusing techniques such as GPRS and EDGE, which can re-use existing GSM infrastructure. The finalUMTS coverage is about 80% of the total population.

Thus, the population of the covered area is some 4 million in the small country and 60 million in thelarge country. The GSM penetration is assumed to saturate in 80% (in terms of mobile phone owners,not in terms of number of terminals) in 2009.

3.1.2 Regulatory environment

The role of the NRA (National Regulatory Authority) is of key importance for several cost parametersin UMTS deployment. Firstly, the choice of assignation mechanism impacts heavily on the cost ofentry, as illustrated by two extreme examples, those of Finland and United Kingdom. In the first case,licenses were awarded through a comparative hearing process to four operators for a minimal annualfee covering administrative costs. In the second case, auctions were held in order to assign fivelicenses, with one license open for bidding by new entrants only. The resulting fees, on the order of 8billions € per operator, are to be paid up front. To focus more strictly on the operators’ infrastructure-related business cases, the case of comparative hearings as the basis of the calculations has beenchosen. In the large country, the fee is 10 € per inhabitant, payable over a 10-year period in equalinstallments (independently of the duration of the license). The yearly fee is thus 60 M€. In the smallcountry (2€/inhabitant), the yearly fee is 800 k€. The case of auctions has been cover through thesensitivity analysis.Certain service obligations following comparative hearings have been also taken into account in thisstudy. In particular, the coverage of 60% of the population within 3 years as the most expectedEuropean practice, is ensured, by scheduling deployment in the suburban areas in 2005. Thisapproach leads to an overall coverage of over 80% of the population. Regarding roaming agreements,it has been assumed that the new entrant is allowed to seek such agreements for calls in areas inwhich it has no infrastructure and pays an interconnection fee of 2€ per call to the incumbent operator.

3.2 Telecommunication Mobile Services

3.2.1 GPRS

For GPRS, an average bitrate of 40 kbits is assumed, on a “best effort” basis. Many of the applicationswill be WAP (Wireless Application Protocol) based, e.g. e-mail and different kinds of location-specificservices. The GPRS technically enables operators to provide packet switched data at rates rangingfrom 14.4 to 115 kbps, depending on the number of channels used and the service quality. However,because using more than two time slots requires more complex terminals, and enhanced channelcoding only works in good radio conditions, it is reasonable to expect average bit rates on the order of40 kbits or lower. The main application is Internet access, and since GPRS uses dynamic allocationschemes, the user can be billed on a per-volume basis rather than according to call duration.

3.2.2 UMTS

The UMTS Forum [7] has listed a broad range of services, which should be made available with theadvent of UMTS systems. They are summarized in Table 1.

4

Type of service Professional customers Residential/SOHOcustomers

Audio & videobroadcasting

Specialized programs (medical, salespromotions...)

digital TV

Asymmetrical exchanges,consultation

Internet, intranet, tele-activities games, VOD, NOD,Internet, e-commerce,tele-learning, drivingassistance

Voice telephone, voice mail telephone, voice mailData e-mail, file transfers e-mailSymmetric imagecommunications

Video telephony, conferencing video telephony

Table 1 - UMTS applications according to the UMTS Forum.

It is also recalled that the data rates available for the different services depend on the zone and speedof the user, as illustrated in Table 2.

Picocell Microcell Macrocell SatelliteZone In-building,

downtownUrban Semi-urban,

ruralRural

Max. bitrate 1-2 Mbps 384 kbps; 1 Mbps, lowmobility

384 kbps 9.6 kbps384 kbps (S-UMTS)

User speed Stationary-pedestrian

120 km/h 500 km/h 120 km/h

Table 2 – Available bitrates in different areas.

According to their requirements in terms of delay, bit error rate or other criteria, these services willutilize bearer services, which have been defined for UMTS. For the model, we deal with a limitednumber of such bearer services, with their associated bit rates, as listed below in Table 3. AlthoughUMTS bearer services have been defined in the literature as real-time (low constrained delay, LCD) ornon-real time (unconstrained delay, UDD), the more restrictive LCD bearer service only hasconsidered. Voice services are assumed to remain on the operator’s GSM network, which is upgradedappropriately.

User data rate64 kbps (symmetric)128 kbps (symmetric)384 kbps (asymmetric)1 Mbps (asymmetric)

Table 3 - Offered UMTS bitrates.

