Fcc Omnibus Broadband Initiative (Obi) Technical Paper Mobile Broadband Benefits of Additional Spectrum

8/3/2019 Fcc Omnibus Broadband Initiative (Obi) Technical Paper Mobile Broadband Benefits of Additional Spectrum

1/27

F C C S T A F F T E C H N I C A L P A P E R

FederalCommunicationsCommission

44512thStreet,SW

Washington,DC20554

MOBILE BROADBAND:THE BENEFITS OF

ADDITIONAL SPECTRUM

OCTOBER 2010

F E D E R A L C O M M U N I C A T I O N S C OM M I S S I O N | M O B I L E B R O A D B A N D : T H E B E N E F I T S O F A D D I T I O N A L SP E C T R U M 1


2/27


FederalCommunicationsCommission

MobileBroadband:TheBenefitsofAdditionalSpectrum

Abstract

The National Broadband Plan recommended that the Commission make available 500

megahertz (MHz) of new spectrum for wireless broadband, including 300 MHz for mobile

flexibleusewithinfiveyears. Inaddition,thePresidentdirectedinaJune28,2010,Executive

Memorandumthat500MHzofnewspectrumformobileandfixedbroadbanduse. Thispaper

provides additional technical analysis to validate the need for additional mobile broadband

spectrumin

the

near

term,

and

estimates

the

value

created

by

making

new

spectrum

available.

Giventhe inherentuncertaintyofanyforecastofthefuture,thegoalofthisanalysisisnotto

reach definitive numeric findings of spectrum need and economic benefit, but to make a

reasonabledemonstrationthatmobiledatademandislikelytoexceedcapacityundercurrent

spectrum availability in the nearterm, and that meeting this demand by making additional

spectrumavailableislikelytocreatesignificanteconomicvalue. Ouranalysissuggeststhatthe

broadband spectrum deficit is likely to approach 300 MHz by 2014, that making available

additionalspectrum formobilebroadbandwouldcreatevalue inexcessof$100B inthenext

fiveyearsthroughavoidanceofunnecessarycosts. Thisestimateofvaluecreationisnarrow,as

it does not account for the broader social value created through mobile broadband, which

some

economists

estimate

as

multiples

of

the

private

value.

Since

making

new

spectrum

availablehashistoricallytakenbetweensixandthirteenyears,andsincemobiledatagrowth

trends are expected to continue beyond the nearterm forecast in this paper, these results

supporttheneedfortimelyactiontofreespectrumformobilebroadband,consistentwiththe

recommendationsoftheNationalBroadbandPlanandthePresidentsdirective.



3/27


TableofContents

I. Introduction_____________________________________________________________ 4

II. MethodologicalOverview___________________________________________________5

III. TrafficDemandandNetworkCapacity _______________________________________ 6

IV. ModelInputs:DataDemand,NetworkDensity,SpectralEfficiency,andBaselineSpectrumUse 8

a. TrafficDemandIndustryWideTrendsThrough2014 ______________________________________9

b. NetworkDensity CellSiteGrowth _____________________________________________________12

c. NetworkCapacity SpectralEfficiency___________________________________________________ 14

d. BaselineSpectrumUse _______________________________________________________________ 15

V. ModelOutput:SpectrumNeedandEconomicValue __________________________________ 17

VI. SensitivityAnalyses _____________________________________________________________ 22

a.

Sensitivityto

Data

Growth

Forecast_____________________________________________________22

b. SensitivitytoCurrentSpectrumUsedforData ____________________________________________23

c. SensitivitytoNetworkDensityCost _____________________________________________________23

VII. Conclusion _____________________________________________________________ 26

AppendixA:MethodologySummaryofMobileDataDemandForecasts__________________27



4/27


I. Introduction

TheNationalBroadbandPlanrecommendedthattheCommissionmakeavailable500

megahertz(MHz)ofnewspectrumforwirelessbroadband,including300MHzformobile

flexibleuse

within

five

years.1

In

addition,

the

President

directed

in

aJune

28,

2010,

Executive

Memorandumthat500MHzofnewspectrumformobileandfixedbroadbanduse.2 Thedrive

tomakeavailablenewspectrumforbroadbandisgroundedinstrongconsumerdemandfor

highspeedwirelessInternetaccess.

Datausageoverwirelessnetworksisrapidlyincreasingasmoreconsumerssurftheweb,check

email,andwatchvideoonmobiledevices. InjustthelatestsixmonthsofFCCreporting,

subscriptionstomobiledataservicesincreasedby40%.3 Andtheamountofdatausedby

wirelessconsumersisincreasingsubstantiallyexhibit1belowshowsanincreaseofover450%

intheamountofdataconsumedperlinebetweenthefirstquarterof2009andthesecond

quarterof

2010.

4

Exhibit1:DataConsumptionGrowthperLine

Average Data Used Per Line

201.8182.5

129.1

98.9102

43.5

0

50

100

150

200

250

Q1 2009 Q2 2009 Q3 2009 Q4 2009 Q1 2010 Q2 2010

DataUsed

(megabytes)

Arangeofrecentwirelessindustrytrendsindicatestronggrowthofmobiledatausage.

42%ofconsumersareestimatedtoownasmartphone,upfrom16%threeyearsago.5

1See

National

Broadband

Plan

Chapter

5,

Spectrum,

at

www.broadband.gov

2PresidentBarackObama,June28,2010,MemorandumfortheHeadsofExecutiveDepartmentsandAgencies,Unleashing

theWirelessBroadbandRevolution. Seehttp://www.whitehouse.gov/briefingroom/presidentialactions/presidential

memoranda.3InternetAccessServices:StatusasofJune30,2009,releasedSeptember2010bytheFCCsWirelineCompetitionBureau.

4ValidasLLCdata,September8,2010. Foundatwww.myvalidas.com.

5PaulCartonandJeanCrumrine,NewSurveyShowsAndroidOSRoilingtheSmartphoneMarket,ChangeWave

Research,Jan.4,2010,availableathttp://www.changewaveresearch.com/articles/2010/01/smart_phone_20100104.html.

Alsofoundatpara.159oftheFCCs14th

MobileWirelessCompetitionReport. DataperiodcitedisbetweenOctober2006and

December2009.

http://www.changewaveresearch.com/articles/2010/01/smart_phone_20100104.htmlhttp://www.changewaveresearch.com/articles/2010/01/smart_phone_20100104.html


5/27


PCaircardusersconsume1.4gigabytes(GB)permonth 56timestheamountofdatausedbyaregularcellphone

6

AT&T,theexclusiveUScarrieroftheiPhone,hasseenmobilenetworktrafficincrease5,000%overpast3years.

7

UsersofClearwiresfourthgeneration(4G)WiMAXserviceconsume7GBpermonth

280times

the

amount

of

data

used

by

aregular

cell

phone.

8

Lookingahead,industryanalystsgenerallysharetheviewthatmobilenetworkdatatrafficwill

continueasignificantupwardtrend. Assmartphones,laptops,andotherdevicesbecome

increasinglyintegraltoconsumersmobileexperiences,mobiledatademandisexpectedto

growbetween25and50timescurrentlevelswithin5years.

TheNationalBroadbandPlan(Plan)recognizestheenormouspotentialofmobiledatagrowth.

Toensurethatadequatewirelesscapacityisavailabletomeetthisdemand,thePlancallsforan

additional500megahertzofnewspectrumtobemadeavailableoverthenexttenyears,

includinganadditional300megahertzofspectrumsuitableformobileflexibleusewithinfive

years.

ThispaperprovidesthetechnicalbasisforthisfundamentalrecommendationofthePlan.

Specifically,byestimatingvariousfactorsaffectingaggregatemobilenetworkcapacityand

mobiledatagrowth,itisclearthat:1)mobiledatademandwilloutstripavailablewireless

capacityinthenearterm;andthat,2)makingavailable300MHzofadditionalspectrumfor

mobilebroadbandislikelytoentaileconomicvalueofatleast$100billioninthenextfiveyears.

Also,beyondthefiveyearforecastperiodinthispaper,mobiledatademandisexpectedto

continueitsstronggrowth,supportingtheneedforactionoverthelongertermtomake

availableadditionalspectrumformobilebroadband.

