Automobile in transition? An economic and environmental Analysis of policies for reducing CO2 emissions from transport

CO₂

a b ility St a n d a r d s : Aver a ge or M a r gin a l E m ission s ? A n E x a m p le of Fer t ilizer

U se a n d C or n E t h a n ol 57

3.1 Int roduction . . . . . 3.2 The Analyt ical Framework . 3.3 A Numerical Example . . . 3.4 Discussion and conclusions .

59 62 71

79

4 A Fr amework t o Evaluat e P olicy Opt ions for Support ing Elect ric Vehicles in

U rban Fr eight Tr ansport 83

4.1 4.2

4.4 4.5

Int roduction . . . Lit erat ur e review

4.3 A framework for policy evaluat ion Numerical Experiment s

Conclusion .

84 87 89 98 107

11

A cknow ledgem ent

T here are a numb er of people who have played p ar t icular ly imp ort ant roles dur ing my P hD degree and deserve t o b e named .

I owe sp ecial t hanks t o my primary sup ervisor , Gunnar S. Eskeland who sp ent countless hour s discussing my resear ch , supp ort ed me for various research act ivit ies, and int roduced me t o int erest ing and helpful p eople in b ot h t he academia and t he business. T h ank you for being a good collab orat or and a firm friend as well. Many t h anks for Linda Nøst bakken and Knut Einar R osendahl who provided t heir supp ort s and professional advice t o my research .

T he collab orat ion , feedback and supp ort provided dur ing my resear ch has b een highly appr e- ciat ed , not ably t hat received from J onas Andersson (Norwegian School of Economics) , Xueqin Zhu (Wageningen Universit y & R esearch) , Edwar d Sm eet s (LEI- Wageningen Universit y &

Research ) , Leif Krist offer Sandal (Norwegian School of Economics) , J oshu a Linn (R esour ces for t he Fut ur e) , Seyed Most afa Mirhedayat ian (Norwegian School of Economics) , F loris Tobias Zout m an (Norwegian School of Economics) , and Vivienne Bowery Knowles (Norwegian School of Economics) .

My friends and P hD colleagues, Bendik , Mai , Yun , Xingyu , Lar s, Ondfej, Aij a , Evan , P at rick , F inn , Xiaoyu , Yuanming, Xunhua , Mario, Chunb o and Andreas, among ot her s: It has b een a pleasur e t o know you . Your friendships and advice have been a great sour ce of mot ivat ion t hese last four years. It were you t hat h ave accomp anied me in t he cold , dar k , raining wint ers and also t he spirit ed , st unning, sunny summer s in Bergen , Norway where I finished my P hD . T he final acknowledgem ent s must go t o my p ar ent s, Qin Chen and Weihong Yan for t eaching me t he value of educat ion , for cheering me up when I feel discour aged and for giving me t he confidence t o chase my dreams.

Shiyu Yan Bergen , Sept emb er 2017

CO₂

CO2

CO2

CO₂ CO₂

CO₂

CO2

CO₂ CO₂ CO₂

CO2

CO₂ CO₂

CO2

CO₂

CO₂ CO₂

CO₂

CO₂

CO₂

CO₂ CO2

CO₂ CO₂

CO₂

CO₂

CO₂ CO₂

CO2

CO₂

CO₂

g CO₂eq./ M J

CO

₂

CO₂ CO₂

CO₂ CO₂ CO₂

CO₂

CO2

CO₂

CO2

CO₂g/ km

CO₂

CO₂

CO₂

CO2 CO2

CO₂

CO2

CO₂

CO2

CO2

CO2

CO2 CO2

CO2

CO2

CO2

CO2

CO2 NOx

CO₂

CO₂

CO2

CO₂ CO₂

CO₂

CO₂

CO₂

CO2

g/ km g/ km

CO2

CO₂ CO₂

CO2

CO₂

CO₂

CO2

CO₂

CO₂

NOx

CO2

CO

CO₂

CO2

CO2

CO₂

CO2

CO2

CO₂

CO₂

CO₂

CO₂ CO₂

CO2

CO₂

CO₂

CO2 CO2

CO₂

CO₂ CO₂

CO₂

CO₂

CO₂ CO₂

Q_H Q_L

U = U(Q_L, Q_H)