3.3 Market forecast

Since the most probable evolutionary path for the incumbent involves GPRS, we first evaluate uptakeof GPRS services starting in 2000 and then we consider uptake of UMTS services starting in 2002 inthe dense urban and urban areas. Roll-out in suburban areas is an input parameter, since it may varyaccording to national regulator obligations. Since regulators are planning to actively promote acompetitive context for next-generation mobile services, a market sharing by 4 operators is

5

reasonable. Regarding market share evolution, the incumbent starts with 31% market share in 2000,and finishes with 25% market share in 2009, while the new entrant starts with 5% market share in2002 and finishes with 25% in 2009, which is a relatively optimistic scenario for the latter.

To evaluate the data rates to be handled by 3rd-generation mobiles systems, it is important to identifythe different types of customers [8], since their calling patterns and usage differ. An example ofsegmentation is the following,

n corporate users, with intensive and almost entirely professional use, primarily during officehours;

n business users, less intensive use than corporate users;

n SOHO/telecommuters, professional and private use, during the day and in the evening;

n advanced residential customers, primarily private use, evenings and weekends;

n mass market, last arrived on the market, low use (i.e. stay within flat-rate limits).

For the purposes of this business case, corporate and business users are combined with professionalsin one segment; advanced residential and SOHO customers in another, and mass market is handledas a segment of its own. Thus we consider three main market segments: professional or businessusers, advanced residential/SOHO, and mass market.

3.3.1 Uptake of GPRS services

The demand curves for GPRS [8], which reflect a synthesis of European market forecasts, are appliedto the defined areas, yielding the year-end customer numbers (the demand curve is illustrated inFigure 2).

3.3.2 Uptake of UMTS

In the time frame 2000-2005, it is expected that operators will initially focus attention on the lucrativeprofessional market segment. UMTS services become attractive to the residential/SOHO market witha slight delay, and then the mass market. The following curves illustrate the central scenario forUMTS, also recalling the demand for GPRS, which flattens as customers migrate from GPRS toUMTS.

GPRS and UMTS penetration

0%

20%

40%

60%

80%

100%

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Year

% o

f tot

al p

opul

atio

n

GPRS

64 kbps symmetric

128 kbps symmetric

384 kbps asymmetric

1 Mbps asymmetric

Voice

Figure 2 - Demand for UMTS and GPRS services as percentages of total population.

3.3.3 Tariffs

Tariff schemes for next-generation mobile services were devised, based on information from differentEuropean countries for fixed telephony, broadband Internet access, and mobile telephony and data.They are summarized below, along with usage assumption [8].

6

These tariffs take into account the competitive context (4 operators) and rapidly falling prices forbroadband wireline services.

3.3.3.1 GPRSFor GPRS, the IP service is comparable to Internet access via cable or ADSL, in the sense that the«attached» mode means the terminal is «always on». However, the data rates offered are similar to orlower than modem rates using the PSTN. Considering an added value for mobility, an initial monthlyflat fee of 20€ appears reasonable. Volume or transaction based billing could also be implemented.Our business case assumptions are summarized Table 4.

Mbytes per day Connection fee Monthly flat rateBusiness 0.5 10 20Residential 0.25 10 20Mass market 0.1 10 20Robot 0.01 10 20

Table 4 - GPRS volumes and tariffs (in €)

Uplink traffic is considered negligible. A 10% reduction per year has been applied for the first twoyears, then 5% for two years, then the tariffs remain stable. Connection fee reduction is 5% per year.

3.3.3.2 UMTSThe Table 5 lists the services considered and pricing proposals, based on the tariffs of existingservices [11] plus a 50% premium for mobility

Mbytes perday

Connection fee Price permonth

Reference

64 kbps symmetric 1 25 30 GPRS + symmetry128/144 kbpssymmetric

2 45 45 ISDN + mobility

384 kbpsasymmetric

4 50 35 GPRS + dataspeed

1 Mbps asymmetric 5 70 75 ADSL+ mobilityTable 5 - UMTS Volumes and Tariffs (in €)

Flat-rate tariff schemes for next-generation mobile services were devised, based on information fromdifferent European countries for fixed telephony, broadband Internet access, and mobile telephonyand data. They are summarized below, along with usage assumptions. These tariffs take into accountthe competitive context (4 operators) and rapidly falling prices for broadband wireline services.

A 10% reduction per year is expected for the first two years, then 5% for two years, then remainstable. The connection fee reduction is 5% per year as well as for GPRS.

