Koldioxidrelaterad skatt på bilar Impacts from CO2 differentiated
Transcription
Koldioxidrelaterad skatt på bilar Impacts from CO2 differentiated
Koldioxidrelaterad skatt på bilar Impacts from CO2 differentiated vehicle taxes on CO2 emissions from passenger cars Malene Sand Jespersen Jørgen Jordal-Jørgensen Nicolai Kristensen Anders Pontus Matstroms 5187 Beställningsadress: Naturvårdsverket Kundtjänst 106 48 Stockholm Tfn: 08-698 12 00 Fax: 08-698 15 15 E-post: [email protected] Internet-hemsida: www.naturvardsverket.se Miljöbokhandeln: www.miljobokhandeln.com isbn 91-620-5187-3 issn 0282-7298 © Naturvårdsverket Tryck: Naturvårdsverkets repro 2002-03 Upplaga: 300 ex Koldioxidrelaterad skatt på bilar Impacts from CO2 differentiated vehicle taxes on CO2 emissions from passenger cars Malene Sand Jespersen Jørgen Jordal-Jørgensen Nicolai Kristensen Anders Pontus Matstroms Förord Naturvårdsverket till uppgift att utveckla styrmedel så att miljömålen om en hållbar utveckling kan nås och Energimyndigheten att driva på utvecklingen för effektivare energianvändning och minskad klimatpåverkan. De genomsnittliga koldioxidutsläppen från nya bilar är högre i Sverige än i övriga EU-länder och för att målet om begränsad klimatpåverkan ska kunna nås, är det viktigt att minska utsläppen av koldioxid från trafiken. Som ett led i detta arbete presenterar Naturvårdsverket och Energimyndigheten här en studie om hur fordons- och försäljningsskatter på bilar kan bidra till minskade koldioxidutsläpp. COWI har genomfört studien. Här analyseras konsekvenserna av två scenarier – att omforma den befintliga fordonsskatten genom att inkludera en koldioxidberoende del och att införa en försäljningsskatt. De samhällsekonomiska effekterna analyseras ur ett kort-, medium- och långsiktigt perspektiv. Studiens resultat baseras i huvudsak på modellbaserade beräkningar. Rapporten har utarbetats på uppdrag av en projektgrupp med deltagare från Naturvårdsverkets enhet för kommunikationer och energi bestående av Reino Abrahamsson och Larsolov Olsson samt Peter Kasche från Energimyndigheten. Medverkande från COWI har varit Malene Sand Jespersen, Jørgen Jordal-Jørgensen och Nicolai Kristensen. Från VTI (Väg- och transportforskningsinstitutet) medverkade Anders Pontus Matstroms. Rapporten är framtagen i ett nära samarbete mellan COWI, Naturvårdsverket och Energimyndigheten. Bedömningar, slutsatser och rekommendationer som framförs i rapporten står COWI för. Dessa delas inte nödvändigtvis av Naturvårdsverket och Energimyndigheten. Stockholm i mars 2002 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars iii Table of Contents 1 1.1 1.2 1.3 1.4 1.5 Sammanfattning Slutsatser Syfte, studiens omfattning och organisation Definition av skattescenarierna Tillvägagångssätt Resultat 1 1 6 6 8 13 2 2.1 2.2 2.3 2.4 2.5 Executive summary Main conclusions Purpose, scope of study and study organisation Definition of the tax scenarios Study approach Results 21 21 26 26 29 34 3 3.1 3.2 3.3 Introduction Background and purpose Approach and study organisation Outline of this report 43 43 44 45 4 4.1 4.2 4.3 4.4 4.5 Model and methodology The overall model framework and methodology The car choice model The scrapping module The car fleet module Socio-economics 49 49 51 56 58 59 5 5.1 5.2 5.3 5.4 5.5 Data and model inputs Taxes and charges Vehicle sales Scrapping Vehicle fleet Socio-economics of car buyers 61 61 61 63 64 68 iv Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 5.6 5.7 Socio-economics assessment Elasticities 72 72 6 6.1 6.2 The tax scenarios Reference level The two main scenarios 77 77 85 7 7.1 7.2 7.3 Results of the scenarios Base scenario and the registration tax scenario Base scenario and circulation tax scenario Registration tax scenario - Sensitivity Analysis 87 87 90 92 8 8.1 8.2 Socio-economic implications Private households Employment effects 101 101 108 9 9.1 9.2 9.3 Cost Benefit Analysis Total cost benefit analysis Decomposed effect from CO2 differentiation Concluding comments 115 115 124 128 10 10.1 10.2 10.3 10.4 Conclusions CO2 reducing potentials Socio-economic implications Policy implications Applicability of study results 129 129 133 134 135 11 References 137 Table of Appendices Annexes A. The Car Choice Model B. Model validation C. Sensitivity Analysis Tables Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 1 Sammanfattning 1.1 Slutsatser 1 Denna studie har undersökt effekterna av att använda fordonsskatter som medel för att reducera personbilars koldioxidutsläpp. Studiens resultat baseras i huvudsak på modellbaserade beräkningar. Beräkningar har gjorts för att analysera den potentiella koldioxidreduktionen och andra effekter i två skattescenarier. I första scenariot antas att en koldioxidberoende skatt på bilköp (en försäljningsskatt) införs och i det senare scenariot analyserar man effekterna av en koldioxidberoende skatt på bilägarskap (fordonsskatt). Innan man redogör för slutsatserna är det dock nödvändigt att redogöra för de grundläggande principerna för de undersökta skattescenarierna. Referensnivå Bestämma referensnivån Valet av referensnivå har varit en avgörande punkt i utformningen av ovan nämnda koldioxidskatter. Referensnivån bestämmer för varje bilstorlek nivån på koldioxidutsläpp som ger oförändrad bilskatt. Koldioxidutsläppen mäts i antal gram koldioxid per kilometer. Om koldioxidutsläppen från en speciell bilstorlek överstiger referensnivån kommer detta resultera i en skatteökning och om koldioxidutsläppen är mindre än referensnivån kommer detta resultera i en skatteminskning. Följaktligen beräknas skattebasen som: koldioxidutsläpp för aktuell bilstorlek minus referensnivån för koldioxidutsläpp och skatten fastställs som ett fast belopp i SEK per gram. Resultaterande skatt kan bli positiv, negativ eller noll beroende på skillnaden mellan utsläppsnivån från aktuell bil och gällande referensnivå. Fordonsskatten består av den nya hypotetiska koldioxidskatten och den existerande fordonsskatten och systemet är utformat så att den totala fordonsskatten för en bestämd bil aldrig kan vara negativ. Den nya hypotetiska försäljningskatten antas bestå av en värdebaserad skattedel och en koldioxidskatt och i likhet med fordonsskatten kan den totala försäljningsskatten inte bli negativ. Med andra ord, systemen tillåter inte subventioner. En ren koldioxidskatt skulle leda till en påtaglig storleksminskning (downsizing)1 vilket skulle göra de mindre och mest koldioxideffektiva bilarna mycket billigare än de större bilarna jämfört med dagens situation. Även om sådan storleksminskning kan utgöra ett effektivt sätt för att reducera koldioxidutsläpp kan 1 Termen storleksminskning kan användas för att beskriva situationen då bilköpare skulle tendera att köpa mindre bilar än vad de gör idag vilket skulle resultera i att genomsnittstorleken för bilarna i bilparken skulle minska. 2 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars detta leda till konflikt med andra samhällsmål. Beträffande modellberäkningarna i denna studie har man utformat en referensnivå som tillåter viss storleksminskning, men endast i storleksordningen 25 %. I praktiska termer innebär detta att referensnivån definieras huvudsakligen som en funktion av bilstorleken, där storleken mäts som bilens markarea. Valet av nivå på storleksminskning i denna studie utesluter inte att man beaktar även andra nivåer av storleksminskning som grund för referensnivån. Försäljningsskatten antas bestå av både en värdebaserad del (10% av bilens värde) och en koldioxidberoende del (880 SEK per antal gram koldioxid över referensnivån och beräknad som ett negativt värde i det fall utsläppsnivån är under referensnivån). Den totala försäljningsskatten som bilköparen skall betala kan inte vara negativ. Försäljningsskattens koldioxiddel kan dock vara negativ. Fordonsskatten definieras som existerande fordonsskatt plus en koldioxidberoende del. Den koldioxidberoende delen är satt till 44 SEK per gram koldioxid över referensnivån och till ett negativt värde när utsläppsnivån är under referensnivån. Den antagna ändringen av fordonsskatten är endast tänkt att gälla för nya bilar och den totala fordonsskatten kan inte vara negativ. Resultat av scenarieberäkningar Undersökningarna av de två skattescenarierna pekar på följande slutsatser: • Den antagna försäljningsskatten ger en årlig reduktion av bilparkens totala koldioxidutsläpp i storleksordningen 5% under 20 år. Sett i ett kortare tidsperspektiv är de årliga reduktionerna mindre dvs. strax över 1% under 5 år. De potentiella reduktionerna är större än de som de antagna ändringarna av fordonsskatten ger, men de uppnås huvudsakligen genom en minskning av antalet bilar i bilparken. Den genomsnittliga utsläppsnivån per bil (gram koldioxid/km/bil) är nästan konstant i detta scenario och bilarnas genomsnittsålder ökar med ca ett halvt år som ett resultat av minskad nybilsförsäljning. • Fordonsskatten skulle ge en årlig reduktion i storleksordningen 2% under 20 år och ca 0,5% under 5 år. Trots att denna reduktion är mindre än den för försäljningsskatten innebär den dock en mer påtaglig reduktion av den genomsnittliga utsläppsnivån per bil. I detta fall skulle de genomsnittliga utsläppen reduceras med 2,5% under 20 år jämfört med obetydlig reduktion genom försäljningsskatten. Dessutom påverkas inte bilparkens storlek. • Försäljningen av svensktillverkade bilar (dvs. Volvo och SAAB) på den svenska marknaden kommer, i båda scenarierna, att påverkas i större omfattning än den totala bilförsäljningen. Detta på grund av att svensktillverkade bilar typiskt är större och mindre koldioxideffektiva än genomsnittsbilen. Denna effekt kommer att vara tydligast för försäljningsskatten som ett resultat av minskningen av bilparken (med ca 4% under ett 20 års perspektiv). • Ingen av de antagna skattesystemen leder till påtaglig ändring av andelen dieselbilar i bilparken. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars • 3 Företagsbilar är mindre priskänsliga än privatbilar. Dock är företagsbilar mer känsliga för ändringar i årskostnader, dvs. bränsleskatt och fordonsskatt. Därför står privatbilar för 70% av koldioxidreduktionerna i försäljningsskattscenariet medan företagsbilar svarar för 60% av koldioxid reduktionerna i fordonsskattescenariet. Beträffande det senare är det viktigt att komma ihåg att de årliga kostnaderna består av en fordonsskatt och bränslekostnader. Man räknar med att en ökning av fordonsskatten med t.ex. 1000 SEK skulle ge samma effekt som en ökning av den årliga bränsleskatten med samma belopp. Vidare räknar man med att ändringar i fordonsskatten helt avspeglas i beslutet av företagsbilköp. Idag avspeglas inte den årliga fordonsskatten som företagen betalar för att ha företagsbil i den anställdes privata beskattning. Om denna policy upprätthålls kommer modellberäkningarna i denna studie att överskatta effekten av fordonsskattscenarierna. Om man antar att 33% av alla företagsbilar är andelen där den anställde själv bestämmer vilken bil han/hon vill ha, då kommer överskattningen uppgå till 20% av den totala koldioxidreduktionen. Införandet av en koldioxidberoende del av fordonsskatten skulle samtidigt göra några av de mest ineffektiva bilarna mycket dyra i form av årlig företagsbeskattning. Därför skulle vissa företag inte längre erbjuda sina anställda att ha företagsbil. Den reella överskattningen skulle därmed förmodligen blir avsevärt lägre än 20%. Eftersom företagsbilar utgör ca 50% av nybilsförsäljningen och de är känsligare för de årliga kostnaderna, är dessa antaganden viktiga. Slutsatser Man bör vara försiktig vid jämförelsen av de två systemen, då de påverkar bilköparnas och bilägarnas skattebörda på olika sätt. Man kan dock dra vissa viktiga allmänna slutsatser. Analyserna visar tydligt på vikten av att skilja mellan skattenivåer och skattedifferentiering. Grovt sett påverkar skattenivån den allmänna prisstrukturen i sig, dvs. den gör bilarna relativt dyrare jämfört med andra konsumentvaror. Därför påverkar skattenivån människors val att ha en bil eller inte. Som jämförelse påverkar diffentiering människors val men har mindre inverkan på den allmänna prisstrukturen (när man jämför bilar med andra konsumentvaror). Försäljningsskatten pålägger bilköparna nya skattebördor då den innebär införandet av en ny skatt. Ökningen av skattenivån påverkar bilparkens storlek negativt eftersom den leder till en generell ökning av bilpriserna i relation till andra konsumentvaror. Detta leder till minskad skrotning och minskad försäljning av nya bilar. Förutom en ökning av skattenivån, består den antagna utformningen av den nya försäljningsskatten även av en skattedifferentierad del, den koldioxidberoende delen av skatten. Denna del medför ett enhetligt pris (skatt) för 4 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars varje g koldioxid som släpps ut per kilometer.2 Denna del av skatten påverkar inte antalet bilar utan endast människors bilval. I fordonsskattescenariot antar man endast förändringar i utformningen av existerande skatt och därigenom påverkas inte den genomsnittliga skattebördan. Därför påverkas inte antalet bilar utan endast människor val av biltyp. Om fordonsskattens genomsnittsnivå ökas skulle detta inverka på antalet bilar och därigenom ge en ytterligare koldioxidreducerande effekt. En ökning skulle även leda till en mer specifik differentiering och därmed även en ökad inverkan på bilvalet. Sammanfattningsvis har studien visat att: • Den antagna fordonsskatten kan ge årliga koldioxidreduktioner i storleksordningen 2% under en 20-års period utan att ge avsevärt skeva effekter på bilparkens struktur. Trots att scenarierna visar på en övergång från svensktillverkade bilar till andra märken så är detta skifte relativt litet. I realiteten kommer antagligen den svenska biltillverkningsindustrin att rätta sig efter denna förändring bland annat genom att utveckla och producera biltyper som är mera anpassade till de nya förhållandena. • Försäljningsskatten kommer att ge större koldioxidreduktion men kommer även att ge mer specifika motsatta effekter såsom reducerad bilpark, minskad försäljning av svenska bilmärken och ökad genomsnittsålder för bilar. Det senare kan visserligen vägas upp av införandet av ett skrotningspremiesystem men detta måste utformas med försiktighet för att garantera dess effekt i relation till koldioxidmålet. För att säkerställa kvaliteten i studiens resultat beträffande försäljningsskatten har känslighetsanalyser använts. Analyserna visar att ovannämnda slutsatser är mycket starka beträffande en mängd antaganden, däribland koldioxidskattenivån, den allmänna skattenivån och skrotningselasticiteten. Däremot är resultaten mycket känsliga för storleken på bilparkens generella priselasticitet. Den använda elasticiteten ligger på –0,6 och bygger på resultaten från andra relevanta studier. Elasticiteten visar att en ökning av bilpriserna med 10% skulle leda till en minskning av bilparken med 6%. En ändring av elasticitetsnivån till – 0,35 skulle ge en koldioxidminskning med 3% istället för 5% när elasticiteten är –0,6. Orsaken till detta är att en lägre elasticitet innebär en mindre inverkan på bilförsäljningen och därmed på bilparkens storlek.. Kostnadsnyttoanalys Försäljningsskattescenariot och fordonsskattescenariots konsekvenser för samhället har analyserats med hjälp av en kostnadsnyttoanalys. Analysen visar att kostnaderna, som följer av införandet av försäljningsskatten, kommer att överstiga fördelarna. Detta framför allt på grund av en minskning i bilparksstorleken. Förutom att den minskade bilparken ger koldioxidreduktioner ger den även en påtaglig välfärdsförlust. Värdet av denna välfärdsförlust utgör i själva verket den huvudsakliga kostnadsdelen av försäljningsskatten. Storleken på den negativa nettovinsten är dock beroende av det pris som används för att värdera 2 ovanför den bestämda referensnivån Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 5 de uppnådda koldioxidreduktionerna. Som jämförelse pekar kostnadsnyttoanalysen på att fordonsskattescenariot kommer att innebära en nettovinst för samhället. Den kommer att ge mindre skadliga effekter samtidigt som den medför påtagliga koldioxidreduktioner. Reduktionerna är dock mindre än de uppnådda i försäljningsskattescenariot. Nettofördelarna i detta scenario kommer att öka över tiden i och med att nya mer bränsleeffektiva bilar kommer att penetrera bilparken över tiden. Kostnadsnyttoanalyserna har även värderat enbart effekterna av den koldioxiddifferentierade delen av de två skatterna, dvs. man bortser från den effekt på priset som enbart är en effekt av försäljningsskattens införande. Analysen visar att koldioxiddiffentieringen, i båda fallen, skulle ge en nettovinst för samhället men effekten är fortfarande störst för fordonsskatten. Befolkningen på landsbygden tenderar att köpa större och mindre bränsleeffektiva bilar jämfört med tätortsbefolkningen. Landsortsbefolkningen kommer därför att påverkas i större omfattning av skattescenarierna än tätortsbefolkningen. Beräkningarna har också visat att barnfamiljer tenderar att påverkas mer än familjer utan barn. Trots att dessa effekter inte utgör en del av kostnadsnyttoanalysen, då de huvudsakligen relaterar till problemställningar beträffande fördelning, visar de dock på att skattescenarierna kommer att ge fördelningseffekter. Beräkningar visar också att aktivitetsnivån i den svenska biltillverkningsindustrin kommer att påverkas negativt. Detta är mest påtagligt för försäljningsskatten och kommer att öka ytterligare om man även tar med i beräkningen effekten på relaterade industrier. Effekterna på den ekonomiska aktiviteten har endast betydelse för lönsamheten i den utsträckningen dessa aktiviteter inte ersätts av andra aktiviteter någon annanstans i ekonomin. Återigen visar det dock på att det också kommer att innebära snedvridande effekter för industrin. Policyeffekter Denna studie har visat att fordonsbeskattning kan påverka koldioxidutsläppen. Den har också visat att en fordonsskatt troligen skulle vara det mest effektiva instrumentet att använda i detta fall. Det finns dock andra sätt att påverka bilparkens koldioxidutsläpp, bland annat genom att påverka körkostnaden såsom drivmedelskostnader och vägavgifter. Dessa alternativ har dock inte analyserats här. Kravet på beräkningarna i denna studie är att skatterna högst får ge en bilstorleksminskning på 25%. Detta begränsar de koldioxidreduktioner som kan uppnås. Om man vill uppnå avsevärt ökade reduktioner kan det krävas mer storleksminskning. Om målet är avsevärt ökade reduktioner av de genomsnittliga koldioxidutsläppen från nya bilar krävs ytterligare differentiering av koldioxidskatten. En högre differentiering kräver antingen en högre skattenivå eller att man tillåter att skatten blir en nettosubvention för de mest koldioxideffektiva bilarna. Policyn vars syfte är att uppmuntra köp och användning av mer energieffektiva fordon kommer att innebära relativt sett högre kostnader för landsortsbefolkningen än tätortsbefolkningen och vara fördelaktigare för familjer utan barn, än för familjer med barn. Detta till följd av att dessa grupper köper större och mer 6 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars bränsleförbrukande bilar. För att kompensera för denna fördelningseffekt kan det finnas önskemål att kompensera detta. Sammanfattning I den resterande delen av sammanfattningen presenteras studiens syfte och den använda metoden. Dessutom presenteras de viktigaste resultaten, både vad gäller utsläpp och samhällsekonomiska effekter, som följer av ett införande av antingen en försäljningsskatt eller en fordonsskatt. 1.2 Syfte, studiens omfattning och organisation Organisation Denna rapport presenterar resultaten av en studie gjord för Svenska Naturvårdsverket och Svenska Energimyndigheten. COWI A/S har utfört studien i samarbete med Statens Väg och transportforskningsinstitut (VTI) i Sverige under perioden september till december 2001. Grunddragen i de antagna ändringarna av skattesystemet - scenarierna- bestämdes i nära samarbete med Naturvårdsverket och Energimyndigheten. På liknande sätt bestämdes definitionerna av de olika tillämpade resultattyperna och de alternativa beräkningar som skulle göras i nära samarbete med kunden. I detta avseende har de tre projektmötena med kunderna och andra berörda myndigheter på Naturvårdsverket varit ytterst värdefulla.3 Syftet med mötena har varit att säkra en gemensam ståndpunkt beträffande projektets omfattning och framskridande. Syfte Syftet med studien är att analysera effekterna på hushållens efterfrågan av nya bilar vid ett införande av koldioxiddifferentierade försäljningsskatter. Under arbetets gång bestämdes det att man även skulle analysera konsekvenserna av att omforma fordonsskatten och inkludera en koldioxidberoende del. Analysen skulle både undersöka effekten på koldioxidutsläppen och konsekvenserna för bilförsäljningen och bilparken. Studien skulle skilja mellan dieselbilar och bensinbilar å ena sidan, och privatbilar och företagsbilar å andra sidan. Studien skulle inte enbart titta på konsekvenserna för nybilsförsäljningen utan även analysera inverkan på utskrotningen av gamla bilar och på bilparken. Studien skulle dessutom analysera de samhällsekonomiska effekterna och beräkningarna skulle gälla för både kort-, medium- och långsiktigt perspektiv. 1.3 Referensnivå Definition av skattescenarierna En viktig del i definitionen av skattescenarierna är definitionen av en referensutsläppsnivå. Koldioxidskatter införs endast för utsläpp som överskrider denna referensnivå och skatten blir negativ när utsläppen är mindre än referensnivån. Koldioxidskattebasen utgör alltid mängden (gram) koldioxid per kilometer minus referensnivån. Referensnivån utgör den mängd (gram) koldioxid per kilometer som inte ger någon skatt. En lika viktig del i definitionen av skattescenarierna är att systemen definieras så att den totala försäljningsskatten (eller fordonsskatten) aldrig kan bli negativ, dvs. nettosubvention är inte tillåtet. Dessutom är fordonsskatten budgetneutral, 3 4 oktober, 15 november och 18 december 7 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars dvs. den skall ge samma intäkt som gällande skattesystemet, varken mer eller mindre. Vid definitionen av referensnivån har man försökt att definiera den så att den både erkänner betydelsen av att tillåta storleksminskning, men samtidigt så att den inte leder till allt för hög storleksminskning. För att illustrera detta ytterligare kan man titta på två ytterligheter. Den ena ytterligheten är ett system där skatten endast är beroende av mängden (gram) koldioxidutsläpp per bil och kilometer. Ett sådant system skulle alltid gynna de små bilarna på bekostnad av de större och oundvikligen leda till påtaglig storleksminskning. Den andra ytterligheten är ett system som endast syftar till att påverka människors bilval inom givna storlekskategorier. Ett sådant system skulle inte leda till någon storleksminskning utan istället leda till en oönskad storleksökning eftersom de minsta bilarna (mest bränsleeffektiva) skulle få en högre skatt i förhållande till bilpriset än större bilar. Följaktligen skulle de större bilarna bli relativt billigare än de var tidigare. Sammanfattningsvis har man definierat en referensnivå som tillåter en storleksminskning med 25%. Detta illustreras i figuren nedan där gram koldioxid per kilometer är ritad i förhållande till bilens markarea. Figur 1.1 Referensnivå för koldioxid, bensinbilar g CO2 per kilometer 500 400 300 200 100 4 5 6 7 8 9 10 11 2 Markarea (m ) Kurvan i figuren visar resultatet av en beräkning som baseras på en viktning på 25% mot storleksminskning och 75% för en skattekurva som inte tillåter någon storleksminskning alls. Denna referensnivå har tillämpats på bensindrivna bilar. Figuren innehåller markeringar för varje biltyp som finns på den svenska marknaden. Bilar som befinner sig ovanför linjen betalar koldioxidskatt och bilar som befinner sig under linjen erhåller en koldioxidrabatt. Om man tillät en storleksminskning på 100% skulle referensnivån endast ge en horisontell linje i figuren. Om man däremot inte tillät någon storleksminskning skulle kurvan i figuren bli brantare. 8 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars En liknande referensnivå har tillämpats för dieselbilar. Dessa referensnivåer används både i fordonskattescenariot och i försäljningsskattescenariot. Tillämpade skattescenarier I utformningen av försäljningsskattescenariot har man varit noggrann med att sätta en nivå som inte ger orealistiskt höga skatter men är tillräckligt hög för att ge utrymme för differentiering och därigenom möjliggör betydande beteendeförändringar som resulterar i en reduktion av koldioxidutsläpp. Skattescenarierna är sammanfattade i tabellen nedan. Det bör noteras att skattescenarierna har baserats på ett antal preliminära beräkningar. Tabell 1.1 Översikt skattescenarier1. Skattedel och huvudsakliga kännetecken Fordonsskatt Försäljningsskatt Icke koldioxidberoende del Existerande skatt: skatten är differentierad efter bilens tjänstevikt enligt en trappmässigt linjärt samband. En ökning av skatten med ca 150 SEK för varje 100 kilo för bensinbilar och med 575 SEK för dieselbilar. Värdebaserad (ny skatt): Koldioxidberoende del 44 SEK per gram ovanför referensnivån och ett liknande negativt värde om utsläppen är under referensnivån 880 SEK per gram ovanför referensnivån och ett liknande negativt värde om utsläppen är under referensnivån. Andra kännetecken Fordonsskattens totala intäkt skall vara oförändrad Den totala försäljningsskatten kan inte vara negativ 10% av bilvärdet 1) Separata funktioner är uträknade för dieselbilar och bensinbilar och referensnivåerna är illustrerade i Figur 1.1 1.4 Tillvägagångssätt 1.4.1 Modellbaserade analyser av skattescenarierna En viktig utgångspunkt i studien är bilvalsmodellen som COWI tidigare har utvecklat och använt vid många andra tillfällen. Modellen innehåller en detaljerad modellering av bilköpares efterfrågemönster. Den ger relativt detaljerade analyser av hur förändrade skattenivåer och skattestrukturer påverkar hushållens önskemål. Denna modell utgjorde utgångspunkten för utformning av bilefterfrågemodulen som har tillämpats i denna studie. Denna modul kombinerades samtidigt med en bilparksmodul och en skrotningsmodul som illustrerat nedan. Figur 1.2 Illustration av ingång och utgång Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Modellbaserat synsätt 9 Denna studie bygger främst på användningen av olika modeller. Figuren ovan visar de viktigaste delarna i detta tillvägagångssätt. Beräkningarna använder tre separata men inbördes beroende moduler: • • • Modul för registrering av nya bilar Modul för bilpark Modul för skrotning Nya bilar Modulen för registrering av nya bilar mäter i vilken utsträckning hushållens val av nya bilar påverkas av förändringar i prisstrukturen som en följd av ändrade eller nya skatter. Modulen bygger på en detaljerad kartläggning av försäljningen av nya bilar i Sverige under året 1999/2000 vilket inkluderar en mångfald av fysiska kännetecken, energieffektivitetsförhållanden och priset på varje biltyp. Vidare innehåller modulen en detaljerad kartläggning av relevanta bilköpares samhällsekonomiska kännetecken och en detaljerad kartläggning av varje grupps preferenser och priselasticitet. Bilparken Modulen för bilparken uppdaterar varje år bilparken med hjälp av resultaten från de andra två modulerna. Bilparksmodulen innehåller också en detaljerad kartläggning av den svenska bilparken som den var i september 2001. Studien använder sig av en kort, medium och lång (20 år) tidshorisont och bilparksmodulen används regelbundet för att registrera förändringar från år till år. Bilskrotning Skrotningsmodulen används för att registrera bilskrotningen. Skrotning bestäms av två faktorer: bilarnas överlevnadskurva och skrotningspriselasticiteten. Generell priselasticitet Om den totala skattebördan som påförs bilköparna eller bilägarna ökar, kommer detta högst sannolikt leda till att antalet bilar i bilparken minskar. Skrotningsmodulens priselasticitet visar på människors tendens att behålla sina bilar längre och därigenom öka bilarnas genomsnittsålder. En generell priselasticitet är dock nödvändig för att ta hänsyn till effekten av den resulterande prisökningen på bilparken, och därigenom effekten på nybilsförsäljningen. Studien använder en långsiktig priselasticitet för bilägarskap på 0,6, dvs. en ökning av priset med 10% kommer att innebära en nedgång i bilparkens storlek med 6% i det långa loppet. Denna elasticitet har uppnåtts i ett flertal svenska studier. Beräkningarna antar att halva denna effekt är gradvis införd efter 7 år och att den totala effekten är gradvis införd efter 30 år. Utveckling av energieffektivitet Erfarenheter visar att bilars bränsleeffektivitet har förbättrats med ca 2% per år det senast årtiondet och beräkningarna antar att denna trend kommer att hålla i sig. ”Vad händer om”synsätt Det ovannämnda modellbaserade tillvägagångssättet är av en "vad händer om"natur. Man har inte tagit med i beräkningen makroekonomisk utveckling eller andra liknande dynamiska drivkrafter. Styrkan i detta tillvägagångssätt är att det ger en tydlig analys av koldioxidegenskaper och relaterade biverkningar av de analyserade skatteinstrumenten. Samtidigt minskar man antalet uppskattningar och komplexiteter i beräkningen. Modellen bör därför inte användas för planering och prognoser. 10 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 1.4.2 Olika typer av resultat Resultaten av de modellbaserade beräkningarna visar på de kort, medium, och långsiktiga effekter som skattescenarierna ger. Resultaten är beräknade utifrån första året då den antagna ändringen inträffade och efter fem, tio och tjugo års implementering. För varje år beräknas en basnivå (nollnivå). Denna basnivå beräknar resultaten av nu existerande skattesystem. Effekter beräknas således som skillnaden mellan basnivån för gällande år och scenarioresultaten för gällande år. Grundnivå Skapandet av en basnivå är nödvändigt för att kunna genomföra de efterfrågade analyserna. Basnivån förutsätter att existerande skatter behålls och att ingen av de antagna ändringarna införs. Tabellen nedan sammanfattar basvärdena för försäljningen av nya bilar för det första året. I bedömningen av beräkningsresultaten av konsekvenserna av de antagna skattescenarierna har resultaten från beräkningarna jämförts med basvärdena. Tabellen ger en kort beskrivning av de olika resultattyperna som sådana jämförelser skulle ge. Tabell 1.2 Resultattyper och basvärden: nybilsförsäljning. Resultattyp Basvärde Motivering Genomsnittligt koldioxidutsläpp (g per km) 198,1 Illustrerar en förbättring av koldioxidresultatet i genomsnitt (koldioxidutsläpp per kilometer per bil) Genomsnittlig livslång skatteintäkt (SEK/bil) 96 237 Illustrerar den genomsnittliga effekten på varje bilägares skattebetalning över fordonets livslängd (från köp till skrotning). Skatten beräknas som försäljningsskatt plus värde per livslängd av fordonsskatt och bränsleskatt Genomsnittsstorlek 3,70 Ändringar av denna indikator illustrerar betydelsen av storleksökning respektive storleksminskning Genomsnittlig dieselandel 6,5% Detta resultat visar huruvida andelen dieselbilar i nybilsförsäljningen ändras som ett resultat av den antagna skatteändringen Genomsnittlig andel svensktillverkade bilar 30,5% Detta resultat visar andelen svensktillverkade bilar i relation till den totala bilförsäljningen Genomsnittlig försäljningsskatt (SEK/bil) - - Genomsnittlig fordonsskatt (SEK/bil/år) 1 566 Illustrerar vad som händer med den genomsnittliga fordonsskatten som ett resultat av den antagna ändringen. Införandet av en försäljningsskatt kan leda till ändringar i försäljningsskatten som ett resultat av en annorlunda sammansättning av bilförsäljningen. Genomsnittligt försäljningspris, SEK exkl. moms 145 825 Används för att ge en indikator på vad som händer med biltillverkningsindustrins omsättning. Måste kombineras med resultatet av sålda bilar för att ge en rättvis bild. Genomsnittlig skatt per livslängd består av Försäljningsskatt Fordonsskatt (för hela livslängden) Visar sammansättningen av den genomsnittliga skatten per livslängd som skall betalas. 24 780 11 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Resultattyp Bränsleskatt (för hela livslängden) Basvärde Motivering 71 456 På liknande sätt illustrerar tabell 1.3 basvärdena för bilparken. Dessa värden omfattar både de kort, medium och långsiktiga perspektiven. Basvärdena illustrerar vad som skulle hända om inga ändringar gjordes av existerande skatter. Som framgår av tabellen minskar de genomsnittliga koldioxidutsläppen (g/km/bil) och de totala koldioxidutsläppen över visad period. Detta är framför allt ett resultat av det senaste årtiondets tekniska utveckling och antagandet att denna trend kommer att hålla i sig. Tidigare erfarenheter visar att tekniska framsteg ger en årlig förbättring av bränsleeffektiviteten med ca 2%. Den gradvisa ersättningen av gamla bilar med nya och mer bränsleeffektiva bilar leder till genomsnittsreduktioner av koldioxidgenomsnittsutsläppen. Tabellen visar även att fordonens genomsnittliga livslängd tenderar att minska lite över tiden medan bilarnas genomsnittstyngd ökar lite. Värt att notera är också att andelen dieselbilar kommer att ligga kvar på en låg nivå men kommer att öka från nästan 5% till 6,5%. Tabell 1.3 Basscenario, total bilpark (1, 5, 10 och 20 år) Bas 1 Antal bilar 4 153 155 Nybilsförsäljning 205 154 Genomsnittligt koldioxidutsläpp (g CO2/km) 235 Totalt koldioxidutsläpp (ton) 11 994 305 Genomsnittsålder 8.9 Genomsnittslängd (cm) 451 Genomsnittsvikt (kg) 1 340 Antal svenska bilar 1 365 726 Andel svenska bilar 32,9% Antal dieselbilar 198 069 Andel dieselbilar 4,8% Tidshorisont (antal år) 5 10 4 153 155 4 153 156 226 536 195 659 221 203 11 364 228 10 458 465 9.0 9.4 449 447 1 375 1 402 1 289 483 1 248 816 31,0% 30,1% 236 867 276 989 5,7% 6,7% 20 4 153 152 240 064 162 8 325 037 9.2 442 1 408 1 255 375 30,2% 270 625 6,5% Som nämnts är denna modell varken en beräknings- eller prognosmodell. Följaktligen antas efterfrågan på bilar fortsatt att vara stabil om inte ändringar införs i systemet i form av t.ex. prisändringar, som följd av ändringar i skattestrukturen och/eller skattenivån. Därför förblir mängden bilar i bilparken oförändrad under hela perioden. I basscenariot antas att det inte blir några förändringar av nuvarande förutsättningar. Enda undantaget är relaterat till energieffektivitet, vilken antas öka med 2% per år som beskrivits ovan. En annan faktor som påverkar resultaten av basscenariot är skrotningsmönstret. Detta mönster bestäms av åldersfördelningen hos nuvarande bilpark. Den nuvarande bilparken fördelas så att ett stort antal bilar är mellan 1 och 3 år gamla respektive 12 och 14 år gamla. Denna fördelning påverkar varje års omsättningssiffror och den resulterade genomsnittsåldern för bilar. 12 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 1.4.3 Känslighetsanalyser Beräkningarna är föremål för ett antal kritiska antaganden. Därför har känslighetsanalyser utgjort en viktig del av studiens ramar för att visa på säkerheten i resultaten. Översikt av känslighetsanalyser Känslighetsanalyser har genomförts för att värdera i vilken utsträckning de uppnådda resultaten är känsliga för dessa antaganden. Följande känslighetsanalyser har genomförts. Känslighetsanalyserna har endast genomförts i försäljningsskattescenariot eftersom försäljningsscenariot medför störst effekter. Tabell 1.4 Översikt, känslighetsanalyser Känslighetsanal yser Motivering och beskrivning Grad av koldioxiddifferentiering Den ursprungliga nivån av differentiering är 880 SEK per CO2 ovanför referensnivån. Den alternativa beräkningen värderar i vilken utsträckning resultat ytterligare accentueras vid dubblering av denna, dvs. genom att använda en kostnad på 1760 SEK per gram. Försäljningsskattenivå Nivån var ursprungligen satt till 10% av värdet. Med tanke på att den totala skatten inte kan vara negativ har den alternativa analysen till syfte att undersöka huruvida en ökning av detta värde (upp till 20%) kan leda till ytterligare minskningar och vad detta skulle ge för ytterligare bieffekter. Begränsningen att nivån aldrig kan vara negativ innebär att nivån på den nivåbaserade skatten i viss utsträckning definierar omfattningen av den differentieringsnivå man kan använda, men nivån påverkar å andra sidan också bilars genomsnittsålder vars effekt verkar i motsatt riktning. Storleksminskningsnivåer I den ursprungliga strukturen tillåts en storleksminskning med 25%. En höjning av detta värde har ingen inverkan på bilparkens storlek (som bestäms av den generella priselasticiteten) men påverkar bilköparnas val av nya bilar. Ju mer man tillåter storleksminskning desto mer kommer människor att välja mindre och mer energieffektiva bilar. Olika skattefunktioner Scenarierna tillämpar olika skattefunktioner (med likvärdiga former) för dieselbilar och bensinbilar. Dessa funktioner kan påverka resultaten. För att kunna värdera detta har de två funktionerna slagits ihop till en. Skrotningspriselasticitet Skrotningspriselasticiteten fastställer delvis hastigheten med vilken bilparken förnyas och hastigheten med vilken den anpassar sig till nya förutsättningar. Därför undersöks hur en halvering av priselasticiteten påverkar skrotningen. Storleken på den generella priselasticiteten Storleken på den generella priselasticiteten bestämmer antalet bilar i bilparken. Ursprungligen bestämdes elasticiteten till –0,6. En minskning av elasticiteten skulle innebära en minskning av påverkan på bilparken i form av antal bilar och genomsnittsålder. Följaktligen kan man förvänta sig att se en mindre minskning av de totala koldioxidutsläppen. Känslighetsanalyserna undersöker de förändringar som uppkommer om man tillämpar en elasticitet på –0,35 i istället för –0,6. Infasning av den allmänna priselasticiteten Priselasticiteten på –0,6 är långsiktig i den bemärkelsen att det tar många år innan full effekt är uppnådd. Ursprungligen antar beräkningarna att halva priselasticiteten är infasad över en sjuårsperiod och resterande hälft under resterande del av hela 30-års perioden. Känslighetsanalyserna antar att halva elasticiteten istället endast är infasad under en fyraårsperiod och varar hela perioden på 30 år. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 13 1.4.4 Samhällsekonomi och lönsamhet Vid värderingen av de samhällsekonomiska effekterna sattes fokus på två aspekter, en eventuell inverkan på landsortsbefolkningens välfärd och eventuell inverkan på biltillverkningen och dess följdeffekter på exempelvis underleverantörsindustrin. Samhällsekonomi För att illustrera de samhällsekonomiska konsekvenserna av skattescenarierna har följande analyser gjorts: ♦ Analys av i vilken utsträckning landsortsbefolkningen påverkas mer eller mindre jämfört med tätortsbefolkningen. Analysen har gjorts genom att använda kvantitativa indikatorer för den relativa skattebörda som påförs lands- och tätortsbefolkningen och på den välfärdsförlust som varje grupp drabbas av som ett resultat av att välja mindre bilar än i ursprungsfallet. I denna analys skiljer man mellan ensamstående och par och mellan de som har barn och de som inte har. ♦ Analys av i vilken utsträckning sysselsättningen kommer att påverkas. Analysen har gjorts genom att använda kvantitativa indikatorer på sysselsättningseffekter av förändringar för biltillverkningen och på försäljningen av fordon, reservdelar och underhåll plus bensinstationer. Kostnadsnyttoanalys Dessutom värderar studien de kostnader och den nytta som samhället kommer att få som ett resultat av skattescenarierna. Faktorer såsom minskade koldioxidutsläpp, välfärdsförluster som resultat av ändring av bilstorlek, minskad sysselsättning, minskad vinst, effekter på andra utsläpp (NOx, partiklar, SO2, HC), buller och olyckor tas med i beräkningen. 1.5 Nybilsförsäljning Resultat 1.5.1 Koldioxidreduktioner, nybilsförsäljning, skrotning och bilparken Tabell 1.5 och 1.6 visar resultaten som försäljningsskatten respektive fordonsskatten ger på nybilsförsäljningen. Som framgår av tabellen, påverkar fordonsskattescenariot sammansättningen av nybilsförsäljningen i högre omfattning än försäljningskatten. Den leder till en minskning ner till 192,8 g/km/bil på utsläppsnivån jämfört med ett genomsnitt på 195,3 för försäljningsskatten. Detta gäller dock endast nybilsförsäljning. Intressant att notera är att båda skattescenarierna knappast medför några förändringar av andelen dieselbilar. En viktig orsak till detta är att man använt olika referensnivåer för dieselbilar respektive bensinbilar för att på så sätt undvika radikala förändringar i fördelningen mellan de två. Effekterna av att slå samman de två till en skattefunktion, dvs. tillämpa samma referensnivå för båda bränsletyperna, visar dock att det endast skulle leda till marginella ändringar. Andelen dieselbilar skulle i detta fall endast öka med 0,5%. 14 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Tabell 1.5 Resultat av försäljningsskattescenario Bas Genomsnittligt koldioxidutsläpp (g per km) Ny försäljningsskatt 198,1 195,3 Genomsnittlig skatteintäkt per livslängd (SEK per bil) Genomsnittsstorlek Genomsnittlig dieselandel Genomsnittlig andel svensktillverkade bilar 96 237 115 227 3,70 6,5% 30,5% 3,66 6,5% 29,2% Genomsnittlig försäljningsskatt (SEK per bil) Genomsnittlig fordonsskatt (SEK per bil per år) Genomsnittligt försäljningspris, SEK exkl. moms Genomsnittlig skatt per livslängd består av (SEK) Försäljningsskatt Fordonsskatt (för hela livslängden) Bränsleskatt (för hela livslängden) 1 566 20 161 1 558 145 825 143 854 24 780 71 456 20 161 24 543 70 523 Tabell 1.6 Resultat av fordonsskattescenario Bas 198,1 Ny årlig fordonsskatt 192,8 Genomsnittlig skatteintäkt per livslängd (SEK per bil) Genomsnittsstorlek Genomsnittlig dieselandel Genomsnittlig andel svensktillverkade bilar 96 237 92 497 3,70 6,5% 30,5% 3,65 6,6% 29,5% Genomsnittlig försäljningsskatt (SEK per bil) Genomsnittlig fordonsskatt (SEK per bil per år) Genomsnittligt försäljarpris, SEK exkl. moms Genomsnittlig skatt per livslängd består av (SEK) Försäljningsskatt Fordonsskatt (för hela livslängden) Bränsleskatt (för hela livslängden) 1 566 1 443 145 825 142 120 24 780 71 456 22 766 69 731 Genomsnittligt koldioxidutsläpp (g per km) Bilpark Vid betraktelse av hela bilparken ser man att det fortfarande är försäljningsskatten som ger störst minskning av koldioxidutsläppen. Detta beror framför allt på att fordonsskatten inte alls påverkar bilparkens storlek eftersom 15 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars fordonsskatten endast är en rekonstruktion av existerande skatt medan försäljningsskatten involverar utformningen av en ny skatt som skall komplettera existerande skatt. Därför påverkar försäljningsskattescenariot bilpriserna i relation till andra konsumentvaror. Det gör inte fordonsskattescenariot. Analyserna visar att en försäljningsskatt som består av en koldioxidskatt på 880 SEK per gram koldioxid per bil plus en värdebaserad del (10% av bilvärdet) kommer att ge koldioxidreduktioner i storleksordningen 5% i långa loppet (20 år). På medellång sikt kommer reduktionerna att vara i storleksordningen 3% och 2% på 10 respektive 5 år. Den motsvarande förändringen i de genomsnittliga utsläppen per bil är dock försumbara, vilket innebär att den huvudsakliga reduktionen kommer av en minskning av bilparken. Detta scenario leder också till en ökning av genomsnittsåldern som ett resultat av minskad skrotning. De modellbaserade beräkningarna visar att genomsnittsåldern skulle ha ökat med ett halvt år på 20 år. Resultaten av de två skattescenarierna för bilparken redovisas nedan. Försäljningen av svenska bilar påverkas i båda scenarierna. Detta återspeglar det faktum att svensktillverkade bilar tenderar att vara större än vad genomsnittbilmärkena är och därmed påverkas mer av de antagna scenarierna. Tabell 1.7 Effekter av försäljningsskatt, total bilpark (1,5, 10 och 20 år) Förändring Antal bilar Försäljning av nya bilar Genomsnittligt koldioxidutsläpp (g CO2/km) Totalt koldioxidutsläpp (ton) Genomsnittsålder Genomsnittslägd (cm) Genomsnittsvikt (kg) Antal svenska bilar Andel svenska bilar Antal dieselbilar Andel dieselbilar 1 -24 846 -20,47% 0,37 -43 669 0,15 0,01 -1,82 -7 920 0,01% -2 225 -0,03% Tidshorisont (antal år) 5 10 -114 289 -183 303 -12,59% -6,81% 1,27 1,07 -203 436 -329 885 0,53 0,60 -0,07 -0,28 -6,80 -8,73 -41 043 -71 310 -0,14% -0,41% -10 372 -15 064 -0,10% -0,07% 20 -253 094 -11,21% -0,10 -410 717 0,53 -0,92 -10,79 -113 248 -0,94% -17 940 -0,04% 16 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Tabell 1.8 Effekter av fordonsskatt, total bilpark (1,5, 10 and 20 år) Förändring 1 Antal bilar Försäljning av nya bilar Genomsnittligt koldioxidutsläpp (g CO2/km) Totalt koldioxidutsläpp (ton) Genomsnittsålder Genomsnittslängd (cm) Genomsnittsvikt (kg) Antal svenska bilar Andel svenska bilar Antal dieselbilar Andel dieselbilar Effekt av försäljningsskattescenariot Resultat 1 0,00% -0,26 -10 322 0,00 -0,06 -0,60 -1 932 -0,05% 151 0,00% Tidshorisont (antal år) 5 10 0 0 0,00% 0,00% -1,31 -2,41 -53 146 -97 978 0,00 0,00 -0,34 -0,65 -3,21 -6,17 -10 312 -19 824 -0,25% -0,48% 805 1 547 0,02% 0,04% 20 2 0,00% -3,90 -158 560 0,00 -1,20 -11,31 -36 356 -0,88% 2 836 0,07% 1.5.2 Känslighetsanalyser Känslighetsanalyser har endast genomförts beträffande försäljningsskattescenariot eftersom detta scenario ger de största påverkningarna. Fordonsskattescenariot påverkar inte bilparkens storlek, antalet nya bilar eller skrotningsmönstret. Fordonsskatten påverkar endast sammansättningen av nybilsförsäljningen. Sex känslighetsanalyser har genomförts och de visar på följande (notera att alla beräkningsresultatgäller tidshorisonten 20 år): En ökning av koldioxidskattenivån inverkar marginellt på resultaten. En dubblering till 1760 SEK/g ger endast en ytterligare minskning av det totala koldioxidutsläppet på storleksordningen 0,5 procent. Största orsaken till denna marginella påverkan är att även om skattehöjningen ger något mer utrymme för differentiering så reduceras eller begränsas denna effekt av kravet på att den totala skattebetalningen aldrig kan vara negativ. Den högre skattenivån på 1760 SEK/g ger oftare situationer där den totala skattebetalningen blir negativ (dvs. det skulle resultera i en subvention), men på grund av att inte ge subvention blir den faktiska differentieringsstyrningen mindre än vad den teoretiskt kan ge. En ökning av skattens värdebaserade del ger vissa ytterligare koldioxidreduktioner men bilparken minskar mer och effekten minskas något av en ökning av bilarnas genomsnittsålder. En dubblering av skattens nivåbaserade del med 20% istället för ursprungliga 10% skulle således leda till en ytterligare minskning i storleksordningen två procentandelar. Ytterligare minskningar av antalet bilar är det viktigaste bidraget till denna ytterligare koldioxidreduktion. Ökningen av skattens nivåbaserade del ger mer utrymme för den koldioxiddifferentierade skatten men effekterna har visat sig vara relativt små. Vidare ökar den högre skattenivån bilarnas genomsnittsålder vilket motarbetar koldioxidreduktionerna. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 17 Ökad grad av storleksminskning kommer att påverka sammansättningen av nybilsförsäljningen även om effekten blir begränsad. Den nuvarande tillåtna storleksminskningsnivån på 25% räcker för att fånga det mesta av koldioxidreduktionen i form av bilars genomsnittsutsläpp. I själva verket visar beräkningar att ca 80% av den potentiella minskningen kommer av denna storleksminskningsnivå. En ytterligare ökning av den tillåtna nivån skulle således ge endast begränsat ytterligare reduktioner och samtidigt ge ytterligare påverkan på bilförsäljningen. Denna effekt skulle bli mest framträdande i fordonsskattescenariot eftersom detta scenario endast anses uppnå koldioxidreduktioner genom ändringar i sammansättningen av nybilsförsäljningen. Att slå samman de två skattefunktionerna för diesel och bensin i en och samma funktion har endast en liten inverkar resultaten. Den viktigaste effekten är en ökning av andelen dieselbilar med 0,5%. Om skrotningselasticiteten är lägre än förutsett ger försäljningsskattescenariot lite mer koldioxidreduktioner. Används en elasticitet hälften så stor som den ursprungliga skulle ytterligare reduktioner i storleksordningen 0,5 procent fås. Denna reduktion är ett resultat av två effekter. För det första talar den lägre priselasticiteten för att, även om människor tenderar att behålla sina bilar en längre tid på grund av ökningen, så minskar denna effekt jämfört med ursprungsscenariot. Därför kommer genomsnittsåldern för bilarna inte att öka lika mycket som i ursprungsscenariot. Dessutom kommer den lägre skrotningsbenägenheten att förbättra försäljningen av nya bilar jämfört med ursprungsscenariot. Om bilparkens priselasticitet är lägre än förutsett ger försäljningsskatten avsevärt mindre koldioxidreduktioner och bilparken påverkas avsevärt mindre. Om man använder en priselasticitet på –0,6 istället för som ursprungligen tänkt på –0,35 kommer försäljningsskatten ge koldioxidreduktioner i storleksordningen 3% och bilparken kommer att minska med 3,7%. Storleken på elasticiteten har således en avgörande betydelse för beräkningsresultaten. En accelererad infasning av priselasticiteten skulle leda till större förändringar de första åren efter implementeringen. Priselasticitetens tidsperspektiv har viktiga konsekvenser för spridningen av effekterna över tid. Säkerheten i resultaten Sammanfattningsvis är analysresultaten relativt säkra vad gäller flertalet huvudförutsättningar. Resultaten påverkas endast i liten utsträckning av förändringar i skattenivån, koldioxidskattenivån och skrotningselasticiteten. Nivån på den generella priselasticiteten får viktiga konsekvenser för resultaten. Det sätt på vilket elasticiteten infasas påverkar storleksordningen på de kort,och medellånga effekterna i synnerhet. 18 Tätorts- och landsortsbefolkning Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 1.5.3 Samhällsekonomi och lönsamhetsanalys Analyserna visar att familjer på landsbygden tenderar att köpa dyrare bilar än familjer i tätorterna. Deras genomsnittsinkomst är lägre och följaktligen använder de en avsevärt större del av sin inkomst till bilköp. Skattescenarierna kommer därför att ha mernegativa effekter på landsortsbefolkningen än tätortsbefolkningen. Tabellen nedan illustrerar den påverkan som skattebördan kommer att ha på tätortsbefolkningen respektive på landsortsbefolkningen. Tabell 1.9 Skattebördeeffekter av försäljningsskattescenariot och fordonsskattescenariot. skattescenario Familjetyp Försäljningsskatt Fordonsskatt Genomsnittlig försäljningsskatt i procent av genomsnittsinkomst Procentmässig förändring av genomsnittlig fordonsskatt Tätort Landsbygd Tätort Landsbygd Ensamstående utan barn 10% 12% -11% -1% Ensamstående med barn 8% 9% -5% -2% Par utan barn 5% 7% -10% 3% Par med barn 4% 7% -8% 5% En annan viktig aspekt av den påverkan som landsorts- respektive tätortsbefolkningen utsätts för är den välfärdsförlust som följer av ändringar i beteende orsakad av prisändringar. Vissa människor skulle på grund av prisökningen, som ett resultat av försäljningsskatten, välja bort bilen. Andra skulle till följd av den koldioxiddifferentierade skatten välja att köpa en annan typ av bil, en bil som skiljer sig från den som de annars skulle köpt. Tabellen nedan summerar de uträknade välfärdsförluster som de olika familjetyperna och tätorts- respektive landsortsområdena skulle drabbas av. Som framgår av tabellen är välfärdsförlusten för landsortsbefolkningen större än den för tätortsbefolkningen i samtliga fall. Vidare ger försäljningsskatten mycket större välfärdsförlust än fordonsskatten. Detta följer av den stora välfärdsförlust som uppstår på grund av minskningen av antalet bilar. Tabell 1.10 Välfärdsförlust för landsort- och tätortsbefolkningen till följd av de två skattescenarierna (SEK/bil - för hela livslängden) Skattescenario Familjetyp Försäljningsskatt Tätort 1 Landsbygd Fordonsskatt Tätort Landsbygd Ensamstående utan barn -21 557 -26 447 -3 275 -4 212 Ensamstående med barn -25 986 -28 826 -4 876 -4 773 Par utan barn -24 913 -32 114 -4 037 -4 404 Par med barn -28 900 -39 159 -5 539 -5 599 1 Detta illustrerar välfärdsförlusten som följer av införandet av skatten i sig liksom av den koldioxiddifferentierade delen. Den senare representerar den dominerande delen. 19 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Biltillverkningsindus tri och anknutna industrier Båda scenarierna kommer att påverka försäljningen av svenska bilar i Sverige. Detta kommer i sin tur påverka den svenska biltillverkningsindustrin. Den totala förändringen framgår av tabellen nedan. Försäljningsskatten ger störst förändring. Detta beror framför allt på att detta scenario involverar införandet av en ny skatt som kompletterar den existerande skatten medan fordonsskatten endast är en rekonstruktion av den existerande skatten. Således leder fordonsskatten till en minskning av försäljningen av svenska bilar och en motsvarande ökning av försäljningen av andra bilar, medan registreringsskattescenariot leder till en allmän minskning av antalet sålda bilar. Tabell 1.11 Ändring av antalet svenska bilar sålda i Sverige. Antal bilar. Tidshorisont/scenario 1 år Försäljningsskattescenario -14 184 -10 409 -5 664 -10 056 -1 932 -2 133 -1 843 -2 261 Fordonsskattescenario 5 år 10 år 20 år Den minskade efterfrågan på svenska bilar på den svenska marknaden leder således till en minskning av aktivitetsnivån i den svenska biltillverkningen. Denna effekt är översatt till en sysselsättningseffekt. Man har beräknat att den resulterande sysselsättningsförlusten i biltillverkningsindustrin kommer att motsvara de nivåer som framgår av tabellen nedan. Dessa bygger på ett antal antaganden som gjorts för att kunna uppskatta sysselsättning per bil och för att justera för produktionsförhållandena mellan Nederländerna och Sverige. Förutom effekten på biltillverkningsindustrin kan andra närbesläktade industrier komma att påverkas. Bland de viktigaste närbesläktade industrierna finner man bilförsäljning, bilreparation och underhåll och bensinstationer. Sysselsättningseffekten inom dessa industrier har därför också inkluderats i beräkningarna. Denna beräkning är baserad på ett antal enkla antaganden beträffande sysselsättning per bil per år och baseras på den totala bilparksstorleken som framgår av tabell 1.7 och 1.8. Tabell 1.12 Sysselsättningsförlust som följd av nedgång i försäljningen av nya svenska bilar. Antalet anställda (inklusive ägare/egenföretagare) Tidshorisont/scenario 1 år Försäljningsskattescenario, därav -1 348 -3 367 -3 514 -5 447 -971 -1 633 -733 -1 608 -377 -1 734 -2 781 -3 839 -1 -2 -2 -4 Fordonsskattescenario, därav -183 -203 -175 -214 - biltillverkning -183 -203 -175 -214 0 0 0 0 >-1 >-1 >-1 >-1 - biltillverkning - bilförsäljning och reparation 5 år 10 år 20 år Effekt i relation till bas, % - bilförsäljning och reparation Effekt i relation till bas, % 20 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Tabellen illustrerar att införandet av en försäljningsskatt som består av en koldioxiddifferentierad del ger störst effekter på sysselsättningen. Detta beror framför allt på minskning av bilförsäljningen och antalet bilar i bilparken. Kostnadsnyttoanalys En kostnadsnyttoanalys har gjorts för att ge en indikation på storleksordningen av effekterna på samhället av de två skattescenarierna. Resultaten av dessa beräkningar visas nedan. Beräkningar som dessa är alltid föremål för en rad antaganden och osäkerheter och analysen som gjorts i denna studie är ganska elementär. Även om det bör påpekas att resultaten av de två scenarierna inte är helt jämförbara eftersom försäljningsskatten innebär införandet av en ny skatt, bidrar resultaten dock till att underbygga de tidigare komparativa slutsatserna. Analysen nedan antar att på sikt kommer en stigande del av dem som blir arbetslösa att finna arbete någon annanstans och att den ekonomiska aktiviteten kommer att kompensera förlusten. Analysen använder sig av ett enhetspris på koldioxidreduktioner på 1,50 SEK. Kostnaden av färre och mindre bilar har räknats som en välfärdsförlust och effekter på energiförbrukning är inkluderat i detta. Effekterna på utsläppen från andra föroreningar bestäms delvis av valet av fordonstyp och delvis av hur fort de nya Euronormerna kan infasas. En minskning av skrotning leder således till en fördröjd effekt av nya normer jämfört med vad som annars skulle vara fallet. Tabell 1.13 Resultat av kostnadsnyttoanalys av de två skattescenarierna (miljon SEK) Kostnads- och vinstfaktorer Försäljningsskattes cenario Fordonsskattescenario År År 1 20 1 20 Kostnader - mindre bilar - färre bilar 57 113 75 1 152 39 0 46 0 - sysselsättning - vinst Totala kostnader Vinst 25 43 239 9 26 1 262 3 4 45 0 1 47 80 -57 46 3 73 -166 -166 768 54 472 31 1 326 64 26 -1,096 20 0 0 0 20 -26 -26 301 -3 0 0 298 251 115 1,767 - koldioxidreduktioner - andra föroreningar - olyckor - buller Total vinst Total nettovinst/år - diskonterad Total ackumulerad nettovinst Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 2 Executive summary 2.1 Main conclusions 21 This study has investigated the implications of using vehicle taxes as a means to reduce CO2 emissions from passenger cars. Model based calculations constitute the core output of the study. Calculations have been done to analyse the CO2 reducing potentials and other effects in two tax scenarios. The first scenario assumes that a CO2 dependent tax on car purchase (a registration tax) is introduced, and the latter analyses the effects from a CO2 dependent tax on car ownership (circulation tax). Prior to summarising the main conclusions to be drawn, it is however necessary to outline the basic principles that governs the design of the assumed new taxes. The reference level The determination of a reference level is a key feature of the approach taken in designing both of the above CO2 taxes. Basically, the reference level determines, for each car size, the level of CO2 emissions that will result in a tax of zero (0). The CO2 emissions are measured in grams of CO2 per kilometre. If the CO2 emissions of a particular car size exceed the reference level, this will result in a tax increase, and if the CO2 emissions are less than the reference level, this will result in a tax reduction. Thus, the tax base is: the CO2 emissions for the car size in question minus the CO2 emissions reference level, and the price is a fixed amount of SEK per gram. The resulting tax may be positive, negative or zero depending on the difference between the emissions level of the car in question and the relevant reference level. The circulation tax is composed of the new, hypothetical CO2 tax and the existing circulation tax, and the system is designed so that the total circulation tax invoked on a particular car can never be negative. The new, hypothetical, registration tax is assumed to be composed of a value based part and the CO2 tax, and similarly to the circulation tax, the total registration tax is not allowed to be negative. In other words, the systems do not allow for subsidisation. Setting the reference level A purely CO2 dependent tax would lead to substantial downsizing4 as it would render the smaller and most CO2 effective cars much cheaper than the larger cars compared to the situation of today. While such downsizing can be an ef4 To illustrate the term downsizing one can say that is used to describe the case where car purchasers would tend to buy smaller cars than today thereby resulting in a smaller average size of the cars in the car fleet. 22 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars fective means to reduce CO2 emissions it can be in the conflict with other objectives of the society. For the present model calculations, a reference level has been defined which allows for some downsizing, but only in the order of 25%. In practical terms, this means that the reference level is defined mainly as a function of the size of the car, where the size is measured as the area of the car. The choice of downsizing in the present study does not exclude other levels of downsizing to be considered as well. The registration tax is assumed to include both a value dependent part (10% of the value of the car) and a CO2 dependent part (880 SEK per gram of CO2 above the reference level, and calculated as a negative value in case the emissions level is below the reference level). The total registration tax to be paid by a car purchaser is not allowed to be negative, but the CO2 part of it may become negative. The circulation tax is defined as the existing circulation tax plus a CO2 dependent element. The latter is set at 44 SEK per gram of CO2 above the reference level, and as a negative value when the emissions level is below the reference level. The assumed change of the circulation tax is only assumed to apply for new cars, and the total circulation tax is not allowed to be negative. Results from the scenario calculations The investigations of these two tax scenarios point to the following conclusions: • The assumed registration tax leads to an annual reduction in total emissions of CO2 from the car fleet in the order of 5% n 20 years. Taking a shorter time horizon, the annual reductions are smaller, e.g. just above 1% in 5 years. The potential reductions are larger than those that would result from the assumed changes of the circulation tax, but they are mainly achieved through a decline in the number of cars in the fleet. The average emission level per car (gram of CO2/km/car) is almost constant in this scenario, and the average age of cars increases by about half a year as a result of reduced turnover rates. • The circulation tax would lead to an annual reduction in the order of 2% in 20 years, and of about 0.5% in 5 years time. While this reduction is smaller than for the registration tax, it nevertheless implies a more significant reduction in the average emission level per car. In 20 years the average emissions would have been reduced by 2.5% in this case compared to virtually no reduction for the registration tax. Furthermore, the redesign of the circulation tax does not lead to a change in the size of the car fleet. • The sales of Swedish car makes (i.e. Volvo and SAAB) on the Swedish market will in both scenarios be more affected than the sales of cars as a whole. This reflects the fact that Swedish car makes are typically larger, less CO2 efficient cars than the average car. This effect will be most outspoken for the registration tax, mainly as a result of the decline in the car fleet (a drop of about 4% in the 20 years perspective),. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 23 • Neither of the assumed tax systems leads to significant changes in the proportion of diesel cars in the car fleet. • Company cars are less sensitive to price changes than private cars. On the other hand, company cars are more sensitive to changes in the annual costs, i.e. fuel and circulation tax. Consequently, private cars stands for 70% of the CO2 reductions in the registration tax scenario while company cars stands for 60% of the CO2 reductions in the circulation tax scenario. In regard to the latter, it is important to note that the annual costs consist of both circulation tax and fuel costs. It is assumed that an increase in the circulation tax of e.g. 1000 SEK would have the same effect as an increase in the annual fuel costs of the same amount. Furthermore, it is assumed that changes to the circulation tax are fully reflected in the car purchase decision of company cars. Presently, the annual circulation tax paid by the company for a company car is not reflected in the employee’s personal taxation. If this policy is maintained then the model calculations referred to in this report overestimate the effect from the circulation tax scenarios. Assuming that 33% of the company cars fall into the group where the employee himself decide exactly which car to have this overestimation amounts to 20% of the total CO2 reductions. However, imposing a CO2 element to the circulation tax would make some of the most ineffective cars very expensive in terms of annual taxation for the company. Therefore, some companies would decide no longer to offer these cars to their employee. Therefore, the real overestimation would probably be substantially lower than 20%. In the light of the higher sensitivity of company cars to the annual costs, and because company cars constitute about 50% of the sales of new cars, these assumptions are important. Conclusions While care should be taken to compare the performance of the two systems, as their impact on the tax burden of car purchasers and car owners differ, there are nevertheless important generic conclusions to be made. Thus, the analyses clearly point to the importance of distinguishing between tax levels and tax differentiation. Roughly speaking, the level of the tax affects the overall price structure per se, i.e. it makes cars relatively more expensive than other consumer goods. Hence, it has an impact on whether people decide to have a car. By comparison, differentiation affects the choice that people make, but has less impact on the overall price structure (when comparing cars to other consumer goods). The registration tax clearly imposes new financial burdens on car purchasers as it involves the introduction of a new tax. This increase in the level of the tax has a negative impact on the size of the car fleet because it leads to overall price increases of cars relative to other consumer goods. It thus leads to a reduction of scrapping and a reduced sale of new cars. In addition to introducing an in- 24 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars crease in the tax level, the assumed design of the new registration tax however also includes a tax differentiation element in the form of the CO2 dependent part of the tax. This part imposes a unit price (tax) on each g of CO2 that is emitted per kilometre5. This part of the tax does not have an influence on the number of cars, but solely on the choice of car that people make. The tax scenario for the circulation tax only assumes design changes in the existing tax thereby not influencing the average tax burden. Consequently, it has no impact on the number of cars, but solely an influence on the type of car that people decide to buy. If the average level of the circulation tax was increased, this would have implications for the number of cars, and thereby a further CO2 reducing impact. An increase in the average level would also allow for a more distinct differentiation and hence, a stronger impact on the choice of car as well. In conclusion, the study has shown that: • The assumed circulation tax can provide CO2 reductions in the order of 2% in 20 years time without significant distorting effects on the structure of the car fleet. While there is a shift from Swedish car makes to other brands, it is relatively small and in reality the Swedish car manufacturing industry will presumably adapt to this change among other things by means of developing and producing car types that are more adequate for the new conditions. • The registration tax will provide more CO2 reductions, but it will also involve more distinct adverse effects in the form of reduced car fleet, reduced sales of Swedish brands and increased average age of the cars. The latter may be counterbalanced by means of the simultaneous introduction of a scrapping premium scheme, which should nevertheless be designed carefully to ensure its efficiency in relation to the CO2 objective. The robustness of the results of the study as regards the registration tax has been investigated through the conduct of sensitivity analyses. These analyses show that the above conclusions are very robust in regard to a variety of assumptions including the level of the CO2 tax, the overall tax level, and the elasticity of scrapping. However, the results prove to be very sensitive to the size of the overall price elasticity of the car fleet. The applied elasticity is -0.6 and has been established based on consultations of relevant other studies. This elasticity implies that an increase in prices of cars of 10% would lead to a 6% decline in the car fleet. Changing the level of this elasticity down to -0.35 would imply that the registration tax would lead to a 3% reduction compared to 5% when the elasticity is –0.6. The reason for the lower effect is that a lower elasticity means that the effect on the car fleet is substantially reduced. Cost benefit analysis The implications to society of the registration tax scenario and the circulation tax scenario have been analysed by means of a cost benefit analysis. The analysis shows that the costs associated with the introduction of the registration tax 5 above the defined reference level Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 25 will outweigh the benefits. This is mainly a result of the decline in the size of the car fleet. While the reduced car fleet provides CO2 reductions, it also involves a substantial welfare loss due to this decline. The value of this welfare loss actually constitutes the dominating share of the costs of the registration tax. The size of the negative net benefit depends however on the price used to value the achieved CO2 reductions. By comparison, the cost benefit analysis indicates that the circulation tax scenario will provide a net benefit to society. It will have less distortion effects while at the same time providing for significant CO2 reductions. But, the reductions are lower than those achieved in the registration tax scenario. The net benefits provided in this scenario will increase over time as the new, and more fuel efficient, cars come to penetrate the car fleet more over time. The cost benefit analyses have also looked at the effects from the CO2 differentiated part of the two taxes alone, e.g. disregarding the effect on prices that will result from the introduction of the registration tax per se. This analysis shows that the CO2 differentiation would provide a net benefit to society in both cases, but the effect is still largest for the circulation tax. The rural population tend to buy larger, and less fuel efficient, cars than the urban population. Consequently, the rural population will be affected more than the urban population by the tax scenarios. Equally, the calculations have also shown that families with children tend to be affected more, in a negative sense, that those without children. While these effects are not part of the cost benefit analysis, as they relate mainly to concerns over distribution, they nevertheless show that there will be distribution effects from the tax scenarios. Calculations also indicate that the level of activity in the Swedish car manufacturing industry will be affected negatively. This effect will be most pre-dominant for the registration tax, and further enhanced if account is taken of the effect on related industries as well. This effect is only of concern to the cost-benefit to the extent that these activities are not replaced by activities elsewhere in the economy, but again, they do illustrate that there will also be distorting effects on industry. Policy implications This study has shown that CO2 emissions can be affected by means of vehicle taxation, and that the circulation tax would most likely be the most appropriate instrument to use in that case. However, there are other options for affecting the CO2 emissions from the car fleet. These options include other measures to affect the cost of driving, such as fuel costs or road pricing. These options have however not been analysed here. The calculations done in this study has been subject to the requirement that the taxes should not result in more than 25% of downsizing. This limit does put a limit to the amount of CO2 reductions that can be achieved. Thus, if significant further reductions are desired more downsizing may be needed. If the target consists of significantly higher reductions in the average CO2 emissions from new cars this would call for more differentiation of the CO2 tax. More differentiation would require either a higher tax level or allow the tax to be become a net subsidy for the most CO2 effective cars. 26 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Policies which aim to encourage the purchase and use of more energy efficient vehicles will tend to impose relatively more costs onto the rural population compared to the urban population, and to be more in favour of families without children compared to families with children. A consequence of these consumer groups preference for bigger and more fuel consuming cars. Consequently, there may be a wish for compensating measures to correct for this impact on welfare distribution. Outline of the executive summary The rest of the executive summary outlines the purpose of the study and presents the applied methodology. Furthermore, all main results are presented – both with respect to emissions and with respect to socio-economic consequences following the introduction of either a registration tax or a circulation tax. 2.2 Purpose, scope of study and study organisation Organisation This report presents the results of a study undertaken for the Swedish Environmental Protection Agency (Naturvårdsverket) and the Swedish Energy Agency (Energimyndigheten). COWI A/S carried out the study in association with the Swedish National Road and Transport Research Institute (VTI) in the period September to December 2001. The features of the assumed changes to the tax system - the scenarios - were established in close collaboration with the Swedish EPA and Energy Agency. Similarly, the definition of the types of results to be produced and alternative calculations to be carried out was also established in close collaboration with the client. In this regard, the three project meetings that have been held with the clients and other governmental stakeholders at the premises of the EPA have been of utmost value6. The meetings served to ensure a common basis of understanding of the project scope and project progress. Purpose The purpose of the study is to analyse the demand effects with regard to household's purchase of new cars as a result of the introduction of CO2 differentiated registration taxes. During the conduct of the study, it was decided to also include an analysis of the implications of redesigning the circulation tax to include a CO2 dependent element. The analysis should look both at the resulting CO2 emissions and at the implications for the sales of cars and the car fleet. The study was to clearly distinguish between diesel cars and petrol cars on the one hand, and between private cars and company cars on the other hand. The study should not only look at the implications for the sales of new cars, but also take into consideration impacts on scrapping behaviour and on the car fleet. Socio-economic implications should be considered and the calculations should apply both a short-termed, medium-termed and a long-termed view. 2.3 Reference level Definition of the tax scenarios In defining the tax scenarios, an essential feature is the definition of a reference emissions level. CO2 taxes are only imposed for emissions that are in excess of 6 4 October, 15 November and 18 December Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 27 this reference level, and the tax becomes negative when the emissions are less than the reference level. The CO2 tax base is always the amount (gram) of CO2 per kilometre minus the reference level. The reference level is the amount (gram) of CO2 per kilometre that will result in a tax of zero. Equally important, the systems are defined so that the total registration tax (or circulation tax) can never be negative, i.e. net subsidy is not allowed for. Furthermore, the circulation tax has been defined to be budget neutral in the sense that it should provide a revenue similar to the revenue that would be generated under the current tax regime - not more and not less. In defining the reference level, care has been taken to define it so that it on the one hand acknowledges the importance of allowing for downsizing while at the same time not leading to high levels of downsizing. To illustrate this further, one may envisage two extremes. At the one extreme, one could imagine