• No results found

Tax-free digital news?

N/A
N/A
Protected

Academic year: 2022

Share "Tax-free digital news?"

Copied!
18
0
0

Laster.... (Se fulltekst nå)

Fulltekst

(1)

ContentslistsavailableatScienceDirect

International Journal of Industrial Organization

www.elsevier.com/locate/ijio

Tax-free digital news?

ØysteinForosa, Hans Jarle Kinda,b, TimWyndhama,

aNHHNorwegianSchoolofEconomics,EconomicsHelleveien30,Bergen5045, Norway

bCESifo,Norway

a r t i c l e i n f o

Articlehistory:

Received14May2018 Accepted14May2018 Availableonline21June2019

JELclassification:

D11 D21 L13 L82

Keywords:

Two-sidedmarkets Multi-homing Digitalmedia Incrementalpricing Valueaddedtax

a b s t r a c t

Newsplatformsarestruggling.Theirprintedreadershipispre- dominantlyold,andtheirdigitalproductstrugglestowinthe attention of the young. For several decades tax reductions havebeenusedinEuropetoincreasethecirculationofprinted newspapers.Wouldextendingthesereductionstodigitalplat- formsstimulatedigitalconsumption?Usingatwo-sidedpric- ingmodelwhereaprintplatformandadigitalplatformcom- peteformulti-homingconsumersandadvertisersweshowthat theanswerisno.Thetwo-sidednessofthemarketmeansthat thedigital pricewouldincrease. Notonly woulddigital cir- culationdecreasebutsotoowouldthefractionofconsumers thataccessnewsfrombothplatforms.Keymediapolicygoals ofreach(circulation)andpluralism(multi-homing)wouldbe harmed.

© 2019TheAuthors.PublishedbyElsevierB.V.

ThisisanopenaccessarticleundertheCCBY-NC-ND license.

(http://creativecommons.org/licenses/by-nc-nd/4.0/)

1. Introduction

For several hundred years newspapers have operated in a two-sided market selling news and commentary to readers and eyeballs to advertisers. The United Kingdom’s

Correspondingauthorat:KingEdward’sSchool,EdgbastonparkRoad,Birmingham,B152UA,UK.

E-mailaddresses:oystein.foros@nhh.no(Ø.Foros),hans.kind@nhh.no(H.J.Kind), tgw@kes.bham.sch.uk(T. Wyndham).

https://doi.org/10.1016/j.ijindorg.2018.05.006

0167-7187/© 2019TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense.(http://creativecommons.org/licenses/by-nc-nd/4.0/)

(2)

Table1

VATRates in selectedEuropean countries. Source:European Commission(2016a) andStatsministerens Kontor(2015).

Standardrate Printednewspaper e-Newspaper

Austria 20% 10% 20%

Belgium 21% 6% 21%

France 20% 2.1% 2.1%

Germany 19% 7% 19%

Spain 21% 4% 21%

Norway 25% 0% 25%(pre1/3/2016)

0%

UK 20% 0% 20%

first regular newspaper, The Daily Courant, was launched in 1702 and consisted of a single page of news with adverts on the back (Williams, 2009). For most of the time afterwards, and certainly since the early 19th century, newspapers have been seen as importantforthehealthofdemocracy(Gentzkowet al.,2011).

Preferentialtaxtreatmentofprintednewspapersiswidespread,particularlyinEurope as illustratedinTable 1.1 IntheUKandNorway,printednewspapers donotpayvalue added tax (VAT) on sales to readers.2,3 The preferential treatment aims to increase circulationandensurethatpeoplegetinformationfromseveraldifferentsources(multi- homing inthetwo-sidedmarketterminology).

The circulationof printednewspapers is shrinking rapidly, leading to questions over theirfuture.Forexample,thereachofnationalprintednewspapersin theUKdecreased by more than a quarter between 2005 and2015 (Ofcom, 2015). Circulation in Norway also fellby more thana quarter between1999 and 2013.4 Inthe UKThe Independent has already ended itsprint edition, believing that it “will be the first of manyleading newspapers toembraceawhollydigitalfuture” (Lebvedev,2016).

Akeydriverofthefallinthecirculationofprintednewspapersisthenewsconsumption habitsoftheyoung.Theyounghavealwaysbeenlesslikelytobuyanewspaperthanthe old,butreadershipofnewspapershasfallenfasterfortheyoung.In2005,nationalprinted newspapers intheUKreachedaround75%ofthoseaged65oraboveandslightlyunder 70% ofthose aged between 15and 34. By 2015,reach had fallen to aroundtwo thirds andonethird respectively(Ofcom,2015).ThecurrentsituationinNorwayisevenmore stark:In2014,82%of67–79yearoldsreadanewspaperonanaveragedaycomparedto 26% ofthoseaged16–24(EFTASurveillanceAuthority,2016).