3.3.4 Customer service, marketing and promotion.

With the introduction of broadband mobile services, we assume additional personnel for customercare (services and billing). Marketing and advertising costs are estimated to be approximately equal to1 €/inhabitant for both GPRS and UMTS. Because subscribers must change terminals when adoptingGPRS, then UMTS services, the operator may find it necessary to subsidize the purchase of terminals.We include a marketing and advertising budget, as well as a subsidy for the purchase of GPRSterminals. A 15% churn figure was taken into account to compute the number of terminals to besubsidized.

3.4 Network architectures and upgrades

GPRS dimensioning is based on the current structure of the GSM network. Hardware and software

7

upgrades have been taken into account, as required by the new technologies and additional user load.In UMTS, the radio dimensioning procedure takes into account first coverage then capacity.Analytically:

3.4.1 Upgrade for GPRS

The basis for GPRS is the current GSM network, comprised primarily of base stations, base stationcontrollers (BSC), mobile switches (MSCs), home and visitor location registers (HLR, VLR), as well asthe transmission links from BS to BSC and from BSC to MSC.

GPRS requires the addition of TRXs to handle the increased traffic load. The TRX’s need to beupgraded in order to enable GPRS services. It is assumed that only software upgrade is needed, andis actually performed at the BSC level but in some systems, hardware upgrades might also berequired. The increased traffic load is assumed to require addition of TRX’s. The cost of a GSM TRX(including installation) is 10 K€. The number of additional TRX’s is computed according to the existingcapacity and additional packet data user load. The GPRS network architecture (upgrade from GSM toGPRS) is illustrated in Figure 3:

GGSN

SGSN

Packetsubsystem

Internet(TCP/IP)Internet

(TCP/IP)BSS

NSS

Landline Ntw(PSTN/ISDN)Landline Ntw(PSTN/ISDN)

SIM card

BSC

MSC

HLR

BTS

Updates in existing network elements

GSM mobile

Figure 3 Upgrade from GSM to GPRS

The packet subsystem is a new part of the network. It consists of equipment providing packet trafficrelated functions. The following elements must be added to the core network in order to reuse theexisting GSM infrastructure. The SGSN Serving GPRS Support Node (for 100,000 customers), whichis connected to one or several BSCs, and handles data call management, traffic and billinginformation. A GGSN Gateway GPRS Support Node is needed (10,000 attached customers), forhandling interconnection with other networks (IP, X25, etc.) as well as firewall, billing, and routinginformation functions. Furthermore we consider one Domain Name Server per SGSN On the otherhand we include to the initial cost the CG Charging Gateway one per 250,000 customers GRPScustomers (100,000 in the attached mode). For the operation and maintenance of the GPRS networkan OMC – G (Operation and Maintenance Center for GPRS) is needed for 1 million subscribers. Inorder to manage the sharing of radio resources and retransmission of faulty data over the radio, weconsider 1 PCU (Packet Control Unit) per BSC initially, and then an additional PCU per 64 Mbps ofthroughput. Upgrading PCU capacity over time is probably not a viable solution, as it demands severalon-site visits. It is also required in most all countries, one Legal Interception Gateway for 1 millionsubscribers. In addition, software upgrades to the HLR are necessary. The model considers thefollowing positioning of these elements (Table 6):

8

GSM structure GPRS hardware GPRS softwareMSC SGSN, GGSN, DNS, CG,

OMC-G, LGIHLR upgrade, MSC upgrade,management upgrade

BSC PCU BSS softwareBTS Additional TRX TRX upgrade

Table 6 – Model of the elements

3.4.2 Upgrade for UMTS

Since UMTS can utilize many of the GSM network's elements, in an incumbent operator's UMTSnetwork there are both kind of elements. Radio equipment will initially accommodate UMTS, EDGE,GPRS and GSM, to ease the migration from GSM to UMTS. Such an upgraded network is illustratedin the Figure 4. A pure UMTS network would not have GSM BTSs and BSCs.