Q_H

T_c < 0 < Q_L T_c

CO₂

CO₂

Ci = (Pi ,t₀+ Ti , t₀) + ^T_{t = 0}^{i Mi t+ a cti t+ ( f p}_{( 1+ )}^{t+ f t}_t ^{t) f eiDi t} Pi , t₀

t0 Ti ,t₀ i

t0 (Pi ,t₀ vr ti ,t₀) acti t

Di t f pt f tt

f ei i

CO₂

CO2

CO₂ CO₂

CO₂

g/ km CO₂

g/ km g/ km

j _i

i t

i

CO2

j

Ti t

CO₂ CO₂

i j t

j t _i

j t

CO₂

CO2

g/ km CO2

CO₂ CO₂

k

ln Q_{i t} = ₁T_{i t}+ ₂T_{i t}g_k+ ₁F C_{i t} + _{j t} + _i + _{i t}

i t 2 i t 2 i t j t i i t

1× 2

CO₂

CO2

R²

CO₂

CO2

g/ km g/ km g/ km g/ km g/ km

g/ km g/ km g/ km g/ km

CO2

R²

R²

R²

g/ km CO₂

g/ km CO2

CO₂

g/ km g/ km g/ km g/ km

± g/ km

R²

CO2 CO2

CO2 CO2

g/ km g/ km

CO₂ g/ km

CO₂

CO₂ CO₂

CO2

CO2 CO2

CO₂ CO₂

CO₂

CO2

CO₂

CO2 CO2

CO₂

CO₂ CO₂

CO2

CO₂ CO₂

× × CO₂

CO₂ CO₂

CO2 CO2

CO₂

CO2 CO2

CO₂

CO₂

CO₂

CO2

CO₂

CO₂

CO2

CO₂

CO2

CO2

CO₂ CO₂

CO2

CO₂ CO₂

CO₂ CO2

CO₂

CO2

CO₂

CO2

CO2

A cknow ledgem ent

We are grat eful for help and comm ent s from Linda Nøst bakken , It ziar Lazkano, J onas An- dersson , F loris Zout man , J ar le Møen , Lasse Friedst rom, Vegar Østli, and ot her colleagues at t he Norwegian School of Economics and discussant s in workshops and conferences. We grat e- fully acknowledge financial supp ort from t he Norwegian Resear ch Council t hrough t he Cen SE S program, and delivery and help wit h dat a from t he Informat ion Office of Road Traffic, ofv .no.

CO₂

CO₂

CO2

$/ kW h $/ kW h

CO₂

CO₂

CO2

CO₂

CO₂

CO₂ g/ km

CO₂ g/ km g/ km

g/ km g/ km g/ km g/ km

CO₂ g/ km g/ km

CO₂

CO₂ g/ km CO₂ CO₂ g/ km

CO₂ g/ km

NOx CO2

CO₂

CO2 CO2

CO₂ g/ km

e g/ km

e

CO₂ NO_x

T C_i^v = p^v_i + vr t^v_i +

l

t = 0

act^v_{i t} (1 + )^t +

l

t = 0

(ep^v_{i t}+ et^v_{i t}) × ee^v× di t

(1 + )^t +

l

t = 0

mc^v_{i t} (1 + )^t

p^v_i v i cev bev i

vr t^v_i l

mc^v_{i t} i t act^v_{i t}

d_{i t} ep^v_{i t}

v = i cev v = bev et^v_{i t} ee^v

N B_i

T C_i ^{i cev} T C_i^bev

N B_i = p^v_i + vr t^v_i +

l

t = 0

act^v_{i t} (1 + )^t +

l

t = 0

(ep^v_{i t}+ et^v_{i t}) × ee^v× d_{i t} (1 + )^t +

l

t = 0

mc^v_{i t} (1 + )^t

i nc_i

i nc_i = (vr t^{i cev}_i +

l

t = 0

act^{i cev}_{i t}

(1 + )^t) (vr t^ev_i +

l

t = 0

act^ev_{i t} (1 + )^t)

R²

i m t

ln si m t = ln i nci m t + ln f ueli m t + ct+ m t + i m t

si m t m i t

f uel_{i m t}

m t

ct

E Bi

E C_{i t}^v i

ecm_it^v ece^v_{i t}

ecn^v_{i t}

E B_i = E C_i ^{i cev} E C_i^bev

E C_i^v=

l

t = 0

(ecm^v_{i t}+ ece^v_{i t} × ee^v× d_{i t}+ ecn^v_{i t} × d_{i t}) (1 + )^t

CO2

CO₂

ece^v_i =

j

r^v_{i j} + c^v_{i j}

r_{i j}^v v i cev bev

r^bev_{i j} j i

r^{i cev}_{i j}

j i c^v_{i j} v i cev bev

ecn^v_{i t}

km

e

l s

Italy Austria Hungary Portugal Germany ci rc ulation tax change - Energy price change

Norway France UK Austria Italy Portugal Hungary Germany - Registra!ion tax change - Circulation tax change

En e rgy tax change En ergy poncechange

Maintenancecostchange

(a) Mit subishi i-M iEV/(mini ) (b) N issan Leaf ( comp act)

(c) B MW i3 (m id size) ( d) T esla (sp orts)

F igur e 2.2: Benefit s of swit ching from an ICEV t o a BEV by car model.

Cross-driver comparison

T he net ben efits of swit ching t o a BEV vary wit h drivers who t r avel different dist ances annually.