1IntheUS“Federal,state,andlocalgovernmentshavetraditionallyprovidedavarietyofspecialeconomic supportstotheindustry,includingexemptionsfromnewspaperandadvertisingsalestaxesandexcisetaxes ontelecommunicationsequipmentusedforinformationgathering” (Picard,2004).

2Valueaddedrefersto“thevaluethataproducer...addstohisrawmaterialorpurchases(otherthan labour)beforesellingtheneworimprovedproductorservice” andcanbecalculatedasthesumofwages andprofits(Tait,1988).DevelopedinFranceinthemiddleofthe20thCenturyVAThadbecomewidespread bytheturnofthemillennium,particularlyinEurope.Formoreonthetheory,historyandpracticaldetails ofVAT,seeTait(1988).

3Since1March2016digitalnewspapersinNorwayalsodonotpayVAT.

4CalculationbyauthorsbasedondatapresentedinStatministerensKontor(2015)

(3)

Giventhefrequentuse ofVAT policyto stimulate consumptionof printednewsand theincreasingproportionofdigitalcontentthatisplacedbehindapaywall,itisnatural forpolicymakerstoaskwhethertheVATexemptionshouldbeextendedtodigitalnews.

MemberStatesoftheEuropeanUnionareprohibitedbytheVATDirectivefromapplying abeneficialratetodigitalnews.5However,theEuropeanCommissionandtheEuropean Councilare seeking to amend the VAT Directive and allowMember States the option to offer e-publicationsin eachcountry the same tax rate as their printed counterparts (SeeEuropeanCommission(2016b,2016c,2016a);EuropeanCouncil(2017).Norway,a memberoftheEuropeanEconomicArea(EEA)6 butnotoftheEuropeanUnion,isnot constrainedbytheVATDirectiveandwasthefirstwithintheEEAtoimplementa zero taxregimefordigitalaswellas printednews.7

IntheStateaidapprovaloftheNorwegianzero-taxregimetowardsdigitalnews,EFTA SurveillanceAuthority(2016) noted(page13):“The mainobjectiveof theproposed zero VAT rate isto supportthe demand anduse of news and current affairs content among consumers, thereby also promoting media pluralism and diversity. This requires instru- mentsaimedat consumers.Reducing the costof electronicnews servicesisadirectand effective means ofensuringhigh newsconsumption andtherebya broadandenlightened publicdiscourse.” Furthermore,thefollowingconcernoftheNorwegiangovernmentwas accentuated(EFTASurveillanceAuthority(2016),page12):“theexistingzeroVATrate fornewspapersprimarilysupportsthemediaconsumptionofthe middle-agedorelderly.”

Againstthisbackdrop,ourresearchquestionis:

DoesreducingVAT ondigitalnewsreduce thepriceofdigitalnewsandstimulate digitalnewsconsumption?

Surprisingly,theanswerisno.IftheVAT exemptionisextendedto digitalplatforms then the price towards digital readers increases. Demand for the digital platform de- creases, as does the fraction of consumers getting information from different sources (multi-homing).

Thedeparturefromthestandardintuitionisdrivenbythepresenceoftheadmarket.

Alower tax rateon thedigitalproduct increases theprofitability of thereadermarket for the platform, but has no direct effect on the advertising market. This means that thedigitalplatformwill place more weighton readermarketprofitsand lessweighton advertisingmarketprofits;itbecomesmoreimportanttosetarelativelyhighsubscription priceandboost readermarginsdespitetheconsequent reductions in readership andad revenues.

Naturally,theprofitofthedigitalplatformincreasesfollowingareductionofVATon digitalsubscriptions. But,surprisingly,theprofitoftheprintedplatformalsoincreases.

5 ThisrestrictionhasnotpreventedsomeMemberStatesfromreducingVATonelectronic newspapers.

AscanbeseeninTable1Franceusesalowerrate.Belgium,LuxembourgandItalyhaveallofferedlower thanstandardVATratesonelectronicnewspapersatsomepoint.

6 TheEEAcomprisestheMemberStatesoftheEuropeanUnionandthreenonmembers:Iceland,Liecht- enstein,andNorway.

7 InIcelandnewspapersaretaxedatareducedrateof11%,whilsttherateinLiechtensteinis2.5%.

(4)

So,tax-freedigitalnewsmayhelpthesurvivalofprintnewspapers,albeitattheexpense ofreducedonlinereadership.

OurmodelisbaseduponHotelling(1929);thedominantworkhorseforanalysingtwo- sided mediamarkets(theseminalcontributionisAndersonandCoate (2005).Weallow forasymmetricplatformswithrespecttotaxratesandmarginalcosts.Thedigitalplat- form haslowermarginal costsand,attheoutset, ahigher(standard) taxratethanthe print platform. Ourmodel incorporatestwo-sided pricing; both platformssellsubscrip- tionsto consumers andeyeballs toadvertisers. Westartbyrecognising therealitythat consumershavelongboughtmorethanonenewspaper.8Thusweallowmulti-homingby both consumersandadvertisers.