SGSNInternet

(TCP/IP)Internet

(TCP/IP)

GGSN

Packet subsystem

IN

GSM mobile

Landline Ntw(PSTN/ISDN)Landline Ntw(PSTN/ISDN)

SIM card

BSC

MSC

HLR

BTSGSM / WCDMAmobile

WCDMAmobile

Co-sited GSM + WCDMA

WCDMA BSS

WCDMA BS WCDMA

RNC

Figure 4 - UMTS Architecture

The following elements are necessary to offer UMTS services: The Base Station Subsystem BSS iscomprised of specific Base Stations (BS: 3 sectors, 5 MHz to 15 MHz duplex each) and RadioNetwork Controllers (RNC: one per 256 BSs). The RNCs are connected to the UMTS SGSN, whichmay be common to GPRS and UMTS in some cases. In our model two different components are usedas 3G SGSN. For the Network SubSystem (NSS) it must be added to the core network a UMTSSGSN one per 100,000 customers, for 12 Mbps of throughput; For the operations and maintenance anOMC as one system for 1 million subscribers. In the case of an incumbent operator, othercomponents, such as the DNS, GGSN, MSC, HLR, and billing system can be shared with GPRS. Thegreenfield operator naturally has to acquire all those network elements. In that case, the dimensioningrules mentioned in the previous paragraphs have been used.

3.5 Radio dimensioning for UMTS

The radio dimensioning procedure takes into account first coverage then capacity. This process isentirely independent from the existing GSM infrastructure. To ensure physical coverage, we start withthe cell radius determined from representative link budgets calculated in the uplink, which is morerestrictive since the transmitter power from the mobile is lower than that of the base station. The cellradii in different area types have been listed in Table 7.

9

Cell radius (km) for 80% indoor coverage probabilityData rate 64 kbps 128 /144 kbps 384 kbps 1 MbpsDense urban 0.39 0.38 0.30 0.28Urban 0.97 0.97 0.73 0.70Suburban 1.56 1.55 1.17 0.94

Table 7 - Cell radius per services and area

Since the highest rate to be offered is 1024 kbps, we consider a cell radius of 0.28 km in the denseurban area, 0.7 km in the urban, and 0.94 km in the suburban area. Note that these cell radius aregreater than those of existing GSM cells. This means that, for the incumbent operator, we assume thatno new site buildout is required for coverage. The new entrant, however, must invest in the sites aswell. It is assumed that full coverage is built in a year before the start of service (2001 in dense urbanand urban areas, 2002 in suburban areas).

According to the data rate required, the existing number of BTS sites and sectors and the maximumnumber of users, we compute the data rate per square kilometer, i.e. the available capacity. Thiscapacity is matched against that computed from the demand curves and the usage patterns, andwhen the user load exceeds the offered capacity, it is necessary to add network components.

4 Techno-Economic Definitions

The objective of a business case for this network is to estimate investments, revenue, operating cost,general administration cost and taxes. The network is expected to generate revenue for the life of theproduct. Deducted from the revenue stream are depreciation, operating cost, general administrationcost and taxes. The business case is assessed according to four conventional criteria, Net PresentValue (NPV), Internal Rate of Return (IRR), cash balance, and payback period.

The Net Present Value (NPV) is today's value of the sum of resultant discounted cash flows (annualinvestments and running costs), or the volume of money which can be expected over a given period oftime. If the NPV is positive, the project earns money for the investor. It is a good indicator for theprofitability of investment projects, taking into account the time value of money or opportunity cost,which is expressed in the discount rate (10% in the present case).

The Internal Rate of Return (IRR) is the interest rate resulting from an investment and income(resultant net cash flow) that occur over a period of time. If the IRR is greater than the discount rateused for the project, then the investment is judged to be profitable. This criterion is especially useful incomparing projects of different type and size. The Internal Rate of Return gives a good indication of“the value achieved” with respect to the money invested.

The Cash Balance (accumulated discounted Cash Flow) curve generally goes deeply negativebecause of high initial investments. Once revenues are generated, the cash flow turns positive and theCash Balance curve starts to rise. The lowest point in the Cash Balance curve gives the maximumamount of funding required for the project. The point in time when the Cash Balance turns positiverepresents the Payback Period for the project.

5 Main Economics Results

The four Scenarios were investigated in order to provide guidelines for incumbent and greenfield (newentrant) mobile operators present in two types of European markets: on one hand, a small country withhigh mobile penetration and a high propensity to adopt 3G technologies, and on the other, a largecountry with a lower penetration of mobile customers and a “slower” take-up rate of mobile terminals.The main techno-economic results are net present value (NPV), internal rate of return (IRR), and pay-back periods. The investment project that is presented here, was studied using the tool which wasdeveloped by European Project ACTS 364/TERA [1], [4]. We also studied the sensitivity of the NPV to

10

market share, and overall service penetration.

5.1 Net present value and Internal rate of return

With four operators sharing the market, the customer base appears sufficient to achieve a positiveNPV over the study period in all four cases. The IRR is higher than the 10% discount rate, whichshows that the investment is sound over this period. Table 8 summarizes the economic results of thetwo basic scenarios (large and small country) for both the incumbent and new entrant.