a system where the tax to be paid was solely determined by the amount (gram) of CO2 per kilometre emitted by a car. Such a system would always reward the smaller cars at the expense of the larger ones, and hence it would inevitably lead to substantial downsizing. At the other extreme, one could imagine a system that solely aimed to affect people's choice of cars within given size categories. Such a system would not lead to any downsizing, but would actually have the unintended effect of upsizing as the result of the smallest and most fuelefficient cars being imposed with a high tax in relation to car price compared to bigger cars. Consequently, it would render the larger, more fuel effective cars relatively less expensive than they were before. In conclusion, a reference level has been defined, which allows for 25% downsizing. This is illustrated in the figure below where gram of CO2 per kilometre is pictured against the area of the car. 28 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 2.1 CO2 reference level , Petrol Cars 600 CO2 per km 500 400 300 200 100 0 0 2 4 6 8 10 12 Area (m2) The curve shown in the figure is the result of a calculation, which attaches a weight of 25% to the downsizing and a weight of 75% to a tax curve that would not allow for downsizing at all. This is the applied reference level for petrol cars. The figure contains indications for each car type on the Swedish market. Cars above the line pay a CO2 tax and cars below get a CO2 rebate. If one allowed for 100% downsizing, the reference level would merely be a horizontal line in this figure whereas no downsizing would imply a much steeper curve than the one that is shown. A similar reference level has been established for diesel cars. These reference levels are used both in the scenario for circulation taxes and the one for registration taxes. The applied tax scenarios In designing the scenario for the registration tax, care has been taken to set a level which does not result in unrealistically high taxes, but which is sufficiently high to provide for sufficient scope for differentiation and thereby for significant behavioural changes resulting in reduced CO2 emissions. The resulting tax scenarios are summarised in the below table. It should be noted that the resulting scenarios shown in the table have been established on the basis of a number of preliminary calculations. Table 2.1 Overview of the tax scenarios1). Tax elements and key features Not CO2 dependent element Circulation tax Existing tax: tax is differentiated according to the kerb weight of the car using a stepwise linear relation. The tax increases by approximately 150 SEK for each 100 kilo for petrol cars and by 575 SEK for diesel cars Registration tax Value based (new tax): 10% of the value of the car 29 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars CO2 dependent element 44 SEK per gram above the reference level, and a similar negative value if the emissions are below the reference level 880 SEK per gram above the reference level, and a similar negative value if the emissions are below the reference level. Other features Total revenue from the circulation tax should be unaffected The total registration tax cannot be negative 1) Separate functions are calculated for diesel cars and for petrol cars, and the reference levels are illustrated in Figur 1.1 2.4 Study approach 2.4.1 Model based analyses of the tax scenarios An important basis for the study is the car choice model, which COWI has previously developed and used on several other occasions. The model contains a detailed modelling of car purchasers' demand patterns. It allows for quite detailed analyses of how households' will react to possible changes in tax levels and tax structures. This model constituted the basis for the construction of the car demand module, which was applied in this study. This module was combined with a car fleet module and a scrapping module as illustrated below. Figure 2.2 The model based approach Illustration of Entry and Exit A model-based approach constitutes the core of this study. The figure above illustrates the key aspects of this approach. Basically, the calculations make use of three separate, but interrelated, modules: • • • A module for new car registrations A module for the car fleet A module for scrapping New cars The module for new car registrations assesses how households' choice of a new car will be affected by changes to the price structure invoked by changed or new taxes. This module builds on a detailed mapping of the sales of new cars in Sweden in 1999/2000, which includes a variety of physical features, energy efficiency ratios, and the price of each car type. Further, the module contains a detailed mapping of the relevant socio-economic features of car buyers, and a detailed mapping of the preferences and price elasticities of each group. The car fleet The module for the car fleet updates the car fleet each year with the results of the two other modules, and it contains, as a starting point, a detailed mapping of 30 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars the Swedish car fleet as of September 2001. The study applies a short- medium and long-termed (20 years) time horizon, and this module is consequently used to keep track of the changes from one year to the other. The scrapping of cars The scrapping module is used to keep track of the scrapping of cars. Scrapping is determined by two factors: the survival curve of the cars and the price elasticity of scrapping. Overall price elasticity If the overall tax burden that is invoked on car purchasers or car owners increase, this will most likely lead to a decline in the number of cars in the fleet. The price elasticity of the scrapping module captures the tendency of people keeping their cars for a longer time thus extending the average age of the cars. An overall price elasticity is however necessary to capture the effect from the resulting price increase on the car fleet thereby also taking account of the effect on the sales of new cars. The study applies a long termed price elasticity for car ownership of 0.6, i.e. a 10% increase in the price will result in a decline of the car fleet of 6% in the longer term. This elasticity has been obtained in several Swedish studies. The calculations assume that half of this effect is phased in after 7 years and that the total effect is phased in after 30 years. Energy efficiency development Experience shows that over the last decade, fuel efficiency of cars have improved by about 2% per year and the calculations assume that this trend will continue. A what-if approach The model based approach outlined above is of a what-if nature. No attempts are made to take account of macro-economic developments and other similar dynamics. The strength of this approach is that it allows for a distinct analysis of the CO2 performance and related side effects of the analysed tax instruments, and that it reduces the amount of assumptions to be made and complexities to be included. On the other hand though it implies that the model should not be used for projections and forecasts. 2.4.2 Types of results The model-based calculations provide results that illustrate the short-, medium and long termed effects from the tax scenarios. Results are calculated for the first year of the assumed change, and after five, ten and twenty years of implementation. For each year, a base situation is calculated. This base calculates the results under the existing tax regime. Changes are thus calculated as the difference between the base situation for the year in question and the scenario results for the year in question. Baseline The establishment of a base situation is critical to carrying out the requested analyses. The base situation is one, which assumes that the existing taxes remain in place, and none of the assumed changes are implemented. The below table summarises the base values for the first year regarding sales of new cars. In assessing the results from the calculations of the implications of the assumed tax scenarios, the results from the calculations are compared to the base values. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 31 The table provides a brief description of the types of results that such comparisons will provide. Table 2.2 Types of results and base values: sales of new cars. Type of result Base values Motivation Average CO2 emissions (g per km) 198.1 Illustrates the improvement in CO2 performance on average (CO2 emission per kilometre per car) Average lifetime tax revenue (SEK/car) 96,237 Illustrates the average effect on each car owner's tax payments over the lifetime of the vehicle (from purchase to scrap). The tax is calculated as registration tax plus lifetime value of circulation tax and fuel tax Average size 3.70 Changes in this indicator illustrate the order-ofmagnitude of up- or downsizing Average Diesel share 6.5% This result shows whether the share of diesel cars in the sales of new cars change as a result of the assumed tax change Average Swedish car makes 30.5% This result shows the share of Swedish car makes in the total car sales. Average registration tax (SEK/car) - Average Circulation tax (SEK/car/year) 1,566 Illustrates what happens to the average circulation tax as a result of the assumed change. The introduction of a registration tax may lead to changes in the registration tax as a result of a different mix of the car sales. Average dealers price, SEK excl. VAT 145,825 Serves to provide an indicator of what happens to the turnover of the car manufacturing industry. Needs to be combined with the resulting amount of cars sold to provide an accurate picture Average (lifetime) tax consists of Registration tax - Shows the composition of the average life time tax to be paid. - Circulation tax (whole life time) 24,780 Fuel tax (whole life time) 71,456 Note: Base values are based in model calculation of the base scenario Similarly, Table 7.2 illustrates the base values regarding the car fleet. These values cover the short, medium and long termed perspective. The base values illustrate what would happen in the case where no changes were made to the existing taxes. As can be seen the average CO2 emissions (g/km/car) and the total CO2 emissions decline over the period. This is a result largely of the observed technological progress that has taken place during the last decade, and the assumption that this trend will continue. Past experience shows that technological improvements have provided an annual improvement in fuel efficiency of about 2%. The gradual replacement of old cars with new 32 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars and more fuel efficient cars lead to average reductions in the average CO2 emissions. The table also shows that over time, the average length of the vehicles will tend to decline a little whereas the cars will be slightly heavier on average. It is noteworthy also that the share of diesel cars will remain at a low level, but increase from almost 5% up to 6.5%. Table 2.3 Base scenario, total car fleet (1, 5, 10 and 20 years) BASE Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,153,155 205,154 235 11,994,305 8.9 451 1,340 1,365,726 32.9% 198,069 4.8% Time horizon (in years) 5 10 4,153,155 4,153,156 226,536 195,659 221 203 11,364,228 10,458,465 9.0 9.4 449 447 1,375 1,402 1,289,483 1,248,816 31.0% 30.1% 236,867 276,989 5.7% 6.7% 20 4,153,152 240,064 162 8,325,037 9.2 442 1,408 1,255,375 30.2% 270,625 6.5% As mentioned, the model is neither a projection nor a forecast model. Consequently, demand for cars is assumed to remain stable unless changes are invoked to the system in the form of for example price changes caused by changes to the tax structure and/or the tax level. Therefore, the number of cars in the fleet remains unaltered throughout the observed period in the base scenario. The base scenario is exactly a scenario, which assumes that there are no changes to the current conditions. The sole exception from this relates to energy efficiency, which is assumed to increase by 2% annually as described above. Another factor, which has implications for the results of the base scenario, is the pattern of scrapping. This pattern is determined by the age distribution of the current fleet. The current fleet is distributed so that a large number of cars are between 1 and 3 years old and between 12 and 14 years old respectively. This distribution has implications for each year's turnover rates, and consequently it affects the resulting average age of vehicles, which is seen to fluctuate. 2.4.3 Sensitivity analyses The calculations are subject to a number of critical assumptions. Therefore, sensitivity analyses have constituted an important part of the study scope in order to illustrate the robustness of results. Overview of sensitivity analyses Sensitivity analyses have been conducted to assess the extent to which the results obtained are sensitive to these assumptions. The following sensitivity analyses have been conducted. The sensitivity analyses have only been carried out for the registration tax scenario, as this is the scenario that carries the largest effects. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 33 Table 2.4 Overview of sensitivity analyses Sensitivity analyses Motivation and description Degree of CO2 differentiation The original level of differentiation is 880 SEK per gram of CO2 above the reference level. The alternative calculation assesses the extent to which results are further accentuated when doubling this, i.e. applying a rate of 1760 SEK per gram. Level of registration tax The level was originally set at 10% of the value. Given that the total tax is not allowed to be negative, the alternative analysis aims to investigate whether an increase in this level (up to 20%) can provide further reductions and what this would imply of additional side effects. The constraint that it can never be negative implies namely that the level of the value-based tax to some extent defines the scope for the level of differentiation that can be applied, but the level on the other hand also has an impact on the average age of cars, and this effect works in the opposite direction. Levels of downsizing In the original set-up, allowance is made for 25% downsizing. Increasing this amount has not implications for the size of the car fleet (determined by the overall price elasticity), but it affects car purchaser's choice of new cars. The more downsizing is allowed for, the more will people tend to choose the smaller and more fuel effective cars. Separate tax functions The scenarios apply different tax functions (with similar forms) for diesel and petrol cars. These functions may have implications for the results. To assess this, the two functions are merged into one. Price elasticity of scrapping The price elasticity of scrapping partly determines the speed by which the car fleet is renewed and the speed by which it adapts to the new conditions. Therefore, the implications of halving the original price elasticity are investigated. Size of the overall price elasticity The size of the overall price elasticity determines the number of cars that will constitute the car fleet. Originally, the elasticity was set at -0.6. Reducing it would imply that the impact on the car fleet would be reduced in terms of number of cars and average age. Consequently, however, one would also expect to see a lower reduction in total CO2 emissions. The sensitivity analyses investigates the changes that would occur if an elasticity of -0.35 is applied instead of -0.6. Phase-in of the overall price elasticity The price elasticity of -0.6 is long-termed in the sense that it takes many years until the full effect is realised. Originally, the calculations assume that half of the elasticity is phased in over a period of 7 years, and the rest in the remainder of the total 30 year period. The sensitivity analysis assumes that half of the elasticity is phased in over only 4 years instead and maintains the whole period as 30 years. 2.4.4 Socio-economics and cost benefit In assessing the socio-economic implications, emphasis was put on two aspects, viz.: the possible impact on the welfare of the rural population and the possible impact on car manufacturing and the derived effect on for example supply industries. Socio-economics To illustrate the socio-economic implications of the tax scenarios, the following analyses have been conducted: 34 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars ♦ An analysis of the extent to which the rural population is affected more or less relative to the urban population. This is assessed by means of providing quantitative indicators of the relative tax burden imposed on rural and urban population and on the welfare loss that each group will suffer as a result of the tendency of choosing smaller cars than in the base case. In this analysis, a distinction is made between singles and couples and between those with children and those without children. ♦ An analysis of the extent to which employment will be affected. This is assessed by means of providing quantitative indicators of the employment impact on car manufacturing and on vehicle sales, repair and maintenance plus service stations. Cost benefit analysis Additionally, the study provides a comprehensive assessment of the costs and benefits that society will incur as a result of the tax scenarios. In this, the study takes into account such factors as saved CO2 emissions, welfare loss due to change in car size, loss of employment, loss of profits, effects on other emissions (NOx, particulates, SO2, HC), noise and accidents. 2.5 Sales of new cars Results 2.5.1 CO2 reductions, sales of new cars, scrapping and the car fleet Tabell 1.5 and Tabell 1.6 show the results on new car sales from a registration tax and a circulation tax respectively. As can be seen, the circulation tax scenario has a more significant impact on the composition of the sales of new cars than the registration tax has. It leads to a decline down to 192.8 g/km/car in the emissions level compared to a resulting average of 195.3 for the registration tax. This is concerned only however with the sales of new cars. It is interesting to note that both tax scenarios result in almost no changes to the share made up by diesel cars. A major reason for this observation is that different reference levels have been applied for diesel and petrol cars to ensure that there would not be a radical shift in the distribution between the two. The implications of merging the two into one tax function, i.e. applying the same reference level for both fuel types, however shows that this would only lead to marginal changes also. The share of diesel cars would in this case only increase by 0.5%. 35 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 2.5 Results registration tax scenario Base Average CO2 emissions (g per km) Average lifetime tax revenue (SEK per car) Average size Average Diesel share Average Swedish car makes Average registration tax (SEK per car) Average Circulation tax (SEK per car per year) Average dealers price, SEK excl. VAT Average (lifetime) tax consists of (SEK) Registration tax Circulation tax (whole life time) Fuel tax (whole life time) New registration tax 198.1 195.3 96,237 3.70 6.5% 30.5% 115,227 3.66 6.5% 29.2% 1,566 20,161 1,558 145,825 143,854 24,780 71,456 20,161 24,543 70,523 Table 2.6 Results circulation tax scenario Base Average CO2 emissions (g per km) Average lifetime tax revenue (SEK per car) Average size Average Diesel share Average Swedish car makes Average registration tax (SEK per car) Average Circulation tax (SEK per car per year) Average dealers price, SEK excl. VAT Average (lifetime) tax consists of (SEK) Registration tax Circulation tax (whole life time) Fuel tax (whole life time) The car fleet 198.1 New circulation tax 192.8 96,237 3.70 6.5% 30.5% 92,497 3.65 6.6% 29.5% 1,566 1,443 145,825 142,120 24,780 71,456 22,766 69,731 When considering the whole car fleet, the registration tax still brings about the largest total decline in CO2 emissions. This is largely because the circulation tax does not at all affect the size of the car fleet. The reason for this being that the circulation tax is merely a reconstruction of the existing tax, whereas the registration tax involves the establishment of a new tax that is additional to the existing circulation tax. Consequently, the registration tax scenario affects the costs of cars relative to other consumer goods, while the circulation tax does not. 36 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The analyses show that a registration tax that consists of a CO2 tax of 880 SEK per gram of CO2 per car plus a value based element of 10% will provide CO2 reductions in the order of 5% in the longer term (20 years). In the medium term, the reductions will be in the order of 3% and 2% in 10 and 5 years time respectively. However, the corresponding change in the average emissions per car is negligible, so the bulk of the reduction is provided through the reduction in the car fleet. This scenario also leads to an increase in the average age, as a result of reduced scrapping. In twenty years, the model-based calculations show that the average age will have increased by half a year. The results from the two tax scenarios for the car fleet are shown below. The sales of Swedish cars are affected in both scenarios. This reflects the fact that Swedish car makes tend to be larger than the average car make, and consequently they will be affected more than the average car by the assumed scenarios. Table 2.7 Registration tax effects, total car fleet (1, 5, 10 and 20 years) Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton/year) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 -24,846 -20.47% 0.37 -43,669 0.15 0.01 -1.82 -7,920 0.01% -2,225 -0.03% Time horizon (in years) 5 10 -114,289 -183,303 -12.59% -6.81% 1.27 1.07 -203,436 -329,885 0.53 0.60 -0.07 -0.28 -6.80 -8.73 -41,043 -71,310 -0.14% -0.41% -10,372 -15,064 -0.10% -0.07% 20 -253,094 -11.21% -0.10 -410,717 0.53 -0.92 -10.79 -113,248 -0.94% -17,940 -0.04% Table 2.8 Circulation tax effects, total car fleet (1, 5, 10 and 20 years) Change 1 Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton/year) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 0.00% -0.26 -10,322 0.00 -0.06 -0.60 -1,932 -0.05% 151 0.00% Time horizon (in years) 5 10 0 0 0.00% 0.00% -1.31 -2.41 -53,146 -97,978 0.00 0.00 -0.34 -0.65 -3.21 -6.17 -10,312 -19,824 -0.25% -0.48% 805 1,547 0.02% 0.04% 20 2 0.00% -3.90 -158,560 0.00 -1.20 -11.31 -36,356 -0.88% 2,836 0.07% Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Sensitivity of results from registration tax scenario Results 37 2.5.2 Sensitivity analyses The sensitivity analyses were conducted only for the registration tax scenario, as this scenario has the most significant impacts. The circulation tax scenario does not affect the size of the car fleet, the number of new cars or scrapping patterns. The circulation tax only has an impact on the composition of the sales of new cars. Six sensitivity analyses were conducted, and they showed the following (note that all references to results from calculations refer to the 20 years time horizon): Merely increasing the level of the CO2 tax only has a small effect on the results. A doubling to a level of 1760 SEK/g only provides an additional reduction in total CO2 emissions in the order of 0.5 percentage points. The main reason for this is that while this allows a little more room for differentiation, this effect is reduced or limited by the condition that the total tax payment can never be negative. The higher tax level of 1760 SEK/g would more often result in cases where the total tax payment should be negative (i.e. a subsidy would result) according to the formula, but due to the constraint it is set at 0 instead. Thus, the realised differentiation power is smaller than it could potentially have been had this constraint not been applied. Increasing the level of the value based part of the tax provides some additional CO2 reductions, but the car fleet declines more and the effect is to some extent crowded out by an increase in the average age of cars. A doubling of the value based part of the tax up to 20% instead of the original 10% would thus lead to an additional reduction in the order of 2 percentage points. Further reductions in the number of cars are the most important contributor to this further CO2 reduction. The increase in the value-based part of the tax provides more room for the differentiated CO2 tax, but the resulting effect from this proves to be quite limited. Furthermore, the higher tax level results in a further increase in the average age of the cars, which counteracts the CO2 reductions. Increasing the level of downsizing further will affect the composition of the sales of new cars, although the impact will be limited. The current level of allowed downsizing of 25% is sufficient to capture most of the potential CO2 reduction in terms of average emissions from new cars. Actually, calculations show that about 80% of the potential reduction are harvested through this level of downsizing. Increasing the allowed level further would thus provide only limited additional reductions, and at the same time it would have further negative repercussions on the car sales. This effect would be most outspoken in the case of the circulation tax scenario, as this scenario is considered solely achieving CO2 reductions through changes to the composition of the sales of new cars. Merging the two tax functions for diesel and for petrol into one function only has a very little impact on the results. The most significant effect is an increase in the share of diesel cars by 0.5%. 38 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars If the real outcome of scrapping is lower than originally anticipated, the registration tax scenario would provide a little more CO2 reductions. Applying an elasticity, which is half of what it was originally would result in additional reductions in the order of only 0.5 percentage points. This reduction is the result of two effects. First, the smaller elasticity would imply that while people would still tend to keep their cars longer as a result of the increase, this effect would be reduced compared to the original scenario. Consequently, the average age would not increase as much as in the original scenario. Second, the lower scrapping would also result in more sales of new cars than in the original scenario. If the price elasticity of the car fleet is lower than originally anticipated, the registration tax would bring significantly less CO2 reductions and the car fleet would significantly less affected. If a price elasticity of -0.35 is applied instead of the original -0.6, the registration tax would provide CO2 reductions in the order of 3%, and the car fleet would be reduced by 3.7% as well. Thus, the size of the elasticity has a determining impact on the results of the calculations. Accelerated phase in of the price elasticity would lead to more significant changes in the first years after implementation. The time perspective of the price elasticity thus has important implications for the distribution of the effects over time. Robustness of results Urban and rural population Concludingly, the results of the analyses are quite robust as regards a number of the key assumptions. The results are affected only very little from changes in the tax level, the level of the CO2 tax, the elasticity of scrapping. The level of the overall price elasticity however has important implications for the results, and the way that the elasticity is assumed to be phased in affects the order-ofmagnitude of the short- and medium termed effects in particular. 2.5.3 Socio-economics and cost-benefit analysis The analyses show that families in rural areas tend to buy more expensive cars than families in urban areas. Their average income is lower, and consequently, they use a significantly higher proportion of their income for car purchase. Consequently, the tax scenarios will have larger negative implications for the rural population than for the urban population. The below table illustrates the impact on the tax burden for the urban and the rural population respectively. 39 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 2.9 Tax burden implications of the registration tax scenario and the circulation tax scenario Tax scenario Family type Single without children Registration tax Circulation tax Average registration tax in percent of average income Percentage change in average circulation tax Urban Rural Urban Rural 10% 12% -11% -1% Single with children 8% 9% -5% -2% Couple without children 5% 7% -10% 3% Couple with children 4% 7% -8% 5% Another important aspect of the implications for rural and urban population is the welfare loss that is suffered as a result of behavioural changes that are invoked by the price changes. Some people will no longer have a car as a result of the price increases invoked by the registration tax, and some will choose to buy one that is different from the one that they would have otherwise as a result of the CO2 differentiated tax. The below table summarises the calculated welfare loss that will be suffered by the different family types and urban and rural areas respectively. As can be seen the welfare loss is higher for the rural population than for the urban population in all cases. Furthermore, the registration tax causes a much higher welfare loss than the circulation tax. This is due to the substantial welfare loss that is suffered as a result of the decline in the number of cars. Table 2.10 Welfare loss to rural and urban population as a result of the tax scenarios (SEK/car - total life time) Tax scenario Family type Registration tax 1) Urban Rural Circulation tax Urban Rural Single without children -21,557 -26,447 -3275 -4212 Single with children -25,986 -28,826 -4876 -4773 Couples without children -24,913 -32,114 -4037 -4404 Couple with children -28,900 -39,159 -5539 -5599 1) This is the welfare loss from the introduction of the tax per se as well as from the CO2 differentiated element. The latter accounts for the dominating share. Car manufacturing industry and derived industries Both scenarios will affect the sales of Swedish cars in Sweden. This in turn will affect the Swedish car manufacturing industry. The total change in the sales of Swedish cars that will result in the two scenarios is shown below. The registration tax results in the largest change. This is due mainly to the fact that this scenario involves the introduction of a new tax that is additional to the tax that is already in place, whereas the circulation tax is merely a reconstruction of the existing tax. Thus, the circulation tax scenario leads to a decline in the sales of Swedish cars and a corresponding increase in the sales of other cars, whereas 40 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars the registration tax scenario leads to an overall decline in the number of cars that are sold. Table 2.11 Change in the number of Swedish cars sold in Sweden in the two scenarios. Number of cars. Time horizon/scenario Registration tax scenario Circulation tax scenario 1 year 5 years 10 years 20 years -14,184 -10,409 -5,664 -10,056 -1,932 -2,133 -1,843 -2,261 The reduced demand for Swedish cars in the Swedish market thus leads to a decline in the level of activity in Swedish car manufacturing. To provide an indicator of the effect from this, this effect is translated into employment effect. It is estimated that the resulting loss of employment in the car manufacturing industry will correspond to the values shown below. These are based on a number of assumptions, which were made in order to derive an estimate of employment/car, and to correct for the distribution of production between the Netherlands and Sweden. In addition to the effect on the car manufacturing industry, other related industries can be affected as well. Among the most important related industries, one finds the car sales, car repair and maintenance and gas stations. The employment effect on this industry has therefore also been sought quantified. This quantification again rests on a number of simplifying assumptions regarding employment/car/year, and this effect is related to the effect that the tax scenario has on the overall size of the car fleet as shown in tables Tabell 1.7 and Table 7.7. Table 2.12 Loss of employment as a result of the decline in the sales of new Swedish cars. Number of employees (including owner/self-employed) Time horizon/scenario 1 year 5 years 10 years 20 years -1,348 -3,367 -3,514 -5,447 - of which car manufacturing -971 -1,633 -733 -1,608 - of which car sales and repair -377 -1,734 -2,781 -3,839 -1 -2 -2 -4 Circulation tax scenario -183 -203 -175 -214 - of which car manufacturing -183 -203 -175 -214 0 0 0 0 >-1 >-1 >-1 >-1 Registration tax scenario Effect relative to base, % - of which car sales and repair Effect relative to base, % The table illustrates that the introduction of a registration tax that includes a CO2 differentiated element has the largest effect on employment. This is a result mainly of the decline in the size of the car fleet that result from this scenario. 41 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Cost benefit analysis A cost benefit analyses has been conducted to provide an indication of the signs and the orders-of-magnitude of the effects on society from the two tax scenarios. The results from these calculations are shown below. Such calculations are always subject to a range of assumptions and uncertainties, and the analysis done in this study is quite rudimentary. While it should also be underlined that the results for the two scenarios are not fully compatible, as the registration tax includes the introduction of a new tax, the results nevertheless serve to underpin the previous comparative conclusions. The analysis below assumes that a certain, and increasing, fraction of those who become unemployed will find other employment elsewhere, and that the economic activity in question will compensate for the corresponding loss of profits. The analysis applies a unit price of CO2 reductions of 1,50 SEK. The costs of fewer and smaller cars have been calculated as a welfare loss, and consequently effects on energy use are included in this. The effects on the emissions of other pollutants is determined partly by the types of vehicles chosen, and partly by the speed by which new Euro norms can be phased in. Reduced scrapping per se thus leads to a delayed penetration of the new norms compared to what would otherwise have happened. Table 2.13 Results of cost benefit analysis of the two tax scenarios (million SEK) Cost and benefit items Registration tax scenario Circulation tax scenario Year Year 1 20 1 20 Costs - smaller cars - fewer cars 57 113 75 1,152 39 0 46 0 - employment - profit 25 43 9 26 3 4 0 1 239 1,262 45 47 80 -57 46 3 73 -166 -166 768 54 472 31 1,326 64 26 -1,096 20 0 0 0 20 -26 -26 301 -3 0 0 298 251 115 1,767 Total costs Benefits - CO2 reductions - other pollutants - accidents - noise Total benefit Total net benefit/year - discounted Total cumulated net benefit 42 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 3 Introduction 3.1 Background and purpose 43 This report presents the results of a study undertaken for the Swedish Environmental Protection Agency (Naturvårdsverket) and the Swedish Energy Agency (Energimyndigheten). COWI A/S has carried out the study in association with VTI (Swedish National Road and Transport Research Institute). The study was launched in August 2001. COWI submitted its proposal in early September 2001, and was awarded the contract later that month. The study was completed on 21 December 2001. The technical specifications The technical specifications for the tender called for a study to analyse demand effects, with regard to households' purchase of new cars as a result of the introduction of various CO2 differentiated registration taxes. Based thereupon, the study would consequently include an assessment of how such demand changes would affect CO2 emissions from passenger cars. In doing this, the study was to consider: • • • How the demand for new passenger cars would change as a result of a possible CO2 differentiated registration tax How the possible CO2 differentiated tax will impact upon the total sales of new cars and the composition of the vehicle fleet with regard to age, size distribution, petrol/diesel shares, average CO2 emissions of new cars and scrapping The impact on annual mileage from the choice of more fuel efficient cars (the rebound effect) The study was to include both company cars and private cars, and it should allow for an assessment of the implications for the proportion of petrol and diesel cars respectively of the total sales of new cars. The Technical Specifications requested that the following effects should be looked at in quantitative terms: • Effects on average CO2 emissions from new cars • Effects on annual vehicle sales and vehicle scrapping 44 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars • Effects on CO2 emissions from the vehicle fleet in 1,5, 10 and 20 years respectively. The results of the study are to be used as background information for possible future designs of CO2 differentiated taxes including registration taxes and encompassing an assessment of their environmental and socio-economic implications. Basis for the study COWI AS has developed a detailed car choice model, which has been used for similar purposes on several other occasions. The model contains a detailed modelling of car purchasers' demand patterns. It allows for quite detailed analyses of how households' will react to possible changes in the tax levels and the tax structures. This model, combined with the knowledge, expertise and relevant previous work with VTI, thus provides a strong basis for the conduct of the present study within the relative short period of calendar time. Additions to the Technical Specifications As stipulated in the Technical Specifications, the further contents and scope of the study has been established during the conduct of the study and in close collaboration with the Swedish Environmental Protection Agency and the Swedish Energy Agency. It was thus decided not only to consider registration taxes but also to include calculations that looked more into CO2 differentiated annual taxes (circulation taxes). 