Toourknowledge, wearethefirsttoassessthereactionto taxchangesbytwo-sided duopolists facing multi-homingconsumers (consumers that are “shared” between each platform). As well as being a significant measure of media pluralism,multi-homing by consumershasimportantimplicationsforcompetitionbetweenplatforms.Whenallcon- sumers buyonlyone product (they eachsingle-home or are“exclusive” to a platform), the“competitivebottleneck” problemof Armstrong(2002,2006)exists: sinceanexclu- sive reader can only be accessed through theplatform from which she purchasesthere isnodirectcompetitionforadvertisersamongplatforms.RecentcontributionsbyAthey et al.(forthcoming),Ambrusetal.(2016) andAndersonet al.(2018)introduce compe- titionforadvertisersbyallowingconsumers tomulti-home.

We combine ingredients from Anderson et al.(2018), which considers multi-homing consumers in a pure ad-financedtwo-sided market, and Anderson et al. (2017), which considersmulti-homingconsumersinaone-sideduserfinancedmarket.Theseingredients areusedtoextendthesimplesingle-homingmodelwithdualsourcefinancing(two-sided pricing)ofAndersonandGabszewicz(2006).Despitethemanycomponentsweconstruct asimplemodeltohighlightthecoremechanismdrivingourresults:thetwo-sidednessof themarket.

There are papers assessingtheimpact of VAT onprice in two-sided marketswith a monopolist platform (Kind et al., 2008), with duopolists (Kind et al., 2013) and with perfectcompetition(Kindetal.,2008).Animportantlimitationofthesemodelsisthat, even when there is more than one platform, consumers are assumed to single-home.

Platforms are also assumed to have symmetric cost structures. We relax both these assumptions.

Wealsoextendourmodel tothepure single-homingconsumercase,for tworeasons.

First, we show that this mightbe theoutcome of a VAT reduction in ourmodel. It is possiblethatreducingVATondigitalnewscouldincreasethedigitalpricetotheextent thatnoconsumerbuysbothproducts.Second,previousinvestigationsofthereactionto tax changesbytwo-sided duopolistsfacingsingle-homingconsumers havebeenlocation games.Inthosepapersreducedtaxrateshaveincreasedreaderprices throughincreased

8Americansurveydatafrom1917to1919showedthat15%ofhouseholdswhoreportedreadingadaily newspaperreportedreadingtwoormore(Gentzkowetal.,2014),butthedigitisationofnewshasincreased theprevalenceofmulti-homing(seeAtheyetal.,forthcomingandPeitzandReisinger(2015)).

(5)

horizontal differentiation. In our pure single-homing model we find the same inverse relationshipbetweentaxratesandpriceswithoutanychangeinhorizontaldifferentiation.

InSections2and3wepresentthefoundationsandfindingsofourmodelwithshared consumers (multi-homing). In Section 4 we investigate the case when there are only exclusiveconsumers (single-homing).InSection 5wediscusstherobustnessofourmain results.Wesummariseanddiscuss ourresultsinSection 6.

2. Themodel

Considertwocompetingmediaplatforms;oneproducingaprintednewspaper,andthe othera digitalnewspaper.Thedigitalplatform(D)hasmarginal costsequalto cD0, whiletheprintplatform(P)facesamarginalcostequaltocP>0.Throughoutweassume cP>cD.9 Inthebasic model weset cD=0, whilein Section 5 we allowforcD>0.We abstractfrom anyfixedcosts.Thetax rate(VAT)onsubscriptionsforeachplatformis τi,wherei=D,P.Theplatformsarelocatedattheextremesofa“Hotellingline” with length 1. Platform D is at the far left and platform P at the far right. Platforms sell subscriptionsatpricepito consumersandeyeballstoadvertisers. Thisimpliesthatthe consumerspaypi andplatformsreceive 1+piτ

i.Wespecifyadprices below.

Consumer (reader) tastes are uniformly distributed along the line. We may inter- pret the horizontal differentiation as age. Young people to the left, old people to the right. Consistentwith empiricalstudies ofthe USnewspapermarket(Gentzkow, 2007;

Fan,2013),andGentzkowet al.(2014))weassumethat consumersaread-neutral.10In Section 5 we consider theoutcome whenconsumers dislike ads.The distance disutility (transportationcosts)is givenbyt.