Incumbent Small IncumbentLarge

GreenfieldSmall

GreenfieldLarge

NPV (M€) 357 4,961 90 1,992IRR 26% 29% 14% 17%Rest Value (M€) 8 112 28 327Pay-back period 6.8 6.7 8.3 8.0Number of customers 2009 1,100,000 14,600,000 800,000 11,500,000Investments (M€) 364 3,773 417 4,194Running costs (M€) 564 10,127 421 7,113Investment per connectedcustomer (€)

340 260 500 360

ARPU per month over thestudy period (€)

42 42 60 60

Table 8 - Summary of the economic results for operators with 25% end market share.The following figures (Figure 5 and Figure 6) illustrate the revenues, investments and cash balance ofall scenarios. Significant investments are required during the second year of the project in order toinstall the UMTS BTSs in urban and suburban areas. The lowest point of the cash balance indicatesthe amount of investment funding required from the operator. Comparing the two cases (incumbentand greenfield) in large and small countries shows that the maximum need for funding is quitesignificant especially for a large country: 4.6 billion € in the larger country for an incumbent, and 5.4billion € for a new operator. Respectively, for a small country the maximum finance needed is 0.44 and0.55 billion €. The average revenue per user (ARPU) during the study period is higher in the case ofthe greenfield operator because the income from voice traffic is not shared with the GSM network as itis in the case of the incumbent. The incumbent also has GPRS subscribers with lower monthlyspending.

11

-10000

-5000

0

5000

10000

15000

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

M E

uros

Revenues GLRevenues IL

Investments IL

Investments GLCash Balance GL

Cash Balance IL

Figure 5 - Economics of the Large Country case ( I=incumbent, G=greenfield, S=small, L=large)

-800

-600

-400

-200

0

200

400

600

800

1000

1200

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

M E

uros

Revenues GS

Revenues IS

Investments ISInvestments GS

Cash Balance GS

Cash Balance IS

Figure 6 - Economics of the Small Country case (I=incumbent, G=greenfield, S=small, L=large)

Since the previous figures have different scales, we present the results for all cases together (Figure7) to more clearly emphasize the differences in cash balances. In all cases (Figure 7), the slopes ofthe cash balance curves at the end of the period indicate good future earning potential. The case oflarge country has much more potential revenues than the same investment in a small country due tothe greater potential market size (60 million instead of 4 million inhabitants covered by a UMTSsystem). The one that has significant importance is to compare the “investment project” in the samearea for an incumbent and a “greenfield” operator. The percentage difference in cash balancebetween the operators in a small country in 2009 exceeds 80%. For a large country the difference isaround 70%.

12

-6000

-4000

-2000

0

2000

4000

6000

8000

10000

12000

14000

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

M E

uros

Cash Balance GL

Cash Balance IL

Cash Balance IS

Cash Balance GS

Figure 7 – Cash Balance Curves for all the cases (I=incumbent, G=greenfield, S=small, L=large).

Figure 8 illustrates economic features for all cases. Taking into account the same tariffs for all cases,the incumbent's payback period is almost 6.8 years; for the greenfield it is more than eight (8.3) yearsin a small country. In a large country, incumbent is still (as in the small country) the “winner”, whilegreenfield demands larger investments and thus longer time for full payback. The fact that the existinginfrastructure can be partially reused, favors the incumbent operator. The payback period for anincumbent is 6.7 years and for a greenfield it is around eight years. However, the 7-year pay backperiod may appear long to certain decision-makers. In a competitive environment, this greater paybackperiod could be the turning point for the failure of a greenfield project. A critical comment could be thatthe investing operator should be powerful enough to survive for eight years without any real profits. Itis clear that, under the conditions described above, the operator will have to either accept a delayedreturn on investment while maintaining attractive tariffs, or raise tariffs, with the risk of reducing itscustomer base.