II. MethodologicalOverview

Thispaperexploreshowmobilebroadbandgrowthwillaffectspectrumneeds. Inaddition,

recognizingthatincreasingtheamountofspectrumavailableisonealternativeamongseveral

toincreasemobilenetworkcapacity,thispaperalsoestimatestheeconomiceffectsofmaking

availablenewspectrumtomeetmobiledatademand.

First,industrywidetrendsandforecastsareusedtopredictspectrumneeds. Industryanalysts

havepredictedtotalmobiledatatrafficrelativetocurrentlevels,andtrendssuchasthe

6ValidasLLCdata,foundinthe14

thMobileWirelessCompetitionReportoftheFederalCommunicationsCommission,at

para.182,releasedMay20,2010.Traditionalhandsetsareestimatedtoconsumeapproximately25MBpermonth.7KrisRinne,Sr.VicePres.ofArchitecture&Planning,AT&T,RemarksattheFCCSpectrumWorkshop1112(Sept.17,2009),

availableathttp://www.broadband.gov/docs/ws_25_spectrum.pdf.Ms.Rinneaddedthatinadditiontoincreaseddatausage,

voiceusagecontinuestorisealso.Id.8http://gigaom.com/2010/03/12/clearwiresbigbetonourbroadbandaddiction/.WenotethatClearwiresubstypicallyuse

theserviceasasubstituteforwiredbroadband.Traditionalhandsetsareestimatedtoconsumeapproximately25MBper

month.

http://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://gigaom.com/2010/03/12/clearwires-big-bet-on-our-broadband-addiction/http://www.broadband.gov/docs/ws_25_spectrum.pdf


6/27


increasingnumberofcellsitesareknownandcanbeextrapolatedforward. Byadjustingthe

expectedgrowthindatademandforoffsettinggrowthinnetworkdensity(whichistheresultof

addingnewcellsites)andspectralefficiency,wecanforecastfuturespectrumneedsrelativeto

abaselineindexofcurrentspectruminuse. Therefore,reasonableapproximationofboththe

baselineandfuturetrendsassociatedwithmobiledatauseandnetworkcapacityenablea

realisticprojection

of

spectrum

need.

While

the

values

of

these

inputs

to

the

model

may

vary

inpractice,theincorporationoftrendsanticipatedbyrelevantexpertsallowsforacogent

forecast.

Theoutputfromthisapproachisthereforeanaggregatenationalprojectionoflikelyspectrum

needs,whichislikelytomaskdifferencesacrossmarkets,butisnonethelessusefulfor

approximatingtheeconomiceffectsofmakingnewspectrumavailableformobilebroadband.

Theprojectionsofmobiledatademandusedinthisanalysisarebasedinpartonhistoric

marketdynamics,suchasallyoucaneatpricingfordata. Theeffectofnewpricingstrategies

onconsumerdatademandisnotyetknown,buthasthepotentialtoimpactdatatraffic

projectionsifwidelyadoptedinthemarket.

Thispaperwilldemonstratethatevenwhenusingconservativeassumptionsaboutthemarket

factorsthataffectspectrumneed,itislikelythatspectrumwillbecomeanincreasinglyscarce

resourceinthenearterm,andthatfreeingspectrumformobilebroadbanduseoverthenext

fiveyearswillentailsignificanteconomicbenefits.

Theestimateofvaluecreatedbyreleasingnewspectrumforbroadbandisnarrow,andlimited

totheavoidanceofunnecessarycostsbutforthenewspectrum. Thispaperdoesnot

undertakeacomprehensiveanalysisofthebenefitstosocietythatmayresultfrommakingnew

spectrumavailable,whichsomeeconomistsestimateasmultiplesoftheprivatevalue.9

These

findings,taken

together,

clearly

support

the

need

for

timely

action

to

make

new

spectrum

availableformobilebroadband.

III. TrafficDemandandNetworkCapacity

Todeterminetheneedforandbenefitsofreleasingadditionalspectrumforbroadband,the

firststepistoanalyzethedriversofmobiletrafficdemandandtotalavailablenetworkcapacity.

AsillustratedinExhibit22below,theleftsideoftheexhibit,demandandgrowthindemand

dependsondataconsumptionbydevicetypeandthenumbersofeachtypeofdeviceinuse.

9Theeconomicbenefitsestimatedhereinrepresentonlythereductionincostofmeetingmobiledatademand;no

approximationofproducerorconsumersurplusisoffered,andnoestimateofexogenouscostsisprovided. Thispaperdoes

notestimatethecoststothirdpartiesofmakingavailablenewspectrum,likelyauctionrevenue,orotherfactors. Fora

discussionofconsumerandproducersurplusinthewirelessmarket,seeHazlettandMunoz,AWelfareAnalysisofSpectrum

AllocationPolicies,RANDJournalofEconomics,V.40,No.3,Autumn2009. AlsoseeGregoryRosston,TheLongandWinding

Road:TheFCCPavesthePathwithGoodIntentions,TelecommunicationsPolicy,V.27,No.7,August2003;and,JerryHausman

"ValuingtheEffectofRegulationonNewServicesinTelecommunications",BrookingsPapersonEconomicActivity,

Microeconomics.Vol.28(1997).



7/27


AnexampleofdemandsidegrowthisapparentinAT&Tsrecentexperienceasthenumberof

iPhoneusershasincreased,mobilenetworktraffichasincreased5000%.10

Networkcapacity(theamountofdataanetworkcancarry),ontheotherhand,dependson

spectralefficiencyofwirelesstechnologies(theamountofspectrumneededtotransmitagiven

amountof

data),

available

spectrum,

and

the

number

of

cell

sites

used

to

provide

service.

11

In

thissense,newspectrumissubstitutable,toapoint,withcapitalexpenditurestobuildnew

cellsitesanddevelopandimplementmoreefficientwirelesstechnologies.

Therefore,onemeansofapproximatingthemarginalbenefitofallocatingadditionalspectrum

formobilewirelessusageisbydeterminingthereductioninincrementalcostsofsubstitutes,

suchasnewcellsitesandtechnologies.Ineffect,weconsideranindifferencecurvebetween

additionalspectrumandadditionalnetworkinvestment. Theindifferencecurveplotsthe

numberofsitesandtheirrespectivecostagainstthespectrumrequiredforanticipatedmobile

datademandin2014. Thismethoddemonstratesdifferentcombinationsofspectrumand

capitalneededtomeetprojectedmobiledatademand.

Exhibit2.Driversofmobiletrafficdemandandmobilenetworkcapacity

Drivers of mobiletraffic demand

Drivers of availablenetwork capacity

Capacity per cell-siteCapacity per cell-site

Usage by device typeUsage by device type

Penetration of devicetypes


Spectral efficiency ofwireless technology


Spectrum allocationSpectrum allocation

Number of cell-sitesNumber of cell-sites

Wirelessnetwork

Drivers of mobiletraffic demand

Drivers of availablenetwork capacity

Capacity per cell-siteCapacity per cell-site

Usage by device typeUsage by device type





Spectrum allocationSpectrum allocation

Number of cell-sitesNumber of cell-sites

Wirelessnetwork

Thisframeworkofmobiledatatrafficandcapacityenablesanunderstandingoftheeffectsof

releasingadditionalspectrum. Assumptionsinthispapercouldbeamendedtoyieldsomewhat

differentnumericalbenefitestimates;however,anyreasonableapproximationofrelevant

variablesis

likely

to

confirm

the

conclusion

that

releasing

additional

spectrum

will

create

substantialeconomicvalue.

10KrisRinne,Sr.VicePres.ofArchitecture&Planning,AT&T,RemarksattheFCCSpectrumWorkshop1112(Sept.17,2009),

availableathttp://www.broadband.gov/docs/ws_25_spectrum.pdf.Ms.Rinneaddedthatinadditiontoincreaseddatausage,

voiceusagecontinuestorisealso.Id.11

SeealsoFCCOBITechnicalPaper,TheBroadbandAvailabilityGap,releasedApril2010.

http://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://www.broadband.gov/docs/ws_25_spectrum.pdfhttp://www.broadband.gov/docs/ws_25_spectrum.pdf


8/27


IV. ModelInputs:DataDemand,NetworkDensity,SpectralEfficiency,andBaselineSpectrumUse

Futurespectrumneedscanbeunderstoodasafunction,ormultiplier,ofcurrentspectrum

usedfor

mobile

broadband

nationwide.