3.2 Model approach Approach and study organisation A model-based approach was applied to address the purpose of the study and the related requirements as stipulated in the Technical Specifications. It basically consists of three modules: • A car demand module which assesses the changes in the demand for new passenger cars as a result of the assumed price changes (that result from the taxes) • A car fleet module, which, at the starting point, consists of the car fleet in Sweden in year 2001. • A scrapping module constructed on the basis of historic data. Each year, the car fleet module is updated with the calculated results from the car demand module and the impacts on scrapping are calculated. Thereby, it is then possible to calculate an estimate of the resulting size of the car fleet. Thus, the assumed change of the tax system is entered into the car demand module (through its impact on car prices and/or car running costs), and the consequent impacts on car demand and scrapping are assessed. An overall price elasticity is used to allow for estimations of the impacts on the total sales volume and on the size of the car fleet. Socio-economic assessment In assessing the socio-economic implications, emphasis was put on two aspects: Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 45 • The possible impact on the welfare of the rural population • The possible impact on employment focusing on car manufacturing and the most important related industry (i.e. car sales, repair and maintenance plus service stations). Setting up scenarios The core features of the assumed changes to the tax system - the scenarios were established in close collaboration with the Swedish Environmental Protection Agency and Energy Agency. Similarly, the definition of the types of results to be produced and the alternative calculations to be carried out was done also in close collaboration with the client. Study organisation In this regard, the three project meetings that have been held with the client at the premises of the Swedish EPA have been of utmost value7. The meetings served to ensure a common basis of understanding of the project scope and project progress. Among the issues that have been discussed at these meetings are the following: the operational scope of the study, how to prioritise the efforts of the consultant, the definition of the base scenario and alternative scenarios and the scope of the socio-economic analyses. In addition, project meetings have been held internally on a regular basis, and a one-day working session was held with the participation also of VTI. To ensure a smooth and effective implementation of the project and in the light of the limited calendar time allocated for the study, a small core team was established. The team consisted of: Malene Sand Jespersen (project manager), Jørgen Jordal-Jørgensen (daily team leader), Nicolai Kristensen (economist) and Anders Pontus Matstroms (numerical analysis and transport research in Sweden). 3.3 Outline of this report This report presents the results of the study. The report is organised as follows: Chapter 2 describes the overall model approach as well as the separate modules in more detail. This includes also an explanation of the method used to derive the CO2 emissions from the vehicle fleet, of how company cars and private cars are treated in the model and how the split between diesel and petrol cars is calculated. Furthermore, chapter 2 describes in more detail the method used to provide an assessment of the socio-economic implications of the scenarios that have been analysed. Chapter 3 provides an overview of the types of data that are used to provide the requested quantitative estimates for the sales of new cars, the vehicle fleet and scrapping. This includes a description of both the data that relates to the vehicles in physical and monetary terms, and the data that relates to the socioeconomics of car buyers. The latter is an essential feature of the car choice model. The chapter explains the overall price elasticity that has been used to 7 4 October, 15 November and 18 December 46 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars estimate the impact on the sales volume and the size of the car fleet. Lastly, the chapter also provides an overview of the data used to assess the socio-economic implications of the scenarios. Chapter 4 explains and motivates the scenarios that have been selected for detailed analyses. Two base scenarios are identified. The first scenario assumes that a registration tax is introduced, which contains a CO2 differentiated element. The other scenario assumes that the existing circulation tax is restructured to include a CO2 differentiated element. In both cases, an upper limit is defined for the tolerable total tax level. The definition and operational scope of the scenarios were supported by a number of initial calculations. The chapter provides an overview of results from these initial calculations that were used as background for establishing the level of tax, the level of differentiation and the type of the tax to use. Further, the chapter provides an overview of the types of results that the model calculations provide. The chapter explains the core features of the base scenario and it outlines the types of alternative calculations (sensitivity analyses) that assist to assess the sensitivity of results in regard to certain specified parameters. Chapter 5 provides an overview of the results that are provided from the model based calculations. The chapter summarises and discusses the resulting output tables which provides information, among other things, on: The resulting CO2 emissions in total and per car, The size of the car fleet, The sales volume (number of new cars sold), The average age, length and weight of cars, Tax revenue and average tax payment per car, The split between petrol and diesel cars, The composition of the car sales and vehicle fleet with regard to car makes (separating also the domestic brands from foreign brands), • The level of downsizing, • The average and total dealer's price, • • • • • • • In addition, the chapter summarises the results of alternative calculations that are used to assess the sensitivity of results with regard to certain parameters and assumptions. The parameters that are analysed are: • • • • The overall price elasticity The price elasticity of scrapping The tax level The level of differentiation Chapter 6 assesses the socio-economic implications of the base scenario focusing on impacts on rural/urban population and on employment effects related to the effects on the Swedish car manufacturing industry and the sector encompasses car sales, repair and maintenance plus service stations. Chapter 7 assesses the overall gains and losses to society from introducing the tax changes that are contained in the two scenarios. This is done by means of a Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 47 cost-benefit analysis that seeks to take into account all the major effects and to quantify those in monetary terms. Chapter 8 summarises the generic conclusions that can be drawn from this study. The chapter thus discusses the amount of CO2 reductions that can be achieved from rethinking the vehicle taxation system within the revenue limits that can be tolerated. In this, the chapter also discusses the appropriateness of the registration tax versus the circulation tax, and the extent to which the level of the tax matters compared to the level of differentiation. The chapter discusses the medium and long termed effects with regard to CO2 emissions from the vehicle fleet. 48 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 4 49 Model and methodology This chapter explains first the construction of the three-tiered model based approach and its underlying thinking. First, the overall approach and the interlinkages between the various modules are explained. Second, each module is described in more detail followed by a description of the role played by the overall price elasticity and other elasticities. Lastly, the chapter summarises the approach taken to provide the socio-economic analysis of the results. The latter includes also the cost-benefit analysis. 4.1 The overall model framework and methodology This section explains the above three modules in overall terms, and provides a thorough description of how they work together and how the relevant policy instruments (the circulation tax and in particular the registration tax) influence the entire model-framework. The more detailed description of the three modules is contained in sections 4.2, 4.3 and 0. Model framework The general idea is to make an explicit modelling of entry (new car registrations) and exit (scrappage) into the car fleet. This idea is illustrated in the following figure. Figure 4.1 Current car fleet Illustration of Entry and Exit At the point of departure, the current stock of cars is the total car fleet in Sweden. The CO2 emissions of the cars in the existing fleet differ according to car age, engine size, diesel/petrol and whether the car is of a Swedish fabricate or not. The set-up however does not take into account that the CO2 emissions may also vary for different variants of cars. This implies for example that the exact emissions from a Ford Focus 1.6 collection 3D is not used as input to the model. The current car fleet is updated sequentially year by year by adding the inflow of new cars and by subtracting the outflow of scrapped cars. For new cars the exact CO2 emissions are known. 50 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Inflow The inflow of new cars to the car fleet is estimated using the car choice model for Sweden, which was previously developed by COWI for the European Commission. This model uses data on car price and operation costs, vehicle characteristics and socio-economic features of car buyers, cf. section 4.2 to derive its outputs. Outflow The outflow of cars from the car fleet is determined in the scrapping module, cf. section 4.3. The car characteristics of the scrapped cars are car age, engine size and whether the car is diesel or petrol driven. Socio-economics and cost-benefit A number of socio-economic characteristics are included in the analysis. The data allow for an analysis of the socio-economic effects from introducing a registration tax in terms of the effects across different family structures, income and the level of urbanisation. Furthermore, the results from the model calculations are analysed with a view to assessing the impact on industry's employment. Lastly, the results of the study are combined in a cost-benefit analysis that aims to assess the costs and benefits from the two tax scenarios that are analysed and to compare these. Elasticities To properly reflect the developments over time, elasticities that affect the inflow and outflow of cars to and from the car fleet are added on cf. section 5.7. These elasticities cover: • • • The rebound effect; i.e. the change in the annual driven distance as a result of people driving more fuel-effective cars (when cars are more fuel efficient, the running cost per kilometre is reduced) The change in the size of the car fleet as a result of a change in the price of cars, e.g. as a result of introducing a registration tax (the more expensive cars become relative to other consumer goods, the less will the demand for cars be) The change in the number of scrapped cars as a result of a change in the price of cars, e.g. as a result of introducing a registration tax (the more expensive cars become to acquire, the longer will it take until the cars are scrapped) The effects described above are illustrated in Figure 4.2. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 51 Figure 4.2 Module description 4.2 Model background and overall approach The EU model compared to this study The car choice model 4.2.1 Model scope and approach The COWI Cross Country Car Choice Model is used in this study to do the model-based calculations. This model is a further development of the Car Choice Model that was originally developed in a Danish context, and which was further developed in connection with the study on "Fiscal Measures to reduce CO2 Emissions from Passenger Cars" undertaken by COWI AS in 2001 for EU DG-ENV. The latter included among other things the development of a specific model for Sweden, and this model constitutes the starting point for the present study. However, in the context of this study, modifications are made to among other things the socio-economic database of the model and the result tables to be provided by the model. These modifications serve to ensure that the model can properly address the specific questions that are raised in this study, cf. section 3.1. For example, the results provided by the EU model were subject to certain conditions, viz.: the proportion of diesel vehicles should not increase, the scenarios should be budget neutral and there should be no downsizing as a result of the scenarios. It is important to be aware of these differences as they do in- 52 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars fluence the scope of action set by the scenario calculations as well as the results per se. Generic features of the model The model can be termed a "what if" model in the sense that it does not perform projections nor forecasts. It is thus not a prognostic model. Rather, the model merely calculates what would happen to today's car demand if certain characteristics of today's tax systems were changed (i.e. the scenario assumptions), all other things being equal. In undertaking the assessments and the scenario analyses, the model simply compares the results of the present situation with the hypothetical situation where the tax structure and/or the tax levels is assumed to be different. The model only includes new cars. Consequently, second hand cars are excluded from the analyses undertaken in this module. Furthermore, the model-based calculations only consider passenger cars with petrol and diesel engines. Other technologies could in principle be included as well, as long as they are in supply today. The major motivation underlying this delineation is the scarce current supply of cars equipped for alternative fuels, which renders estimations on these cars less robust. Car purchase decisions are based on a number of parameters. Apart from those that relate to the car (technical and financial parameters including taxes) the model also considers the income, household structure and age of the purchaser of the car. These socio-economic features of car buyers are accounted for at the model calibration stage. Compared to the EU model, the socio-economic features of car buyers are extended here to include singles with children explicitly and to distinguish between rural and non-rural population. Being a what-if model, the model does not take into account that demand patterns may change in the future as a result of for example changes in fashion, changes in production costs and pricing policies of manufacturers and innovation. The model includes both private cars and company cars. The demand for private cars and for company cars are calculated separately and then combined taking the proportion of company cars into account. Company cars are less sensitive to price changes relative to private cars, while they are more sensitive to annual cost (circulation tax and fuel costs) than private cars. These features are taken account of in the model calculations and they are very important, because company cars constitute about 50% of the sales of new cars in Sweden. The most important parameters of the model are the parameters related to the car price on the one hand and the annual cost on the other hand. The annual costs consist of the annual tax and the annual fuel cost. Consequently, it is assumed that the car purchasers consider fuel cost and circulation tax in the same way when making his purchase decision. In other words, an increase of 1000 SEK in the annual tax is assumed to have the same effect as an increase in the annual fuel cost of 1000 SEK. Furthermore it should be noted that the effect from changes to the annual tax for the company cars rests on the assumption that such tax changes are reflected in the personal taxation of company cars. Model output Important outputs from the model are the structure of demand for new passenger cars in Sweden and the associated average energy consumption and Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 53 CO2-emissions. New fiscal measures (e.g. a registration tax) can be introduced into the model and the consequential effects on the structure of demand for new cars can be calculated acknowledging the above limitations. 4.2.2 Car choice model framework The overall model set-up is illustrated in Figure 4.3. Figure 4.3 Car choice model Four parts of the model As shown, the model structure consists of four parts: The input data, the database, the model and the results/outputs: • The input data are the variables to be analysed in terms of their effect on car purchase (e.g. a registration tax). • The database contains observed data from the market, such as prices and technical characteristics of sold cars, plus assumptions made outside the model (relating to for example annual mileage). The database also includes 54 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars socio-economic information enabling a characterisation of car buyers according to income level, household structure, urbanisation, children or no children and age). • The model combines the estimated relations between the objective conditions (database and input data) and subjective consumer behaviour (the elasticities derived from the parameter estimates, cf. Annex A). • The results/outputs are the calculated car purchases given the above input data, database and model. The output includes information on the resulting tax revenues, energy use, average car prices, composition of the resulting sales of new vehicles, the consequent CO2 emissions from new cars, etc. Input The input data for the model are the relevant taxes. This includes the existing circulation tax and the fuel tax, as well as the possible taxes or tax changes that are assumed in the scenarios. Private cars and company cars The model includes both private cars and company cars. The demand for cars is estimated separately for private cars and company cars and then combined taking the company car share into account. It should be noted that, compared to private cars, company cars are less sensitive to price changes. This observation is included in the model calculations. Outputs The model outputs provide information on for example the distribution of the total purchase of new passenger cars and the associated average CO2-emissions given the established assumptions about the fiscal measures (expressed by the input data). The outputs may be decomposed in various dimensions as illustrated in Figure 4.3. The exact scheme for presentation of the outputs in this report is described in chapter 7. 4.2.3 Database for the car choice model The database includes three types of information on the purchases of new cars: • • • Socio-economic features of buyers of new cars Information on car price and the running cost Other, physical, vehicle characteristics including CO2 emissions Car sales Comprehensive and consistent sets of data on cars and car sales are available for the last half of 1999 and the first half of 2000. Marketing Systems GmbH provided these data on car sales and the associated prices and car features. The data were originally acquired in the context of the EU study, and they are assessed to be adequately up-to-date to justify their use also in this study. Socio-economic data Socio-economic data were constructed in the above EU study as well. These data constitute the starting point for establishing the socio-economic database for this study, but it is further modified in order to take into account also singles with children and to allow for a differentiation between rural and non-rural car purchasers. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 55 Company cars Company cars constitute approximately 50% of the total number of purchases of new cars in Sweden (although it depends on the business cycle).8 The factors that affect decisions on the purchase of company cars differ from those that affect the decisions on private cars. To improve the accuracy of the model information on the purchases of company cars and the taxation systems that govern this part of the market was collected by means of a questionnaire in the context of the mentioned EU study. The applicability and accuracy of this information was assessed, and based thereupon it was decided to use these data also in this study. Car prices and costs Among other things, the data on vehicles include the car prices. This information, combined with information on relevant taxation including VAT, is used to calculate the dealer's price (i.e. the price exclusive of VAT) and the associated car taxes. The exact derivation of the purchase price is done in a separate module. Another separate module derives the running costs per 100 km. The car costs and the vehicle characteristics together with the socio-economic data form the "consumer car characteristics", which are used as a basis for the car choice model calculations. 4.2.4 The logit model The logit model combines the information about the consumer car characteristics with the observed market data for car sales in order to establish reliable relations (and thereby elasticities) for consumer behaviour on the car market. Hence, the first step of the model calibration is to establish a model set-up that provides calculated results/outputs for today's situation which are reasonably similar to the situation as it is actually observed. Three sub-models The car choice model consists of three sub-models. The first model determines whether the consumer buys a private car or a company car. Two models are used to determine the choice of car for the private car buyers and the choice of car for the company car buyers, respectively. Thus, there are separate modules covering private cars and company cars. In the private car model, consumers are assumed first to choose the type of car and then, subsequently to choose the specific car. This sequential discrete choice structure is estimated using a nested logit model, cf. Annex A. In the company car model, the choice of the car buyers’ (companies) is modelled in one step (a more simple non-nested logit model). This step determines both the car type and the specific car at the same time. The logit models build on the assumption that the consumer maximises his or her utility. Outputs The output of the model calculations is a large number of individual car choices, summarised as the relative (probability-based) change in the purchase of each car, e.g. as a result of introducing a registration tax. The relative change 8 Source: Statistiska meddelanden SSM 01:1, Fordon enligt bilregistret, fjärde kvartalet och hela året 2000. The definition of a company car is: All cars owned by private or public companies. 56 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars will be applied to the present pattern of car purchases in order to arrive at the estimated number of specific cars purchased after the change of input data. Among the most interesting results in the context of this study are the CO2 emissions that result from the alternative tax scenarios and the consequent effect on car purchases. This overall result may be decomposed and illustrated in various ways and dimensions, as shown in Figure 4.3. The model set-up and model considerations are described in more detail in the Technical Annex. 4.3 The scrapping module The scrapping module serves to keep track on the effects that changes to tax system have on scrapping behaviour, and to ensure that scrapping is taken account of in general. Car owners tend to keep their cars longer when the price of new cars increases. For the registration tax this means that the scrap rate has to be elastic towards changes in the car price. The scrapping module thus needs to have information on survival curves for the cars and on the effects that price changes have on these curves. Two sources The scrapping module is therefore based on two sources: • • Circulation tax does not affect scrapping Survival curves for the Swedish car fleet 1979-1995 Price change simulations in a model prepared for DETR, UK The way that the circulation tax scenario is framed (described in chapter 6) imply that it will have no impact on the average car price, and hence it will not affect neither scrapping (as expressed by the survival curves) nor the amount of sold new cars. Consequently, there is no need for price change simulations in the circulation tax scenario. 4.3.1 Survival curves VTI supplied historical data on the survival rates of the Swedish car fleet across different age groups. The Swedish survival curves provide aggregated survival rates for the entire car fleet, i.e. across different types and sizes of cars as well as for both petrol and diesel. The average life length (age) of cars increases over time. However, trends in survival rates are not taken into account, because the technology is assumed constant. A deviation from the constant technology assumption is that the CO2 effectiveness is assumed to follow the historical trend over time, cf. section 5.4, but this is assumed not to affect the average age of cars, but merely their (energy) performance. The applied survival rates are from 1995. This is the most recent year for which we have observations for each age group, cf. Figure 4.4. 57 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 4.4 Conditional survival rates, Sweden (1995) Conditional survival rate 1.2 1 0.8 0.6 0.4 0.2 24 22 20 18 16 14 12 10 8 6 4 2 0 0 Age Along the vertical axis is the so-called conditional survival rate. This indicates the probability that a given car will survive T+1 years given it has already survived T years. 4.3.2 Price change simulations A registration tax will imply that the consumer prices of new cars will increase, and the scrap rate of old cars will be expected to decrease, because car owners will hold on to their old cars for longer than before. This effect is captured in the scrapping module by applying an English simulation model from Department for the Environment, Transport and the Regions (DETR). The model allows simulations of price increases and the resulting scrapping. Hence, elasticities can be obtained by comparing the scrapping before and after a simulated price change. The model is estimated on English data. It could therefore be potentially relevant to adjust other factors of the model in order to mirror Sweden as much as possible. The model allows for changes in such factors as for example, the speed limit, a possible scrapping subsidy and vehicle running costs (including circulation tax). However, the model uses a constant elasticity. Therefore, it is not relevant to introduce such changes in the other variables. The constant elasticity implies that the changes would apply both prior to and after a price change, and that the effect would therefore net out. The model allows for a split between diesel and petrol cars. For the latter group, the model also applies different elasticities across engine sizes. The increase in survival rates per percent price increase is shown in Table 4.1. 58 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 4.1 Change in Survival per percent change in price, petrol and diesel cars Engine Size, Petrol cars Age of Vehicle < 1500 1500-2000 >2000 1 0 0 0 2 0 0 0 3 0 0 0 4 0 0 0 5 0 0 0 6 0 0 0 7 0.0000 -0.0001 -0.0002 8 0.0001 0.0000 -0.0002 9 0.0004 0.0002 -0.0004 10 0.0008 0.0008 0.0002 11 0.0015 0.0014 0.0004 12 0.0019 0.0019 0.0006 13 0.0025 0.0028 0.0014 14 0.0035 0.0034 0.0017 15+ 0.0041 0.0048 0.0038 Diesel cars All 0 0 0 0 0 0 0.0000 0.0001 0.0004 0.0008 0.0014 0.0020 0.0021 0.0024 0.0023 One incorporates the price effect by multiplying the elasticity per percent with the percentage increase in prices. In order to calculate the size and composition of the car fleet after the price change, one adds the resulting increase in the survival rate to the survival rates obtained from the historical survival curves. 4.4 The car fleet module Method of calculation The car fleet module keeps track of the number of cars in the fleet at any point in time. The starting point is the entire Swedish car fleet as of September 2001. During a year, some of these cars are scrapped and the remaining cars become one year older. To this new “temporary fleet” we add the new cars that have been sold during the year. This information is entered from the Car Choice Model. These cars have the age of 0. In this manner the car fleet for 2002 is modelled. Cars are then scrapped again during the following year, the remaining become another year older, etc. Categories of cars The fleet (as well as the cars that enters it or the cars that are scrapped) is characterised by the following variables: • • • • Age (0 to 25+) Engine size (3 groups) Diesel or petrol Swedish fabricate or not Swedish Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 59 These 4 categories give 312 different groups of cars.9 For each of these groups the fleet module also contain information on: • • • • 4.5 Average CO2 emissions with the group Average length Average weight Number of cars in each group Socio-economics 4.5.1 Effects on urban and rural car purchasers Changes to the car tax structure have different implications for different families. This is because the different types of families prefer different types of cars. Focus on distribution effects The assessment here focuses on the distribution effects. For example, it aims to assess how the tax burden is distributed between different types of families, and to analyse how the invoked shifts in the demand for new cars are distributed between different types of families. To this end the Danish model has been re-estimated to allow for different demand preferences in the relevant family types. Furthermore, a new set of data for Swedish car purchasers has been set up to support this dimension of the model calculations. 4.5.2 Car manufacturing Changes in the level of economic activity of the car manufacturing industry have repercussions both on the car manufacturing industry itself as well as on other related sectors, upstream and downstream. Focus on most important sectors The assessment here centres on the task of identifying the most important sectors that can be affected and the extent to which they will be affected by changes in level of economic activity of the car manufacturing industry. For the most important sectors, the employment effects that will result from the scenarios are assessed in quantitative terms. 9 26 x 3 x 2 x 2 = 312. (26 is the life span of the cars, 3 covers the three engines groups, 2 covers the options of either petrol or diesel driven, and 2 expresses that the car may be of a Swedish fabricate or not. 60 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 61 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 5 Data and model inputs 5.1 Taxes and charges Table 5.1 below provides an overview of the relevant taxes and charges that are currently in effect in Sweden. The information is derived from the EU study. Table 5.1 Existing taxes and charges in Sweden relevant for this study Type of tax Level and features VAT on car purchase 25% Fuel tax, petrol 4.6 SEK/litre Fuel tax, diesel 3.0 SEK/litre Circulation tax Differentiated according to the kerb weight of the car using a stepwise linear relation. The tax increases by approximately 150 SEK for each 100 kg for petrol cars and approximately 572 SEK per 100 kg for diesel cars. Annual tax on company cars Value based tax: 10,980 SEK plus 9% of the value of the car. 5.2 Vehicle sales Marketing Systems GmbH has supplied the Consultant with monthly records of all new registrations (from January 1999 to the latest available data for 2000). The set of data refers to all passenger cars. Thus, it includes both those that are purchased and used privately, and those that are purchased under a company scheme, but used also or solely for private purposes. Technical variables and vehicle characteristics All new registration entries (for example, Audi TT 1,8 Roadster) are linked to the various technical variables and vehicle characteristics, as shown in Table 5.2 overleaf. 62 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 5.2 Descriptive vehicle variables Variable Definition Number of doors 2, 3, 4 or 5 Transmission manual / automatic Air-conditioning Body type Price (incl. All taxes) Driven wheel front / rear / all Space index Length millimetres Width millimetres Wheel base millimetres Turning diameter metres Luggage capacity Fuel type super unleaded 95 RON / diesel Acceleration (0 to 100 km/h) seconds Fuel consumption ECE litres per 100 km CO2 emission per km Driver airbag standard / optional Passenger airbag standard / optional Side airbags standard / optional Engine size cubic centimetres HP Top speed miles per hour Weight kerb weight and GVW (gross vehicle weight) Engine type in-line / V-type Number of cylinders Number of valves per cylinder 2, 3, 4 or 5 Price inclusive of tax and VAT NOTE: Features of new cars in Sweden the above variables are inventoried for each vehicle type (for example, Audi TT 1,8 T Roadster), which is given a unique identification code (in this example, aud-tt0010). The following table provides a summary of how selected vehicle characteristics apply to new registered cars in Sweden. 63 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 5.3 New car characteristics Car type Registrations Average engine size 2% 1,109 147 81,550 Small 13% 1,414 162 108,020 Lower medium 26% 1,661 181 148,685 Medium 29% 1,938 200 200,279 Upper medium 25% 2,399 223 270,626 Luxury 0% 3,905 293 660,352 Sport 1% 2,177 218 302,315 Off-road 2% 2,651 275 263,560 MPV 2% 2,591 254 250,818 Mini Average CO2 emissions g/km Average price It is assumed that the CO2 emissions from new cars that will enter the fleet from year 2001 and onwards are reduced by 2% annually. Based on historical data, this order-of-magnitude is assessed to properly reflect the technological progress in regard to energy efficiency. As can be seen from Table 5.4, this order-of-magnitude is a reasonable reflection of past developments. The table is reproduced from VTI report 386, table 8. Table 5.4 Annual rates of increase in fuel effectiveness Year 1.3% 2.0% 1.8% 1.9% 1.9% 1980 1984 1988 1990 1992 5.3 Yearly increase in effectiveness Scrapping The data sources for the scrapping module are described in section 4.3. 64 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 5.4 Available data Vehicle fleet Vägverket has supplied data on the total fleet of cars in Sweden. The data disregards cars that are not driving on the roads on a regular basis. The data set, which comprises 4,057,244 cars, is from September 2001. It includes a host of different variables of which the most important are: • • • • • • • Age of the car (vintage) Fabricate Model Fuel type (petrol or diesel) Effect (motor size) Weight Length However, the set of data does not contain information on emissions and on the engine size. Consequently, estimations are necessary to provide such data. Approach 5.4.1 Estimation of CO2 emissions CO2 emissions were estimated by applying a model from VTI.10 This model provides estimates on the CO2 emissions by using the following car characteristics to explain CO2 emissions: • • • • • Weight Motor effect/weight Dummy for automatic gear Vintage Model The model, however, needs various extensions. It is based on data from the 1980s and up to 1992, and with a linear extrapolation from 1992 and onwards. The necessary extensions differ for petrol and diesel cars. Amendments of relevance to petrol cars The CO2 emissions for petrol cars that are calculated in the VTI model are amended for two reasons: • • The model is based on a US driving cycle The model does not capture the introduction of catalysts in petrol cars from the late 1980s The first shortcoming, the change from a US driving cycle to a European driving cycle, necessitates that the calculated amount of CO2 per km is multiplied by a factor of between 5% and 10%. The exact factor would depend on the car type. However, no detailed knowledge is available on the specific factor that would apply to the various car types. 10 VTI report 386: “Rimliga mål för genomsnittlig bränsleförbrukning för nya personbilar. Written by Henrik Jönsson (1993). 65 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The second shortcoming is that catalytic converters were introduced from around 1987. Catalytic converters reduce the energy efficiency and consequently, they lead to higher emissions of CO2. This effect is not captured by the extrapolation. Consequently, the emissions become underestimated. In the extended model the effect from catalysts has been introduced to new cars gradually, with 20% in 1987, 80% in 1988 and 100% from 1989. Motivated in these two shortcomings, the estimated CO2 emissions have been calibrated to fit the actual CO2 emissions for petrol cars by adding 20% to the CO2 emissions from the VTI model (before 1987 only by 10% since catalysts were not introduced before 1987). The resulting CO2 emissions over time are shown in Figure 5.1. Figure 5.1 Calculated CO2 emissions from petrol cars, 1970 to 2001 320 g CO2 per km 300 280 260 240 220 01 99 20 97 19 95 19 93 19 91 19 89 19 87 19 85 19 83 19 81 19 79 19 77 19 75 19 73 19 19 19 71 200 The introduction of catalysts causes a small upwards shift in the level of CO2 emissions around 1987-1990. From Vägverket, Environmental Report 2000, it is however known that the CO2 emissions were actually stable from 1987 to 1990. This shortcoming from comparing the extended model with the actual state of affairs gives rise to a small inaccuracy. It is however not important for the overall results of the model. The reason why the emissions proved to be stable in the period 1987-1990 is that cars became significantly heavier in this period. Heavier cars are less energy effective than lighter ones and consequently, this effect outbalanced the effect from the trend of 2% annual improvements in energy efficiency. Since 1997 the trend has again been increased efficiency along the historical trend. 