Remark(one-sidedmarket):Marginalcostsareapproximatelyequaltozerofordigital goods(e.g. e-books). Itis wellknown fromthetax literaturethatsucha coststructure impliesthatVAT actsasa puresurplustaxwith noimpactonconsumerprices in one- sidedmarkets.Toseethis, considerthezeromarginalcostprofitfunctionπ= 1+pτx(p), where x(p)is thedemand function. The tax rate τ clearly drops outof the first-order condition∂π/∂p=0, so that it only affects the ‘profit-split’ between the firmand the government.This insightprovides us with a clearbenchmark in thetwo-sided markets weanalyse.

2.1. Consumerdemand

Lettheconsumerutilityofbuyingfrom onlyplatformD oronlyplatformPbe

uD=vD−tx−pDand (1)

uP =vP−t(1−x)−pP, (2)

9 WealsouseDandPtorefertothedigitalandprintedproduct,respectively.

10 SeeChandra andKaiser (2015) fora comprehensive surveyof theliterature on consumers’attitude towardsadsinnewspapermarkets.

(6)

Fig.1. Sharedandexclusivereaders.

respectively, where vi isthe verticalquality ofplatformi,pi isthesubscription feeand xisthelocationoftheconsumer.

The utility ofa consumerwho buys both products is thesum of individual utilities lessanyutilityloss duetooverlap:11

u(D+P)=uD+uP−d. (3)

WefollowtheessenceofAndersonetal.(2017)andsaythatconsumerswithuD>uP

perceive D as theirprimarygood andP as theirsecondary good. Aconsumer willbuy bothproductsif theincremental utility of multi-homingis positive,12 where herincre- mentalutilityofmulti-homingisspecifiedasu(D+P)max[uD,uP].Tofindthelocation of the consumer indifferent between buying only D and buying both D and P we set u(D+P)−uD=0 andsolveforx,yielding

xDP =1−vP −pP−d

t (4)

where weimplicitlyassumethat t>vP −pP−d>0.

Likewise thelocation ofthe consumerwhoisindifferent between buying onlyP and buying bothPandD is

xPD=vD−pD−d

t . (5)

Fig. 1 identifies those consumers who buy only D, those who buy both D and P and those whobuyonlyP.

Under multi-homing,consumerdemandat eachplatform is XDM H = xDP

D’s exclusive readers

+(xPD−xDP)

shared readers

=xPD (6)

XPM H = 1−xPD P’sexclusive

readers

+(xPD−xDP)

shared readers

=1−xDP. (7)

11We are grateful toPaul Belleflamme and theEditor forspecifying this formulation.Our qualitative findings hold forother formulations, includingwhen the utility“loss” from the incremental purchaseis proportionaltothesumofverticalqualitiesandwhenthe“loss” isproportionaltotheverticalqualityof thesecondaryproduct(asinAndersonetal.(2017)).

12In ourmodel consumers whobuy bothproducts are thosein the middle ofthe Hotellingline, their distasteforeitherpublicationisnot“too” strong.

(7)

Demandforeachnewspaperisstrictlydecreasinginownprice(∂XiM H/∂pi=1/t).13 Thenumberofexclusivereadersforeachnewspaperis,however,independentoftheprice itcharges.This isbecause wehave assumedthat allconsumers read atleast one news- paper.Anewspaper’snumber ofexclusivereadersis thereforedeterminedby,and more precisely,isinverselyrelatedto, thedemand fortheothernewspaper.Since anincrease inpidoesnotaffecttheincrementalvalueofnewspaperj(leavingdemandforthatnews- paperunchanged),itcannotaffectnewspaper´ısnumberofexclusivereaderseither.This partlyreflectsthepeculiaritiesoftheHotellingmodelandourspecificassumptions,but doesnotqualitativelyaffecttheresultswederivebelow(seeSection5.1foradiscussion).

2.2. Platforms andadvertisers

Bothplatformscan costlessly place commercialsin theirnewspaper.Asin Anderson etal.(2018)weassumethatplatformssetapriceperad,andthatadvertisersonlyplace one advertper platform. We assume a perfectly elastic demand curve for ads, with a massA ofhomogenousadvertisers.

The expected value for an advertiser of reaching a reader who sees the ad on one and only one of the platforms is equal to β. This value reflects the advertiser’s profit marginand theproportion of consumers who, on seeing the ad for thefirst time, buy theproduct.Asin AndersonandCoate (2005) theplatformsareabletoextractallthe advertisers’surplusfrom exclusiveconsumers.Weallowforthesecondimpressionto be worthlessthanthefirst.Thustheexpectedvaluetoanadvertiserofaconsumerseeing hisadverttwiceisβ(1+σ),withσ(0,1].

Tospecifytheequilibriumpriceperadvertweusetheprincipleofincrementalpricing asdevelopedinAndersonetal.(2018),i.e.pricesatagivenplatformwillbedetermined bytheincrementalvaluetoanadvertiserofadvertisingonthatplatform.This prevents either platform from charging more than σb for its shared consumers. The price per adatplatform iwillthereforebe ai=βXie+σβXs,where Xie isplatform i’s exclusive consumers and Xs represents the number ofconsumers that purchaseboth products.14 TotaladrevenuesatplatformiwillbeA(βXie+σβXs)orbXie+σbXs,usingtheidentity b≡Aβ.