The

multiplier

used

in

our

model

is

based

on

an

averageofreputableindustryanalystmobiledatademandforecasts,adjustedtoaccountfor

additionalnetworkdensityviacellsitegrowthandimprovementsintechnologyresultingin

increasedspectralefficiency.

Byassumingthatspectrumrequiredtosupporttraditionalvoicewillremainatcurrentlevels,12

thetotalspectrumrequiredinfutureyearsiscalculatedbyapplyingthemultiplieronlyto

spectrumusedfordataduringthe2009baselineyear. Currentvoicespectrumisthenaddedin

eachyeartoreachthetotal. Thefollowingflowchartdepictstheanalyticstepsinthismodel:

Exhibit3.TopDownForecastFlowchart

DataTraffic

DemandForecasts

Data Trafficper SiteGrowth

SiteGrowth

Forecast

SpectralEfficiencyForecast

Data Trafficper Site growth

adjusted fortechnology

Average DataTraffic

Demand

2009SpectrumUtilization

2009 Voice

Spectrum

2009 DataSpectrum

Future DataSpectrumRequired

Total FutureSpectrumRequired

12Estimatingtherateofvoicegrowth,therateofvoicetrafficmigrationfromcircuitswitchedtoVoIP,andtheassociated

spectralefficiencygains,ascomparedwithcountervailingconsiderationssuchastheproliferationofunlimitedtalkplans,are

beyondthescopeofthispaper. Theactualspectrumrequiredtosupporttraditionalvoiceinthefuturemaybemoreorless

thanassumedhere;however,giventhemagnitudeofprojectedmobiledatatrafficgrowth,anyvariancefromtheconstant

usedinthismodelwillnotsignificantlychangetheresultingprojectionoffuturespectrumneeds. Inparticular,mobilecarriers

arenotexpectedtofullytransitionvoiceservicesto4GInternetProtocolwithinthefiveyearforecastperiod.



9/27


a. TrafficDemandIndustryWideTrendsThrough2014

AnalystsandindustryprofessionalshaveforecastedsignificantfuturegrowthinUSmobiledata

usage,thoughtheprecisemagnitudeofprojectedgrowthvariesbetweenforecasts. Acertain

degreeofuncertaintyisexpectedinanyprojectionofthefuture. Sincetheforecasts

collectivelyprojectarangeoffuturedemand,takinganaverageofthreeinformedindustry

analysesnormalizes

the

variance

across

projections.

The

analyses

used

in

this

paper

are

from

therespectedindustrysourcesofCiscoSystems,CodaResearch,andtheYankeeGroup. Their

quantitativeprojectionsoffuturemobiledatademandareshowninthechartbelow,expressed

relativetoactual2009datatraffic.13

Exhibit4.IndustryMobileDataForecasts

Industry Forecasts of Mobile Data Traffic

0X

5X

10X

15X

20X

25X

30X

35X

40X

45X

50X

2009 2010 2011 2012 2013 2014

TrafficRelativeto

2009

Cisco

Coda

Yankee Group

Average

Theaverageofthesethreeprojections,asseeninExhibit4,revealsstrongexpectedgrowthin

mobiledatatrafficfrom2009levelsbyafactoroffiveby2011,morethan20timesby2013,

andreaching35times2009levelsby2014. Inallthreeforecasts,thetrendremainsupwardin

2014,implyingcontinuedgrowthbeyondtheforecastperiod.

Mobiletrafficdemandisdrivenbythedatausagecharacteristicsofeachdevicetypeandthe

numbersofeachtypeofdeviceinuse.Webrieflyreviewtheseinthesectionsbelow.

Usagebydevicetype

Deviceswithenhancedfunctionalitytendtoconsumemoredata. Thisisdemonstratedinestimatesofmobiledatauseduringthesecondhalfof2009byindependentindustryanalyst

Validas. Blackberry,withsuperioremailutility,consumestwicetheamountofdatamonthlyas

anormalmobilehandset. TheiPhone,usefulformobilewebbrowsingandapplications,

13CiscoVisualNetworkingIndex:GlobalMobileDataTrafficForecastUpdate,20092014,February9,2010. CodaResearch

Consultancy,USMobileTrafficForecasts:20092015,2009. YankeeGroup,SpectrumRichPlayersAreintheDriversSeatfor

MobileBroadbandEconomics,June2009. SeeAppendixAofthispaperforamethodologicalsummaryofeachforecast.



10/27


consumesfivetimesthedatamonthlyasBlackberry. Andaircards,facilitatingafullmobile

computingexperience,consumefivetimesmoredatamonthlythantheiPhone.14

These

devices,aggregatedacrossthenetwork,leadtosignificanttotalmobiledataconsumption.

CiscoSystemshasproducedanestimateofaggregatemobiledataconsumptionbydevicetype

between2009

and

2013,

shown

in

Exhibit

5below.

15

Consistent

with

Validas

estimates,

traffic

generatedfromnetbooksandnotebooksishighest,especiallywhenthewirelessnetworkis

usedasasubstituteforwiredbroadbandconnection.

Exhibit5.EstimateofmobiletrafficbydevicetypeandforecastsProjections of usage by device type

MB/month/device

0

5,000

10,000

15,000

20,000

20132012201120102009

Portable* (Complement to Fixed Broadband)

Portable* (Substitute for Fixed Broadband)

Smartphones

3G Feature phones

* Portable refers to computing devices (netbooks and notebooks), tablets, handheld gaming consoles, e-readers, digitalcameras and camcorders, digital photo frames, and in-car entertainment systems

ThecorrespondinggrowthrateintheamountofdatausedbydevicetypeisshowninExhibit6.

Notetherapidusagegrowthprojectedforalldevicetypes,whichsuggestsadoublingofusage

almosteverytwoyearsorlessacrossalldevices.

14ValidasLLC,http://www.myvalidas.com/industryanalytics.aspx. Validasisacompanythatofferswebbasedwirelessbill

analysisandoptimizationservicestoconsumersandbusinesses,usingdatagainedfromsuchanalysistoprovideinsightto

industrytrends.15

ExPartefilingofCiscoSystemsinGNdocket0951,filedMarch23,2010.



11/27


Exhibit6.ProjectedgrowthrateoftheamountofdatausedbydevicetypeProjected usage growth rateby device type, 2008-2013

Percent

4734

84

67

3G Feature phones

Smartphones



* Portable refers to computing devices (netbooks and notebooks), tablets, handheld gaming consoles, e-readers,digital cameras and camcorders, digital photo frames, and in-car entertainment systems

Whenportablessuchasnetbooksandnotebooksareusedexclusivelyonthewirelessnetwork

(i.e.,thewirelessnetworkisusedasasubstituteforthewirelinenetwork),growthissimilarto

whatisseenonfixednetworks.16

TheaveragemonthlyusagepersubscriberontheClearwire

network,which

many

consumers

use

as

asubstitute

for

wired

broadband,

is

already

7GB.17

Continuedgrowthofthisdevicesegmentislikelytocontributesignificantlytothegrowthof

mobiledatatraffic.

Penetrationbydevicetype

Thenumbersofeachtypeofdeviceinuse(penetration),aswellastheexpectedgrowthin

penetration,isshownintheexhibitbelow.Smartphonesincorporatethefeaturesoftraditional,

voiceonlyhandsets. So,increasingpenetrationofsmartphonessuggestsadecreasing

penetrationoftraditionalhandsets.

Notethe

rapid

growth

in

penetration

expected

for

portables

the

device

category

also

correspondingtohighestusage(seeExhibit7). Further,withrelativelymodestpenetration

levelsforportablestodayandampleroomforgrowth,rapidpenetrationgrowthcouldcontinue

beyond2014.

16CiscoVisualNetworkingIndex:GlobalMobileDataTrafficForecastUpdate,20092014,February9,2010.

17RemarksofClearwireCEOBillMorrowattheInternationalCTIAWireless2010show,ClearwireExtends4GLeadershipInthe

UnitedStates,March23,2010.