13

NPV - IRR - Payback Period

35790

1992

4961

26

17

14

29

88.3

6.76.8

0

1000

2000

3000

4000

5000

6000

Incumbent S Incumbent L Greenfield S Greenfield L

M E

uro

s

0

5

10

15

20

25

30

35

IRR

(%)

Pay

-bac

k p

erio

d (y

ears

)

NPVIRRPayBackPeriod

Figure 8 - NPV – IRR – Payback Period (All cases)

The final techno-economic results concerning the NPV (Figure 8 and Table 8) and the IRR show theoverall difference between the two different operators. The NPV for an incumbent operator is almost300% (small country) and 150% (large country) greater than the NPV for the greenfield operator. TheIRR in the two cases for the incumbent is 26% and 29%, and for the greenfield 14% and 17%respectively. This is the crucial point for strategic decisions. The lowest accepted IRR value for atelecommunication operator varies proportionally according to its investment strategy and potentialeconomic situation. A general conclusion that could be acceptable concerning the IRR is that theinvestment of an incumbent operator is not as risky as that of the greenfield operator, in absolutevalues. In the small country the discounted investment per connected customer for the incumbent is344€ and 498€ for the greenfield, and in the large country it is 259€ for the incumbent and 364€ for thegreenfield (Table 8).

In general, large countries cover more subscribers with high-capacity demand as well as a highwillingness to pay. Even if the large country is more cost-sensitive, the investment strategy will dependon the demand for high capacity. This is especially true for narrowband speeds but could be profitable(and more interesting for the customer) at broadband capacities as well within the next few years. Itcan generate good revenues mainly because it is offering not only one service (multimedia UMTSservices), but also a wide palette of already existing Internet services. In addition, UMTS offer theoperators the ability to present new service sets to their customers and potentially to the wholeInternet community, based on the Internet platform. However, the combination of a higher price persubscription for a higher capacity service and a potential increase in penetration compared to thepenetration of a lower capacity service can increase the total revenue in both cases (incumbent ornot). The appearance of new services can increase the penetration of new customers.

5.2 Sensitivity analysis of Market share

In the actual performance of the analysis telecommunications network operators very often resort toplain discounted cash flow analyses and simple sensitivity analyses when assessing their businesscases [12]. These traditional approaches are often unable to capture the flexibility of the decision-makers to adapt and revise later decisions in response to unexpected market developments.

Sensitivity analysis consists of study the impact of single parameter without any correlation to otherparameters. NPV and IRR, as financial ratio, have been used as criterion for the sensitivity analysis.The sensitivity results are presented after the identification and the ranking of the four most possibleparameters (with 95% confidence) with greatest impact on NPV and IRR, which are the tariff factors,the market share, the license fee and the BTS cost.

14

The sensitivity calculations have been performed on NPV and IRR indicators for each scenario (largeand small country, incumbent and newcomer operators). Figure 9 and Figure 10 illustrate indicativelyfour of these eight combinations. Table 9 presents the parameters, which have the most impact onNPV or IRR based on one of the sensitivity analysis. Our purpose is to present a ranking of fourparameters, which give the biggest changes on the NPV or the IRR.

The NPV or IRR middle value corresponds to the base value of the parameter. For instance in Figure9, NPV base value is 4,961 M€ (with 25% Market share in 2009) and for Market share in 2009 equal to30% (High value) the NPV reaches the value of 4,916+4,121=9082 M€. Furthermore, multipliers areused for the variation of values of UMTS tariff and BTS cost by a convenient way (i.e. 0.4 means 40%of the base value).

Uncertain parameter Low Base HighMarket Share in 2009 10% 25% 30%UMTS tariff multiplier 0.4 1 1.6UMTS license fee [€] (large) 0 10 150UMTS license fee [€] (small) 0 2 150UMTS BTS cost multiplier (84 – 156 k€) 0.7 1 1.3

Table 9 Sensitivity Ranges

-5,000

-4,000

-3,000

-2,000

-1,000

0

1,000

2,000

3,000

4,000

5,000

Low value -4,277 -3,425 200 857 -3,188 -3,866 204 911

High value 4,121 1,142 -2,961 -857 3,035 1,188 -3,301 -911

UMTS tariff multiplier I

Market Share in 2009 I

UMTS licence fee I

UMTS BTS cost

multiplier I

UMTS tariff multiplier G

Market Share in 2009 G

UMTS licence fee

G

UMTS BTS cost

multiplier G

Figure 9 Sensitivity analysis results. Change in NPV compared to nominal situation (4,961 M€) in case oflarge country both for Incumbent (I) and Greenfield (G) case.