Table 2 compares the values, incremental values and prices for a platform’s single- homingandmulti-homingconsumers.

Ourmodel set-upallows formulti-homingconsumers,asymmetricplatformswithre- specttomarginalcostsandtaxrates,aswellastwo-sidedpricing(platformschargeboth

13 Forinstance,ahigherpriceonthedigitalnewspapermovesxPDtotheleft(whichimpliesthatalarger shareoftheconsumerswillreadonlytheprintednewspaper).

14 Tofixideasconsideranumericalexample.Letthevalueofafirstimpression(β)be0.9andthevalue ofasecondimpression(σβ)be0.36.Themassofconsumersisnormalisedto1.Furthermore,let20%of consumersbuyexclusivelyfromtheprintedfirm,30%exclusivelyfromthedigitalfirmand50%fromboth.

Thentheperadvertpricesunderincremental pricingwill beaD=0.90.3+0.360.5=0.45andaP= 0.90.2+0.360.5=0.36. Asubscriptionfeedecrease attheprintedfirmthatledto60%ofconsumers buyingfrombothfirmsandtheremaining40%exclusiveconsumerssplitequallywouldleadtothesymmetric peradvertprices:aD=aP=0.90.2+0.360.6=0.396.

(8)

Table2

Advertisingvaluesforexclusiveandsharedconsumers.

Exclusiveconsumer Sharedconsumer

Valuetoadvertiser β β(1+σ)

Incrementalvaluetoadvertiser β σβ

Incrementalprice β σβ

AdvertiserSurplus 0 β(1+σ)2σβ=β(1σ)

consumers and advertisers). Given this complex set up, for the sake of simplicity, we search forNashequilibriain asettingwhereplatformssetprices forreadersandforads simultaneously.

3. Multi-homingconsumers

When some consumers buy both products, the platforms face the following profit functions:

πD= pD

1+τDXDM H+bXDe +σbXsand (8) πP =

pP 1+τP −cP

XPM H+bXPe +σbXs, (9)

where the first terms represent reader market profit and the second and third terms represent ad marketprofits. Note that in a one-sided market(b=0), τD would act as a puresurplus tax sincemarginal costson thedigitalnewspaper arezero (seeRemark above).

Using(6)wecanrewrite(8)toseparateprofitfromexclusiveandsharedconsumers πD=

pD 1+τD

+b

XDe + pD

1+τD

+σb

Xs,

andwrite thefirstorder conditionforthedigitalplatformas D

dpD =

XDe +Xs

1+τD + pD

1+τD

∂Xs

∂pD +σb∂Xs

∂pD =0 (10)

Raising pD has the standard effect on reader market profitability. It increases the profit margin but reduces sales. If there were no ads in D (in which case we would have a one-sided product), profit maximization dictates that the term in the square bracketof(10)should beset tozero(marginalrevenueequal tomarginal cost,whichis zero).However, the termoutsidethebracketis negativewhen second impressionshave a positivevalue,showingthat Dsoptimaltwo-sidedpriceislowerthanitsoptimalone- sided price.15 This isdue to thefact that the increasedadvertisingprofits gainedfrom selling eyeballs (alongside the additional reader sales) exceed the foregone margin on existing consumers.

15Since ∂X∂ps

D <0andσb>0.Toseetheformersubstitute(4)and(5)intoXs=xP DxDP.

(9)

Fig.2. Effectofadigitalpriceriseonreaderdemands.

NotethatDsmarginal consumersare theones itshareswithP.Whileasmall price risewill reduce thesurplusof infra-marginalcustomers(includingexclusiveconsumers) theonlyconsumers thatwill stopbuying Darethosebuyingitas asecondary product, in other words for its incremental value, as Fig. 2 shows. Mathematically, raising pD

reducesxPDbut hasnoeffect onxDP.

Alowertaxrateonthedigitalproductincreasestheprofitabilityofthereadermarket fortheplatform,buthasnodirecteffectontheadvertisingmarket.Thismeansthatthe digitalplatformwillplacemoreweightontheterminthesquarebracketof(10)compared to the term outside; it becomes more important to set a relatively high subscription priceandboost readermarginsdespitetheconsequent reductions in readership andad revenues. Incontrastto typicalresultsin one-sidedmarkets,wemight thereforeexpect the consumerprice to be decreasing in thetax rate, other things being equal. This is confirmedbysolving (10)to findthedigitalplatform’sreactionfunction

pD(◦)=vD−d−σb(1+τD)

2 (11)

from which we immediately see that ∂p∂τDD()=σb2. Furthermore, we observe that the sizeof thepricechange dependsonthe(incremental)valueoftheshared consumerson theadvertisingmarket.Thereasonforthisis,asnotedabove,thattheplatformcanonly affectthenumberofsharedreaders-andnotthenumberofexclusivereaders-through itspricingbehaviour.