Table 2.2 present s t he requirement s of annual dist ance t r avelled t o achieve breakeven condit ions in which t he t ot al ownership cost s of BEVs and ICEVs ar e equal. Drivers wit h higher annual dist ance t ravelled are more likely t o buy BEVs, because more energy cost s will be saved by swit ching from an ICEV t o a BEV . In addit ion , t ravelling longer dist ances means lower pur chase cost of a BEV per kilomet re. In Norway, Fr ance, and t he UK , where huge t ax incent ives are provided , drivers can benefit from swit ching t o BEVs wit hin a norm al r ange of annual dist ance t ravelled . Ot her count ries require drivers t o have an ext remely high annual dist ance t r avelled in order t o b enefit from driving BEVs. However , owing t o t echnical limit at ions, BEVs are not able t o fulfil t he necessary annual dist ance t r avelled sufficient ly. In fact , t his conflict leads t o low sales of BEVs.

46

km

e / km

R²

R²

CO₂

kg CO₂

CO2 e / kg

CO₂ e / kg

CO₂

CO₂

BEVs. It is imp ort ant t o p oint out t hat key prerequisit es for t he success of BEV adopt ion are t he availability of clean elect ricity and t he significant reduct ion of BEV product ion cost s.

Fur t hermore, futur e resear ch on t he deployment of new vehicle t echnology ( e.g. BEVs) needs t o t ake int o account local incent ives ( e.g. regional dist ribut ions of char ging st at ions, commut ing rout es, access t o bus lane, and free par king for elect ric vehicles) in order t o b et t er assess t heir sp ecific merit s relat ive t o alt ernat ive vehicle opt ions ( e.g. ICEVs) .

A cknow ledgem ent

I am grat eful for help and comm ent s from Gunnar S. Eskeland , Linda Nøst bakken from t he Norwegian School of Economics, and Knut Einar Rosendahl as well as Eirik Romst ad and St åle Navrud from t he Norwegian Universit y of Life Scien ces and discussant s in workshops and conferences.

52

A pp endix A

Table 2.5: Summary st at ist ics.

Variable Observat ion M ean St d . D ev . M in M ax

B EV sales numb er 746 128.2386 513.4582 0 5970

B EV sales share 746 0.000463 0.002614 0 0.033155

Incent ive (5% discount rat e ) 746 3811.982 10373.92 0 128947.7 Incent ive ( 10% discount rat e ) 746 3699.247 10278.77 0 127761.5 Vehicl e regist rat ion t ax 746 3128.461 9886.945 0 12 1963.5

A nnual ci rculat ion t ax 746 84.10644 190.4441 0 1899

Fu el sav ing 746 6319.933 3598.464 2091.94 19438.61

Table 2.6: Vehicle pairs of BEV and ICEV models. NissanRenaultRenaultTeslaMitsubishiCitroenPeugeotSmartVolksw BEVLeafFlucncc Z.E.ZoeModel SBMWi3i-MiEVC-ZeroiOnFortwo EVc-up! Basic price26400287842400044925310131800018000180001875022 Engine power80706626012547474755 Energy efficiency0.1730.140.1460.1810.1290.1350.1350.1350.151 Weight15211543146820901195108010801080900 NissanRenaultRenaultBMWBMWMitsubishiMitsubishiMitsubishiSmartVolksw ICEVNoteFlucncc Z.E.Clio740iScricslImiragcImirageImirageFortwo ICEup Basic price12000200001300058000280001040010400104001000015000 Engine power64826623512559595952 Engine size1198159889829791598119311931193999 Energy efficiency0.0490.0680.0430.1060.0480.0370.0370.0370.043 Cr J Weight15461747100918251315845845845780 CO2 intensity1391559918413796969697

ln si t = ln i nci t+ ln Xi t+ pr oi+ yeart + i t

si t i t i nci t

X_{i t}

pr o_i

i yeart i t pr oi

Chapt er 3

How t o Measure Greenhouse Gas Emissions by Fuel Ty pe for B inary Sust ainability St andards : Average or Marginal Emissions? A n Ex ample of Fert ilizer U se and Corn Et hanol

Xueqin Zhu

Environment al Economics and Nat ur al Resources Group , Wageningen University, Hollandseweg l , 6706KN, Wageningen , t he Net herlands

Shiyu Yan

Depart ment of Business and Management Science, Norwegian School of Economics, N-5045 Bergen , Norway

Edward Smeet s

LEI Wageningen UR , The Hague, t he Net herlands

Siemen van Berkum

LEI Wageningen UR , The Hague, t he Net herlands

kgCO₂ eq./ M J

kg CO2 eq./ GJ kg CO₂ eq./ GJ

t he m arginal emissions of a biofuel ( e.g. emissions from t he last unit of biofuel produced from t he last unit of corn) for evaluat ion of emission savings.

Fur t hermore, t he biofuel sect or is closely connect ed t o ot her economic sect ors. T here is a close economic linkage b etween t he input s of corn et hanol (e.g. nat ur al gas) t hrough commodity m ar ket s and government p olicies (T yner et al. , 2012; Bab cock, 2013) . P rice changes in t he oil m ar ket have a direct impact on t he dem and for et hanol and t he price of gasoline t hrough en ergy m ar ket s (Tyner and Taherip our , 2007; Serra et

al.,

2011) , whi ch influences t he use of input s for et hanol product ion ( corn and fert ilizers) and t here by G H G emissions. Cur rent biofuel policies (e.g. t ax for fossil fuels and t ax credit s for biofuels) change t he economic incent ives of economic agent s t o choose t heir energy product s, which have impact s on GHG emissions.