66 Amendments of relevance to diesel cars Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The CO2 emissions for diesel cars calculated in the VTI model are amended for the following reasons • • • The model is based on a US driving cycle (like for petrol cars) The model does not provide for a split between diesel and petrol cars Technological development for diesel cars has recently flourished The change in the driving cycle has the same impact on diesel cars as on petrol cars, cf. above. The VTI model is estimated across both petrol and diesel cars. However, there were very few diesel cars in Sweden in the 1980s. Consequently, the determination of the parameters becomes highly dominated by petrol cars. Generally, diesel cars have 10% lower CO2 emissions than petrol cars. In recent years, there has been a surge in the technology for diesel cars with the introduction of the TDI technology (for some models called HDI). Cars with this new technology have much lower CO2 emissions than older diesel cars without TDI. The new technology entered the market around 1997 and is gradually becoming more outspread. The TDI technology has been phased in from 1997 with the following shares. Table 5.5 TDI share in Diesel cars Year 1997 1998 1999 2000 2001 TDI share 20% 40% 60% 80% 100% Given the changes described above, the calculated CO2 emissions evolve over time as shown in Figure 5.2. 67 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 5.2 Calculated CO2 emissions from diesel cars, 1970 to 2001 380 360 g CO2 per km 340 320 300 280 260 240 220 200 Validity of calculated CO2 emissions Approach 01 99 20 97 19 95 19 93 19 91 19 89 19 87 19 85 19 83 19 81 19 79 19 77 19 75 19 73 19 19 19 71 180 The above estimations of CO2 emissions are assessed to be valid and sufficiently accurate for the purposes of this study. The VTI model probably estimates CO2 emissions from cars that entered the fleet before 1992 quite accurately. CO2 emissions from new cars are estimated in the Car Choice Module, which provides the specific emission level for the specific car. For the intermediate period 1992-2000 the applied corrections result in a perfect match for the year 2000 (on average) and probably a very good fit for the remaining years. 5.4.2 Estimation of engine size The engine size is not included in the data set from Vägverket either. It has therefore been estimated in an OLS regression model. The model weights each group of explanatory variables by the number of cars in each group. The explanatory variables are the following: • • • • • weight (“tjänstevikt”) motor effect length g CO2 per km a dummy to indicate whether the car is Swedish In the model set-up, the engine size is grouped into one of three groups, cf. Table 5.6. Table 5.6 Number of cars in each engine size group, diesel and petrol Engine Size <1500 ccm 1500-2000 ccm >2000 ccm Total Petrol 576,042 1,156,305 2,139,169 3,871,516 Diesel 502 17,727 167,499 185,728 68 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The table shows that the cars in Sweden are generally quite large and that the diesel cars are relatively larger than petrol cars. 5.5 Socio-economics of car buyers This section describes the socio-economic set of data that has been applied for the car choice model calculations. Approach The car choice model requires socio-economic data for car purchasers. Such data are not directly available for Sweden. To set up an approximate data set for car purchasers the study combines the socio-economic data for Sweden with car purchase ratios from Denmark, and takes into account the larger number of cars per inhabitants in Sweden and the higher company car share in Sweden. The method is explained stepwise below. Source The main source for the socio-economic data is a database, which contains information for all individuals aged 17 years or more11. The database is constructed based on data material from SCB. The information provided in this database covers: • • • • • • • • • 11 Municipality Family ID number Year of birth Sex Car ownership Personal income Family income Family status Number of children See "Modeller och prognoser för regionalt bilinnehav i Sverige" Pontus Matstoms, VTI for a more detailed description of the database. 69 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars For the purpose of this study, the families in the above database have been aggregated into the following dimensions: Table 5.7 Overview of the socio-economic dimensions of the car choice model Feature Dimensions Family type Single Couple Children No children One or more children Urbanisation Rural municipalities All other municipalities Net family yearly income Low income (below 151,800 SEK) Medium income (between 151,800 and 453,200 SEK) High income (above 453,200 SEK) Based on the above data set one can set up the following socio-economics table for the total population of families in Sweden. Table 5.8 Socio-economics for Sweden. Total population (Families) Family type/income group Single No children Urban Low Medium High Total 1,096,937 469,330 5,885 1,572,152 74,916 139,390 947 215,253 123,511 45,172 357 169,040 10,553 18,374 71 28,998 Couple No children Urban 404,017 210,080 14,020 628,117 Couple Children Urban 225,671 354,581 17,416 597,668 Couple No children Rural 58,256 24,325 960 83,541 Couple Children Rural 39,192 47,665 1,180 88,037 2,033,053 1,308,917 40,836 3,382,806 Single Children Urban Single No children Rural Single Children Rural Families total The proportion of the total population that purchase a car in a given year is calculated based on similar data for Denmark. Furthermore, the proportion is adjusted for differences between Denmark and Sweden as regards the proportion of company car purchases and the total number of cars per family. This is illustrated below. Distribution of sales using Danish purchase ratios Firstly, we calculate the expected distribution using the Danish car purchase ratios. This calculation would result in the following table. 70 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 5.9 Socio-economics for Sweden, car buyers, uncorrected Family type/income group Medium High 13,895.70 22,856.35 1,049.84 37,802 949.02 6,788.29 168.94 7,906 1,564.60 2,199.87 63.69 3,828 133.68 894.81 12.67 1,041 Couple No children Urban 27,975.08 36,346.94 3,805.59 68,128 Couple Children Urban 15,625.99 61,347.74 4,727.40 81,701 Couple No children Rural 4,331.86 3,686.22 260.82 8,279 Couple Children Rural 2,914.28 7,223.18 320.59 10,458 67,390 141,343 10,410 219,143 Single No children Urban Single Children Urban Single No children Rural Single Children Rural Total car purchase Relatively more cars in Sweden than in Denmark Low Total Next, there are more cars per person in Sweden relative to Denmark (355 per 1000 inhabitants in Denmark and 428 per 1000 inhabitants in Sweden). This means that more cars are purchased in Sweden compared to Denmark. Taking into account that there were 8,859,429 inhabitants in Sweden in 1999 and that the average lifetime for the cars is approximately 15.8 years, this implies an additional 40,933 car purchases per year compared to the above table.12 It is assumed that it is solely the medium and high-income groups that purchase these additional 40,933 cars. Thus, the correction only affects the number of car purchases of these two groups. The adjustment is done in proportion to the number of households in these two groups. Thus, if the high-income group constitutes 10% of medium and high-income groups, then the group is allocated 10% of the additional car purchases. The resulting car purchases corrected for the difference in the number of cars per family between Sweden and Denmark are shown in Table 5.10. 12 Calculated as 8,859,429/1,000*(428-355)/15.8 = 40,933 71 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 5.10 Socio-economics for Swedish car buyers, corrected for number of cars per family Family type/income group Low Single No children Urban High Total 13,896 29,021 1,333 44,250 949 8,619 215 9,783 1,565 2,793 81 4,439 134 1,136 16 1,286 Couple No children Urban 27,975 46,151 4,832 78,958 Couple Children Urban 15,626 77,895 6,003 99,524 Couple No children Rural 4,332 4,681 331 9,344 Couple Children Rural 2,914 9,172 407 12,493 67,390 179,468 13,217 260,076 Single Children Urban Single No children Rural Single Children Rural Total car purchase Company cars are more frequent in Sweden Medium Company cars share constitute 50% of the car purchases in Sweden compared to only 27% in Denmark. This means that private families purchase relatively fewer cars in Sweden compared to Denmark. Simply applying the Danish car purchase share to Sweden would therefore result in an overestimation of private cars in Sweden. The total number of car purchases in Sweden was 354,64913 in 2000. Consequently, the overestimation amounts to 354,649 * (50% - 27%) = -82,751 cars per year. It is expected that company cars are only relevant to the medium and high-income groups. Similarly to the above adjustment for the number of cars, the adjustments here is also done proportionate to the number of households in these two income groups. Table 5.11 shows the socio-economic data applied in the model calculations. Table 5.11 Socio-economics for Swedish car buyers corrected for company cars and number of cars per family Family type/income group Single No children Urban Low Medium High Total 13,896 16,558 761 31,214 949 4,918 122 5,989 1,565 1,594 46 3,204 134 648 9 791 Couple No children Urban 27,975 26,331 2,757 57,063 Couple Children Urban 15,626 44,442 3,425 63,493 Couple No children Rural 4,332 2,670 189 7,191 Couple Children Rural 2,914 5,233 232 8,379 67,390 102,393 7,541 177,325 Single Children Urban Single No children Rural Single Children Rural Total car purchase 13 Source: SIKA, Statistiska Meddelenden SSM 01:1 72 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 5.6 Socio-economics assessment 5.6.1 Data on urban and rural population Data on urbanisation are the same as described in section 5.5, i.e. the source is Statistics of Sweden (SCB). 5.6.2 Employment The necessary data to estimate the employment effects from the scenarios were obtained from a range of sources. Data on car manufacturing and derived industries have been obtained from SCB (input-output tables for 1998). This information was used to identify the most important related industries. Information supplied by an abstract from "Bilismen i Sverige 2000" (Swedish National Road and Transport Research Institute) was used. This information assisted to assess the relative distribution of sales between the various types of vehicles (passenger cars and heavy vehicles), to assess the export share, and to estimate the proportion of the production that is effectuated in the Netherlands. Online data from SCB provided employment data. Information from the Statistical Yearbook of Sweden 2001 was used to assess the amount and distribution of vehicles serviced by the Swedish car sales and car repair and maintenance sectors, and to get employment data and overall economic data for the Swedish economy. 5.7 Three additional effects from taxation that are external to the model Some effects are not included directly in the model since that would be by far too cumbersome and very time consuming. Instead, the calculations make use of existing studies that offer information on aspects needed to supplement the modules in the model framework. In particular, we include three additional effects: • • • Rebound effect Elasticities Rebound effect Changes in the size of the car fleet (number of cars) as a result of a price increase Changes in scrapping (number of cars) as a result of a price increase People choose to drive longer distances when the price per km decreases. This effect is taken account of by the introduction of a measure of this price elasticity. The elasticity reveals how much the driven distance changes when the price per kilometre changes, e.g. as a consequence of more energy effective cars. Johansson and Schipper state an elasticity of –0.2.14 This estimate, which we apply in the tax scenarios, is based on panel regressions, and the authors’ knowledge about limitations in the data material and statistical methods as well as experience. The estimate of –0.2 is a long run elasticity estimate. 14 See Oluf Johansson and Lee Schipper (1997) “Measuring the long-run fuel demand of cars”, Journal of Transport Economics and Policy, September, pp. 237-252. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 73 Another possible implication of a reduced number of cars in the fleet is that the annual distance driven per car is likely to increase. Thus, if the price increase that results from a registration tax gives rise to a reduced number of cars, it is likely that the annual distance driven per car will increase. The reason is that the marginal car owner (who chooses not to remain a car owner) has lower utility out of the car compared to car owners who are not marginal.15 The model does not capture this possible effect on mileage, and hence on total emissions. Price elasticities The two price elasticities, which supplement the modules of the model approach are directly related to the price increase that result from the introduction of a registration tax. Car owners will keep their old cars for longer, because the opportunity costs of buying a new car increase. The price increase will cause the number of cars in the car fleet to decline. Consequently, the sales of new cars will decrease more than scrapping to more than outweigh the decline in the latter, and hence to provide for the resulting overall decline of the car fleet. The price elasticity in the scrapping model is obtained by simulating the DETR model, cf. section 4.3.2. The elasticity varies across age and engine size and fuel type. The long run price elasticity for car ownership determines, as described above, the volume of car sales. This price elasticity is set to 0.6. Thus, a price increase by 10% will imply a reduction of the car fleet by 6% in the long run. The value of 0.6 is obtained from consulting several international studies for various countries as well as for Sweden.16 Long run effects As described, price increases will lead to decreases of the car fleet. In turn, this will have implications for among other things the emissions of the car fleet. Consequently, it is important to consider the concept of the "long run", as the price elasticities apply to the long termed perspective. In other words: “How long is the long run”? In this case there are two dimensions that determine the long run: • • Humans life cycle The life cycle of human beings The life cycle of cars Some car owners may recently have bought a house in a rural area with the intention of living there for many years, say 30-40 years. The introduction of a registration tax could in principle mean that they would have chosen to live in another area where they could do without a car had they known about the registration tax. However, they will not change their decision as regards settlement. 15 E.g. families with two cars may choose to sell car number two – and this will increase the mileage of car number one for that family. 16 For Sweden, see VTI Report 301 (1986): “Personbilsinnehavet i Sverige 1950-2010” by J.O Jansson, P. Cardebring and O. Junghard. 74 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars For this reason, it can be argued that the entire long run effect is fully phased in after 30-40 years.17 Cars life cycle However, a large proportion of the effect will depend on car owners that will choose not to be car owners when their current car is scrapped and they are faced with the new and higher tariffs. Hence, much of the long run effect is phased in the first years following a price increase (registration tax), cf. Figure 5.3. Figure 5.3 Phase in of the effect from an increase in car price as a result of a registration tax 4200000 4100000 4000000 3900000 Initial 3800000 RegTax 3700000 3600000 3500000 1995 2000 2005 2010 2015 2020 2025 2030 2035 In the analyses, we thus assume that half of the effect is phased in after 7 years, while the whole effect is phased in after 30 years. Sensitivity analysis All elasticities are subject to some uncertainty. Thorough sensitivity analyses serve to investigate the robustness of the calculations against these uncertainties. Method Basically, each sensitivity analysis consists of comparing the results of an alternative model run, which includes the revised elasticity, with the original results of the registration tax scenario. 17 The very last part is phased in after around 40 years but the vast majority is phased in after 30 years. 75 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The following sensitivity analyses constitute part of the study. • • • The price elasticity for the size of the car fleet lowered from –0.6 to –0.35 The phase in of the price elasticity of the size of the car fleet is speeded up so that half of the effect is introduced after 4 years instead of 7 years. The period of 30-40 years for total phase remains unaltered. The price elasticity of scrapping is reduced to 50% of the original values. As regards the former of the three, the selected alternative value of the price elasticity (-0.35 instead of 0.6) is motivated in a comparative analysis of a number of European countries. This analysis looked at the average car price relative to average income levels and compared this to the number of cars per inhabitant in the same countries. The analysis indicated the trend value is in the range of -0.35. This is illustrated in Figure 5.4. Figure 5.4 The relation between cars per 1000 inhabitants and car prices relative to income for 12 European countries 6.6 ln(cars per 1000 inhabitants) Issues covered 6.4 6.2 6 5.8 5.6 5.4 5.2 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 ln(car price/incom e) Note: The countries included in the analysis are Denmark, Finland, France, Netherlands, Ireland, Italy, Portugal, Spain, Sweden, germany, UK, Austria. The estimation is set up in a log-log framework. The results of all of the sensitivity analysis are discussed in section 7.3. 76 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 6 77 The tax scenarios The analysis centres around two main tax scenarios. The first scenario involves the introduction of a registration tax. The registration tax is partly value based and partly differentiated according to the level of the CO2 emissions. The second scenario assumes that the existing circulation tax is redesigned so that it is supplemented with a CO2 differentiated component in addition to its original features. Initial calculations Prior to finally deciding on the types of scenarios to analyse and their specific design, initial calculations were undertaken. These initial calculations served to provide quantitative inputs into the task of designing scenarios that fulfil two important pre-conditions, namely that there should be a reasonable expectation that the chosen design of scenario should provide some CO2 reductions while at the same time also being considered politically feasible. The initial calculations were done by means of a preliminary version of the model, which was ready at that time. Reference level A key feature of the applied method of design is the establishment of a reference level. The reference level is defined so that any cars with emissions above the reference is imposed with a CO2 tax payment whereas cars with emissions below obtain a CO2 rebate. It is a condition however, that the total value of the tax in question can never be negative. Thus, for the registration tax, the sum of the value based part and the CO2 rebate (if relevant) should be positive or zero. The following section explains the reference level and its derivation. 6.1 Two possible tax functions or combinations of the two Reference level Intrinsically, one can apply one of two tax functions or a combination of the two. The first function relates to the size of the car. The taxation level is consequently determined by the fuel efficiency for any given car size. This implies that such a function would not allow for downsizing. In other words, there is a strong focus on the fuel efficiency for a given car size. The reference will in this case be different for different car sizes, and it will increase with car size. The second function levies the CO2 tax independently of the size of the car. This function will have a fixed reference level, which is set at 199 grams of 78 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars CO2 per km. Contrary to the above, this function will allows for 100% downsizing, and will not take into account the fuel efficiency for a given car size. The first of the above functions was estimated by OLS regression. The regression took account of the fact that there are a different number of cars for each size. Consequently, different weights were attached as a reflection of this. The second tax function is simply given as a straight line, cf. Figure 6.1. The figure shows both of the functions as well as the actual emissions for various car sizes sold in Sweden. The scattered points in Figure 6.1 illustrate the latter. Figure 6.1 Two functional forms with different weight on fuel efficiency. 600 CO2 per km 500 400 300 200 100 0 0 2 4 6 8 10 12 Area (m2) Initial calculations In order to establish the reference level to be applied, a series of initial calculations were undertaken. These initial calculations were undertaken separately for the registration tax scenario and for the circulation tax scenario. A description of the most important results of these initial calculations is included below. 6.1.1 Registration tax scenario - initial calculations As mentioned above, the possible registration tax to be analysed should consist of two parts, viz.: • • A percentage of the value of the car (this part is independent of the car CO2 emissions) A CO2 differentiation element 79 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The CO2 differentiation element will be negative for cars with CO2 emissions below the reference emission (rebate) and positive for cars with emissions above the reference emission. Scope of the initial calculations The initial calculations allowed for 50% downsizing. Furthermore, the initial calculations looked at a large number of different tax scenarios. These different scenarios had different combinations of the following: • • • The CO2 differentiation level (The calculations looked at a range of 440 SEK – 2,640 SEK per g CO2/km above reference) The value based part of the registration tax (The calculations looked at a range of 10% - 50% of car price) The level of the fixed reference (The range of between 179 to 239 g CO2/km, (the average of the two is 199 g CO2/km)) The results from these calculations are shown in the figure below. Figure 6.2 Initial calculations, possible CO2 reductions across different scenarios CO2 reductions 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% -1.0% 0 20000 40000 60000 80000 100000 Average registration tax per car (SEK) These calculations clearly illustrate that the level of the registration tax sets an upper limit for what can be achieved in terms of CO2 reductions. This is because the level of the registration tax limits the level of CO2 differentiation. For example, the figure shows that a registration tax level of 40,000 SEK per car sets an upper limit of the CO2 reductions of approx. of 3.5%.18 18 Comparing to countries with high registration tax it could be noted that Finland has a registration tax of approx. 70,000 SEK, The Netherlands has a tax of 35,000 SEK and Denmark has a registration tax of 141,000 SEK. 80 Tax levels of 11,000 SEK and 35,000 SEK respectively Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Based on the above scan of the CO2 effect of different possible tax scenarios, two scenarios were chosen for further analysis • • Tax level of 11,000 SEK Tax level of 35,000 SEK The 11,000 SEK tax scenario consists of a value based part, which is 5% of the car price, and a CO2 differentiation element, which is of 440 SEK per gram of CO2/km. The reference level is defined to allow for 50% downsizing. The CO2 reductions from this scenario were calculated to be 1%, and the average registration tax was calculated to be 10,250 SEK per car. The other tax scenario assumes a CO2 differentiation of 1,760 SEK per grams CO2/km. Other features were similar to the first calculation. This scenario would result in the new taxes shown in the following figure. Figure 6.3 CO2 reductions from a tax level of 35,000 SEK Registration tax (SEK) 400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000 0 50 100 150 200 250 300 350 400 g CO2 per km 20% registration tax 20% registration tax plus CO2 differentiation The CO2 reduction from this scenario would amount to 3.1%, and the average registration tax would be approximately 35,000 SEK per car. More progressive CO2 differentiation The above calculations are based on a fixed tax (440 SEK and 1,760 SEK per gram of CO2 above the reference level respectively) per gram of CO2/km i.e. a linear CO2 differentiation. A more progressive taxation scheme where the price per gram of CO2 increases with the distance from the reference showed that it is not possible to gain more CO2 reductions for a given level of registration tax by applying a more progressive CO2 differentiation. Figure 6.4 shows the exponential function type that was applied to analyse this. 81 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 6.4 Progressive CO2 tax Tax (SEK) 600,000 500,000 400,000 300,000 200,000 100,000 -100,000 -150 -100 -50 0 50 100 150 200 250 g CO2 above (below) reference The calculations varied the shape of the curve (the progressiveness) in order to assess the additional gains to be achieved from enhancing the progressiveness. The calculations showed that a progressive CO2 differentiation would not increase the CO2 reductions to be achieved given the acceptable level of registration tax, cf. Figure 6.5. Figure 6.5 CO2 reductions when progressive CO2 differentiation is applied CO2 reductions 4.5% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% 0 20000 40000 60000 80000 100000 Average registration tax per car (SEK) The reason why a progressive tax would not provide significant further reductions is likely to lie in a comparison of the way that the two resulting tax functions are positioned with regard to the CO2 emissions. The linear relation is above the progressive relation for cars that are not extreme in their CO2 emissions, whereas the progressive one exceeds the linear one mainly in the extreme 82 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars cases. While the extreme cases causes substantial CO2 emissions at the individual level, i.e. for the individual car, the extreme cases cover a limited number of cars. The extreme cases would be cars with emissions that emit more than 150 grams above the reference level, cf. Figure 6.4). This argument is further supported by the fact that the price elasticity for very expensive cars, which are also often the cars high above the CO2 reference, is relatively low. This means that the increased tax burden for these cars has relatively less impact on purchase behaviours. Scope of the initial calculations 6.1.2 Circulation tax - initial calculations The initial calculations for the circulation tax scenario assume that the circulation tax is to consist of the existing tax plus a CO2 differentiation element. Furthermore, the circulation tax is restricted by the requirement that it should result in non-negative values. In other words, while a CO2 rebate is allowed, the total circulation tax should never become negative. This condition is similar to the one applied for the registration tax scenario. Figure 6.6 shows the CO2 reductions that would result from applying different levels of CO2 differentiation (ranging from 9 to 90 SEK per g CO2/km above the reference level). Similarly to the registration tax scenario, allowance is made for 50% downsizing. Figure 6.6 CO2 reductions from applying different levels of CO2 differentiation CO2 reductions 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% 0 20 40 60 80 100 SEK per g CO2 above reference Figure 6.7 illustrates the tax change that would result from applying a CO2 differentiation of 44 SEK per g CO2/km above the reference. 83 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 6.7 Circulation tax with and without CO2 differentiation Circulation tax (SEK) 12000 10000 8000 6000 4000 2000 0 0 100 200 300 400 500 g CO2 per km Existing circulation tax Existing plus CO2 differentiation Applying a CO2 differentiation of 44 SEK/gram above the reference level would reduce the CO2 emissions by 3%. 6.1.3 Downsizing and lower tax limit Inspired by some of the key features that underlie the approach taken in the Netherlands in their investigations of how to introduce CO2 differentiation into their tax scheme, the applied tax function attaches a weight of 25% to the horizontal tax function. This implies that 25% downsizing is permitted, and it leaves 75% weight to the tax function, which relates the CO2 emissions to the area of the car.19 19 The estimated OLS regression for petrol cars resulted in the following 2. order polynomial: CO2 per km = 199.42 – Area*31.27 + (Area)2 *4.04. For diesel cars the regression analysis resulted in the following 2. order polynomial: CO2 per km = 968.46 – Area*245.12 + (Area)2 *18.13. 84 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 6.8 Weighted functional form applied, Petrol Cars 600 CO2 per km 500 400 300 200 100 0 0 2 4 6 8 10 12 Area (m2) A separate but similar form is applied for diesel cars, were 25% downsizing also is allowed. As mentioned the taxes are not allowed to become negative (in which case they would be subsidies). This condition applies to all cars, also those with extremely low emissions. However, the CO2 differentiation part of the tax may become negative (assume the form of a rebate) and in that way reduce the nonCO2 differentiated part of the tax, cf. Figure 6.9. Figure 6.9 Truncated tax structure (the “no-subsidy”-characteristic) Tax Area 85 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 6.2 The two main scenarios The section describes the tax functions that were ultimately decided upon for the two tax scenarios to be analysed. 6.2.1 Reference level calculation The reference CO2 level is calculated according to the following formula: CO2Reference = ADS * C0 + (1-ADS) * (C1 + C2 * A + C3 * A2) Where ADS is a parameter to control the level of downsizing. The value of ADS is set to 0.25. A is the area of the car (length * width in meter) The parameters for petrol cars and diesel cars are shown below. Table 6.1 Parameters in CO2 reference calculations Parameter C0 C1 C2 C3 Value for petrol cars 199.7 199.4185 -31.2686 4.04179 Value for diesel cars 174.7 968.4628 -245.127 18.1254 Based on the CO2 reference and the actual CO2 emissions of the car the CO2 differentiation element is calculated as follows: CO2 differentiation tax = (Actual CO2 per km - CO2Reference) * SEK per g CO2 6.2.2 Registration tax scenario The initial calculations showed that in order to avoid upsizing of the car fleet it was necessary to allow for a certain degree of downsizing. Hence, it was decided to allow for 25% downsizing as mentioned in the preceding paragraph. Upsizing occurred in the initial calculations because of the fact that all cars were to pay a tax, including the small and energy effective ones. The latter could be imposed a tax of zero, but not less than that. Consequently, the small and fuel efficient cars actually experienced the highest price increase in relative terms. The design of the registration tax in this scenario consists of two elements as shown below: 86 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 6.2 Design of the registration tax in the registration tax scenario Value based element 10% of the value of the car (corresponds to an average taxation level of 20,000 SEK/car CO2 differentiated element 880 SEK per gram of CO2 above the reference level and negative below the level. The negative value of the CO2 tax can never more than outweigh the value-based part. In other words, the registration tax can never be negative. The registration tax is levied separately for petrol and diesel cars (different reference levels are applied). 6.2.3 Circulation tax The circulation tax to be applied in the circulation tax scenario is defined as the current tax plus a CO2 differentiated element. The features of the existing circulation tax are shown in Table 5.1. In addition to this, the scenario assumes a CO2 differentiated element of 44 SEK per g CO2 above the reference level, and counted as a negative value (rebate) below the reference level. Similarly to the registration tax, the CO2 tax may thus reduce the non-CO2-differentiated part of the circulation tax; i.e. it may work as a subsidy and the average total tax burden and average total tax revenue remains constant. Also in this case, a separate circulation tax is levied for diesel and petrol cars. 87 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 7 Results of the scenarios 7.1 Base scenario and the registration tax scenario The main features of the registration tax in this scenario are shown in Table 6.2. 7.1.1 New car sales The results of the model calculations for sales of new cars are provided in Table 7.1. Table 7.1 Results for new car registrations, Base and Registration tax scenario Outputs Base calculation New registration tax 198.1 195.3 96,237 115,227 3.70 3.66 6.5% 6.5% 30.5% 29.2% - 20,161 1,566 1,558 145,825 143,854 - 20,161 Circulation tax (whole life time) 24,780 24,543 Fuel tax (whole life time) 71,456 70,523 Average CO2 emissions (g per km) Average lifetime tax revenue (SEK per car) Average size Average Diesel share Average Swedish car makes Average registration tax (SEK per car) Average Circulation tax (SEK per car per year) Average dealers price, SEK excl. VAT Average (lifetime) tax consists of (SEK) Registration tax Base and scenario calculations The table shows the results both from the base scenario and from the registration tax scenario. The base scenario models the results under the current tax structure, i.e. with no registration tax and with the existing circulation tax. 88 The introduction of a registration tax is seen to reduce the average CO2 emissions from new cars by 1.4%, which is equivalent to 2.8 g CO2/km/car. Part of this reduction comes about through a smaller size of the new cars. The average size decreases from 3.70 to 3.66. Related to this downsizing effect, the share of Swedish cars in the sales of new cars decreases from 30.5% to 29.2%. Swedish cars tend to be larger than the average car. A decline in the total number of new cars sold (not shown in Table 7.1, but in Table 7.4) further accentuates this drop. The average registration tax, which is set to 20,000 SEK, results in an increase in the average life time tax revenue per car of around 20%. 7.1.2 Scrapping The age distribution of the Swedish car fleet is shown in Figure 7.1. The figure illustrates how business cycles among other factors have effected the age distribution. Consequently, the car fleet has a bulk of 1-3 year old cars and of 12-14 year old cars. Figure 7.1 Age distribution of the Swedish car fleet, September 2001 24 22 20 18 16 14 12 10 8 6 4 2 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 0 Frequency Results Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Age all se not se Note: “se” signifies “Swedish cars” This pattern that characterises the Swedish car fleet’s age distribution is also reflected in the level of scrapping over time, cf. Figure 7.2. This figure also shows that the level of scrapping is higher in the base scenario than in the registration tax scenario. The reason is that the registration tax causes the number of cars to decline. This is further discussed in the next section. 89 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 7.2 Scrapping over time, Implementation year 2001 Number of scrapped cars 300000 250000 200000 Reg tax 150000 Base 100000 50000 97 89 81 73 65 57 49 41 33 25 17 9 1 0 Years from implementation Results from base calculation and from scenario calculation 7.1.3 Total car fleet The results for the total car fleet are given in Table 7.2, Table 7.3, and Table 7.4. The tables show the results from the base scenario, the registration tax scenario, and the difference between the two scenarios. The tables present the results over the different time spans, viz. 1, 5, 10 and 20 years after the introduction of a registration tax, which is first introduced into the calculations in 2001. In Table 7.2 the average age can be seen to fluctuate over time. These fluctuations are caused by the age distribution of the Swedish car fleet, cf. Figure 7.1. Average emissions The most striking result when comparing the base scenario and the registration tax scenario (i.e. Table 7.4) is that the average CO2 emissions/car are constant over time. Although the average CO2 emissions for new cars decrease by 2.8 gram of CO2 per km (cf. Table 7.1), this effect is counteracted by an increase in the average age of the car fleet of about half a year. The latter involves an increase in the average CO2 emissions and consequently, the net effect on average CO2 emissions is virtually zero. Therefore, there is no rebound effect either. Total emissions However, the total number of cars in the car fleet declines as a result of the price increase caused by the introduction of the registration tax. Therefore, the total CO2 emissions from the car fleet also declines. Other effects From the tables, it can also be seen that the proportion of Swedish cars (i.e. Volvo and Saab) in the car fleet is reduced from 30.2% to 29.3% (year 20). The drop in the total car fleet amplifies the implications of this decline. The cars also tend to be smaller than in the base scenario. This can be seen to apply both to the sales of new cars per se, and also to the total car fleet. With regard to the latter, both the average length and the average weight of the cars decrease. The longer the time span the larger this decrease. 