12/27


Exhibit7.Forecastednumbersofeachtypeofdeviceinuse(penetration)andcompound

annualgrowthrate(CAGR)

Projected penetrationby device type, 2014

Millions

10.7

127.1

153.7

51.8

Basic and feature phones

Smartphones



CAGR of penetration bydevice type, 2009-14

Percent

-6

14

61

37

* Portable refers to computing devices (netbooks and notebooks), tablets, handheld gaming consoles,

e-readers, digital cameras and camcorders, digital photo frames, and in-car entertainment systems

Mobiledatatrafficisakeydriverofspectrumneed,andastheseforecastsshow,theconsensus

amongindustryexpertsisthatdemandforsuchtrafficislikelytoincreasesignificantlyoverthe

nextfiveyears.

Thefollowingsectionsdescribeotherfactorssuchascellsitegrowthandspectralefficiency

thatarerelevanttoestimatingspectrumneed.

b. NetworkDensity CellSiteGrowth

Wirelessoperators

continue

to

invest

capital

in

building

new

cell

sites.

Cell

sites

can

be

engineeredfordifferentpurposes,includingincreasingcapacity,toimprovesignalquality,and

toexpandcoverage. Thelatteremphasisexpandstheoperatorsfootprint;theprimary

purposeofothersocalledinfillsitesistoincreasecapacity,andinthisrespectcanbe

conceivedasanalternativetonewspectrum. Theamountofcapacityanyparticularnewsite

addstoawirelessnetworkisdifficulttocharacterize;therefore,wefocusonnationalaverages

andtrends.

AccordingtodatapublishedbyCTIA,thenumberofcellsitesintheUnitedStateshasbeen

growingatabouta7%compoundannualgrowthrate(CAGR)overthepastfiveyears.18

Thisis

seenin

the

following

table:

18CTIA



13/27


Exhibit8.USCellSiteGrowthsince1997

Date TotalCellSites

Year

over

Year

Growth

5Year

Running

Avg

Growth

1997 38,650

1998 57,674 49%

1999 74,157 29%

2000 95,733 29%

2001 114,059 19%

2002 131,350 15% 28%

2003 147,719 12% 21%

2004 174,368 18% 19%

2005 178,025 2% 13%

2006 197,576 11% 12%

2007 210,360 6% 10%

2008

220,472

5%

8%

2009 245,912 12% 7%

AsExhibit8shows,cellsitegrowthyearoveryearhasbeenuneven,buttheoveralltrendinthe

fiveyearaverageisthatthegrowthhasbeenslowing.19

Thistrendispartlyafunctionof

saturationofwirelessavailabilityasanestimated99.6%ofthepopulationnowhaswireless

access,20thereforenewcellsitegrowthwillemphasizeincreasedcapacity.

Projectingnewcellsitegrowthisanimportantinputtospectrumdemandforecasts. Forthe

purposeofthisanalysiswewillassumeflatgrowthatthe2009levelof7%overthenextfive

years. Thisisamoreconservativeapproachthanprojectingintothefuturetheflattening

growthseen

in

recent

years,

since

new

infill

cell

sites

that

add

capacity

are

asubstitute

for

newspectrum. Additionally,weassumeallfuturecellsitegrowthwillemphasizecapacityover

coverage,whichisalsoconservativefromthestandpointofestimatingspectrumneedssince

newcoveragesitesdonotaddresscapacityconstraints.21

Theassumptionofconstantcellsitegrowthatcurrentlevelsenablesestimationofspectrum

demandaswellasthebenefitsofreleasingadditionalspectrum. Werecognizethatsincenew

cellsitescanbeasubstitutefornewspectrum,onemightforecastrenewedgrowth

accelerationofcellsiteconstructionasmobiledatatrafficincreases,absentconstraintsrelating

tolocationscarcity. Indeed,therateofcellsitegrowthincreasedfrom2008to2009,andone

19Theslowdowninsitegrowthislikelyasignofamaturingindustry,butmayalsoreflecttheincreasingdifficultyofacquiring

sitesduetoscarcityoflocations,zoning,andotherconstraints.TheFCCrecentlytookactiontoalleviatesomeoftheregulatory

barrierstositeexpansion.SeeNovember18,2009DeclaratoryRulinginWTDocket08165.20

FCCestimatebasedonAmericanRoamerdata. FoundintheFCCs14th

MobileWirelessCompetitionReport,atpara.42.21

Sincethispaperisfocusedonmobiledatatraffic,strategiestooffloadtraffictofemtocellsandWiFiisnotdirectly

considered. Inaddition,therolloutofsuchnetworkarchitecturestrategieshasbeenslowtodate,anditseffectsareunclear.

Foradiscussionofthisissue,seeforexample,RysavyResarch,MobileBroadbandCapacityConstraintsandtheNeedfor

Optimization,updatedFebruary24,2010.



14/27


mightexpectthistocontinueifnoadditionalspectrumwastocomeavailable. However,by

holdingcellsitegrowthconstant,weareabletocomparethemarginalcostofmeetingmobile

datademandbyaddingnewcellsitesrelativetoaddingnewspectrum. Doingsodemonstrates

thataddingnewspectrumwillyieldsubstantialeconomicvalue.

c. NetworkCapacity SpectralEfficiency

Aswirelesstechnologieshaveevolved,sohasperformance. Witheveryevolution,theindustry

hasachievedhigherpeakdatathroughputs,improvedspectralefficienciesandlowerlatencies

(delays). Forexample,4Gusesanative,allIParchitecture,thusbenefittingfromthe

innovationandcostefficiencyofapacketswitchednetwork,relativetolegacycircuitswitched

technology.

Themostimportantdimensionofwirelessnetworkperformanceisspectralefficiency,typically

measuredinbitspersecondperHertz. Thismetricreflectstheamountofdataasectorcan

transmit

on

a

normalized

time/bandwidth

basis.

As

such,

spectral

efficiency

drives

average

downlinkdatacapacityofacellsitelinearly.Exhibit9showstheevolutionoftheaverage

downlinkdatacapacitiesofasinglesectorina3sectorcellsitefortheGSMfamilyofwireless

standards.22

While4Gtechnologieshaveincreasedspectralefficiencyrelativetopriorgenerationair

interfacestandards,suchgainsarelikelytobecomesmallerinthefuture.Indeed,4Gstandards

achieveimproveddatathroughputbythecombinednoveluseofseveraltechnologies,andwe

donotassumethatfuturegainsinthroughputwillberealizedviafurtherinnovationontheair

interfaceonaperdevicebasis,particularlywithinthefiveyeartimehorizonthatistheprimary

focusofthispaper.23

However,as4Gcarriersincreaseasaproportionoftotalcarriersoverthe

nextfive

years,

it

is

likely

that

overall

spectral

efficiency

will

continue

to

improve.

Exhibit9.Evolutionofdownlinkspectralefficiency24

0.720

0.480

0.0

0.5

1.0

1.5

LTE

1.36-1.5

HSPA,

Rel 7

1.08-1.29

HSPA,

Rel 6

HSDPA,

Rel 5

WCDMA

0.16-0.24

EDGE

0.09

GPRS

0.03

Downl ink spectral efficiency by technologyBps/Hz

22QualcommExPartefiling,SpectralEfficiency,GNDocketNo.0951

23SeealsoOBITechnicalPaper,TheBroadbandAvailabilityGap,foradditionaldiscussion.

24Calculatedfora(paired)2x10MHzspectrumallocationforalltechnologies.WCDMAperformancebasedon1x1and1x2

antennaconfigurations;HSDPARel5andHSPARel6resultsbasedon1x1and1x2configurations,respectively.HSPARel7

performanceassumes1x2and2x2configurationswhileLTEresultassumes2x2.Performanceof(3G)EVDO,whichisnot

showninthechart,iscomparableto(3G)HSPA.



15/27


Exhibit9aboveshowsthespectralefficiencyofairinterfacestandardsastheyhaveevolved

from2G(GPRS)to3G(HSDPA)to4G(LTE). Theoverallspectralefficiencyofallwireless

networkscanbeviewedasaweightedaverageofthetechnologymix. Overthenextfiveyears

themixislikelytoshiftfrom2Gand3Gtechnologiestoprimarily3Gand4Gtechnologies.25

Ouranalysisassumesthatthiswilleffectivelydoubletheaveragespectralefficiencyofwireless

networks,from

about

0.625

bps/Hz

in

2009

to

1.25

bps/Hz

in

2014.