T he aim of t he resear ch present ed in t his chapt er is t o develop a modelling framework for evalu- at ing t he impact of t he economics of nit rogen fert ilizer use on GHG emissions, based on average and m arginal GHG emissions, considering t he int eract ions of en ergy and agricult ur al market s.

T he model is applied t o t he case of corn et hanol product ion in t he US, and it p ar t icular ly concerns t he impact s of oil price development s and et hanol p olicies, especially t he Volumet ric Et h anol Excise Tax Credit (VEET C) , on average and marginal GHG emissions. T he novelty of t his resear ch is t hat we ar e able t o calculat e b ot h average and m argin al GHG emissions of biofuels, which can b e comp ar ed t o t hose of fossil fuels, and provide useful insight s on t he act ual emission savings of biofuels.

T he rest of t his chapt er is st ruct ur ed as follows. Sect ion 3.2 pr esent s an analyt ical framework for calculat ing t he average and m ar gin al GHG emissions of corn et h anol, t aking int o account energy and agricult ur al market int eract ions. T he framework consist s of an economic model t hat links t he oil price t o t he prices of gasoline, corn et h anol, and corn, t he price of fert ilizers used for t he product ion of corn, and t he price of nat ur al gas for t he product ion of corn et hanol. GHG emissions from corn product ion dep end , among ot hers, on t he applicat ion rat e of fert ilizers.

An economically maximized rat e was recommended in t he Corn Belt in t he US (Sawyer et al. , 2006) . T herefore, we det ermine t he economically opt im al nit rogen applicat ion rat e based on profit m aximizat ion . To calculat e t he average and m argin al emissions of corn et hanol, t he economic model is combined wit h t he dat a on t he GHG balance of corn et hanol from t he Greenhouse Gases, Regulat ed Emissions, and Energy Use in Transp ort at ion (GRE ET ) life

out put s in t his st age influences t he nit rogen applicat ion rat e, which definit ely impact s t he GHG emissions from corn cult ivat ion . In t he st age of et hanol product ion , in which corn is convert ed t o et hanol, nat ur al gas is t he second import ant input aft er corn. T herefore, any exogenous forces which influence t he price of nat ur al gas and t hus, t he cost of product ion of corn ( e.g. a change in oil price) impact t he price of et hanol. T his has a feedback effect on corn product ion and fert ilizer use in t he first st age, and t hereby result s in changes in GHG emissions. In t he st age of blending wit h gasoline, exogen ous changes ( e.g. oil price changes) have implicat ions for t he price of gasoline and et hanol t hrough t he en ergy m ar ket . T his again has a feedback effect on t he product ion of corn as well as fert ilizer use, and influences final GHG emissions. Below, we present t he quant it at ive relat ionships t hat describ e t he opt imal applicat ion rat e of nit rogen and t he market int eract ions of input s and out put s, which finally det ermine t he GHG emissions of corn et hanol.

ransporto fdry

·I

^{GHGemi s sions}

Fuel com

Nat ural gas GHGemissions

Corn-ethano l conversion

Transport of GHGemi ssions

Fuel ethanol

Ethanol blending OHO emissions

Gasoline (distrib utio n)

Biofuel

po licy Ethanol

prod uel

---·

F igur e 3.1: T he analyt ical fr amework for calculat ing GHG emissions from corn et h anol produc- t ion .

3 .2 .1 E co no mi c analy sis

In t his subsect ion , we elaborat e in a m at hem at ical model t he opt imal applicat ion r at e of nit rogen fert ilizer in corn cult ivat ion , t he price relat ionship of input s and out put s in et hanol product ion , and t he price relat ionship bet ween et hanol and gasoline under biofuel policies. T he opt im al applicat ion rat e of nit rogen fert ilizers is det ermined by t he profit m aximizat ion of corn farmers, t aking int o account t he yield resp onse t o t he nit rogen input . T he price relat ionship of corn, nat ur al gas as an input , and et hanol as an out put in et hanol product ion is det ermined by t he equilibrium condit ion in which no posit ive profit of et hanol product ion is earned under const ant-ret urn-t o-scale t echnology. As for t he price relat ionship of et hanol and gasoline , energy

64

kgN / ha $/ kgN

cor n = Qcor nPcor n N Pn i t r og en Cot her

Q_{cor n} kg/ ha P_{cor n} Pn i t r og en

$/ kg $/ kgN N

kgN / ha C_{ot her}

$/ ha

Qcor n = m + n × N kN²

m n k N

N = [(Pn i t r og en/ Pcor n) n]/ ( 2k)