15

-20.00%

-15.00%

-10.00%

-5.00%

0.00%

5.00%

10.00%

15.00%

Low value -13.32% -11.89% 0.15% 7.44% -11.72% -17.26% 0.13% 4.77%

High value 10.27% 2.90% -10.46% -5.54% 7.48% 3.44% -10.79% -4.18%

UMTS tariff multiplier I

Market Share in 2009 I

UMTS licence fee I

UMTS BTS cost multiplier

I

UMTS tariff multiplier G

Market Share in 2009 G

UMTS licence fee G

UMTS BTS cost multiplier

G

Figure 10 Sensitivity analysis results. Change in IRR compared to nominal situation (14%) in smallcountry both for Incumbent (I) and Greenfield (G) case.

6 Conclusion and Discussion

The techno-economic prospects for a newcomer and especially for an incumbent operator planning todeploy the UMTS technology are found to be quite positive according to this study. Specifically, thefollowing elements were identified to have major consequences on the profitability of this newbusiness:

• Regulatory decision to promote competition: By deciding to open the UMTS market to at leastfour competing operators, regulators are hoping that the competitive dynamics will work to offerthe widest range of services to the most customers possible at least cost. Our results show thatmarket share at the end of the study period is of major importance. Net Present Value of the largecountry business case is improved by 1,200 M€ (or 3.5% for IRR) for an increased of 5% in themarket share in 2009.

• Cost of licenses: The sensitivity analysis on this parameter show that license cost and thereforethe license assignation mechanism (auction or comparative hearings) can seriously deteriorate thebusiness case since the payback period can be delayed by more than a year, together withsignificantly decreasing NPV. License fee increasing from 10 to 150 €/inhabitant (based on theUMTS auction in United Kingdom, where the average cost per UMTS license has reached thelevel of 122 €/inhabitant), decreases the NPV by 66% for incumbent. Therefore, the regulatoryframework can affect the strategy for the UMTS deployment. A highly competitive environmentfollowing the assignation procedure will lead to a real battlefield for profitability in the lucrativeurban areas.

• Tariffing of voice and data services: The business case used fixed network data tariffs as abasis for UMTS tariffs. It was shown that the overall tariff level ranks first over service penetrationand market share as the most significant factor for UMTS profitability. This result is logical since,in NPV calculation, the tariff level directly impacts total revenues, whereas all other parameters,like the service penetration level and market share, affect the number of customers, and hencealso the costs. Nonetheless, it must not be construed that operators are free to hike tariffs asrequired to achieve a positive result. Indeed, the competitive contexts and dropping prices for fixednetwork services will severely limit their room to manoeuvre in this area.

• Investment schedule: Due to the fact that operators deploy overdimensioned equipment in acompetitive market situation, using a coverage rather capacity approach the cost of BTSequipment incurs a heavy financial burden to be supported for 8-10 years. This coverageapproach is a limited factor and leads to larger investments in the pre-service year.

16

In conclusion, UMTS operators will have not very much latitude to roll out their networks. Heavyinvestments are required early on in order to cover the most dense areas, and then once again for thesuburban areas. Competitive pressure will keep tariff levels low, and operators will need to consolidatetheir market assumptions with extreme care in order to evaluate the payback period. Lastly, they musthave enough financial resources to stay in debt for a long period of time.

ACKNOWLEDMENTS

The authors would like to thank all the partners of ACTS 364/TERA for their useful support anddiscussions.

7 Reference

[1] http://www.telenor.no/fou/prosjekter/tera/index.htm

[2] L.A. Ims et al, “Multiservice Access Network Upgrading in Europe: A techno-economic analysis”,IEEE Comm. Mag. , vol. 34, no 12, December 1996, pp. 124- 134

[3] A. Zaganiaris et al, “A methodology for achieving life-cycle costs of optical access networks-fromRACE 2087/TITAN,” Proc. 11th Annual Conference European Fibre Optic Communications andNetworks, June 1993, The Hague, the Netherlands.

[4] L.A. IMS, “Broadband Access Networks Introduction strategies and techno-economic evaluation,Telecommunications Technology And Applications Series, Chapman & Hall 1998, ISBN 0 412 828200

[5] D. Katsianis et al, “The economics of next generation mobile systems”, XIIIth International Symposiumon Services and Local accesS (ISSLS 2000), Stockholm, Sweeden June 18-23, 2000.

[6] Donal O’ Mahony, “UMTS: The Fusion of Fixed and Mobile Networking”, IEEE Internet ComputingJanuary-February 1998, pp. 49-56

[7] UMTS Forum, Report #5, 1998.

[8] Anne Cerboni et al, “Evaluation of demand for next generation mobile services in Europe”, ThirdEuropean Workshop On Techno-Economics For Multimedia Networks And Services, Aveiro 14 –16December 1999.