26

We

should

continue

to

seeincreasedmigrationfrom3Gto4Gtechnologiesbeyond2014,resultinginanoverall

improvementinnetworkspectralefficiency. Thesespectralefficiencygainsservetoreduce

networkcapacityaugmentationachievedbyaddingspectrumbandwidthtomeetmobiledata

demand. Evenwhenaccountingforthisfactor,however,itisclearthatadditionalspectrumwill

beneededtomeetmobiledatademand.

d. BaselineSpectrumUse

Thisforecastoffuturespectrumneedsisbasedonmobiledatademandgrowthexpectedfrom

2009forward,withoffsettingadjustmentsmadefortheeffectofnewcellsitesandincreasing

spectral

efficiency.

Therefore,

we

must

estimate

how

much

spectrum

is

used

in

2009,

including

theproportioncurrentlyusedtosupportcircuitswitchedvoicerelativetotheamountusedfor

data.

TheFCCs14th

annualMobileWirelessCompetitionreportdescribedindustrytrendsrelatedto

wirelessvoiceservice. Ofnote,the14th

reportnotedthenearsaturationofthewirelessvoice

marketandtheflatteningofwirelessvoiceminutesofuse,perhapsduetosubstitutionbydata

services,whichareincreasing. Atthesametime,prepaidplanswithunlimitedvoiceminutes

areincreasingandlandlinephoneserviceislosinggroundtowirelessservicetrendsthatcould

offsetfurtherdeclineinwirelessvoice. Therefore,ourpresentmodelwillassumethatthe

spectrumneededforvoiceservicesremainsatcurrentlevels. Thespectrumimpactsofany

futuregrowth

in

voice

services,

driven

via

prepaid

services,

population

growth,

substitution

of

wirelessserviceforwireline,orotherfactors,islikelytobeoffsetinthelongtermbymore

efficienthandlingofvoicetrafficthrough4GVoiceoverInternetProtocol,thoughwedonot

expectasignificanttransitionofaggregatevoicetrafficto4Gwithinthefiveyearforecast

period.27

Inestimatingthespectrumbaselinewehave,again,takenaconservativeapproach. The

NationalBroadbandPlannotedthat547MHz,intotal,iscurrentlylicensedunderflexibleuse

rules,whichallowsformobilebroadbandandvoiceservices. Ofthisamount,the170MHzthat

comprisethecellularandPCSbandsisthemostheavilyused. Themajorityoftheremaining

25See,forexample,forecastsbyWiseHarborathttp://www.wiseharbor.com/forecast.html

26Whiletechnologywillcontinuetoimprove,spectralefficiencyofcurrentOFDMbased4Gsolutionsisapproachingmaximum

expectedlimits. Whileabsoluterealizedvaluesofspectralefficiencymayvary,theprojectionofincreasingefficiencyis

generallyaccepted,anditsincorporationintothepresentmodelappropriatelyadjustsexpectednetworkcapacityforthisfactor.27

BasedondatafromCTIA,theFCCs14th

WirelessCompetitionreportshowsthatvoiceminutesdeclinedbetween2007and

2009,whileprepaidsubscribershavegrownasapercentageofthetotalsubscriberbaseoverthepastseveralyears, andalso

notedanationwidewirelesssubscriberrateofnearly90%. Asnotedinthereport,thedeclineinvoiceminutesofusecouldbe

relatedtosubstantialincreasesintextmessaging,asconsumerssubstitutealternativemethodsofcommunicationforstandard

voice.

http://www.wiseharbor.com/forecast.htmlhttp://www.wiseharbor.com/forecast.htmlhttp://www.wiseharbor.com/forecast.html


16/27


377megahertzhasbeenmadeavailablewithinthepastsixyearsandisjustnowcoming

online.28 Ofcourse,manyofthelargestmajormetropolitanareasarecurrentlyusingmore

than170MHzofspectrumtosupportexistingvoiceanddataservicesforexample,TMobile

hasexpandedits3GserviceinmajormarketsusingtheAWSband,andSprinthasrecently

launchedits4Gserviceusingthe2.5GHzband.

Sincespectrumisusedmostintensivelyinmajorurbanareas,wirelesscarriersgenerally

prioritizesuchareaswhenbringingnewspectrumonline. However,tomodelspectrumneeds

nationwide,wewilluse170MHzasourcurrentbaseline,sincethatamountbestreflects

frequencyuseacrossthecountry. Thisisaconservativebaselinefromtheperspectiveof

projectingfuturespectrumneeds,sinceasmallerstartingpointeffectivelyincreasesthe

availabilityofcurrentlylicensedbutnotyetbuiltoutfrequenciestoaccommodatefuture

mobiledatagrowth,therebyreducingtheadditionalspectrumthatwouldbeneededto

accommodatesuchgrowth.

Finally,becauseourforecastfocusesongrowthindatatraffic,thecurrentspectrumuse

estimatesmustdistinguishbetweenspectrumcurrentlyusedforvoiceversusspectrum

currentlyusedfordata. Informationaboutprovidersprivateallocationofspectrumbetween

voiceanddataservicesisnotsystematicallyreportedandthereforedifficulttoobtain.

Discussionswithvendorsandcarrierssuggestthatthesplitforsomeoperatorsinmajor

marketsisapproximately50%/50%. Forpurposesoftheforecast,wehaveestimatedthatonly

33%ofcurrentspectrumisusedtosupportdataservicesandthat67%iscurrentlyusedto

supportvoice. Thisapproachisconservativeintworespects. First,byassumingalarger

proportionofvoicespectrumuse,weprotectthemodelfromanyunanticipatedgrowthin

voiceusewhichisforecastedtobeflat. Second,byassumingasmallerproportionofdata

spectrumuse,wedecreasetheeffectofthemobiledatademandmultiplierthatisusedto

projectfuture

spectrum

needs.

Since

the

projected

growth

of

mobile

data

demand

is

much

greaterthananypossibledecreaseinvoiceuseorspectralsavingsrelatedto4Gvoicemigration

inthefiveyearforecastwindow,thisapproachensuresthatwedonotuseuncertaintyin

baselinespectrumusetoinflateourprojections.

Bymakingconservativeassumptionsontheamountofspectrumcurrentlyinuse,aswellasthe

proportioncurrentlyusedforvoicerelativetodata,wecangainareasonableprojectionof

futurespectrumneeds. Totheextentthatthisforecastisdifferentthanactualuseinthefuture,

itismorelikelytounderestimatefuturespectrumneedthantooverestimateit. Thefollowing

sectionofthepaperdescribesthefindingsofthismethodology.

28SeeNationalBroadbandPlan,atChapter5,Spectrum. Bandsrecentlymadeavailableforflexibleusethatareexcluded

fromthemodelscurrentbaselineincludeAWS,2.5GHz,and700MHz,amongothers.



17/27


V. ModelOutput:SpectrumNeedandEconomicValue

a. SpectrumDeficit

Usingthemethodologydescribedabove,weestimatethatanadditional275MHzofspectrum

willberequiredtomeetmobiledatademandin2014.

TheoutputofthemodelisshowninExhibit10below. Asdetailedintheprecedingexplanation,

thekeyforecastassumptionsandoutputshighlightedinExhibit6include:

Totaldatademandgrowthrelativeto2009asanaverageofthethreeanalystforecasts(line4),whichisakeydriveroffuturespectrumneeds.

Datademandpersiteadjustedtoaccountfora7%CAGRinthenumberofcellsites(line8),whichasasubstitutefornewspectrum,servestooffsetsomewhattheamountof

spectrumneededinthefuture.

Datademand

per

site

adjusted

further

to

account

for

technology

improvements,

specificallyanincreaseofspectralefficiencyovertime(line11),whichisanotheroffset

toadditionalspectrum.

Totalspectrumrequiredtomeettheadjusteddemand(line16),takingintoaccounttheabovethreeelements.

Thefinaloutputiscontainedinline18,whichshowsthespectrumdeficitprojectedforeachyearthrough2014basedonthecurrentmobilebroadbandspectrumallocationof

547MHz.