Q_{cor n} = [4mk + (Pni t r og en/ P_{cor n})² n²]/ ( 4k)

$/ m³

et han ol = P_{et hanol} P_{cor n} Pnat ur al g as+ P_{D D GS} c₀

P_{ethan ol} $/ m³ P_{cor n} $/ kg P_{nat ur al}

$/ m³ PD D GS $/ kg c0

m³ kg m³

kg m³

P_{D D GS} = xP_{cor n} x

ethanol = P_{et han ol} ( x)Pcor n Pnat ur al g as c0

Pcor n = ( Pet hanol Pnat ur al g as c0)/ ( x)

m³

km m³

P_{et hanol} Pg asol i n e

P_{et hanol} = (Pg asol i n e+ t) t

P_{et hanol} = Pg asol i ne (1 )t + t_c

tc

$/ m³

Pn at ur al g as = a₁+ a₂P_{cr udeoi l}

Pg asol i ne= b₁+ b₂P_{cr udeoi l}

Pnat ur al g as Pg asol i ne Pcr udeoi l

$/ m³ a₁ a₂ b₁ b₂

Pf er ti l i z er = c₁+ c₂P_{cr udeoi l}

c₁ c₂

CO₂ M J

M J CO2

g CO2 eq./ M J

E_et

kg / m³

kg/ ha m³/ ha

Q_{cor n} = Q_{et hanol}

ha

T E (N , Q_{et hanol}) = Ecc(N , Q_{et hanol}) + Ect( Q_{et hanol}) + Eep(Q_{et hanol}) + Eet(Q_{et hanol})

m³ Eav er ag e

Eav er ag e= T E (N , Q_{ethan ol}) Q_{ethan ol}

m³ M J

T E Q_{ethan ol}

T E Q_{et hanol}

T E Q_{et han ol}

N N N N N

Qcor n(= Qcor n + Qcor n)

Q_{et hanol}(= Q_{et han ol}+ Q_{et han ol}) T E N

Qethan ol

Em ar g ai n l

Em ar g inal = T E (N , Q_{et hanol}) Q_ethanol

T E (N , Q_{ethan ol}) T E (N , Q_{et hanol}) Q_{et hanol} Q_{et han ol}

lb kg

P_{cr udeoi l}

$/ bar r el

$/ bar r el

g CO₂ eq./ M J

m n k

B tu bushel gallon l b.

c0 $/ bushel

x

t $/ gal lon

tc $/ gal lon

a1

a2

b1

b2

c1

c2

$/ bar r el $/ bar r el

$/ m³

$/ kg

$/ kgN

kgN/ ha kg/ ha

$/ ha M J/ ha

g CO2 eq./ M J g CO2 eq./ M J

$/ bar r el

$/ bar r el

$/ bar r el

$/ m³

$/ m³

$/ ha

$/ ha

kg

bar r el kg hectar e

$/ bar r el

$/ bar r el

g CO₂ eq./ M J

c₁ c₂

CO₂

$/ bar r el $/ bar r el

$/ m³)

$/ kg

$/ kgN

kgN / ha kg/ ha

M J/ ha

g CO2 eq./ M J

g CO2 eq./ M J

saving p ot ent ial of biofuels. For example, t he product ion and use of nit rogen fert ilizers account s for one-t hird or more of t he GHG emissions of corn et hanol p roduct ion in t he US. T hus, changes in fert ilizer use can have a large imp act on t he G HG-saving p ot ent ial of corn et hanol. T herefore, we apply t he modelling framework developed t o evalu at e t he impact of t he correlat ion b et ween oil m ar ket s and t he market s for et hanol and corn in t he US on nit rogen fert ilizer use and on t he GHG emissions of corn et hanol.

T he result s show t hat a higher oil price result s in higher gasoline, et hanol, and corn prices.

T he profit-m aximizing b eh aviour of far mer s result s in an increase in t he use of fert ilizers t o increase t he product ion of corn. T he effect is t hat t he average GHG emissions p er unit of corn et hanol remain fairly const ant , but t hat t he m arginal emissions increase somewhat (5%) , m ainly as a result of decreasing m arginal yield wit h resp ect t o fert ilizer use. T he conclusion is t hat alt hough higher corn yields result in higher GHG emissions, increasing corn et h anol product ion for fuel reduces GHG emissions on average compar ed t o t he alt ernat ive of increasing gasoline product ion .