90 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The registration tax will lead to a decline in the number of new cars sold. The size of this effect varies from a drop in the order of 7% and up to 20%. The variation is caused by the age distribution of the Swedish car fleet, which is shaped by business cycles. Table 7.2 Base scenario, total car fleet (1, 5, 10 and 20 years) BASE Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,153,155 205,154 235 11,994,305 8.9 451 1,340 1,365,726 32.9% 198,069 4.8% Time horizon (in years) 5 10 4,153,155 4,153,156 226,536 195,659 221 203 11,364,228 10,458,465 9.0 9.4 449 447 1,375 1,402 1,289,483 1,248,816 31.0% 30.1% 236,867 276,989 5.7% 6.7% 20 4,153,152 240,064 162 8,325,037 9.2 442 1,408 1,255,375 30.2% 270,625 6.5% Table 7.3 Registration tax scenario results, total car fleet (1, 5, 10 and 20 years) Reg. tax scenario Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,128,309 163,167 235 11,939,719 9.1 451 1,338 1,357,806 32.9% 195,844 4.7% Time horizon (in years) 5 10 4,038,866 3,969,853 198,023 182,333 223 204 11,109,933 10,046,108 9.5 10.0 449 447 1,368 1,393 1,248,440 1,177,506 30.9% 29.7% 226,495 261,925 5.6% 6.6% 20 3,900,058 213,160 161 7,811,640 9.7 441 1,397 1,142,127 29.3% 252,685 6.5% Table 7.4 Change from registration tax, total car fleet (1, 5, 10 and 20 years) Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 7.2 1 -24,846 -20.47% 0.37 -43,669 0.15 0.01 -1.82 -7,920 0.01% -2,225 -0.03% Time horizon (in years) 5 10 -114,289 -183,303 -12.59% -6.81% 1.27 1.07 -203,436 -329,885 0.53 0.60 -0.07 -0.28 -6.80 -8.73 -41,043 -71,310 -0.14% -0.41% -10,372 -15,064 -0.10% -0.07% 20 -253,094 -11.21% -0.10 -410,717 0.53 -0.92 -10.79 -113,248 -0.94% -17,940 -0.04% Base scenario and circulation tax scenario The design of the circulation tax in the circulation tax scenario is described in section 6.2.3. 91 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Results 7.2.1 New car sales The average CO2 emissions per car for new cars decrease by almost 3% as a result of the redesign of the circulation tax. There are two causes for this effect. First, the proportion of diesel cars increases as a result of the tax change. However, the increase is quite small. Second, the average size of the new cars is reduced, and this is the major reason. As a consequence of these changes the proportion of Swedish cars in the total sales of new cars drop from 30.5% to 29.5%. The average circulation tax decreases. This is a result of the redesign of the tax and the behavioural changes this redesign brings about. Hence, the reduction in the average CO2 emissions comes solely from the introduction of a CO2 differentiated element into the circulation tax, and not from an increase in the tax level. Table 7.5 Results for new car registrations, Base and Circulation tax scenario Outputs Average CO2 emissions (g per km) Base calculation New circulation tax 198.1 192.8 96,237 92,497 3.70 3.65 6.5% 6.6% 30.5% 29.5% - - 1,566 1,443 145,825 142,120 - - Circulation tax (whole life time) 24,780 22,766 Fuel tax (whole life time) 71,456 69,731 Average lifetime tax revenue (SEK per car) Average size Average Diesel share Average Swedish car makes Average registration tax (SEK per car) Average Circulation tax (SEK per car per year) Average dealers price, SEK excl. VAT Average (lifetime) tax consists of (SEK) Registration tax Results 7.2.2 Total car fleet The effect on the total car fleet from the introduction of a CO2 differentiated element into the existing circulation tax is a drop in the total CO2 emissions from the car fleet. The drop is in the order of 0.5% in the first year, and it increases to around 2% in year 20, cf. Table 7.2, Table 7.6 and Table 7.7.20 This reduction stems from a decrease in the average CO2 emissions – an effect, 20 60,000 out of 12 mio in year 1 and 160,000 out of 8,3 mio in year 20. 92 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars which is stronger the longer the time span. This is because this effect is provided solely through the effect from the tax on the sales of new cars. No effect on scrapping and car fleet The absolute level of the annual circulation tax remains unaltered in this scenario. Therefore, there is no change in the number of cars, as there is no general price increase. The same goes for the average age of cars. The proportion of diesel cars in the car fleet remains virtually unaffected while the proportion of Swedish cars is reduced from 30.5% to 29.4% (in year 20). Table 7.6 Circulation tax scenario results, total car fleet (1, 5, 10 and 20 years) Circ. tax scenario Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,153,156 205,154 234 11,981,403 8.9 451 1,339 1,363,794 32.8% 198,220 4.8% Time horizon (in years) 5 10 4,153,155 4,153,156 226,536 195,659 220 200 11,297,796 10,335,992 9.0 9.4 449 446 1,372 1,396 1,279,171 1,228,992 30.8% 29.6% 237,672 278,535 5.7% 6.7% 20 4,153,154 240,064 158 8,126,837 9.2 441 1,396 1,219,019 29.4% 273,461 6.6% Table 7.7 Change from circulation tax, total car fleet (1, 5, 10 and 20 years) Change 1 Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 7.3 Types of results 1 0.00% -0.26 -10,322 0.00 -0.06 -0.60 -1,932 -0.05% 151 0.00% Time horizon (in years) 5 10 0 0 0.00% 0.00% -1.31 -2.41 -53,146 -97,978 0.00 0.00 -0.34 -0.65 -3.21 -6.17 -10,312 -19,824 -0.25% -0.48% 805 1,547 0.02% 0.04% 20 2 0.00% -3.90 -158,560 0.00 -1.20 -11.31 -36,356 -0.88% 2,836 0.07% Registration tax scenario - Sensitivity Analysis This section investigates the implications for the above results of changing a number of underlying key assumptions. Each of the analyses presented in this section reports the registration tax scenario results that would apply under the changed assumptions in terms of comparing these results to the base scenario results. Annex B provides he specific results from the registration tax scenario under the changed assumptions. Sensitivity analyses are only carried out for the registration tax scenario. The registration tax scenario has an effect on scrapping, on the size of the car fleet and the number of new cars sold, whereas the circulation tax only has implications for the composition of the sales and the fleet, and not for the absolute values. Sensitivity analyses are therefore mainly of relevance to the registration tax scenario. 93 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Set-up 7.3.1 CO2 differentiation of 1,760 SEK per g CO2 Here, the tax per gram of CO2 is assumed to be 1,760 SEK instead of the original 880 SEK. Thus, the sensitivity analyses here involve a doubling of the CO2 tax, i.e. the CO2 differentiation. The value-based part of the tax is maintained at the original 10%. The latter implies that the average tax level remains (virtually) unaffected, and consequently, the number of cars that result from this alternative formulation of the registration tax scenario remains quite similar to the number of cars that resulted from the original registration tax scenario. Table 7.8 Change from base scenario when CO2 differentiation is 1,760 SEK / g CO2 Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Results 1 -25,206 -20.78% 0.35 -45,135 0.15 0.01 -1.86 -8,171 0.00% -2,248 -0.03% Time horizon (in years) 5 10 -116,057 -186,263 -12.80% -6.94% 1.17 0.86 -211,090 -343,864 0.54 0.61 -0.08 -0.30 -7.01 -9.09 -42,449 -74,015 -0.16% -0.45% -10,483 -15,205 -0.10% -0.07% 20 -257,270 -11.39% -0.48 -432,078 0.54 -0.96 -11.39 -118,151 -1.04% -18,040 -0.03% The doubling of the CO2 differentiation has only very little impact on the average CO2 emissions compared to the results of the original registration tax scenario. The reason for this insensitivity is the requirement that the total registration tax to be paid by an individual car purchaser can never be negative. Consequently, the doubling of the CO2 differentiation implies that the lower level becomes binding more often than before (i.e. the cases become more frequent, where the actual calculation of the tax would result in a negative one, but this requirement means that it is set at zero instead).21 Consequently, this sensitivity analysis shows that either an increase in the tax burden (i.e. a value-based tax of more than 10%) or a loosening of the constraint that the registration tax may never become negative would be required to provide more CO2 reductions from the registration tax. Motivation 7.3.2 Level of registration tax The preceding paragraph illustrated the relatively little effect that the price of CO2 emissions per gram (the unit tax) has on the total emissions. This (lack of) effect is further illustrated in Figure 7.3. 21 Increasing the price per g CO2 to 2,640 (200% increase from main scenario) does not have more impact on average CO2 emissions than increasing by 100%. 94 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Reduction in total CO2 emissions Figure 7.3 Reduction in total CO2 emissions: 880 SEK (main scenario) per g CO2 and “extra” effect from 1,760 SEK per g CO2 500000 450000 400000 350000 300000 250000 200000 150000 100000 50000 0 1 5 10 20 Time horizon 880 SEK Extra reduction for 1,760 SEK Note: For a time horizon of 1 year the effect from 1,760 SEK is lower (because the immediate age effect is higher when the price is higher). This is not shown in the figure. The reason for the limited effect of increasing the price of emissions (the unit tax) is similar to the reason for the limited effect of the original registration tax scenario. There is namely, in the case of the registration tax a trade off between the effects on average CO2 emissions from the CO2 differentiation (via its impact on new car sales) on the one hand, and the effect on the average age of the car fleet on the other hand. The latter results from the price increase that is caused by the introduction of a registration tax. Set-up To explore this idea of this trade off further, the level of the value-based part of the registration tax is varied in this sensitivity analysis. The original scenario set the level at 10% of the value of the car. In addition to this level, these alternative scenario calculations set the level at three alternative levels, viz.: 5%, 15% and 20% of the value of the car. For each of these levels, including the original level of 10%, the total effect on CO2 emissions is decomposed into three effects. These three effects, which stem from • • • Results The CO2 differentiation effect on the sales of new cars The increase in the average age of the car fleet as a result of the more expensive cars The decrease in the size of the car fleet as a result of the more expensive cars The results are shown for the 20-year time span in Table 7.9. As can be seen, increasing the tax level leads to the expected further reductions in CO2 emissions from new cars. This is due to the enhanced incentive to buy more fuel efficient cars relative to a given size. However, the increasing the tax level also involves increasing the price level. Consequently, the incentive to 95 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars keep the old cars for longer increases resulting in an increase in the average age of the cars in the fleet. This effect actually crowds out the effect from the higher tax level. Furthermore, the increase of the car prices leads to a smaller car fleet, as cars become more expensive than other consumer goods. The effect on the car fleet is however independent from the fact that the above effects work in opposite directions. Table 7.9 Decomposition of the total effect on CO2 emissions, 20 years horizon 20 years horizon CO2 diff. from new cars Age Number of cars Total effect Motivation 5% -0.94% 0.67% -2.89% -3.15% 10% -1.14% 1.12% -4.40% -4.41% 15% -1.25% 1.74% -6.09% -5.60% 20% -1.25% 2.57% -7.88% -6.56% 7.3.3 Different levels of downsizing The original registration tax scenario set the allowed level of downsizing at 25%. Alternating this level will have no implications for the number of cars. This number is determined solely by the price elasticity for the size of the car fleet. Still, the level of downsizing has important implications for the types of the new cars that will be sold. If allowance is made for a high level of downsizing, car buyers will be given a stronger incentive to buy smaller cars. Hence, the level of the CO2 emissions will be reduced more than in the original scenario. Figure 7.4 serves to illustrate this. (The details are presented in Annex B, Table 11.7). The figure illustrates that the original level of 25% downsizing captures a significant share of the potential effect. Figure 7.4 shows a steep curve in the range of downsizing from 0% to 25%, and a less steep curve for higher levels of downsizing. The effect from the level of downsizing on the proportion of Swedish cars is fairly limited, cf. Figure 7.5. This figure shows the effect from the registration tax on the proportion of Swedish cars both for the circulation tax and the registration tax. 96 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 7.4 The relationship between average CO2 emission reductions from new car registrations and the level of downsizing allowed, Registration tax 2.0% 1.8% CO2 reductions 1.6% 1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% 0% 20% 40% 60% 80% 100% 120% Allow downsizing Figure 7.5 Effects of downsizing on the share of Swedish manufactured cars - both from a registration tax and a circulation tax Share of Swedish cars 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 0% 25% 50% 75% 100% Level of downsizing Reg. tax Results Circ. tax As can be seen, for the registration tax scenario, the level of downsizing has very little implications for the proportion of Swedish cars, whereas the effect is a little more pronounced in the case of a circulation tax. The reason for this is that the effect from circulation tax comes solely from new cars, and this is also where the effect from downsizing adheres. The effect from the registration tax on the other hand is to a large extent attributable to the impact it has on the number of cars in the fleet. 97 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Motivation and setup Results 7.3.4 Petrol and diesel merged in tax function estimation The original scenario estimated the second order polynomial tax functions separately for petrol and diesel cars respectively resulting in two different functions. In order to investigate the importance of this approach of separating the two, an alternative estimation has been carried out where only one polynomial is estimated for the entire car fleet. In other words, here petrol and diesel cars are merged in the estimation. Compared to the original scenario results, (Table 7.4), this alternative calculation provides virtually identical results. This is in line with what could be expected. The two original tax functions proved to be quite similar for the relevant intervals (in terms of area of car). However, merging diesel and petrol cars into one tax function results in a minor increase in the proportion of diesel cars; an increase of 0.5% in the long run (20 years and more). Table 7.10 Change from base scenario when petrol and diesel cars are merged in the estimation of the tax function Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Set-up and results 1 -25,381 -20.93% 0.37 -43,692 0.15 0.01 -1.84 -8,141 0.00% -1,299 0.00% Time horizon (in years) 5 10 -116,914 -187,696 -12.91% -7.01% 1.28 1.06 -203,246 -328,888 0.55 0.61 -0.07 -0.28 -6.88 -8.80 -42,279 -73,615 -0.15% -0.43% -5,133 -4,412 0.04% 0.20% 20 -259,287 -11.48% -0.16 -407,557 0.55 -0.93 -10.76 -117,196 -1.00% 2,715 0.50% 7.3.5 The effect from the price increase on scrapping is halved The smaller the price elasticity on scrapping, the smaller is the effect from the assumed tax change on the average age of the cars in the fleet. This is illustrated in Table 7.11, which shows the results from the registration tax scenario assuming that the elasticity of scrapping is halved compared to the original calculation. As can be seen, the average age is only reduced by 0.23 years compared to around 0.5 years in the original registration tax scenario. As a direct consequence of this, the average CO2 emissions also decline in this calculation, although the order-of-magnitude of the decline is limited. The reason why the scenario however results in a decline in this calculation is that the decrease in the emissions from the new cars now dominates the effect from cars becoming older. The sales of new car decline less than originally because the scrap rate is reduced less than in the original calculation. The rest of the effects are almost unchanged, and the overall importance of the scrap price elasticity is generally limited, albeit not entirely negligible. 98 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 7.11 Change from base scenario when scrap price elasticity is halved Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Set-up 1 -25,977 -16.84% 0.26 -50,415 0.10 0.01 -1.32 -8,863 -0.01% -2,002 -0.02% Time horizon (in years) 5 10 -117,241 -185,680 -10.64% -6.01% 0.81 0.41 -226,700 -359,206 0.36 0.38 -0.05 -0.28 -4.87 -6.43 -43,153 -72,852 -0.17% -0.43% -9,095 -13,597 -0.06% -0.03% 20 -254,464 -9.32% -1.14 -452,535 0.23 -0.95 -9.14 -111,468 -0.89% -17,233 -0.02% 7.3.6 Phase in of price effect speeded up Originally, it was assumed that half of the price elasticity of the size of the car fleet was phased in during the first 7 years. Alternatively, it could be assumed to take only 4 years before half of the effect was phased in. Such a shorter period of time for the first half of the elasticity can be expected to lead to larger short-termed effects than in the original scenario. Indeed, this is also what happens. In fact, in the first year the effects almost doubled. The longer the time span we look at, the less pronounced is the effect from changing this assumption of the original scenario. This is because the total effect is still introduced over a 30-40 year time span. Table 7.12 Change from base scenario when phase-in of price elasticity is speeded up Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Results 1 -46,408 -30.98% 0.59 -84,706 0.19 0.06 -2.14 -14,266 0.02% -3,626 -0.03% Time horizon (in years) 5 10 -184,516 -251,460 -15.40% -5.50% 1.92 1.36 -333,112 -456,261 0.66 0.63 0.08 -0.17 -7.35 -8.84 -61,711 -91,369 -0.11% -0.40% -14,936 -19,493 -0.11% -0.07% 20 -281,493 -12.86% -0.31 -464,491 0.46 -0.91 -10.78 -121,442 -0.94% -19,818 -0.04% This result shows that the “phase-in-speed” of the elasticity can have implications for the socio-economic consequences, since these may be more important in the short term. The key result of the registration tax scenario, namely that there is no effect on the average CO2 emissions is not affected by changing the main scenario in this manner, cf. Table 7.12. Set-up 7.3.7 Price elasticity of –0.35 The registration tax scenario assumes a price elasticity of –0.6. This means that an increase of the price of cars by 10% will lead to a decline in the size of the 99 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars car fleet will by 6%. When this elasticity is changed to –0.35, the decline in the number of cars will be reduced compared to the original results. In fact, the drop in the number of cars is almost halved. Results Over a 20-year time span the number of cars will decline by around 147,000 using the elasticity of -0.35, which compares to 253,000 when the elasticity was –0.6. This directly affects the total CO2 emissions. They drop proportionately less. In this respect the size of this elasticity has very high importance for the models result on the total car fleet. The effects on the other variables are much smaller. The alternative price elasticity leads to a small decrease in the average age compared to the original scenario as a result of more new cars being bought under the alternative price elasticity. Table 7.13 Change from base scenario when price elasticity is –0.35 Change Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 -14,708 -15.52% 0.26 -24,377 0.12 -0.02 -1.66 -4,937 0.00% -1,566 -0.02% Time horizon (in years) 5 10 -67,058 -106,948 -8.89% -4.40% 0.82 0.57 -117,006 -195,412 0.45 0.51 -0.17 -0.39 -6.43 -8.61 -27,142 -48,838 -0.15% -0.41% -7,303 -10,102 -0.09% -0.07% 20 -147,427 -8.27% -0.17 -244,428 0.52 -0.94 -10.75 -82,276 -0.94% -11,042 -0.04% 100 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 101 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 8 Socio-economic implications 8.1 Private households In order to assess the implication of the tax changes for private households the households has been segmented according to family type, if they have children or not and according to urbanisation. The following table shows the families segmented in eight groups. Table 8.1 Number of families Family type Single without children Urban Rural 1,572,152 169,040 Single with children 215,253 28,998 Couple without children 628,117 83,541 Couple with children 597,668 88,037 Rural has been defined as families living in the 25% municipalities with lowest population density. The average income varies considerable between different income groups. Generally couples have higher income relative to single adult families. The main reason for this naturally that the income for couples most often consists of two incomes. Furthermore, it is worth noting that families living in rural areas in general have lower income relative to people living in urban areas. Finally, families with children have higher income relative to families without children. Table 8.2 Average income (disposal income, 2000) Family type Urban Rural Single without children 149,773 141,608 Single with children 201,884 197,731 Couple without children 333,152 309,406 Couple with children 397,009 369,054 102 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.3 Average car price initially Family type Urban Rural Single without children 148,000 152,400 Single with children 158,800 152,900 Couple without children 154,700 179,000 Couple with children 158,200 190,700 Families in rural areas on average buy more expensive cars relative to families living in urban areas. Since the average income is lower in rural areas this means that families living in rural areas are using significantly higher share of the income for car purchase. Finally, people in rural areas buy cars with higher CO2 emissions per km relative to people living in urban areas. Table 8.4 Average CO2 emissions per km for new cars purchased by the families Family type Urban Rural Single without children 186.1 192.6 Single with children 189.1 193.1 Couple without children 193.5 203.4 Couple with children 194.4 205.5 8.1.1 Implications from the registration tax scenario The following table shows how the tax burden of the new registration tax is distributed among the families. Families in urban areas will experience a higher tax burden relative to people living in urban areas. Table 8.5 Average registration tax in tax scenario Family type Urban Rural Single without children 14,467 17,380 Single with children 16,183 18,207 Couple without children 16,210 21,666 Couple with children 17,107 24,473 103 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.6 Tax share of average income Family type Urban Single without children Rural 10% 12% Single with children 8% 9% Couple without children 5% 7% Couple with children 4% 7% The tax burden shows how the households are affected by the tax changes in financial terms. This measure of the impact however disregards the effects that occur, because households adjust to the new taxes. Tax increase effect As already mentioned the registration tax consists of two elements. One element is the value-based tax, which adds 10% to the car price. The effect from this increase is close to proportional with the car price and will affect households who buy expensive cars relatively more than to people who buy cheaper cars. The welfare effect from the increase in the tax level is equal to the general welfare loss (the triangle in standard welfare economics) plus the taxes that people pay after they have adjusted to the new tax level. This loss is calculated for the households in the table below. Table 8.7 Welfare loss due to registration tax level increase. Single without children Single with children Couple without children Couple with children Urban 17,939 20,067 20,100 21,213 Rural 21,551 22,577 26,865 30,346 The welfare losses shown in the above table clearly shows that an increase (from zero) in the registration tax level will result in substantial re-distributions of welfare in society, as car purchasers would suffer a substantial loss in welfare. This welfare loss will be in the order of 10-15% of the income. One should note though that the welfare loss only occurs in the year of the purchase of a new car. CO2 differentiation effect The other element of the registration tax is the CO2 dependent element. Due to this tax element, households who buy cars with small CO2 emissions will benefit from the CO2 differentiation element. On the other hand, households who buy cars with high CO2 emissions will experience a loss in welfare. The following figure illustrates the situation for households who buy cars with high CO2 emissions. 104 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 8.1 Changes in consumer surplus due to tax increase Car Price C Demand curve B P1 P0 Supply curve D A Q1 Q0 Number of cars Initially these people face the price P0 and demand the amount of Q0 cars. Their consumer surplus amounts to the area of the triangle P0 A C. The introduction of a CO2 differentiation into the registration tax will imply that these people will experience a price increase. Having adjusting their demand to the new price they will demand the quantity of Q1 cars to the price P1. Their new consumer surplus has now been reduced to the area of the triangle P1 B C. Consequently, these people will suffer a loss in welfare corresponding to the area P0 A B P1. One should note that for society as a whole, there will only be a welfare loss corresponding to the area A B D, since the area P0 D B P1 is collected as taxes and could be used (redistributed) to increase welfare elsewhere in the economy. So for the society as a whole there will only be a welfare loss corresponding to the area A B D. This area is the welfare loss that has been used in the cost benefit analysis to reflect the loss of welfare due to CO2 differentiation. For the households buying cars with low CO2 emissions the situation is the opposite. These households face a decrease in the price and will therefore demand more (small) cars. This situation is illustrated in the figure below. 105 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 8.2 Changes in consumer surplus due to tax increase Car Price C Demand curve B P0 P1 E D A Q0 Q1 Supply curve Number of cars Initially these people face the price P0 and demand the amount of Q0 cars. Their consumer surplus amounts to the area of the triangle P0 B C. The introduction of the CO2 differentiation element into the registration tax will imply that these people experience a price decrease. Having adjusting their demand to the new price they will demand the quantity of Q1 cars to the price P1 (= P0 - CO2 tax). Their new consumer surplus will thus have increased to the area of the triangle P1 A C. Consequently, these people experience a gain in welfare corresponding to the area P1 A B P0. One should note though that for society as a whole there will be a welfare loss, which corresponds to the area A B D. The area P0 A E P1 is paid in tax rebates and needs therefore to be collected elsewhere in the economy where it will consequently reduce welfare. So for the society as a whole there will be a welfare loss corresponding to the area A E B. which is equal to the area A B D. This area is the welfare loss that has been used in the cost benefit analysis to reflect the loss of welfare due to CO2 differentiation. The following table shows the welfare impact for the consumers. The net welfare effect is calculated as the welfare gain for the car purchasers who buy cars with low CO2 emissions minus the loss of welfare for the car purchasers who buy cars with high CO2 emissions. Note that this table only refers to the CO2 differentiation changes. The changes to the tax level are not considered here. 106 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.8 Welfare effects from CO2 differentiation of the registration tax (SEK per car, total lifetime) Family type Urban Rural Single without children -3618 -4896 Single with children -5919 -6249 Couple without children -4813 -5249 Couple with children -7687 -8813 The average effect for private households amounts to a loss in welfare of 5,100 SEK. By comparison, the net welfare effect related to company cars is calculated to be a loss of 6,100 SEK per car. Thus, the new car purchasers will suffer a loss from the CO2 differentiation. The loss is approximately 15% higher for families in rural areas compared to families in urban areas and about 50% higher for households with children compared to households without children. 8.1.2 Implications for the annual circulation tax scenario The following tables show the tax burden from the annual circulation tax before and after the introduction of a CO2 differentiated element into it. Families in rural areas are seen to carry a higher tax burden relative to families in urban areas. Table 8.9 Average circulation tax initially Family type Urban Rural Single without children 1,271 1,374 Single with children 1,578 1,666 Couple without children 1,345 1,588 Couple with children 1,522 2,116 Table 8.10 Annual circulation tax after introduction of a CO2 differentiation element Family type Urban Rural Single without children 1,129 1,354 Single with children 1,504 1,628 Couple without children 1,209 1,637 Couple with children 1,405 2,230 107 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.11 Average difference before and after, anually Family type Urban Single without children Rural -142 -20 -74 -37 Couple without children -136 49 Couple with children -117 114 Single with children The tax burden is only one aspect of the effect on car purchasers from introducing tax changes. The other aspect relates to the welfare change that families will experience due to the CO2 differentiation. The following table shows this change in total welfare. It is measured in a way similar to the one applied for the registration tax. Table 8.12 Welfare effects from CO2 differentiation of the circulation tax (SEK per car, total lifetime) Family type Urban Rural Single without children -3275 -4212 Single with children -4876 -4773 Couple without children -4037 -4404 Couple with children -5539 -5599 8.1.3 Effects on private households from the tax scenarios The overall conclusion from the above calculations is that both tax scenarios will lead to a substantial redistribution of welfare from the car purchasers and to the rest of society. This effect is most predominant for the registration tax scenario, because it includes a general tax increase. This increase leads to a decline in the total car fleet. Furthermore, it should be noted that the heaviest tax burden increase fall on the families in rural areas. This is because families in rural areas purchase cars with higher emissions compared to families in urban areas. Consequently, families in rural areas also suffer a higher welfare loss relative to families in urban areas. This observation also applies to both tax scenarios. The reason is that families in rural areas tend, on average, to buy bigger cars (and also more CO2 emitting cars). Furthermore, they are more dependent on a car, which is reflected in less sensitivity to prices compared to the urban population. Families with children will suffer a higher loss relative to families without children. Again, the reason is that these families purchase cars with higher CO2 emissions relative to families without children. 108 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 8.2 Employment effects Car manufacturing in the Swedish economy The car manufacturing industry is an important contributor to Swedish production and exports. The sector contributes 4%22 of the total production value in Sweden, and accounts for about 2.3%23 of wages. The latter can indicate that the industry is relatively more capital intensive than the average Swedish production and/or that the average level of salaries is below the overall average Swedish level. The industry accounts for about 11.5% of total Swedish export24 and it is estimated that more than 80% of the passenger cars produced in Sweden are exported. The industry employs about 70,00025 people, which corresponds to about 2% of the total employment and about 10% of employment in industry26. Relevant outputs from the model calculations The table below illustrates the key outputs of the model calculations that are of relevance in assessing the socio-economic impacts from the tax scenarios related to the car manufacturing industry. The table clearly shows that the registration tax scenario will have the largest impact on car sales and on the car fleet, while the impact from the circulation tax is lower. Assessment of employment effect In order to assess the employment effect from the above changes in sales of Swedish cars, it is necessary to have an indicator of the number of employees per car. Ideally, one would aim to estimate this at the margin, but the available data allows merely for an estimate of the average. Based on the data and assumptions outlined in Table 8.14, it can be calculated that the production of a car in Sweden requires almost 0.1 employee. Assuming that all employees are full time employed, it corresponds to saying that 1 employee produces 10 cars per year27. It should be noted that the figures contained in the table relate to 1999, as this is the most recent year for which most of the information could be obtained in a consistent manner. Correcting for production in the Netherlands The Netherlands share of production is used to in the assessment of the extent to which a decline in the sales of Swedish brands (shown in the above table) affects the Swedish employment. The model's calculations of the effect on sales of Swedish cars do not distinguish between those manufactured in the Netherlands and those produced in Sweden. Consequently, it is necessary to have an estimate of the share made up by Netherlands manufacture in order not to overestimate the employment effect in Sweden. 22 Input-output table 1998 Input-output table 1998 24 Input-output table 1998 25 SCB online 26 Statistisk Årsbok för Sverige 2001 27 ((385044/(385044+108776*3)*70966/385044) 23 109 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.13: Number of cars sold and number of cars in the car fleet. Base situation and the tax scenarios. Cars New cars Year 1 Car fleet 5 10 20 1 5 10 20 Base 205,154 226,536 195,659 62,206 68,689 59,326 163,167 198,023 182,333 48,022 58,280 53,662 205,154 226,536 195,659 60,274 66,555 57,484 70,530 All cars -41,987 -28,513 -13,325 -26,904 -24,846 -114,289 -183,303 -253,094 Swedish cars -14,184 -10,409 -5,664 -10,056 -7,920 -41,043 -71,310 -113,248 0 0 0 0 0 0 0 0 -1,932 -2,133 -1,843 -2,261 -1,932 -10,312 -19,824 -36,356 All cars -20% -13% -7% -11% -1% -3% -4% -6% Swedish cars -23% -15% -10% -14% -1% -3% -6% -9% 0 0 0 0 0 0 0 0 -3% -3% -3% -3% 0% -1% -2% -3% All cars Swedish cars 240,064 4,153,155 4,153,155 4,153,156 4,153,152 72,791 1,365,726 1,289,483 1,248,816 1,255,375 Registration tax All cars Swedish cars 213,160 4,128,309 4,038,866 3,969,853 3,900,058 62,735 1,357,806 1,248,440 1,177,506 1,142,127 Circulation tax All cars Swedish cars 240,064 4,153,156 4,153,155 4,153,156 4,153,154 1,363,794 1,279,171 1,228,992 1,219,019 Change from base, number of cars Registration tax Circulation tax All cars Swedish cars Relative change from base, % Registration tax Circulation tax All cars Swedish cars Assumptions In addition to the key assumptions listed in the table, the calculations of the socio-economic effects are subject to a number of additional simplifying as- 110 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars sumptions, the most important of which is the what-if nature that is inherent in the analysis. Table 8.14 Overview of key data and assumptions. Data Value (1999) Number of employees in car manufacturing (34) 70,966 Number of passenger cars produced in Sweden 385,044 Number of passenger cars produced and sold in Sweden 66,613 Number of busses and trucks produced in Sweden 108,776 Share of employment that is related to production of passenger cars (see assumption 1 and 2 below) 0,54 Share of production of Swedish brands that is carried out in Netherlands (see assumption 3 below) 0,28 Number of employees in vehicle sales, repair and maintenance plus service stations 77,639 Composition of vehicle fleet passenger cars 3,890,159 trucks and busses 369,162 motorcycles 119,872 Share employment that is related to sales and servicing of passenger cars (see assumption 4 below) 0,56 Assumptions 1. It is assumed that it takes three times more manpower to produce a bus/truck than it does to produce a passenger car. The analysis disregards any other vehicle type. This assumption is used to derive the 0.54 above. If it takes more than that, the employment effect will be over-estimated and vice versa. 2. It is assumed that the average and the marginal effect is the same. In reality, a share of the employment would be involved in administration, marketing, research and sales. Consequently, the employment effect is somewhat overestimated. 