[9] http://www.infowin.org/ACTS/RUS/PROJECTS/FRAMES/index.html

[10] http://www.infowin.org/ACTS/RUS/PROJECTS/ac015.htm

[11] Kjell Stordahl, Leif Aarthun Ims, Marianne Moe, “Broadband market - the driver for networkevolution”, Proc Networks 2000, Toronto, Canada, September 10-16, 2000.

[12] Investment Decisions: The certainty case in Copeland, T E, Weston, J F. Financial Theory andCorporate Policy. Addison-Wesley Publishing Company, Reading Massachussets, 1992.

8 Acronyms

BSC Base Station Controller NSS Network SubsystemBSS Base Station Subsystem OMC-R Operations & MaintenanceBTS Base Transceiver Station RNC Radio Network ControllerEDGE Enhanced Data for GSM Evolution RNS Radio Network SubsystemETSI European Telecommunications Standards Institute SGSN Serving GPRS Support NodeFDD Frequency Division Duplex TDD Time Division DuplexGb Interface Gb between BCS and SGSN TRX Transmitter-receiverGGSN Gateway GPRS Support Node UMTS Universal Mobile Telecommunication ServiceGPRS General Packet Radio Service UTRAN UMTS Terrestrial Radio Access NetworkGSM Global System for Mobile communications VLR Visitor Location RegisterHLR Home Location Register WAP Wireless Applications ProtocolHSCSD High Speed Circuit Switched Data W-CDMA Wideband Code Division Multiple AccessMSC Mobile Switching Center WRC World Radio Conference

17

9 BIOGRAPHIES

D. KATSIANIS (d.katsianis@di.uoa.gr) received the Informatics degree and the M.Sc. in Signal Processing andComputational Systems from the University of Athens. He is a research fellow with the Optical CommunicationsGroup, University of Athens, participating in European R&D projects in Photonic as well as Horizontal domains.He has worked as an expert scientific advisor for OTE S.A. (Hellenic Telecom) in networking aspectsconcerning simulation of ATM networks, Management in an Optical network and TelecommunicationInvestment Analysis, participating in several Eurescom projects. He is also a consulting partner with severalfirms in the field of techno-economic & network design studies, participating in ACTS projects AC226OPTIMUM and AC364 TERA as well as IST 25172-TONIC. His research interests are broadbandcommunications and methodology of network design with techno-economic aspects.

D.VAROUTAS (arkas@di.uoa.gr) received the Physics degree and the M.Sc. in electronics and radio-communications from the University of Athens. He is a research fellow with the Optical CommunicationsGroup, University of Athens, participating in numerous European R&D projects in the RACE, ACTS,Telematics and IST framework. He is also a consulting partner with several firms in the field of techno-economic studies and R&D projects. He is an adviser in several organizations including EETT (Greek NRA fortelecommunications) in the fields of licensing, spectrum pricing and legislation. His research interests arecomponents modeling, optical and microwave communications and techno-economic evaluation of networkarchitectures and services.

T.SPHICOPOULOS (tsfikop@cc.uoa.gr) received the Physics degree from the University of Athens in 1976, theD.E.A. degree and Doctorate in electronics from the University of Paris VI in 1977 and 1980 respectively, theDoctorat Es Science from the Ecole Polytechnic Federal de Lausanne in 1986. From 1977 to 1980 he was withThomson-CSF and at 1980 he joined the Electromagnetic Laboratory of EPFL. Since 1980, he is with theOptical Communications Laboratory of the University of Athens. Currently is a professor at Department ofInformatics at the same university. He is author or co-author of almost 70 scientific papers in the fields ofTelecommunications and Optical Communications. He has participated in a number of European and Nationalprojects and he has been auditor and evaluator of RACE and ACTS projects. He is an adviser in severalorganizations including EETT (Greek NRA for telecommunications).

Mr. ILARI WELLING (ilari.welling@nokia.com) received his M.Sc. degree in telecommunication engineeringfrom the Helsinki university of technology in 1996. He joined the Nokia Research Center in 1995, where he hasbeen studying the techno-economic aspects of telecommunication networks. Recently he has been working onthe European Union projects AC364 TERA and on IST 25172 TONIC in which he is the project manager. Mr.Welling has authored/co-authored several international journal articles and has presented papers in papers inconferences in field of telecom techno-economics. He has co-authored a book Broadband Access Networks:Introduction strategies and techno-economic evaluation" published by Chapman&Hall in 1998 (ISBN 0-412-82820-0).