It should be not ed t hat our analysis is based on an economically opt imal applicat ion rat e of fert ilizers. In reality, risk-averse farmers might overuse fert ilizers owing t o lack of knowledge ab out decreasing marginal yields wit h respect t o nit rogen fert ilizers. T hus, our calculat ion based on economically opt im al applicat ion rat es might underest im at e real emissions. Next , we do not include t he ot her indirect effect s of et hanol product ion , such as land-use change. Hence, our num erical result s on t he margin al emissions reflect only t he lower bound of real emissions relat ed t o t he last unit of et hanol product ion . Fur t hermore, t he use of a linear relat ionship for t he market int eract ions of energy and agricult ur al product s based on hist orical dat a b efore 2007, wit hout considering t he recent development of shale gas, might lead t o over est im at ion of t he economic response of higher oil prices. T he obj ect ive of t he exercise present ed in t his chapt er is not t o produce a t horough calculat ion of GHG emissions of corn et hanol product ion in t he US, which requires est im at ing t he act ual applicat ion rat es in different r egions. R at her , t he modelling framework pr esent ed in t his ch apt er aims t o illust rat e how different effect s can b e

80

t aken int o account when calculat ing emissions. T he novelty is t hat t he diminishing product ivity of corn wit h r esp ect t o nit rogen fert ilizers can h ave profound impact s on t he m arginal emissions of biofuels. T he result s show t h at t he m argin al emissions of corn et hanol product ion in t he US can b e subst ant ially higher t han average emissions, t her eby quest ioning t he efficien cy and effect iveness of biofuel p olicies t o reduce GHG emissions.

An imp ort ant limit at ion of t he modelling framework applied in t his st udy is t hat it represent s only short -t erm economic corr elat ions. Our numerical example t akes t he relat ionship b et ween oil price and nat ur al gas based on hist orical dat a b efore 2007. T herefore, we should be aware t hat t he huge increase in shale gas supplies in r ecent year s might have changed t his quant it at ive relat ionship . Increasing t he use of fert ilizers is, in t he short t erm , a logical and simple way t o increase yields and t o opt imize economic r et urns in resp onse t o higher corn prices. In t he long run , high er corn prices might induce higher corn yields t hrough t echnological changes, such as t he development and use of improved seeds and t he increased use of irrigat ion and agricult ur al m achinery. In t hat case, t he increase in GHG emissions will b e reduced owing t o t he use of improved corn product ion t echnologies and higher corn yields.

Anot her limit at ion of t he research is t hat t he numerical example for applying t he modelling framework does not include ot her indirect effect s, alt hough t he m ar ket int eract ions of input s and out put s are considered . However , we can calculat e t he margin al emissions of corn et hanol, which provides useful insight s for environm ent al m anagement . Economic inst rument s, such as emission t ax , are based on m argin al emissions in order t o det ermine t he opt im al product ion level. T her efore, ident ifying marginal emissions creat es t he basis for p olicy int ervent ion . Moreover , t he prices of oil, nat ur al gas, and agricult ur al commodit ies (including corn) have fluct uat ed subst ant ially dur ing t he t imeframe of t his st udy. T his means t h at t he empirically observed corr elat ions and paramet er values considered in t his st udy are par t ially uncert ain . More det ailed analyses t hat consider longer t imeframes ar e needed t o improve t he accur acy of t he paramet ers used in our modelling framework. Fur t her r esearch is also en cour aged t o model t he economic int eract ions b etween oil, et h anol, and corn m ar ket s in more det ail. T herefore, t his st udy should b e regarded as a first -order assessment t hat , despit e it s uncert aint ies, clear ly shows t he p ot ent ial imp act of t he economic correlat ions between energy and agricult ur al market s on t he average and margin al GHG emissions of corn et hanol. To include ot her indirect effect s of

82

Chapt er 4

A Framework to Evaluate Policy Opt ions for Support ing Electric Vehicles in U rban Freight Transport

Seyed Most afa Mirhed ayat ian

Depart ment of Business and Management Science, Norwegian School of Economics, N-5045 Bergen , Norway

Shiyu Yan

Depart ment of Business and Management Science, Norwegian School of Economics, N-5045 Bergen , Norway

A bst ract

Elect ric vehicles (EVs) ar e considered as a feasible alt ernat ive t o gasoline/ diesel vehicles. Few st udies have addressed t he impact s of p olicies for EVs in ur ban freight t r ansp ort . To cast light on t his t opic, we est ablished a framework combining an opt imizat ion model wit h economic analysis t o det ermine t he opt im al b eh avior of an individual delivery service p rovider company and social impact s (e.g., ext ernalit ies and welfare) in response t o p olicies for supp ort ing EVs, such as pur chase subsidy, limit ed access (zone fee) t o congest ion / low-emission zones wit h exempt ions for EVs, and vehicle t axes wit h exempt ions for EVs. Numerical exp eriment s showed t hat t he zone fee can increase t he comp any 's t ot al cost s but improve t he social welfar e. It great ly reduced t he ext ernal cost inside t he congest ion / low-emission zone wit h a high p opulat ion , dense pollut ion , and heavy t raffic. Alt hough t he vehicle t axes and subsidy have alm ost t he same influences on t he company and societ y, t hey p erform different ly at low t ax/ subsidy rat es due t o t heir different effect s on vehicle rout ing plans. F inally, we performed sensit ivity analyses, which shows t hat local fact ors at t he company and city levels ( e.g., typ es of vehicle and t r ansp ort network) are imp ort ant t o designing efficient p olicies for supp ort ing EVs in t he ur ban freight t ransp ort . K ey words : elect ric vehicle, social welfare, congest ion / low-emission zone, ur ban freight t ransp ort , logist ics, het erogeneous vehicle rout ing problem

•

•

•

CO₂

4 .2 Lit erat ure rev iew

In t his sect ion , we review relevant research on b ot h economic and logist ics research for t he use and evaluat ion of EV p olicies in t he cont ext of ur ban freight t r ansp ort .