Note also that pD() is independentof pP. The intuition is that a consumer whois consideringpurchasingDasasecondaryproduct,willonlyconsiderthepriceofD.Prices arethusstrategicallyindependent.SeeAndersonetal.(2017)forafurtherdiscussionof thisissue.

Thereaction functionoftheprintedplatformisqualitatively similar:

P

dpP =

XPe +Xs 1+τP +

pP

1+τP −cP ∂Xs

∂pP +σb∂Xs

∂pP =0 yieldingthebestresponsefunction

pP(◦)=vP−d−(bσ−cP)(1+τP)

2 . (12)

Thisshowsthat thesubscriptionpriceofplatformPis decreasingin itsowntax rateif σb>cPorwhenthevalueontheadmarketofreachingamulti-homingconsumerexceeds themarginal costofproducinganextracopy.

(10)

Since prices are strategically independent, reaction functions (11) and (12) are also equilibrium values.16 Afurther importantimplicationof priceindependenceisthat the tax rateonone platformhasnoeffect onthepriceoftheotherplatform.

It is noteworthy that for sufficiently valuable ad markets, equilibrium subscription pricesateitherorbothplatformscouldinprinciplebenegative.Werestrictourattention tocaseswherepricesarepositive.17Animportantreasontoabstractfromnegativeprices is that pure negative prices are rarely observed in practice, although often there may be complimentary gifts or other exclusive offersfor subscribers. It is possible that the platformswouldprefertohavenegativepricesirrespectiveoftheVATratetheyfacebut areunableto feasiblyimplementthis.Inthissituation thepricewouldremainstableat zero.

Summing uptheabove analysiswecanstate:

Proposition1. Thepriceoftheprintplatformisindependentofthetaxrateonthedigital platform, andvice versa. Supposethatthere isadecreaseinthe taxrate on

a) The digital platform. Thenown pricewill increase if secondimpressions haveany incremental value(σb>0).

b) Theprintplatform.Thenownpricewillincreaseiftheincremental valueofsecond impressionsisworthmoreperconsumerthanthemarginalcostofprintinganextra copy(σb−cP >0).

This proposition couldalso be worded that own prices decrease in own tax ratesas longastheadvertisingvaluetotheplatformofasharedconsumerexceedsthatplatform’s marginal cost.

Onemightexpectthat dueto priceindependence,thetaxrateonone platformdoes notaffecttheprofitsoftheother.Interestingly,thisisnottrue.SupposethatτDincreases.

This will not affect the price (pP) or total demand (XPM H =1−xDP) for the print platform,butwillaffectitscompositionofexclusive(XPe =1−xPD)andsharedreaders (Xs=xPD−xDP).Inserting for(4), (5) and (12)into (8)and differentiating πP with respect toτD yields

P D

=bdXPe D

+σbdXs D

.

The digital platform will charge a higher subscription price if its tax rate, τD, de- creases. As Fig. 2 illustrates, the total demand for P is unaffected, but some con- sumers who previously bought both products will now only buy P. The increase in the digital price has “converted” some of Ps shared consumers into exclusive con- sumers.18 This conversion will not affect the print platform’s reader market profit, as its reader priceand total demand are unchanged but its advertisingmarket profitwill

16Thesecond-orderconditionis ddp2π2i

i =t(1+2τi) <0.

17SpecificallyweassumevD>d+σb(1+τD)andvP>d+(cP)(1+τP).

18Theincreaseinexclusiveconsumersis dX

e P D = σb2t.

(11)

increase.Exclusiveconsumers areworthmoreontheadmarketthanshared consumers so the print platform will increase the price of its ads, aP. We consequently find that

P

D =σb2t(b−σb)=b2σ(21t−σ)<0.Alowerdigitaltaxrateincreasestheprintedplat- form’sprofit.

For the digital platform we likewise find D

P =b(bσ−c2Pt)(1−σ) which is negative if bσ >cP. Under this conditiona lower tax rate τP increases pP and we have the same mechanism.Wecanstate:

Proposition2. The printplatform’sprofitdecreases inthe taxrate ofitsrival (PM H

D <

0).The digitalplatform’sprofit decreasesin therival’s taxrateif bσ >cP.

Froma mediapluralismperspective,a majorrationaleforpreferentialtax treatment of newspapers has been to increase their circulation and to ensure that people get in- formationfrom several different sources(multi-homing, in our terminology).This may beimportantfore.g.democraticprocesses,knowledgespilloversandanti-biasmeasures.