18/27


Exhibit10.SpectrumNeedForecast TableofResultsYear: 0 1 2 3 4 5

Line Description 2009 2010 2011 2012 2013 2014

1 Data Growth Relative to 2009 - Cisco 100% 242% 598% 1253% 2577% 4722%

2 Data Growth Relative to 2009 - Yankee 100% 266% 631% 1189% 1770% 2332%

3 Data Growth Relative to 2009 - Coda 100% 251% 539% 1154% 2200% 3464%4 Data Growth Relative to 2009 - Average 100% 253% 589% 1199% 2182% 3506%

5 Cell Sites 245,912 263,126 281,545 301,253 322,340 344,904

6 Absolute Growth 100% 107% 114% 123% 131% 140%

7 CAGR 7%

8 Traffic per Site - Growth 100% 236% 515% 978% 1665% 2500%

9 Avg. Spectral Effi ciency (Mbps/MHz) 0.625 0.75 0.88 1.00 1.13 1.25

10 Absolute Growth 100% 120% 140% 160% 180% 200%

11 Tech-Adjusted Traffic per Site - Growth 100% 197% 368% 612% 925% 1250%

12 Spectrum req'd for data (MHz) 57 112 208 346 524 708

13 Percent all oca ted for data 33% 50% 65% 75% 82% 86%

14 Spectrum req'd for voice (MHz) 113 113 113 113 113 113

15 Percent allocated for voice 67% 50% 35% 25% 18% 14%

16 Spectrum - In Use (MHz) 170 225 322 460 637 822

17 Spectrum - Currently Allocated (MHz) 547

18 Surplus/Deficit (MHz) 377 322 225 87 -90 -275

Exhibit11depictsthetrafficgrowthinputstothemodelandspectrumutilizationprojections.

Ofparticularnoteistheinsightthatmobiledatademandwillexceedavailablecapacityby2013,

andwillreachanearly300MHzdeficitby2014. TheTrafficGrowthchartbelowshowsthe

totalassumed

data

demand

from

industry

forecasts,

data

demand

adjusted

for

capacity

from

additionalcellsites,anddatademandfurtheradjustedforspectralefficiencyimprovements.

TheSpectrumUtilizationchartshowstheforecasteddataspectrum,thespectrumassumed

forvoiceservices,andthespectrumsurplus/deficitovertime.



19/27


Exhibit11.TrafficGrowthandSpectrumUtilization

Traffic Growth

0%

500%

1000%

1500%

2000%

2500%

3000%

3500%

4000%

2009 2010 2011 2012 2013 2014

Data Grow th Relative to 2009 -

Average

Traff ic per Site - Grow th

Tech-Adjusted Traff ic per Site -

Growth

Spectrum Utilization

-400

-200

0

200

400

600

800

2009 2010 2011 2012 2013 2014

MHz

Spectrum req'd for data (MHz)

Spectrum req'd for v oice (MHz)

Surplus/Deficit (MHz)

Evenwith

the

conservative

set

of

assumptions

used

in

this

model,

it

is

apparent

that

the

nation

facestheprospectofaspectrumshortagewithinthenextfiveyears. Alatersectionofthis

paperundertakesasensitivityanalysistoshowhowalternativeassumptionsreinforcethis

conclusion. Whilethemodeldoesnotprojectbeyondthisperiod,areasonableextrapolationof

availabledatapointstotheneedforongoingactionoverthelongertermtomakeadditional

spectrumavailableformobilebroadband.



20/27


b. EconomicEffects

Asnotedearlier,increasingnetworkdensitythroughtheadditionofcellsitesistheprimary

substitutetonewspectrumforaddingbroadbandcapacitytothenetwork. Thatis,inthe

absenceofnewspectrum,carriersmaybeexpectedtoincreasethegrowthrateofcellsitesas

ameans

to

meet

data

traffic

demands.

Therefore,

we

can

use

the

results

of

the

spectrum

analysisabovetostudytheeconomicbenefitofnewspectrumrelativetothissubstitute. This

isaccomplishedthroughanindifferenceanalysisinwhichthenumberofsitesandtheir

respectivecostisplottedagainstthespectrumrequiredforanticipatedmobiledatademandin

2014. Ineffect,weareablegenerateanindifferencecurveshowingdifferentcombinationsof

spectrumandcapitalneededtomeetprojectedmobiledatademand.

Thisindifferencecurverequiresanapproximationofthecostofacellsite,whichisthemain

capitalsubstituteforspectruminmeetingmobiledatademand. Forthepurposesofour

analysis,wehavenotincludedthecostofnewtowerconstructionbecauseitislikelythatthe

majority

of

new

cell

sites

will

leverage

existing

tower

infrastructure

to

add

capacity.

This

is

consistentwithourpriorassumptionthatallnewcellsitegrowthwillberelatedtoinfillsites,

ratherthanthoseengineeredtoexpandcoverage.29

Ourestimateoftheaveragecostofacell

siteis$550,000,whichincludesinitialsitedevelopmentcostsaswellasoperatingexpenses.30

Thus,forexample,thecostofanetworkexpansionthatinvolvesahundredadditionalcellsites

isapproximately$55millionplusthecostofthenetworkequipment.

BasedonthesecostassumptionsandtheoutputoftheTopDownforecast,theindifference

curveshowsthat,by2014,releasinganadditional275MHzofspectrumsavesapproximately

$120billionincapitalexpensestoaccommodatemobiledatademand.

Theseestimated

savings

assume

7%

CAGR

cell

site

growth

investment,

as

described

earlier.

Sincethismodelmakesconservativeassumptionsonthecostofmarginalcellsitesubstitutes,

andassumesthattherearenoexogenouslimitationstoaddingsuchsitessuchaslocalzoning

laws,itispossiblethatthesavingsrelatedtonewspectrumcouldbeevenhigherthan$120

billion.

Whilemodelingpreciseconsumereffectsisoutsidethescopeofthispaper,itisverylikelythat

theseunnecessarycostswillaffectconsumerpricesifnewspectrumisnotmadeavailable. And,

recognizingthattherearelikelytobeexogenouslimitationstotheadditionofnewcellsites,

mobilebroadbandservicequalityisalsolikelytosufferifnewspectrumisnotmadeavailable.

29Notethatincludingtowerconstructioncostswillonlyincreasethecostofasite,therebyincreasingtheincrementalcostof

networkcapacityexpansion.Consequently,costsavingsfromadditionalspectrumallocationwillonlyimproveifthisdimension

isaddedtotheanalysis.30

.SeeSectionVI(c)foranexplanationofcellsitecosts.



21/27


Exhibit22.Capitalvs.SpectrumIndifferenceCurvefor2014

Capital vs. Spectrum Indifference Curve - 2014

0

100

200

300

400

500

600

$0

$20

$40

$60

$80

$100

$120

$140

$160

$180

$200

Capital Required to Build and Maintain Capacity ($ billions)

SpectrumR

equired(MHz)

$120B savings

InExhibit12above,thebluediamondrepresentsthepointontheindifferencecurve

correspondingtothe7%CAGRsitegrowthassumptionsdescribedabove. Thisrepresentsa

capitalinvestmentbythemobilebroadbandindustryof$54Binadditionto275MHzof

additionalspectrum. Thecurveimpliesthatifnoadditionalspectrumisreleased,thecostto

buildenoughcapacitysitestohandlethedemandwillbe$174B. Thedifferencebetweenthese

costsrepresentsthevaluecreatedbyadditionalspectrumin2014,whichis$120B.31

ThismodelofnationalwirelesstrendsclearlysupportstheNationalBroadbandPlans

recommendationtomakeavailableanadditional300MHzofspectruminthenearterm.

Reasonableindustryestimatesofmobiledatademandgrowthandoffsettingcapitalinvestment

toincreasecapacity,alongwithconservativeestimatesoftheamountofspectrumcurrently

usedformobiledata,revealastronglikelihoodthatdemandfordatawillexceedspectrum

capacityoverthenextfiveyears. Thevaluethatwillbecreatedbyreleasingspectrumtomeet

thisdemandissignificantandlikelytoexceed$100billion.