Tradit ionally, evaluat ion of t r ansp ort p olicies in ur ban freight t ransp ort involves social and economic issues (Lagorio et al ., 2016) . For example, Hosoya et al . (2003) , And erson et al.

(2005) , Qu ak and De Kost er (2006) , and Holguin-Verasand et al . (2010) p erformed general assessment s of policies t hat affect ur ban freight t ransp ort . Hosoya et al . (2003) st udied Tokyo and used a sur vey t o evaluat e a numb er of freight p olicies: bans on lar ge t rucks, road pricing, and t he const ruct ion of logist ic cent ers. Anderson et al. (2005) provided an ex ant e assessment of regulat ion measur es in UK cit ies, in cluding t ime windows and char ging. Quak and De Kost er (2006) addressed regulat ions based on t ime windows. T hey reviewed pract ices in Dut ch cit ies and assessed p ossible changes t o cur rent p olicy. Holguin-Verasand et al. (2010) evaluat ed t he impact s of p olicy incent ives for encour aging off-hour deliveries on carriers, receivers and societ y.

In part icular , t hey used t he Discret e Choice Model and t he Comprehen sive Modal Emissions Model wit h GP S b ased dat a t o simulat e t he consum er choice of delivering t ime and changes of emissions.

However , t his is st ill an evolving field of research b ecause of t he great er sensit ivity t o envi- ronm ent al issues, new policy measur es, and int roduct ion of new t echnologies . In t he case of promot ing t he pur chase and use of EVs, sever al t ypes of p olicies are involved ( e.g., access t o low-emission zones, exempt ions from vehicle t axes, and pur chase subsidy) . Taefi et al. (2016) reviewed policy m easur es direct ed at emission-free ur ban road freight t r ansp ort . T hey assessed and compar ed p olicies against ot her prosp ect ive opt ions by mult i-crit eria analysis. In t he previ- ous economic resear ch , evaluat ion of EV-supp ort ing p olicies m ainly focused on ex post analysis based on empirical dat a and economet ric approaches, such as t he consumer choice model (Lee et al ., 2016; Greene et al ., 2014) , t he fixed effect model (Chandra et al ., 2010; Gallagher and Mu ehlegger , 2011), and ot her ordinary least squares models (Sierzchula et al. , 2014; Diamond , 2009; J enn et al ., 2013; J imenez et

al.,

2016; Yan and Eskeland, 2016) .

From t he p er sp ect ive of logist ics, t he lit erat ur e on ur ban freight t ransp ort does not yet provide an ample discussion of sp ecific p olicy measur es t o support EVs in ur ban freight t ransp ort . T he

a limit ed or unlimit ed fleet of vehicles wit h different at t ribut es ( e.g., cap acity, fixed cost , and driving range) in order t o serve a set of cust omers wit h given demand . T he obj ect ive is t o decide t he vehicle fleet composit ion and rout es while minimizing t he vehicle rout ing and usage cost s.

J uan et al. (2014) ext ended t he het erogeneous vehicle rout ing problem t o consider mult iple driving ranges for vehicles. T he mult iple driving range variant implies t hat t he t ot al dist ance t raveled by each type of vehicle is limit ed and is not necessarily t he same for all vehicles. T his problem arises in rout ing of EVs (Schneider et al. , 2014; Goeke and Schneider , 2015) and hybrid elect ric vehicles for which t he driving range is limit ed due t o limit ed capacity of bat t eries. Sassi et al. (2014) int roduced a new real-life het erogeneous vehicle rout ing problem where t he mixed fleet consist s of ICEVs and het erogeneous EVs wit h different bat t ery capacit ies (i.e., driving range limit ) and fixed cost s. P art ial recharging for EVs at available recharging st at ions dur ing t rips is allowed , as well as int ermit t ent recharging at t he dep ot . T he m ain challenges facing use of EVs are t heir limit ed driving range and considerably long charging t ime. T he limit ed driving range will probably remain t he m ain obst acle t o using EVs in t he medium t erm as long as t here is no global infrast ruct ur e for replacing bat t eries or direct power induct ion t o EVs dur ing t heir t rip .²

Alt hough t he driving range limit of EVs makes t hem less pract ical for use in real life, advant ages such as free or cheap access t o a congest ion zone, provide an incent ive t o use t hem as an alt ernat ive fleet . T he zone-dep endency asp ect of t he problem t hat we discuss in t his pap er , is similar t osit e dep en den cyin t hesit e-dep endent vehicle routing p roblem int roduced by Nag et al.