Wewillnotgointothese rationales,but notethat thenumberofmulti-homers isequal to

Xs=xPD−xDP = (vP+vD)+bσ(1+τD)(cP−bσ)(1+τP)2d

2t 1, (13)

fromwhichitimmediatelyfollowsthat:

Proposition3. Reducingthetaxrateonthedigitalplatform(τD)decreasesthenumberof multi-homingconsumers. Reducingthetaxrateon theprint platform(τP)decreases the number of multi-homers if the incremental value of a multi-homing consumer is larger thanthe printplatform’smarginalcost(bσ >cP).

Wealsoobservefrom(13)thatthecomparativestaticsofthenumberofsharedreaders are intuitively reasonable. The number of shared readers is increasing in the value of second impressions (bσ) and is decreasing in the strength of horizontal preferences(t), theamount ofoverlap(d)andtheprintedplatform’smarginal cost(cP).

Fig.3showsanumericalexamplewherewesetτP =0andvaryτD.19Withataxrate of25%,as Norwayusedtohave,2.0%ofreadersareshared.Ataxrateof20% asin the UK,implies 1.3%of readersare shared.Reducing thetax rate forthe digitalplatform below10%impliesthattherewouldonlybeexclusiveconsumers.Intheabsenceofshared consumers,the nature of competition between the two platforms changes significantly.

Weinvestigate thisin Section4.

Thelogicoftwo-sided markets,as describedabove,clearlyindicates thatsubsidising newspapersthroughreducedvalue-addedtaxesmightbe anineffectiveorevencounter- productivemeanstoincreasenewspapercirculation.

19 TheotherparametersarevD=vP=0.9,cP=0.3,t=0.55,b=0.55,σ=0.26,d=0.35.

(12)

Fig.3. Impactofthedigitaltaxrateonthenumberofsharedconsumers.

Before we proceed to a single-homing environment, we note a more positive insight from theanalysis above. Apublic policywhichcontributesto highermedia quality(an increasein vi)couldbeaneffectiveway toincreasemulti-homing(despitehighernews- paperprices) aswellasbeing apoliticalgoalin itsown right.Moreprecisely,from Eqs.

(11),(12)and(13)weobserve:

Proposition4. Assumethatmediaqualityimproves(vDandvP increase).Thensubscrip- tionpricesandthe extent ofmulti-homing increase.

Ratherthanloweringthetaxrateondigitalplatforms(whichwouldleadto ahigher digitalpricebut lowercirculation),governments couldforinstancesubsidisejournalism to ensureboth highermediaquality andlargernewspapercirculation.

4. Single-homing consumers

In the previous section we observed the possibility that no consumers multi-home (e.g.duetolowVATrates,asillustratedinFig.3).Furthermore,intheintroductionwe notedthatthepreviousliteratureontaxintwo-sidedmarketshasfollowedtheHotelling convention of assuming that eachconsumer buys a maximum of one product. We now extendthe“puresingle-homing” literaturebyassessingasymmetricplatforms.

Supposethat the marketisshared and eachconsumerbuys one andonly one ofthe media products.20 Consumerdemands resemble the standard Hotelling set-up: XiSH=

1

2+vi2tvj pi2tpj.21UsingasimilarmethodologytothatusedinSection3wecanderive

20Consumers will choose D or P to maximise (1) or (2). The market is covered iff t13(vD+vP+ 2b+b(τD+τP)cP(1+τP)). The market sharing condition is t>max[13(vDvP+b(τDτP) +cP(1+τP)),13(vPvDb(τDτP)cP(1+τP))].

21SeetheAppendixforfulldetails.

(13)

thedigitalplatform’sreaction function:

pD(◦)=t+vD−vP−b(1+τD)

2 +pP

2 . (14)

Eq. (14)showsthat D’sreaction functionshifts upifits taxrate isreduced. Asin the multi-homing case it is optimal for D to shift profit from the advertising side to the consumer side by increasing the reader price. As in the multi-homing case, the price increasewill begreaterthegreatertheper-readeradvertisingrevenue.22

Followingthesame processfortheprintedplatformyieldsthebestresponse function

pP()= t+vP−vD(b−cP)(1+τP)

2 +pD

2 . (15)

Alsotheprint platformwill respond toa low taxrate onitsreader revenueswith high readerpricesifthevalueontheadvertisingmarketofanextrareaderisgreaterthanits marginalcost.

From the response functions we note that prices are strategic complements, so that theytendto moveintandem inresponse to changesin exogenousvariables (e.g.in tax rates).Combining(14)and(15)wefindtheequilibrium prices:

pSHD =t+(vD−vP)2b(1+τD)(1+τP)(b−cP)

3 and (16)

pSHP =t+(vP−vD)2(1+τP)(b−cP)−b(1+τD)

3 . (17)

Summingup:

Proposition5. Single-homing. Supposethat thereisareduction inthe taxrateon a)Thedigital platform. Thenbothplatforms willincrease consumerprices.

b)The printplatform. Thenbothplatforms willincrease consumerpricesifb>cP. Bothwilldecrease consumerpricesifb<cP.