Theseconclusionsarereinforcedbytestinghowpossiblevariancesofthemodelsinput

assumptionsleadtosubstantivelysimilarconclusions,asdiscussedinthefollowingsection.

31Thisvalueislikelytobesharedbetweenwirelessconsumersandcarriers;determiningtheproducerandconsumersurplus

effectisnotwithinthescopeofthispaper. Alsonotethattheestimateof$120billionineconomicvaluefromthereleaseof

newspectrumislikelytobeoffsetsomewhatbythecostsofacquiringandmakinguseofnewspectrum. Spectrumacquisition

costsvaryasafunctionofmanyfactorsandarenotestimatedinthispaper. Capitalcostsofupgradingasitetosupportnew

spectrumbandsarelikelytobesmallrelativetototaleconomicvalueestimatesperhapsseveralbilliondollarsanddonot

impacttheconclusionofsubstantialeconomicvalueresultingfrommakingnewspectrumavailable.



22/27


VI. SensitivityAnalyses

Adjustingtheinputassumptionstothemodelallowsustounderstandthesensitivityofthe

resultstopotentialvarianceofcriticalinputdata. Inparticular,theoutputispotentially

sensitivetothreecriticalassumptions: thedatademandforecast,thebaselineamountof

spectrumused

to

serve

data

traffic,

and

the

cost

of

meeting

data

demand

through

network

density.

a. SensitivitytoDataGrowthForecast

Exhibit13belowshowstheeffectofadjustingthetrafficgrowthforecastbyplusorminus10%

andplusorminus20%:

Exhibit13.DataGrowthSensitivity

Data Growth Sensitivity - Spectrum Surplus/Deficit

-500

-400

-300

-200

-100

0

100

200

300

400

500

2009 2010 2011 2012 2013 2014MHz

Surplus/Deficit (MHz)

Data Growth - 10%

Data Grow th +10%

Data Growth - 20%

Data Grow th +20%

Ifthedatatrafficforecastusedisoverestimatingfuturedemandby20%,wewillstillhavea

spectrumdeficitinexcessof100MHzby2014. Further,itisreasonabletoexpectthatdemand

growthwillcontinueinto2015andbeyond,suchthatthespectrumdeficitwillcontinuetogrow.

Forexample,ifthecurrentdeficitcurvefor20%overestimateofdatatrafficisextrapolated

into

the

future,

it

appears

that

the

need

for

200

MHz

will

follow

shortly

after

2014,

and

300

MHzwillbeneedednomorethanoneortwoyearsafter.

Ontheotherhand,thissensitivityshowsthatifthedatatrafficforecastisunderestimating

futuredemandby20%,thespectrumdeficitwillreach300MHzsixto12monthsearlierthan

currentlyforecasted,and400MHzwillbeneededby2014.



23/27


b. SensitivitytoCurrentAmountofSpectrumUsedforData

RecallfromtheBaselineSpectrumUsesectionthatouranalysisassumes170MHzofspectrum

iscurrentlyinuseinthelargestmarketsthatdrivetheneedforspectrumandthat33%ofthis

(or57MHz)isusedtoservedatatraffic. Thisassumptioniscriticalbecauseforecasted

spectrumneedisamultiplierofthecurrentspectrumusedfordata,basedontheinputstoour

model.

Sincethe

multiplier

in

2014

is

about

12.5X,

it

means

that

every

MHz

of

input

spectrum

affectstheoutputby12.5MHz. Therefore,theoutputisverysensitivetothisassumption.

Giventhefactthat3GtechnologiessuchasHighSpeedPacketAccess(HSPA)canserveboth

voiceanddata,itisincreasinglydifficulttoseparatespectrumusedforvoicefromspectrum

usedfordata. However,givenpublicreportsthatAT&ThasdeployedtwoHSPAcarriers32

in

mostmarketsandisaddingathirdinmarketssuchasNewYorkandSanFrancisco.33

Since

eachHSPAcarrierrequires5MHzuplinkand5MHzdownlink(10MHztotal),itisapparentthat

atleastoneoperatorrequirescloseto30MHztoservethegrowingappetitefordata. Given

thatAT&Talsooperatesa2GGSM/EDGEnetworktoservevoice,itisreasonabletoassume

thatmost

of

this

30

MHz

is

needed

for

data.

If

that

is

the

case,

then

one

can

conclude

that

asingleoperatoraccountsforabouthalfoftheassumed57MHzofcurrentspectrumrequired

fordata. Giventhatalllargemarketsareservedbyatleastfouroperators,andmanyare

servedbyfive,the57MHzinputassumptionappearsconservative.

However,ifwebasetheanalysisonthemajorityoflargemarketsastheyweredeployedin

2009,ratherthanthehightrafficmarketssuchasNewYorkthatarejustreceivingathirdHSPA

carrierinearly2010,thenwecansafelyassumethatin2009AT&Taccountedforcloseto20

MHzofdataspectrum. Althoughitisnotunrealistictoassumethattheotherthreeoperators

canaccountfortheother37MHz,34

weshouldalsoconsiderthepossibilitythatspectrum

deployedtoservedatawasnotfullyutilizedin2009. Ifweassumethatonaverage,data

spectrumwas

running

at

70%

of

full

capacity,

then

the

resulting

assumption

is

that

roughly

40

MHzofspectrumwasrequiredtosupportdataservicesin2009,insteadofthe57MHz

assumedinourprimarymodel.

Torevealtheeffectofthislowerdataspectrumbaseline,Exhibit14showstheindifference

curvefor2014,andalsoprojectsoutto2016,and2018basedonthelowestofthethree

growthextrapolationsshowninExhibit3. Inaddition,the7%CAGRcellsitegrowthisassumed

tocontinue,andspectralefficiencycontinuestoimprove,to1.5b/s/Hzin2016andto1.75

b/s/Hzin2018.

32Carrierinthisusereferstotransmissioninfrastructure,ratherthanthealternateuseofthewordtodescribeawireless

serviceprovider.33

Macquarie(USA)EquitiesResearch,AT&TInc.NetworkDinnerTakeaways,March22,2010at

http://macq.wir.jp/e.ut?e=4rzyJFhQhM54SSBH75Yr4rBt9Kt(lastcheckedMarch30,2010)34

Forexample,itsnotunreasonabletoassumethattwoofthethreeremainingoperatorsrequiretheequivalentofoneHSPA

carrierandthelargestofthethreerequirestheequivalentofnearlytwo. Inaddition,afifthcellularoperatorisavailablein

manymarkets,andClearwireoffers4Gdataservicesinseveralmarkets,buttobeconservative,thisspectrumhasnotbeen

considered.

http://macq.wir.jp/e.ut?e=4rzyJFhQhM54SSBH75Yr4rBt9Kthttp://macq.wir.jp/e.ut?e=4rzyJFhQhM54SSBH75Yr4rBt9Kt


24/27


Exhibit34.2014,2016,&2018IndifferenceCurvesfor40MHzInputAssumption

Capital vs. Spectrum Indifference Curves

2009 Data Spectrum = 40 MHz

0

100

200

300

400

500

600

700

$0

$2

5

$5

0

$7

5

$1

00

$1

25

$1

50

$1

75

$2

00

Capital Required to Build and Maintain Capacity Sites

Spectrum

Required

2014

2016

2018

$195B

$121B

$29B

Inthissensitivityanalysis,theeconomicbenefitofnewspectrumisfarsmallerin2014thanin

ourprimarymodel,at$30billioninsteadof$120billion. However,thesebenefitscontinueto

increaserapidlybeyondthefiveyearforecastperiodasmobiledatademandcontinuestogrow,

evenwithsignificantongoinginvestmentofcapitalinnewsites.

c. SensitivitytoNetworkDensityCost

Theestimatedcosttodeployandoperateacellsiteisanimportantfactorinouranalysis,sinceaddingnetworkdensityinthismanneristheprimaryalternativetonewspectruminmeeting

mobilebroadbanddemand. Differentcellsitecostestimateswillyieldchangestoourestimate

ofeconomicbenefitsfrommakingavailablenewspectrum.