(1988) . In t heir problem , different types of vehicles could only visit t heir preassigned cu st omers;

t hat is, no vehicle t raveled from one cust omer t o anot her cust omer unless bot h cust omers were assigned t o t he same typ e of vehicle. T he difference between t he sit e-depen den t vehicle routing p roblem and our problem is t hat , in t he lat t er , t he cust omers are not preassigned t o each type of vehicle. T here are two types of cust omers wit h regard t o t heir geographic locat ion : inside or out side congest ion/ low- emission zones. ICEVs are charged a zone fee when t hey cross t he congest ion/ low- emission zone. Hence, cust omers of bot h t ypes can be p ot ent ially visit ed by

h t t p :/ / www.isoe.de/ eng lish / pr oj ect s/ fut ur efleet. ht m

88

•

G(V, A ) V

A V = D J D J

V^O i V

A i V j V

i j i , j D A = { (i , j )|i , j V \ { (i , j )|i , j D } } di j

(i , j ) (i , j ) A A^Z A (i , j ) A

i , j V^O i j

(i , j ) A

Dj j J K_{E V} K_{I C E V}

Q_k k K_t, t { E V, I CE V } f_k

k Kt, t { E V, I CE V } Sk

k K_t, t { E V, I CE V } C_k

k K_t, t { E V, I CE V } P_k T_k

k K_t, t { E V, I CE V } E_k

W h/ km L / km k Kt, t { E V, I CE V } Fk

k K_{I C E V}

k K_{I C E V} (i , j ) A \ A^Z F_k

k K_{E V} R_k

x_{i j k} { 0, 1}

(i , j ) k, k K , (i , j ) A

y_k { 0, 1} k K

A A (i , j ) A i , j V

Dj j j J

di j (i , j ) (i , j ) A

i j (i , j ) (i , j ) A

KE V

KI C E V

K = KE V KI C E V

Qk k Kt, t { E V, I CE V }

fk

k Kt, t { E V, I CE V }

Sk k Kt, t { E V, I CE V }

Ck k Kt, t { E V, I CE V }

Fk k KI C E V

Rk k KE V

Pk

k Kt, t { E V, I CE V } Tk

k Kt, t { E V, I CE V }

Ek k Kt, t { E V, I CE V }

M

xi j k { 0, 1} (i , j )

k k K , (i , j ) A

yk { 0, 1} k K

min

k K (i ,j ) A

(P_k + T_k)E_kd_{i j}x_{i j k} +

k K i D j J

(f_k+ C_k S_k)x_{i j k} +

k KI C E V

F_ky_k

i D j J

x_{i j k} 1 k K

i V k K

x_{i j k} = 1 j J

i V

x_{i j k} =

i V

x_{j i k} j V, k K

i S j S

x_{i j k} |S| 1 S J, k K

j V

i V j J

Djx_{i j k} Q_k k K

(i ,j ) A

d_{i j}x_{i j k} R_k k K_{E V}

i D j V

xi j k

i D j V

xi j k 1 k Kt = K_t¹, t { E V, I CE V }

k

k 1 K_t ¹ K_t, t { E V, I CE V }

x_{i j k} { 0, 1} (i , j ) A, k K ; y_k { 0, 1} k K

W = S + R E C

W

S CS P S

S = CS + P S

W P P_c

CS = W P P_c

P_p T C

P S = P_p T C

L E Z

R = P S + V T + F T + L E Z

10 × 365

P S =

k K i D j J

S_kx_{i j k} V T =

k K i D j J

(f_k + C_k)x_{i j k} F T =

k K (i ,j ) A

T_kE_kd_{i j}x_{i j k} L E Z =

k K

Fkyk

E C

e^j_k, k K_t, t { E V, I CE V } , j { I , O}

I

O _{i j} (i , j ) A

E_k k K_t, t { E V, I CE V }

E C =

k K (i ,j ) A

i je^I_kE_kdi jx_{i j k}+

k K (i ,j ) A

(1 i j)e^O_kE_kdi jx_{i j k}

e₁^j_k e₂^j_k e₃^j_k

e^j_k = e₁^j_k+ e₂^j_k+ e₃^j_k k K_t, t { E V, I CE V } , j { I , O}

• CO2

CO₂ kW h

CO₂

CO₂ e₁^I_k

e₁^O_k, k K_t, t { E V, I CE V }

• NO_x SO₂ CO₂

3 NO_x SO₂

P M2.5 CO2 kW h

e₂^I_{I C E V} >

e2O I C E V

I O

E_k L / km kW h/ km

S_k f_k C_k F_k

% % d_j U(0.15 Q_{I C E V}, 0.25 Q_{I C E V})

kg

CO₂ CO₂ g/ km

W h/ km L / km

%

%

kW h kW h L L

CO2 e kW h

CO2 e L

e kW h e L e km

k Kt (i ,j ) A i jE_kd_{i j}x_{i j k} t { E V, I CE V }

k K (i ,j ) A(1 _{i j})E_kd_{i j}x_{i j k} t { E V, I CE V }

R E C S

W