It is straight forward to see that profit is strictly decreasing in own tax rate (see Appendix).Interestingly,themulti-homingresultinProposition2,thatevenacompletely non-altruistic newspapermight findit optimal to lobbyfor a reduction of thetax rate paid byits rival, survives also under single-homing. In themulti-homing casethis was duetocompetitionintheadvertisingmarket,whileitisduetocompetitioninthereader market under single-homing. More precisely, if platform i responds to a tax reduction byincreasing its priceunder single-homing, platform jwill capture a larger number of readers(andchargeahigherprice,sincepricesarestrategiccomplements)andthusmake higherprofitsfromboth readerandadvertisermarkets.

22 Forclarity,notethatunderpuresingle-homingallreadersareexclusiveandsoallreadershavethesame valueontheadmarket.

(14)

Wecanstate:

Proposition6. Single-homing.Profitisdecreasinginowntaxrate.Theprofitoftheprint platformis,moreover,decreasinginthetaxrateofitsdigitalrival.Theprofitofthedigital platform isdecreasing inthe print platform’staxrate if both pricesdecrease inthat tax rate.

Therehavebeensomeconcernsthatdifferencesintaxratesbetweenprintanddigital platforms have led to artificial differences in circulation. This might be correct, but perhaps notin thegenerallyperceiveddirection.Eqs. (16)and (17)show thatreducing the digital tax rateleads to higher prices at eachplatform but that the priceincrease is larger at the digital platform. The relative price increase has the intuitive effect of increasingprintedsalesat theexpenseofdigitalsales:23

Proposition 7. Single-homing. Reducing the tax rate on the digital platform D) will increase salesofprintednewspapers andreduce salesofdigital newspapers.

5. Robustness

5.1. Uncovered markets

Forsimplicity,wehavechosenaframeworksuchthatthenumberofexclusivereaders in themulti-homingcase isindependentof ownprice(c.f. thediscussion belowFig.1).

Inamoregeneralmodel,wherethesupplyofexclusivereadersiselastic,apricedecrease at firm i could win it some exclusive consumers in addition to converting some of js exclusive consumers into shared ones. Our qualitative resultof a negative relationship between VAT rates and reader prices would be unaffected by sucha change.The core requirementforthisfindingisthataplatform’sadrevenuesareincreasinginreadership;

whether readershipincreasescomefrom exclusiveor sharedconsumersdoesnotmatter.

Ifthispositiveindirectnetworkexternalityispresent,thena firmcanrespondtoa VAT increaseonthereadermarketbyreducingthesubscriptionfee,increasingreadershipand increasingadmarketprofits.

5.2. Disutility of ads

Abovewe assumedthatconsumerutilityis unaffectedbythevolume ofads.Nowwe allowforconsumerdisutilityofads;platformsneedtoweightheadmarketbenefitsofan additionaladvertagainstthenegativereadermarketimpactofalessattractiveproduct.

Wenormalisethemassofadvertisersto1,andplatformichoosesanadlevelAi [0,1].

23FulldetailintheAppendix.Astheprofitabilityofdigitalnewspapersincreaseswithadigitaltaxreduc- tion,itispossiblethatsomeofthereductionindigitalcirculationwouldbeoffsetbynewentryofdigital newspapers.Thisisnotconsideredinourmodel.

Referanser

RELATERTE DOKUMENTER

In April 2016, Ukraine’s President Petro Poroshenko, summing up the war experience thus far, said that the volunteer battalions had taken part in approximately 600 military

Based on the above-mentioned tensions, a recommendation for further research is to examine whether young people who have participated in the TP influence their parents and peers in

Overall, the SAB considered 60 chemicals that included: (a) 14 declared as RCAs since entry into force of the Convention; (b) chemicals identied as potential RCAs from a list of

An abstract characterisation of reduction operators Intuitively a reduction operation, in the sense intended in the present paper, is an operation that can be applied to inter-

Ut ifra denne studien har vi kommet frem til at deltakelsen i plattformen Foodora vil kunne svekke merkevarene til sterke luksuriøse restauranter hvor opplevelsen er en stor del

There had been an innovative report prepared by Lord Dawson in 1920 for the Minister of Health’s Consultative Council on Medical and Allied Services, in which he used his

The ideas launched by the Beveridge Commission in 1942 set the pace for major reforms in post-war Britain, and inspired Norwegian welfare programmes as well, with gradual

On the first day of the Congress, on Wednesday 3 June, 2009, we will organize a Pre Congress Workshop on topics related to museums of the history of medicine, addressing the