The$550,000unitcellsitecostusedinthemodelisbasedoninitialcapitalcostsof

approximately$130,000(whichexcludestowercostsastheyaregenerallyexogenousto

wirelesscarriersandmaynotberelevantforalltypesofcellsites),andincludesthenetpresent

value(NPV)ofapproximately$50,000annualoperatingexpensesovera20yearperiodata

10%discountrate. Thesecalculationsarebasedondiscussionswithwirelessandinfrastructure

firmsand

approximate

the

lifecycle

investment

cost

as

the

substitute

for

new

spectrum.

Thirdpartyestimatesvalidatethisassumption. BernsteinResearchrecentlyestimatedthecost

todevelopanewsiteat$300,000,whileRysavyResearchmakesasimilarestimateof$250,000

to$350,000. Bothestimatesareofinitialcapitalexpendituresthatincludetowercostsand

excludeoperatingexpenses. Taking$300,000inassumedupfrontcapitalandadding20year

operatingexpensesof$30,000peryear(whichexcludestowerleasingcosts)witha10%



25/27


discountrate,toaccountforthetotalcostofspectrumsubstitutes,theNPVofasiteis

$550,000.

Exhibit15belowshowsthecapitalspectrumindifferencecurvein2014fordifferentcellsite

costestimates,including$350,000persite,$450,000persite,andourprimarymodelestimate

of$550,000

per

site.

While

the

primary

model

estimate

appears

reasonable,

this

sensitivity

analysisenablesprojectionsofunexpectedcircumstances,suchassignificantamountsof

infrastructuresharingamongcarriers,deflationarysupplycosts,alternativeandlesscapital

intensivenetworkarchitectures(e.g.,microcells,picocells,and/ordistributedantenna

architectures),orothereconomicconditions.

Inaddition,whilethenetpresentvalueofoperatinglikelyoperatingexpensesisincludedinour

primarymodelestimateof$550,000,thelower$350,000numbercanbeviewedasastrict

comparisonofcapitalexpenditures,thoughitisreasonabletoassumethatoperatingexpenses

wouldalsofactorintoafullaccountingofeconomicvalueresultingfromnewspectrum.

Exhibit15.CapitalVs.SpectrumIndifferenceCurveatDifferentCellSiteCostLevels

Capital vs. Spectrum Indifference Curve - 2014

0

100

200

300

400

500

600

$0

$20

$40

$60

$80

$100

$120

$140

$160

$180

$200

Capital Required to Build and Maintain Capacity ($ billions)

SpectrumRe

quired(MHz)

$550k per cell site $450k per cell site $350k per cell site

$350k/site = $77B savings



Inthese

circumstances,

and

holding

all

else

in

our

primary

model

constant,

the

economic

benefitofmakingavailableanadditional275MHzofspectrumcouldbeaslowas$77billionat

verylowcellsitecosts. Atacostof$450,000persite,thebenefitofnewspectrumisnearly

$100billion. Whilelowerthanthe$120billionestimateinourprimarymodel,thissensitivity

analysisdemonstratesthattheeconomicbenefitsofnewspectrumapproach$100billionby

2014evenwithreducedcostofnetworkdensityalternatives.



26/27


Together,thesesensitivityanalysesshowthatalthoughmodelinputvariablesmayvary,a

spectrumdeficitwillverylikelybecomesignificantbymiddecade,andthevalueofreleasing

morespectrumappearsverylikelytosurpass$100Binthelatterhalfofthedecadeatthelatest.

Giventhecorrespondingpossibilityofanunderestimationinthemodelofdatagrowthand

currentspectrumusedfordata,thespectrumdeficitcouldberealizedsooner,andthevalueof

releasingadditional

spectrum

could

be

even

greater.

Since

making

new

spectrum

available

has

historicallytakenbetweensixandthirteenyears35

,andsincemobiledatagrowthtrendsare

likelytocontinuebeyondtheneartermforecastinthispaper,theseobservationssupportthe

needfortimelyactiontofreespectrumformobilebroadband.

VII. Conclusion

Mobilebroadbandservicesareexperiencingsignificantgrowth,drivenbyconsumerdemand

formobiledata. Industryanalystsexpectthisgrowthtocontinue,callingintoquestionthe

capacityofcurrentmobilenetworkstokeepup. Evenwithsubstantialinvestment,itislikely

thatmobile

data

demand

will

exhaust

spectrum

resources

within

the

next

five

years.

The

NationalBroadbandPlanrecommendedthatnewspectrumbemadeavailabletoenable

continuedgrowthofmobilebroadband.

Thispaperdemonstratesthataspectrumdeficitapproaching300MHzislikelyby2014,and

thatthebenefitofreleasingadditionalspectrumislikelytoexceed$100billion. Themodel

usedinthispaperisgeneratedfromexplicitandreasonablemarketassumptions,whichproject

substantialbenefitfromreleasingadditionalspectrumformobilebroadbandinthenearterm.

Thispaperdoesnotundertakeananalysisofnetsocialbenefitsresultingfrommakingnew

spectrumavailable. Theeconomicbenefitsestimatedhereinrepresentonlythereductionin

costofmeetingmobiledatademand. Thisestimateofvaluecreatedbyreleasingnewspectrumforbroadbandisnarrow,andlimitedtotheavoidanceofunnecessarycosts. This

paperdoesnotundertakeacomprehensiveanalysisofthebenefitstosocietythatmayresult

frommakingnewspectrumavailable,whichsomeeconomistsestimateasmultiplesofthe

privatevalue.36

Themethodusedinthisanalysistoforecastfuturespectrumneeddemonstratesthelikelihood

ofenormouseconomicvaluebeingcreatedbyreleasing300MHzofadditionalspectrumto

meetgrowingdemandformobiledata. Giventhatnewspectrumhashistoricallytaken

betweensixandthirteenyearstomakeavailable,andsincemobiledatagrowthtrendsare

likelyto

continue

beyond

the

near

term

forecast

in

this

paper,

these

results

support

the

need

foractiontoimplementtheNationalBroadbandPlansrecommendationandthePresidents

directivetomakenewspectrumavailableformobilebroadband.

35SeeNationalBroadbandPlanatCh.5,Spectrum.

36Thispaperdoesnotestimatethecoststothirdpartiesofmakingavailablenewspectrum,likelyauctionrevenue,orother

factors.



27/27


AppendixA:MethodologySummaryofMobileDataDemand

Forecasts

Thethreeindustryanalysesusedinthispaperasabasisforestimatingfuturemobiledata

traffic

arrive

at

their

forecasts

through

analysis

of

the

likely

mix

of

devices,

applications,

and

servicesinthemobilemarket. Eachanalysishasadistinctmethodology,summarizedbelow.

CiscosVNIforecastmethodologybeginswiththenumberandgrowthofconnectionsand

devices,appliesadoptionratesforapplications,andthenmultipliestheapplicationsuserbase

byCiscosestimatedminutesofuseandkilobytesperminuteforthatapplication. Ciscouses

uniquedatasources,andprovidesapplication,segment,geographic,anddevicegranularity,

andadjustsforfactorssuchasthelikelihoodofoffloadingtrafficthroughmeanssuchasWiFi.

Codasforecastusesdatafromavarietyofsources,includingpublishedandunpublished

forecastsandsurveydata.Forecastsofmobiledatatrafficentailprojectingbehavioracross

traffictypes

and

connections

and

data

service

take

up,

as

well

as

data

volumes,

data

speed

and

hourspermonthaccessingandgeneratingtraffic.

TheYankeeGroupreportusesmodelsofmobiledatatrafficdevelopedthrough20primary

researchinterviewswithtechnologyvendorsandserviceproviderstoidentifytrendsinthe

marketbroadbandmarket.Theyforecastdevicemigrationanddatatrafficdemandsacrossa

varietyofmarketsandwithseveralassumedservicedeliverymodels.

Itshouldbenotedthattheseanalysesforecasttechnologyandmarkettrendsbasedinparton

historicexperience. Thereareunknownfuturevariablesthatcouldaffectprojections,suchas

generaleconomic

conditions,

and

the

potential

for

new

pricing

strategies

such

as

tolled

data

(i.e.,alacarte,ratherthanallyoucaneat)thatcanaffectmobiledatademand.

Documents

Fcc Omnibus Broadband Initiative (Obi) Technical Paper Mobile Broadband Benefits of Additional Spectrum