3. When calculating the employment effect is it assumed that the Netherlands and the Swedish production of Swedish cars are affected similarly. In reality, one could envisage that the Swedish industry would be more affected, since the Netherlands production is dominated by small cars whereas the production in Sweden is dominated by the larger cars. 4. It is assumed that it takes three times more manpower to sell and service a bus/truck than for the passenger cars, lorries and motorcycles. This assumption is used to derive the 0.56 above Source: Bilismen i Sverige 2000, Input-Output tables for Sweden, Statistisk Årsbok för Sverige, 2001 This approach disregards for example any adjustment efforts done by the Swedish industry. Thus, the sales of Swedish cars will not only decline, but it will decline more than the total sales of cars. In other words, the Swedish car manufacturing industry will loose market shares to other brands. However, it is unlikely that the Swedish industry will not react to this, for example through more intensive marketing efforts in Sweden and/or abroad and through an intensified effort towards the development and production of smaller and energy effective cars. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 111 Further, it is assumed that one can apply the labour input of 0.1 for any car. Given the fact that the taxes are likely to affect the sales of larger cars the most, this could imply that the employment effect becomes underestimated. Larger cars would tend to require more labour than smaller cars. From the table it can be seen that the most important contributor (among the other industries) to the production in car manufacturing is: • renting of machinery and equipment without operator and of personal and household goods, computer and related services and other business activities • wholesale and retail sale, repair of motor vehicles and motorcycles, retail sale of automotive fuels • manufacture of fabricated metal products, expect machinery and equipment The remaining industries contribute less than 2% each. The above three sectors are thus likely to be most affected by a decline in the level of activity in the car manufacturing industry. Table 8.17 provides an overview of the most important related industries. In providing quantitative assessments, the analyses look only at the sector that covers vehicle sales, repair and maintenance of vehicles and service stations. The motivation for this delineation being manifold: it is a sector where the outputs of the model based analyses (number of cars and sales of cars) has a direct and significant impact. It is a sector that employs a fairly high number of people, almost 80,000, and it is a sector where substitution activities (e.g. increased exports, increased focus on other lines of activity) are likely to be few - at least in the short run. Based on the data and assumptions outlined above, the employment per car amounts to 0.01528 for car sales and car servicing (sector 50). Resulting employment effects Table 8.15 provides an overview of the calculated quantified employment effect. The effects shown in the table only includes those that relate to car manufacturing and to the sales and servicing of cars. Other sectors are likely to be affected as well such as those who supply the industry with raw materials and intermediate goods and services. The table shows that car manufacturing will see a decline in the order of between 1% and 2% in employment as a result of the registration tax, whereas the circulation tax will result in a decline that is significantly less than 1%. Car sales and servicing will only be affected by the registration tax, as it is the only scenario that affects the size of the car fleet. In total the registration tax may result in an employment decline in these two sectors of between 1% and 4%. 28 (3,890,159/(3,890,159+119,972+369,162*3))*77,639/3,890,159 112 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The order of magnitude of the effect increases over time reaching 4% in 20 years. In regard to the quantitative estimates of the employment effects, the following should be borne in mind: • The average age of the cars increases in the registration tax scenario. This may call for higher demand for repair and maintenance services and this may in turn also increase the demand for spare parts also from the Swedish manufacturing industry. • While the sales of new cars decline, and the size of the car fleet decreases in the registration tax scenario, this need not necessarily result in proportional declines in the level of activity in the car sales sector. • The calculations rest on the assumptions regarding the relative employment inputs/vehicle for passenger cars relative to heavy vehicles, on the assumed share of Netherlands production in the sales of Swedish cars, and on the assumed linear relation between cars produced and number of employees. • The calculations ignore possible reactions from the car manufacturing industry as well as other ongoing or planned developments that will change the current position of Swedish car manufacturing in the market. 113 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.15 Employment effects from the two tax scenarios Years after implementation 1 5 10 20 Employment in car manufacturing -971 -1,633 -733 -1,608 Employment in sales and servicing of cars -377 -1,734 -2,781 -3,839 -1,348 -3,367 -3,514 -5,447 - car manufacturing -1 -2 -1 -2 - car sales and servicing -1 -2 -4 -5 - total -1 -2 -2 -4 -183 -203 -175 -214 0 0 0 0 -183 -203 -175 -214 >-1 >-1 >-1 >-1 >-1 >-1 >-1 >-1 Registration tax Total Relative to base year employment, % Circulation tax Employment in car manufacturing Employment in sales and servicing of cars Total Relative to base year employment, % - car manufacturing - car sales and servicing - total The table below provides a more detailed overview of the most important related industries. Table 8.16 Overview of the employment effects in car manufacturing and in car sales and servicing as a result of the two tax scenarios. From the table it can be seen that the most important contributor (among the other industries) to the production in car manufacturing is: • renting of machinery and equipment without operator and of personal and household goods, computer and related services and other business activities • wholesale and retail sale, repair of motor vehicles and motorcycles, retail sale of automotive fuels • manufacture of fabricated metal products, expect machinery and equipment The remaining industries contribute less than 2% each. The above three sectors are thus likely to be most affected by a decline in the level of activity in the car manufacturing industry. 114 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 8.17 Contribution from car manufacturing to other industries and other industries' contribution to car manufacturing. Share of wages in production value and share of imports. 1998 Industry Composition of production value of car manufacturing industry, % Contribution by car manufacturing to production value of the industry, % Share of wages in production value, % Share of imports in production value, % Manufacture of rubber and plastics products (25) 1,18 5,78 18,51 28,45 Manufacture of basic metals (27) 1,58 3,42 11,5 28 Manufacture of fabricated metal products expect machinery and equipment (28) 3,22 6,28 21,69 15,98 Manufacture of machinery and equipment (29) 1,3 1,55 18,69 25,67 31-32: Manufacture of electrical machinery and apparatus n.e.c. and Manufacture of radio, television and communication equipment and apparatus (31-32) 0,92 0,92 11,32 31,86 Manufacture of motor vehicles, trailers and semi-trailers (34) 13,99 13,99 12,17 31,58 Manufacturing n.e.c.(36-37) 0,97 4,50 27,79 21,82 Maintenance and repair of motor vehicles (50.2) 0,05 0,59 23,49 22,99 Wholesale and retail trade; repair of motor vehicles and motorcycles; retail sale of automotive fuel (50-52 rest) 4,31 2,54 35,58 5,57 Other land transport and Transport via pipelines (60.2-3) 1,4 2,28 20,41 5,37 Financial intermediation, except insurance and pension funding and Activities auxiliary to financial intermediation (65+67) 0,76 1,53 24,23 3,39 Renting of machinery and equipment without operator and of personal and household goods, Computer and related activities and Other business activities (71-74) 6,92 3,64 25,06 8,6 Other (including e.g. salaries, taxes and subsidies 63,4 n.a. n.a. n.a. Total 100 n.a. n.a. n.a. Note: Industrial NACE code in parenthesis. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 9 115 Cost Benefit Analysis External benefits When cars pollute the pollution enters the ecological system. This means that the environmental costs associated with car pollution accrues to the society as a whole and not only to the motorist. This adverse effect on others is in the economic literature referred to as an externality because the motorist by driving a car imposes external costs on all others. Introducing a CO2 differentiated tax can “internalise” and hence diminish such costs by altering the behaviour of individuals in the society. Private costs However, the induced change in behaviour is associated with a cost, which is upheld exclusively by the single individual while the society at large remains unaffected.29 The content of this chapter is as follows. The first section includes the main cost benefit analysis and the main results of the total long run costs and benefits. In the second section the total effect is decomposed and focus is exclusively on effects arising from CO2 differentiation and its implications for the composition of the car fleet whereas the change in the size of the car fleet is left out. The second section is included to allow for closer comparison between the circulation tax differentiation and the registration tax differentiation. 9.1 Total cost benefit analysis This section describes the cost benefit analysis of the long run effects from the changes to the tax system. In practice we compare year by year the new tax system with the base situation and assess the costs and benefits from the difference between the base and the tax scenario. The following benefits are included: • • Saved CO2 from change in CO2 efficiency Saved CO2 from reduction in car fleet • Reduction in other air emissions (NOx, Paticulate matters, SO2, HC) due to reduction in car fleet Reduction in car accidents due to reduction in car fleet • 29 Such costs are often referred to as “private” costs as opposed to the external costs, which are “public”. 116 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars • Reduction in noise due to reduction in car fleet The following costs are included in the analysis: • • • • • Welfare loss due to demand shift (CO2 differentiation) Welfare loss due to reduced car fleet Reduced profit for Swedish car producers Reduced employment Increase in other emissions (NOx, Particulate matters, SO2, HC) due to older cars (both age and mileage effect) 9.1.1 Key assumptions The key assumptions are listed in the tables below Table 9.1 Basic general assumptions in the cost benefit analysis Parameter Value 1 Yearly wage car manufacturing 223,200 Yearly wage car sales and maintennance2 216,000 3 Discount rate 4% 4 New car sales 2002 205,154 1 Source: SCB: Lønstatistisk Årsbok 2000, average for "Finmekanikere", "Maskinoperatører". 2 Source: SCB: Lønstatistisk Årsbok 2000, average for "Finmekanikere", "Maskinoperatører" og "Sælgere m.fl.". 3 Source: SIKA Report 2000:3 (abridged version of "Review of principles for social economic estimates and estimates in the transport sector", SIKA Report 1999:6, 4 This figure refers to the number that should be brought into the car fleet in order to keep the total car fleet constant as assumed in the model calculations. The actual figure for 2000 was 354,649 (Source: SIKA: Statistiska Meddelanden 2001). Table 9.2 Basic car assumptions in cost benefit analysis Parameter Distribution of new cars Average car life time Average mileage Average fuel consumption (l/100 km) 1 Fuel price (SEK/l 10/12_01) Petrol Diesel Average 93.5% 6.5% 15.8 15.8 15.8 12,036 17,765 12,408 8.4 6.5 8.3 8.2 6.5 8.1 1 Source: EU Oil Bulletin Petrolier 10/12-01 (http://www.np.no/ "Fakta i Tal" -> "International statistics" -> "Drivstoffpriser i EU, ukentlig oppdatert" ) Shadow price on CO2 emission costs The reduced CO2 emissions are an external benefit that accrues to society at large. No official price exists for CO2 reduction so a shadow price is applied. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 117 Determination of shadow prices is always associated with some degree of uncertainty and arbitrariness. Hence two different shadow prices are used in order to obtain a span in which the “actual” price is. The two CO2-prices applied are 1.5 SEK per kg CO2 and 0.53 SEK per kg CO2 respectively.30 Employment effects The tax scenarios result in a reduced employment in the Swedish car manufacturing industry and in the related service and repair businesses. From a welfare economic perspective, this has two types of effects. First, it will result in a welfare loss for the people that loose their jobs. Whereas some of the people will find employment elsewhere and the rest will stay unemployed. The latter group will suffer a welfare loss, since they must be expected to gain a welfare surplus from being in employment. Second, society as a whole can obtain a welfare gain because the free labour capacity (that used to work in car manufacturing and the related industries) can be employed somewhere else, whereby they will produce goods or services, and these goods and services will have a value to society. This effect reduces the welfare loss related to the loss of profit in the car industry, because the profit in the car industry will be substituted by profit in other industries. The size of these two welfare effects depends on the ability of the free labour to become employed somewhere else. If a relatively large fraction moves to other places of employment, the welfare loss for the workers will be relatively small and the welfare loss due to reduced profit in the car industry will be small. If, on the opposite, a relatively large fraction of the free labour come to stay unemployed the welfare loss for these people will be relatively large and the welfare loss due to reduced profit in the car industry will be large. The cost benefit analysis assumes that 25% of those that become unemployed in a given year will have stayed unemployed for the whole of this year. In the period after, it is assumed that, each year, 10% of those will find a new job. Furthermore, the analysis is supplemented with sensitivity analysis. One of these assumes that all those, who become unemployed, find a new job already during their first year of unemployment, and another one assumes that only 50% find a new job during the first year. The adjustment process is illustrated in the figure below. Welfare loss for unemployed workers The welfare loss for the unemployed workers is estimated as follows: The unemployed worker originally supplied his labour according to his supply curve illustrated in Figure 9.2. In the initial situation, the individuals work L0 hours gaining the wage W0. The labour supply is an increasing function of the wage. Once the wage is equal to the reservation wage (Wres) their supply of labour will be zero. 30 Source: SIKA report 2000:3 118 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 9.1 Reallocation of unemployment Still unem ployed 60% 50% 40% 30% 20% 10% 0% 0 5 10 15 20 25 Years Low unemployment Central assumption Figure 9.2 Labour supply W0 Wres L0 High unemployment 119 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars In the initial situation, the workers obtain a welfare gain from working, corresponding to the triangle marked by the thick line in the figure. Correspondingly, they will loose this area once they are unemployed. Thus the welfare loss for the group of people, which loose their jobs and stays unemployed is: ½*L0*(W0-Wres). Fuel savings Saved fuel due to the reduced car fleet is not included as benefit in the analysis. The reason is that this reduction in fuel is counterbalanced by a corresponding reduction of the transport benefit from cars. In practical terms, when individuals decide to skip their car, they will have a reduction in their fuel expenses. However, in order to maintain mobility they will have to use the saved money for other means of transportation, e.g. train or bus. Even if they decide to move less around this would mean that their benefit from transport would be reduced and crowd out the major part of the fuel savings this way. Fuel savings due to the increases in energy efficiency are not included either. This gain is included (taken into account) in the car purchaser’s choice when he/she adapts to the new car prices. It is therefore included in the loss of welfare due to the car price change. Other emissions The changes to the demand for new vehicles will not only affect the CO2 emissions, but it will also affect the level of other emissions. This analysis includes the effects on the emissions of major pollutants, i.e. NOx, Particulate matters, HC and SO2. NOx, Particulate matters and HC are regulated with a set of maximum emissions standards. Since the standards are reduced (tightened) over time the average emissions will decline over time. However, if the renewal of the car fleet is delayed, the emissions will decrease slower than they would have done otherwise. Therefore, increases in the average age of the car fleet will lead to higher average emissions. Furthermore, reductions in the size of the car fleet will lead to a decline in the emissions of CO2 as well as the other pollutants. The applied emission factors were calculated in the Danish TEMA2000 model, and are summarised in the following tables for petrol and for diesel cars. Table 9.3 Average emissions per km for petrol cars > 1.4l (g/km) EURO norm Vintage EURO0 EURO1 EURO2 EURO3 EURO3 - 86 87-96 97-00 01-05 06- 0.02107 0.01042 0.01042 0.01042 0.01042 NOx (g) 2.4014 0.2789 0.12551 0.0753 0.03603 SO2 (g) 0.00724 0.00714 0.00714 0.00714 0.00607 HC (g) 1.85067 0.1958 0.11962 0.07328 0.04005 PM10 (g) Note: Calculation settings: default travel pattern, 40 km trip including cold start emissions, average mileage = 60000 120 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 9.4 Average emissions per km for diesel cars > 2.0l (g/km) EURO norm EURO0 Vintage EURO1 EURO2 EURO3 EURO3 - 86 87-96 97-00 01-05 06- PM10 (g) 0.1402 0.0477 0.0373 0.0342 0.0324 NOx (g) 0.7207 0.6596 0.4698 0.3431 0.1904 SO2 (g) 0.0082 0.0062 0.0062 0.0062 0.0052 HC (g) 0.1608 0.0804 0.0563 0.0394 0.0193 Note: Calculation settings: default travel pattern, 40 km trip including cold start emissions, average mileage = 60000 The valuation of cost of these emissions is based on SIKA Report 2000:3 referred in the following table. Table 9.5 Valuation of other emissions Pollutant SEK per Kg NOx 109 SO2 122 HC 50 Particulate matters 3,329 Source: SIKA Report 2000:3 Accidents Fewer cars lead to fewer accidents. On the other hand, older cars are likely to be involved in more accidents, due to a higher risk of mechanical failure. The calculations only include the effect from the reduced car fleet. No data are available to assess the effect from car age on car accidents. However, regular car inspections and the fact that only a fraction of the accidents are caused by mechanical failure makes it reasonable to assume that the effect from car age on accidents is relatively small. Based on the number of accidents with passenger cars involved31, and a valuation of the resulting fatalities and injuries32, the external cost of accidents can be calculated to be 1,866 SEK per car. Noise The reduction of the car fleet is assumed to result in reduced external cost from noise. It is assumed that the relative noise reduction corresponds to the relative reduction of the car fleet. The external cost of 1 km is calculated based on a study on the marginal change in the number of households affected by noise in 31 32 SIKA Statistiska Meddelanda 2000 SIKA Report 2000:3 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 121 Denmark33 and an adjustment to the Swedish valuation of noise34. The result is a cost of 0.01 SEK per km. 9.1.2 Cost benefit analysis of the registration tax scenario The registration tax scenario results in a reduction of the car fleet, an increase in the average age and an increase in the average CO 2 emissions per car. In the cost benefit analysis, the effect from the increased average age almost exactly offsets the CO2 savings that can be harvested due to the CO2 differentiation. Consequently, the total benefit from the registration tax scenario is reduced substantially. Table 9.6 shows the results from the cost benefit analysis. Note that the figures in the main body of the table show the values per year. For instance, the welfare loss due to the reduced car fleet is 1,152 MSEK in year 20. This loss comes from a 6.1% reduction of the car fleet corresponding to a reduction in the number of cars of 253,094. The bottom line of Table 9.6 summarises the cumulated net benefit from the analysis. In a 10 years time horizon there is a loss of approximately 1.3 billion SEK. In a 20 years time horizon there is a loss of 1.1 billion SEK. Taking a more optimistic view on the employment effect, assuming that 95% find other employment already in the first year, this would still result in a net loss of welfare from the registration tax scenario. In this case however, the cumulated loss is reduced down to 0.7 billion SEK in a 10 years horizon and 0.3 billion in a 20 years time horizon. The cost benefit analysis thus shows that it is costly to reduce the CO2 emissions by the introduction of a registration tax. The resulting reduction in the number of cars imposes substantial welfare losses onto the car owners. This welfare loss is of the same order of magnitude as the benefit from the CO2 reductions. On top of this welfare loss for the car owners, there will be a substantial negative impact on the car industry and in the service and repair sector. Even though there is a benefit from fewer accidents and reduced local emissions and noise, these benefits are not sufficient to outweigh the negative impacts from the registration tax. Factors not included The impact on service and repair sector is assumed to be proportional to the number of cars. In reality this effect will to some extent be outweighed by the ageing effect of the car fleet since older cars require more maintenance. This effect is not included in the analysis. Likewise, the impact on car producers excludes efforts to adapt to the new situation. Both of these would reduce the cost of registration tax. The other assumptions underlying the calculations of the employment effects are shown in 33 Samfundsøkonomisk omkostningseffektivitet i transportsektoren, Arbejdspapir 1, Marts 1997 34 8330 SEK per SBT, SIKA Report 2000:3 122 Assumptions Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars In addition to the key assumptions listed in the table, the calculations of the socio-economic effects are subject to a number of additional simplifying assumptions, the most important of which is the what-if nature that is inherent in the analysis. Table 8.14. On the other hand, the welfare loss that motorists will suffer due to the driving in older cars is not included either. This would increase the cost of the registration tax. Table 9.6 Cost benefit analysis of introducing a CO2 differentiated registration tax (1,000,000 SEK per year) Cost and benefit items Year 1 5 10 20 Costs Smaller cars 57 70 64 75 113 520 834 1,152 Employment (Cost) 20 10 3 2 Profit 26 12 4 2 5 15 14 7 17 53 50 24 239 679 969 1,262 53,615 250,300 408,722 512,020 CO2 reductions SEK (0.53) 28 133 217 271 CO2 reductions SEK (1.50) 80 375 613 768 -57 -115 -54 54 46 213 342 472 3 14 23 31 73 488 924 1,326 Total net benefit per year -166 -192 -46 64 Discounted -166 -164 -14 26 Cumulated net benefit -166 -908 -1326 -1096 Fewer cars Car manufacturing Car repair Employment (Cost) Profit Total cost per year Benefits CO2 reductions (ton) Other emissions (NOx, PM10, SO2, HC) Accidents Noise Total benefit per year 123 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 9.1.3 Cost benefit analysis of the circulation tax scenario The above calculations show that it is costly to save CO2 emissions by means of reducing the car fleet. The circulation tax scenario however solely relies on CO2 differentiation. Thus, the size of the car fleet remains unaffected by this scenario. The outcome of this cost benefit analysis is therefore more favourable in this case compared to the registration tax scenario. Table 9.7 Cost benefit analysis of introducing a CO2 differentiation in the annual circulation tax (MSEK per year) Cost and benefit items Year 1 5 10 20 Costs Smaller cars 39 43 37 46 0 0 0 0 Employment (Cost) 3 2 1 0 Profit 4 3 1 1 Employment (Cost) 0 0 0 0 Profit 0 0 0 0 45 48 40 47 13 68 124 201 CO2 reductions SEK (0.53) 7 36 66 106 CO2 reductions SEK (1.50) 20 102 186 301 Other emissions (NOx, PM10, SO2, HC) 0 -1 -2 -3 Accidents 0 0 0 0 Noise 0 0 0 0 20 100 184 298 Total net benefit per year (1.5 SEK/kg) -26 53 145 251 Discounted -26 45 108 115 Cumulated net benefit -26 52 574 1,767 Fewer cars Car manufacturing Car repair Total cost per year Benefits CO2 reductions (ton) Total benefit per year (1.5 SEK/kg) 124 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars While the car industry will suffer some costs induced by a reduced market share of Swedish car makes, the net benefit from introducing a CO2 differentiated element into the annual circulation tax is 1,767 mill SEK (20 years). Taking a less optimistic view on the employment effect in terms of assuming that only 50% will find other employment already in the first year of their unemployment would still result in a net welfare gain from the circulation tax scenario. In this case, the gain is reduced a little, namely down to 1.720 billion in a 20 years time horizon. 9.2 Decomposed effect from CO2 differentiation The results of the cost benefit analyses of the registration tax scenario and the circulation tax scenario are not directly comparable. The registration tax scenario involves a general tax increase, while the circulation tax scenario is merely a restructuring of an existing tax, which renders the total revenue effect neutral. Therefore, this section decomposes the total effect and compares the effect from the CO2 differentiation part of the two tax scenarios. Thus, the effect on the size of the car fleet from the registration tax is disregarded in this analysis. The focus of this analysis is therefore the cost and benefit that occur when car buyers buy smaller and more energy efficient cars. To focus on the CO2 differentiation, it was chosen to look at one generation of new cars. Costs Following this idea the cost from the introduction a CO2 differentiation is the welfare loss that occurs when car purchasers are forced to buy other cars that what they would prefer in the absence of the CO2 tax. This loss is valued in monetary terms by the comparing the car price net of registration tax before and after the CO2 tax. Benefits The benefit is the reduction of CO2 emissions from this generation of cars, which occurs exactly because of the tendency to buy smaller and more energy efficient car. These savings are valued by the two alternative CO2 values (0.53 and 1.5 SEK per kg) respectively. Impacts on industry The CO2 tax will affect the Swedish car manufacturing industry. As has been shown previously, the market share of Swedish car makes will decline. This will lead to reduced profits and reduced employment of the industry. To the extent that the lost production is not compensated for by increased economic activity elsewhere in the economy, this will result in a net cost to society. For the sake of simplicity, other emissions and effects are left out of this analysis. These other effects are to a large extent related to changes in the car fleet size and to the age of cars, and consequently, their omission is likely to have only minor implications as these factors are left out of this analysis as well. 125 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 9.2.1 The CO2 differentiated element of the registration tax The introduction of a CO2 differentiated registration tax implies that car purchasers who buy energy efficient (and to some extent small) cars will see a reduction in the car price. By comparison, car purchasers who buy inefficient cars will experience see an increase in the car price. This observation is however only valid for this analysis, which focuses exclusively on the CO2 differentiation part of the tax. Costs A CO2 differentiation of the registration tax leaves the car buyers better of who buy cars with low emissions, while those who buy cars with high CO2 emissions will see a loss. However, in both cases, society as a whole will experience a welfare loss, which is caused by the distortion of the market equilibrium, which apply in the absence of the tax. The total welfare loss, which results from the CO2 differentiation part of the registration tax, is 72 MSEK. This corresponds to 352 SEK per new car sold. Benefits At the same time, the CO2 differentiation will lead to a decline in CO2 emissions of 7,135 tonnes. Thus, this decline arises solely as a result of more fuel-efficient cars. The monetary value of this reduction is in the order of 60 169 MSEK depending on applied unit value of CO2 (0.53 and 1.5 SEK respectively). If future CO2 savings are discounted with discount factor of 4%, the value of the CO2 savings will be reduced to between 44 and 125 MSEK. The impacts from the reduced market share of Swedish cars include both profit and employment. The reduced profit is calculated as the average price net of tax multiplied by the number of cars sold, and assuming a profit rate of 5%. These figures imply that the registration tax differentiation would reduce the profit in the Swedish car industry by 15 MSEK. The CO2 differentiation reduces the market share of Swedish cars by approximately 1 percentage-point. This corresponds to a reduction of about 2000 cars in 2002. However, 28% of the car production is assumed to take place in the Netherlands, and therefore only 72% of the reduction affects the employment in Sweden. Furthermore, it is assumed that half of the redundant employees find alternative employment. Valuing half of the labour power to the welfare loss from unemployment described above would cost 7 MSEK. The tables below summarise the key figures from this cost benefit analysis. Table 9.8 Costs associated with the CO2 differentiation element of a registration tax (MSEK) Cost item Cost Welfare loss due to car demand shift 72 Reduced profit in car industry 15 Reduction in employment 7 126 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 9.9 Benefits associated with the CO2 differentiation element of a registration tax (MSEK) Cost item Benefit CO2 savings (ton per year) Net benefits 12,333 CO2 savings (SEK, 0.53 SEK per kg) per year 11 CO2 savings (SEK, 1.5 SEK per kg) per year 7 Table 9.10 summarises the net benefits for one generation of new cars for the total lifetime of the cars under alternative assumptions. Table 9.10 Net benefits associated with a the CO2 differentiation element of a registration tax (MSEK) Assumption CO2 value CO2 value 0.53 SEK/kg 1.5 SEK/kg Excluding employment effect -43 38 Half of employees find other employment - 50 30 The net benefit from the CO2 differentiation can thus be seen to vary from a small loss of approximately 50 MSEK (-244 SEK per car) to a gain of 38 MSEK (185 SEK per car). The former applies the low unit value of CO2 and assumes that half of those, who become unemployed, stay unemployed. The latter figure applies the high value of CO2 excludes any employment effect, i.e. assumes that all find other employment. The benefits accrue to society and are not associated with the individual car owner per se. Therefore, it is not entirely correct to calculate the net benefit per car. However, this calculation serves to illustrate the relative achievements. 9.2.2 The CO2 differentiated element of the circulation tax scenario The same calculations were carried out for the scenario that introduces a CO2 differentiation into the annual circulation tax. Costs The welfare loss related to the CO2 differentiation of the circulation tax is calculated to be 39 MSEK. This corresponds to about 190 SEK per new car. Benefits At the same time there will be a reduction in CO2 emissions of 213,000 tonnes. This reduction arises solely as a result of the more fuel-efficient cars. The value of this reduction is between 113 and 320 MSEK, using 0.53 and 1.5 SEK as the unit price of CO2 respectively. If the future CO2 savings are discounted by a 127 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars discount factor of 4%, the value of the CO2 savings is reduced to83 – 235MSEK. Also in this case will there be a decline in the market share of Swedish cars, and this results in a reduction of the turnover of the Swedish car manufacturers. Again, this effect will spill over into reduced profit to the Swedish car producers and reduced employment in the Swedish car industry. The calculations show that introduction of the CO2 differentiated element into the circulation tax will reduce the profit in the Swedish car industry by around 15 mill SEK. The CO2 differentiation will reduce the market share of Swedish cars by about 1 percentage-point. This corresponds to a reduction by 2000 cars in 2002. Under the same assumptions as in the previous section, this translates into a welfare loss from unemployment of a value of about 8 MSEK. Overview Table 9.11 and Table 9.12 summarise the key figures of the cost benefit analysis. Table 9.11 Costs associated with the CO2 differentiation element of a circulation tax ( MSEK) Cost item Cost Welfare loss due to car demand shift 39 Reduced profit in car industry 15 Reduction in employment 8 Table 9.12 Benefits associated with the CO2 differentiation element of a circulation tax (MSEK) Cost item Benefit CO2 savings (ton per year) Net benefits 13,485 CO2 savings (SEK, 0.53 SEK per kg) 7 CO2 savings (SEK, 1.5 SEK per kg) 20 Table 9.13 summarises the net benefit for one generation of new cars for the total lifetime of the cars under alternative assumptions. Table 9.13 Net benefits associated with the CO2 differentiation element of a circulation tax (MSEK) Assumption CO2 value CO2 value 0.53 SEK/kg 1.5 SEK/kg 128 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Excluding employment effect 30 182 Half of employees find other employment 22 174 The net benefit from the CO2 differentiation varies from a net gain of approximately 22 mio (107 SEK per car) to a net gain of 182 mio (146 SEK per car). The latter is based on the high value of CO2 and assumes that all who become unemployed find alternative employment. The former applies the lower price of CO2 and assumes that only half of the unemployed finds alternative employment. 9.3 Concluding comments The costs of introducing a registration tax clearly outweigh the benefits. The redesign of the circulation tax to include a CO2 differentiated element provides a positive net benefit in the medium and long run (5 - 20 years), while there is a net cost in the short term (1-2 years). The reason for this major difference in the costs and benefits of the two tax scenarios is to be found in the fact that the registration tax gives rise to an increase in the consumer price of cars. This increase is very distorting, mainly because of the welfare loss it inflicts on the (former) car owners. The circulation tax scenario was framed so that the average tax level remains unchanged compared to the level of today. Thereby, the size of the fleet remains unaffected, and consequently, the distortions caused by this scenario are much less. Total CO2 emissions are less reduced in the circulation tax scenario compared to the registration tax scenario. However, this has no implications for the overall result. This is determined mainly by the difference in costs. Thus, the overall conclusion, namely that it is costly to reduce CO2 emissions by means of a registration tax, appears to be quite robust despite the fact that the above cost benefit analysis is somewhat rudimentary. A comparison of the costs and benefits of the CO2 differentiation alone, i.e. omitting any inclusion of the effect on the car fleet, shows that the circulation tax differentiation has more effect than the registration tax differentiation. The net benefit from the circulation tax CO2 differentiation is higher relative to the net benefit from the registration tax differentiation. Measured in terms of benefits per kg of CO2, the circulation tax differentiation also provides larger net benefits than the registration tax scenario. It is 0.8 SEK per kg CO2 in the circulation tax scenario, and 0.3 SEK per kg CO2 in the registration tax scenario. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 10 Conclusions 10.1 CO2 reducing potentials 129 It is important to distinguish clearly between issues related to the tax level versus those that relate to tax differentiation. In the context of this study, the level and the differentiation can be said to play two different roles. One of the major merits of the tax level is that it can be used to control the number of cars. The tax differentiation on the other hand, has its strength in its ability to influence people's choice of which car to buy, once they have made the decision to buy a car. The registration tax scenario as it is framed here includes both aspects, whereas the circulation tax scenario has been defined to contain only the differentiation element. The average level of the circulation tax is maintained at the current level. This section first summarises the key results from each of the scenarios separately. Thereafter, the section compares the results from the two scenarios. Key results New car sales 10.1.1 The registration tax scenario The registration tax scenario provides a reduction in total CO2 emissions from the car fleet in the order of 5% in 20 years. Taking a shorter time horizon, the reductions are smaller, e.g. just above 1% in 5 years. The reductions come mainly through a decline in the number of cars in the fleet. The average emission level per car (gram of CO2/km/car) is almost constant in this scenario, and the average age of the cars increase by about half a year as a result of reduced scrapping. It should be noted that the resulting decline in the car fleet is a result mainly of the assumed price elasticity of -0.6. Care should thus be taken not to interpret this decline too directly, as the result is highly sensitive to this elasticity. For example, an elasticity of only –0.35 would provide a reduction in total CO2 emissions of 3% in 20 years. While this figure is lower than the original one, it also involves a smaller reduction in the number of cars, namely 2.5% compared to the original decline of 4.2%. The sales of new car is affected in two ways. First, there is a decline in the sales of new cars, which adheres from the decline in the total car fleet. Second, this 130 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars decline is accentuated by reduced scrapping, which also leads to a reduction in the sales of new cars. Tax level When the value based component of the registration tax is set to 20% instead of the 10%, the total tax burden increases. However, it has only a limited effect on the total CO2 reductions. The reason for this limited effect lies in the trade off between two effects. On the one hand fuel efficiency of the new cars improves as a result of the tax increase, and on the other hand car owners tend to keep their cars for longer as a result of the price increase. The latter has a significant effect on the CO2 emissions since older cars have higher CO2 emissions. The net effect from a higher tax level comes therefore exclusively from a further decrease in the size of the car fleet and the higher the tax level the lower is the extra potential reduction to be achieved this way. CO2 differentiation Doubling the price per g CO2 emitted and maintaining the same average tax level has very little impact on CO2 emissions. The reason is that the tax can never turn into a net subsidy. A doubling of the CO2 tax would therefore imply that more car owners would hit the lower ceiling of zero registration tax. Swedish car makes The sales of cars manufactured in Sweden will drop for two reasons. First of all because Swedish car makes typically are larger, less CO2 efficient cars than the average car, and secondly as a result of the overall drop in the car fleet, which is accentuated by the drop in the share of Swedish cars among new cars sold. 10.1.2 The circulation tax scenario The average tax burden is kept constant in this scenario. Consequently, this scenario has no impact on the size of the car fleet. Hence, the entire effect on CO2 emissions is provided through the differentiation alone. The assumed circulation tax leads to a reduction in the total emissions of CO2 from the car fleet. The reduction is the order of 2% in 20 years. Taking a shorter time horizon, the reductions are smaller, e.g. just above 0.5% in 5 years. The reductions are provided through a decline in the average emissions per car (gram of CO2/km/car). The average emissions decrease by 2.5% over 20 years. As there are no effects on the size of the car fleet and scrapping, the average age of the cars also remain constant. The reason why the relative reduction in total CO2 emissions is lower than the average CO2 reduction per car is that the average mileage increases. This is because cars become more fuel-efficient (the rebound effect), whereby running costs are reduced. Swedish car makes The sales of new cars produced in Sweden is estimated to decline from 30.5% to 29.5%. Over a 20-year time horizon, Swedish cars will constitute 0.88 percentage points less of the car fleet as a result of the redesign of the circulation tax. The reason is that Swedish car makes are typically larger and less CO2 efficient than the average car. 131 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 10.1.3 Comparison of the two scenarios Care should be taken to compare the performance of the two scenarios. They fulfil different roles (cf. below) and they impact differently on the tax burden of car purchasers and car owners. Nevertheless, there are important generic conclusions to be made. Thus, the analyses clearly point to the importance of distinguishing between tax levels and tax differentiation. Roughly speaking, the level of the tax affects the overall price structure per se, i.e. it makes cars relatively more expensive than other consumer goods. Hence, it has an impact on whether people decide to have a car. By comparison, differentiation affects the choice that people make on which car to buy, but has less impact on the overall price structure (when comparing cars to other consumer goods). As mentioned above, the introduction of a value based registration tax plus a CO2 differentiation element would reduce total emissions from passenger cars by 5% in the long run. By comparison, redesigning the circulation tax to include a CO2 differentiated element would reduce CO2 emissions by 2% in the long run. This difference in total CO2 reductions over time is illustrated in Figure 10.1. Figure 10.1 CO2 reduction over time (in percent) - registration and circulation tax CO2 reduction 6.0% 5.0% 4.0% 3.0% 2.0% 1.0% 0.0% 2000 2005 2010 2015 2020 2025 2030 2035 Year % CO2 reduction, circulation tax % CO2 reduction, registration tax It is seen that the effect from the circulation tax is almost fully implemented after 20 years while the effect from the registration tax takes more than 30 years to be fully realised. This is because the effect on the size of the car fleet is phased in gradually over a long time horizon. 132 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The registration tax scenario provides higher CO2 reductions than the circulation tax scenario. This is, however, only a result of the resulting decline of the car fleet. Looking solely at the effects from the CO2 differentiated element of the tax, the differentiating power is highest for the circulation tax. This is seen from the effect on new car sales. By differentiation power is means the reductions provided by the differentiation alone. The fact that the sales of new cars are more affected by the circulation tax than by the registration tax points to the conclusion that people buy annual energy savings at too high a price.35 The sales of Swedish car makes (i.e. Swedish brands) on the Swedish market will in both scenarios be more affected than the sales of cars as a whole. This reflects the fact that Swedish car makes are typically larger, less CO2 efficient cars than the average car. As a result of the decline in the car fleet (a drop of about 4% in the 20 years perspective), this effect will be most outspoken for the registration tax. Comparisons between the differentiating power of registration tax and the circulation tax shows that the effect of the circulation tax seems higher relative to the registration tax. A CO2 differentiation of 880 SEK per g CO2 in the registration tax provides approximately a reduction of 1.4% of the average CO2 emissions from new cars. A CO2 differentiation of 44 SEK in the annual circulation tax provides a reduction of 2.5% in the average CO2 emissions from new cars. With the average lifetime of 15.8 years, the annual amount of 44 SEK would correspond to 700 SEK. The major reason for the different differentiation powers is that company cars are less sensitive to price changes relative to private cars, but on the other hand company cars are more sensitive to the annual cost relative to private cars. The following table shows the differentiating power of the two taxes for private cars and company cars respectively. Table 10.1 Differentiating power of registration and circulation tax Tax Circulation tax Registration tax Private cars Company cars 2.2% 3.1% 2.1% 0.8% As can be seen from the table the CO2 differentiating powers of the circulation tax and the registration tax are almost identical in the case of private cars. Company cars, on the other hand, are much more sensitive to annual cost and therefore more sensitive to the circulation tax relative to the registration tax. It should be noted that the effect from changes to the circulation tax for the com- 35 This phenomenon also occurs in the choice between petrol and diesel vehicles where diesel vehicles often are “too” expensive compared to their relative fuel efficiency. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 133 pany cars rests on the assumption that such tax changes are reflected in the personal taxation of company cars. In regard to the latter, it is important to note that the annual costs consist of both circulation tax and fuel costs. It is assumed that an increase in the circulation tax of e.g. 1000 SEK would have the same effect as an increase in the annual fuel costs of the same amount. Furthermore, it is assumed that changes to the circulation tax are fully reflected in the car purchase decision of company cars. Company cars may be split into two types. Company cars used for company purposes and company cars used by employee for private use (like private cars). Exactly which car is purchased of the latter type may be decided either by the company or by the employee. If the company decides which cars to offer to the employee then the annual tax is taken into consideration when deciding which cars to offer. On the other hand, if the choice is left completely to the employee which car to have as company car, then the annual tax is not taken into consideration by the employee unless the annual tax is included in the personal taxation. Presently, the annual tax paid by the company for a company car is not reflected in the personal taxation of company cars. If this policy is maintained in the new tax system then the model calculations referred to in this report overestimate the effect from the circulation tax scenarios. Assuming that 33% of the company cars fall into the group where the employee himself decide exactly which car to have as a company car this overestimation amounts to 20% of the total CO2 reductions. However, imposing a CO2 element to the circulation tax would make some of the most ineffective cars very expensive in terms of annual taxation for the company. Therefore, some companies would decide no longer to offer these cars to their employee. Therefore, the real overestimation would probably be substantially lower than 20%. In the light of the higher sensitivity of company cars to the annual costs, and because company cars constitute about 50% of the sales of new cars, these assumptions are important. 10.2 Socio-economic implications The cost benefit analyses showed that it is costly to reduce CO2 emissions by means of a registration tax with a CO2 differentiated element. This scenario results in net costs to society. By comparison, the redesign of the circulation tax can provide CO2 reductions which are not of a similar size, but nevertheless significant, and without the similar adverse effects as those that result in the registration tax scenario. Employment effects are less crucial in this case, and thus there is less call for adjustments in the economy as a whole, as well as from the perspective of the industry. In regard to the latter though, the redesign of the circulation tax will imply that the Swedish car makes loose market shares 134 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars to other brands. Consequently, the industry might gain from adjustment to the new tax conditions by means of e.g. development and marketing of new or modified car models, and an enhanced effort in regard to marketing and sales. While there is a net benefit to society from redesigning the circulation tax to include a CO2 differentiated element, it is important to note that there will still be distribution effects from such an initiative. Thus, rural population will suffer the highest loss in welfare, and be affected relatively more by changes in the tax burden compared to the rural population, and so will also families with children compared to those without children. It should be noted that the similar distribution effects will occur also in the case of the registration tax scenario, but in that case the order of magnitude of the welfare loss and the tax burden effect will be larger. The main reason is that people in rural areas tend to buy larger, and less CO2 efficient cars and they tend to have a stronger preference for the car, i.e. the price elasticity is smaller. 10.3 Policy implications The analysis points to the following policy implications: • It would not be efficient to reduce CO2 emissions from the car fleet by means of reducing the number of cars. The cost benefit analysis shows a significant net cost in this case. This calls for the use of other means if the aim is to reduce CO2 emissions from passenger car transport. The total CO2 emissions from the car transport equals: Total CO2 = number of cars * Km driven per car * average g CO2 per km Thus, other means can seek to influence the annual mileage and/or the average emissions per car. The latter is the idea in the CO2 differentiation of the circulation tax, which does not at all aim to influence the number of cars. Another means, which has not been considered in this study, would be to seek to control the annual mileage. This could be done by making it relatively more expensive to drive the car. Taxes on vehicle fuels or road pricing would be possible options in this case. • There is no special need for a registration tax if the wish is to shift demand towards more energy efficient cars. In fact, adding a CO2 differentiated element to the existing annual circulation tax seems more efficient both with respect to CO2 reductions and with respect to the distortions caused by the tax (the latter is illustrated in the cost benefit analysis). • Substantial reductions in total CO2 emissions in the order of 2-5% can be achieved even if the level of downsizing is limited to 25%. However, it is important to be aware that there is a limit to what can be achieved through differentiation alone, even if the allowed level of downsizing is increased. The level of 25% actually appears to capture quite a substantial fraction of the potential reductions that can be harvested through differentiation. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 135 • If significantly higher reductions in the average CO2 emissions from new cars are desired, there are two options to pursue this aim. Either, one must allow the tax to become a net subsidy or one must accept a higher tax level (also of the circulation tax). The latter implies that the result would also be a drop in the size of the car fleet. • Increases of the tax level for new cars also leads to increases in the average age of the car fleet, because car owners wait longer before they scrap their old car. This leads to higher emissions. To avoid this effect it could be considered to apply some other measure to speed up the scrapping of older cars – a higher circulation tax for old cars or a scrap premium. • There are distributional effects from both the registration tax and the circulation tax scenario. They are however most outspoken for the registration tax as a consequence of the decline in the fleet. Consequently, there may be a need for compensating measures, in particular in the case of the registration tax. Thus, the registration tax scenario affects the population in rural areas and families with children more than the other car purchasers. 10.4 Applicability of study results The "what if" approach that is inherent in the model implies that the model provides a fairly transparent means of analysing the effects of changes in the level and structure of relevant fiscal measures. In other words, the "what if" model based calculations provide a means of undertaking quite isolated analyses of the CO2 effectiveness of existing taxation systems and fiscal measures scenarios. Results achieved from the calculations can be (quite) directly related to the applied input data and to the assumptions made. They are not affected by the inclusion of for example macro-economic projections, modelling of future car production costs, fuel prices forecasts and car pricing policies and modelling of future consumer car preferences. This merit arises exactly because of the "what-if" nature of the model. However, the complementary feature of this approach is that the results of the model can in no way be interpreted as projections of what would actually happen in 2020. In reality, actual developments will namely be affected by a number of factors that are not taken into account in the model calculations. These factors include, but are not limited to, the above mentioned issues. Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars 11 137 References DETR (2001): Final Report on the Factors that Influence Car Owners Decisions when they Scrap Cars and HGV's Directorate-General for Environmental Protection, the Netherlands (2001): Study on the potential effects of possible fiscal measures to reduce CO2emissions from passenger cars in the Netherlands, Final Report, December 2001 Energy Research Programme (EFP), Denmark (1998): Personbilers energieffektivitet, muligheder for forbedring gennem afgiftstrukturen, November 1998. EU Oil Bulletin Petrolier (2001): 10/12-01 (http://www.np.no/ "Fakta i Tal" -> "International statistics" European Commission's Directorate-General for Environment (2002): Fiscal Measures to Reduce CO2 Emissions from New Passenger Cars, Main Report, Forthcoming 2002 European Commission's Directorate-General for Environment (2002): Fiscal Measures to Reduce CO2 Emissions from New Passenger Cars, Annex report: Sweden, Forthcoming 2002 Johansson, Oluf and Lee Schipper (1997): “Measuring the long-run fuel demand of cars”, Journal of Transport Economics and Policy, September, pp. 237-252 Jong, Gerard de et al (2001): Vehicle Scrappage, Literature Survey and new Stated Preference Survey. Paper presented at the Association of European Transport conference, Cambridge, 2001. Ministry of Transport, Denmark (1997): Samfundsøkonomisk omkostningseffektivitet i transportsektoren, Arbejdspapir 1, Marts 1997 Ministry of Transport, Denmark (2000): TEAM2000, Et værktøj til at beregne energiforbrug og emissioner i Danmark, Teknisk Rapport 138 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Sandström, Mikael (1998): Ekonomiska styrmedel på vägtrafikområdet, Rapport för trafikbeskattningsutredningen SCB (1998): I/O table SCB(2000): Lønstatistisk Årsbok 2000 SCB (2001): Statistical Yearbook, 2001 SIKA (2000): ASEK kalkylvärden i sammanfattning, Report 2000:3 (abridged version of "Review of principles for social economic estimates and estimates in the transport sector", SIKA Report 1999:6, Report 2000:3) SIKA (2001): Fordon enligt bilregisteret fjärde kvartalet och hele året 2000, Statistiska Meddelanda, 01:1 SIKA (2001): Polisrapporterade vägtrafikolyckor med personskada (Road traffic accidents with personal injury reported by the police), Statistiska Meddelanda, 01:7 VTI (1986): Personbilsinnehavet i Sverige 1950 - 2050, VTI rapport 301. VTI (1993): Mål för nya personbilars genomsnittliga bränsleförbrukning. VTI Rapport 386. VTI (1995): Modeller och prognoser för regionalt bilinnehav i Sverige VTI (2000): Bilismen i Sverige, 2000. Annexes Annex A 1 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars A. The Car Choice Model The car choice model estimates the effect of a registration tax on CO2 emissions from new passenger cars. This is done by calculating the demand for new passenger cars in Sweden and by comparing a base scenario with alternative scenarios. The registration tax is used to alter the incentives that face each individual when he/she has to choose between different vehicles. The car choice model consists of three sub models. The first sub model allocates the cars into private and company cars. Subsequently, the other two sub models allocate the demand for each specific type of private and company car respectively. In each sub model consumers’ choices are modelled under the assumption that they are economically rational and maximise their individual utility. B.1 The choice between a private or a company car The choice between buying a private or a company car is modelled by calculating the benefit of having a specific car as a company car relative to having it as a private car. This is equivalent to modelling the cost of having the car as a private car minus the cost of having it as a company car. These costs (and hence the choice) is closely linked to the taxation rules for acquisition and use of company cars. The first decision-problem facing each individual car buyer is illustrated in Figure 11.1. Figure 11.1 Decision tree for the choice between choosing a private or company car With each alternative there is associated a utility function Vi = β1*C1i + β2*C2i + ...+ εi (Equation 1) where Vi is the utility of alternative i (i = private or company car), the β’s are parameters, C is a vector that contains characteristics of the car and ε is a stochastic error term. The deterministic part contains the explanatory variables pertaining to the consumer and to the choice. The stochastic part of the utility function can be interpreted as representing the unobserved factors that influences the consumer's utility of each alternative. Annex A 2 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars The first choice facing each car buyer in the model is a discrete (binary) choice: either he/she buys a company car or a private car. Therefore the choice is estimated by applying the logit model. The consumer is assumed to make the choice associated with the highest utility. Since the utility function for each choice contains an unknown stochastic component, the logit model states the probability that a given choice is made. The probability for choosing a company car is based on the observed explanatory variables P(companycar ) = eV eV + 1 (Equation 2) The parameters for this model have been estimated based on data from Denmark 1997 and Germany 1999 – 2000 and the appropriate taxation rules in these two countries. After the estimation of the choice between a private and a company car the two other sub models are invoked to determine the exact type of car bought by the consumer. B. 3 Private car choice model The private car model allocates the total demand for private cars on specific cars. The observable characteristics, which are the driving force in the private car choice model are the following • • • • • Price of the car (inclusive tax and VAT) Running cost (fuel and circulation tax) Size of the car (length) Luggage capacity Acceleration Furthermore the model include a “home-market” parameter to account for the fact that consumers tends to choose cars produced in their own country, i.e. Volvo or Saab. The parameters (elasticities)36 in the private car choice model have been estimated based on a full-scale data set from all new registrations in Denmark in 1997. In order to mirror Swedish conditions and the Swedish socio economic distribution Swedish parameters have been calibrated based on the estimated Danish parameters. The private car model has 45 “agents”. These agents differ with respect to family type, household income and age. Separate sub models have been estimated for 24 types of car buyers, where the car buyers are not divided into different age groups, cf. Table 11.1. 36 Parameter estimates and elasticities are not the same, but the size of the parameter estimates determines the size of the elasticities. Annex A 3 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 11.1 Car buyer family type and income in basic data Single female Single male Female living with parents Male living with parents Couple without children (female buyer) Couple without children (male buyer) Couple with children (female buyer) Couple with children (male buyer) Total Low Income 2173 3044 747 3705 3618 11233 3618 3149 31287 Medium income 6055 7148 632 2204 7901 22277 7901 22071 76189 High income 109 180 21 69 707 1513 707 1419 4725 Total 8337 10372 1400 5978 12226 35023 12226 26639 112201 Furthermore, car buyers aged 18 to 29 have special preferences for acceleration. Therefore these individuals have their own parameter for acceleration. The private consumer behaviour is modelled in a two-step procedure. It is assumed that individuals first determine what type (Estate, Hatchback, Saloon, etc) of car he/she would like to buy. Then, secondly, individuals determine what exact car (Volvo S70 2.5, Saab 9-5 2.0T, VW Polo 1.6i, etc) he/she wants, given the choice of type, cf. Figure 11.2.37 Figure 11.2 Choice structure in a multinomial nested logit model The illustrated choice structure is modelled by applying a nested logit model. The nested logit model is also a discrete choice model, which in this case combines the probability of choosing a specific type of car with the probability of choosing a specific car. The probability of choosing a given combination of type and car is in the nested logit framework given as P ( i, j ) = P ( j ) ∗ P ( i | j ) 37 (Equation 3) Each car has been allocated to one of nine different types of car: Cabriolet, Coupé, Estate, High Volume Estate, Mono High volume, Hatchback, MPV, Off-Road Vehicle, Roadster, Saloon or Targa. Annex A 4 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars θ log = e å åe i∈ J θ log e j∈J Vi åe i∈ J Vi * eVi å i∈J eVi The parameters of the utility functions (the β’s) are estimated using maximum likelihood. The summation is running over the index j, where j = Estate, Hatchback, Saloon, etc. and index i, where i = Volvo S70 2.5, Saab 9-5 2.0T, VW Polo 1.6i, etc. To estimate the parameters in the model each agent has been given a random set of alternative cars to choose from (63 cars for each choice). The estimation procedure then estimates the set of parameters to the utility functions that gives the highest utility to the car that was actually chosen. The 63 car options differ across consumers from a pool of more than 2000 specific cars After the parameters are estimated it is straightforward to calculate the total demand for each car by first calculating the probabilities of choosing a certain car for each agent and then sum up across all individual probabilities for each specific car. B. 3 Company car model The company car model allocates the total demand for company cars on specific cars. The observable characteristics, which are the driving force in the company car choice model are the following • • • • • • Cost of acquisition (Personal taxation rules) Running cost (Personal taxation rules) Size of the car (length) Luggage capacity Acceleration Horse power Furthermore, this sub model also includes a home market parameter to account for the fact that consumers tends to choose cars produced in their own country. As was the case for the private car choice model the parameter estimates from the Danish company car model have also been compared with similar parameter estimates from a model based on data for 11 EU Member States for the period 1999 – 2000. Again, the comparison showed that the parameter estimates (elasticities) from the Danish model are very close to the same parameter estimates from the EU model. The company car model has 6 “agents” (companies). The model distinguish between three aggregate sectors: Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars • • • Annex A 5 Primary sector: Agriculture, fishery and manufacturing industries (NACE38 1 – 2) Secondary sector: Energy and construction (NACE 3 – 4) Tertiary sector: Service (NACE 5 – 9) Each of these sectors is further divided into two, to take into account that a company car choice generally is determined in two different ways depending on the type of company: • • Company where the company managers decide which car to offer to the employee based on considerations regarding price and running cost. Company where the employee decides which car to have based on considerations regarding his/her cost (benefit tax payments). The company car model uses the same model approach as the private car choice model, i.e. the probability of choosing a given car is estimated in a multinomial logit model.39 However, for the company car sub model the choice is not modelled as a sequential choice as in the private car model. Instead both the type and specific car is chosen in one single step, cf. Figure 11.3. Figure 11.3 38 Choice structure in a multinomial logit model Nomenclature generale des Activitiés économique dans les Communautes Européennes, Industrial Classification Code prepared by EU in 1970 and adopted by all member states. 39 “Multinomial” because the consumer has many different choice options of which he/she shall choose one. In a binary logit model (such as the model that determines whether the consumer will buy a private or a company car) the consumer only has two options available. Annex A 6 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Annex B 1 B. Model validation A series of tests has been carried out in order to validate the model and its use in Sweden. The tests compare results from model based calculations with observed data in order to validate the ability of the model to calculate the actual demand within a tolerable level of uncertainty. The validation has considered the following three parameters: • • • Estimates of parameters (elasticities). Levels of CO2 emissions Registrations of new cars Parameter estimates The validity of the parameter estimates was tested at detailed level combining demand data with car characteristics for Sweden. These tests showed that Swedish car purchasers prefer larger cars than the average car purchaser in EU (and Denmark). To adjust for this difference the parameters (elasticities) in the model have been adjusted as follows: • • • Car Price parameter reduced by 39% Yearly cost parameter reduced by 9% Car length parameter reduced by 29% Furthermore, the model has been calibrated to fit the average company car share (50%), diesel share (6.5%) and overall average CO2 emissions (198 g CO2 per km) CO2 emission levels Figure 11.4 below show the results of the validation tests for average CO2 emissions. The calculations have been done separately for company cars and private cars because of the difference in emission levels among these two groups. Annex B 2 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 11.4 Modelled CO2 emissions compared to observed CO2 emissions g CO2 per km 250 200 150 100 50 0 Petrol Model forecast, Private Diesel Observed Model forecast, Company As shown in Figure 11.4 the model calculates larger CO2 emissions from company cars relative to private cars. This correlates very well with the fact that company vehicles generally tend to be larger than private cars. The observed CO2 emission levels lie between the calculated level for company cars and the calculated level for private cars. This applies both petrol and diesel, and is in line with what could be expected. In Sweden, company cars constituted 50% of the total number of registrations of new cars in 2000. The figure below shows the results of the validation tests for average engine size. Annex B 3 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 11.5 Validation tests for average engine size Engine capacity, ccm 2500 2000 1500 1000 500 0 Petrol Model forecast, Private Diesel Observed Model forecast, Company Both tests show good resemblance between the model forecast and what was actually observed in the market in 1999/2000. New car registrations Figure 11.6, Figure 11.7 and Figure 11.8 show the results of the more detailed validation tests. The figures compare model results with actual data for CO2 emission levels, engine size volumes and engine size classes respectively. At the overall level, the number of registrations calculated by the model corresponds well with the observed data. However, there are also minor deviations from the observed data that cannot be explained by the model. For example, Figure 11.6 shows that the model calculates too few registrations in the region of 210g CO2 per km. Such unexplained deviations are largely caused by country-specific preferences that the model cannot capture. However, the effects from such country specific preferences are assessed to be minor in this study. Annex B 4 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 11.6 Number of registrations as a function of CO2 emissions for petrol cars (g per km) Num ber of registrations 25% Forecast 20% Demand 15% 10% 5% 0% 50 100 150 200 250 300 350 400 CO2 Figure 11.7 Number of registrations as a function of engine size volume (ccm) for petrol cars Num ber of registrations 20% Forecast 15% Demand 10% 5% 0% 500 1000 1500 2000 2500 cm3 3000 3500 4000 Annex B 5 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Figure 11.8 Number of registrations as a function of engine size class for petrol cars Number of registrations 40% Forecast 30% Demand 20% 10% 0% 1 2 3 4 5 Category 6 7 F M Annex B 6 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars C. Sensitivity Analysis Tables Table 11.2 Scenario where the CO2 differentiation is 1,760 SEK per g CO2 CO2 diff 200 Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,127,949 162,522 235 11,937,886 9.1 451 1,338 1,357,555 32.9% 195,821 4.7% Time horizon (in years) 5 10 4,037,098 3,966,893 197,530 182,077 223 204 11,100,366 10,028,634 9.5 10.0 449 447 1,368 1,393 1,247,034 1,174,801 30.9% 29.6% 226,384 261,784 5.6% 6.6% 20 3,895,882 212,710 161 7,784,939 9.7 441 1,396 1,137,224 29.2% 252,585 6.5% Table 11.3 Scenario where the petrol and diesel cars are merged in the estimation of the tax function Diesel and petrol merged Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,127,774 162,210 235 11,939,690 9.1 451 1,338 1,357,585 32.9% 196,770 4.8% Time horizon (in years) 5 10 20 4,036,241 3,965,460 3,893,865 197,292 181,952 212,500 223 204 161 11,110,170 10,047,355 7,815,590 9.5 10.0 9.7 449 447 441 1,368 1,393 1,397 1,247,204 1,175,201 1,138,179 30.9% 29.6% 29.2% 231,734 272,577 273,340 5.7% 6.9% 7.0% Table 11.4 Scenario where the scrap price elasticity is halved Scrap 50% Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars 1 4,127,178 170,606 235 11,931,287 9.0 451 1,338 1,356,863 32.9% 196,067 4.8% Time horizon (in years) 5 10 4,035,914 3,967,476 202,434 183,906 222 203 11,080,853 10,009,457 9.3 9.7 449 447 1,370 1,395 1,246,330 1,175,964 30.9% 29.6% 227,773 263,392 5.6% 6.6% 20 3,898,688 217,701 160 7,759,368 9.4 441 1,398 1,143,907 29.3% 253,392 6.5% Annex C 2 Impacts from CO2 differentiated sales tax on CO2 emissions from passenger cars Table 11.5 Scenario where the phase-in of the price elasticity is speeded up Speed up 1 4,106,747 141,605 235 11,888,423 9.1 451 1,337 1,351,460 32.9% 194,443 4.7% Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Time horizon (in years) 5 10 3,968,639 3,901,696 191,639 184,894 223 204 10,947,838 9,888,138 9.6 10.0 449 447 1,368 1,393 1,227,772 1,157,447 30.9% 29.7% 221,931 257,496 5.6% 6.6% 20 3,871,659 209,192 161 7,744,423 9.7 441 1,397 1,133,933 29.3% 250,807 6.5% Table 11.6 Scenario where the price elasticity is –0.35 Elasticity -0.35 1 4,138,447 173,306 235 11,963,834 9.1 451 1,338 1,360,789 32.9% 196,503 4.7% Number of cars New car sales Average CO2 emissions (g CO2/km) Total CO2 emissions (ton) Average age Average length (cm) Average weight (kg) Number of Swedish cars Share of Swedish cars Number of diesel cars Share of diesel cars Time horizon (in years) 5 10 4,086,097 4,046,208 206,407 187,053 222 203 11,217,971 10,214,199 9.4 9.9 449 446 1,368 1,393 1,262,341 1,199,978 30.9% 29.7% 229,565 266,887 5.6% 6.6% 20 4,005,725 220,217 161 8,019,502 9.7 441 1,397 1,173,099 29.3% 259,583 6.5% Table 11.7 New car sales, details for various levels of downsizing Base Avearge CO2 emissions (g per km) 198 0 196 New registration tax Downsizing 0.25 0.5 0.75 195 195 195 1 195 Average lifetime tax revenue (EURO per car) Average size Average Dieselshare Average swedish car makes 10,936 3.70 0.07 0.30 13,134 3.69 0.06 0.30 13,094 3.66 0.06 0.29 13,117 3.64 0.07 0.29 13,203 3.63 0.07 0.29 13,309 3.62 0.07 0.29 Average registration tax (EURO per car) Average Circulation tax (EURO per car per year) Average dealers price, EURO excl. VAT Average (lifetime) tax consists of Registration tax Circulation tax (whole life time) Fuel tax (whole life time) 178 16,571 2,290 178 16,474 2,291 177 16,347 2,339 176 16,260 2,439 176 16,205 2,557 175 16,159 2,816 8,120 2,290 2,805 8,039 2,291 2,789 8,014 2,339 2,779 8,000 2,439 2,773 7,991 2,557 2,769 7,984 RAPPORT 5187 Koldioxidrelaterad skatt på bilar Impacts from CO2 differentiated vehicle taxes on CO2 emissions from passenger cars naturvårdsverket har till uppgift att utveckla styrmedel så att miljömålen om en hållbar utveckling kan nås och Energimyndigheten att driva på utvecklingen för effektivare energianvändning och minskad klimatpåverkan. För att klimatmålet ska kunna uppnås är det viktigt att minska utsläppen av koldioxid från trafiken. Under 2001 har ett projekt om styrmedel för miljöanpassning av transporter genomförts. Som ett delprojekt har COWI på Naturvårdsverkets och Energimyndighetens uppdrag genomfört en studie om koldioxiddifferentierade fordons- och försäljningsskatter. Utgångspunkten har varit att undersöka effekterna av att använda fordonsskatter som medel för att reducera personbilars koldioxidutsläpp. I studien analyseras konsekvenserna av två alternativ – att omforma den befintliga fordonsskatten genom att inkludera en koldioxidrelaterad del och att införa en försäljningsskatt. Rapporten är skriven på engelska med en svensk sammanfattning. De övriga rapporterna i projektet är System för bättre framkomlighet i stockholmsregionen, Ekonomiska styrmedel och Vägavgifter – en kartläggning av svenska och internationella erfarenheter. isbn 91-620-5187-3 issn 0282-7298