Project finance or corporate finance for renewable energy? an agent-based insight

Baldau , Thomas; Jochem, Pa ick
A icle — Published Ve sion
P ojec inance o co po a e inance o enewable ene gy?
an agen -based insigh
Jou nal o Economic In e ac ion and Coo dina ion
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Baldau , Thomas; Jochem, Pa ick (2024) : P ojec inance o co po a e inance o
enewable ene gy? an agen -based insigh , Jou nal o Economic In e ac ion and Coo dina ion, ISSN
1860-7128, Sp inge , Be lin, Heidelbe g, Vol. 19, Iss. 4, pp. 759-805,
h ps://doi.o g/10.1007/s11403-024-00425-z
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/315630
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
Vol.:(0123456789)
Jou nal o Economic In e ac ion and Coo dina ion (2024) 19:759–805
h ps://doi.o g/10.1007/s11403-024-00425-z
REGULAR ARTICLE
P ojec inance o co po a e inance o  enewable ene gy?
anagen ‑based insigh
ThomasBaldau 1 · Pa ickJochem1
Recei ed: 3 Janua y 2023 / Accep ed: 19 Augus 2024 / Published online: 9 Sep embe 2024
© The Au ho (s) 2024
Abs ac
S a e-o - he-a mac oeconomic agen -based models (ABMs) include an inc eas-
ing le el o de ail in he ene gy sec o . Howe e , he possible inancing mechanisms
o enewable ene gy a e a ely conside ed. In his s udy, an in es men model o
powe plan s is concep ualized, in which ene gy in es o s in e ac in an impe ec
and decen alized ma ke ne wo k o c edi s, deposi s and p ojec equi y. Agen s
engage in new powe plan in es men s ei he h ough a special pu pose ehicle in
a p ojec inance (PF) s uc u e o ia s anda d co po a e inance (CF). The model
po ays he g ow h o new powe gene a ion capaci y, aking in o accoun echno-
logical di e ences and in es men isks associa ed wi h he powe ma ke . Di e en
scena ios a e con as ed o in es iga e he in luence o PF in es men s on he ansi-
ion. Fu he , he e ec i eness o a simple g een c edi easing (GCE) mechanism is
discussed. The esul s show ha a ying he composi ion o he PF and CF s a e-
gies signi ican ly in luences he ansi ion speed. GCE can eco e he pace o he
ansi ion, e en unde d as ic educ ions in PF. The model se es as a ounda ional
amewo k o mo e in-dep h policy analysis wi hin la ge agen -based in eg a ed
assessmen models.
Keywo ds Agen -based modeling· Ene gy inance· Financial ne wo k in e ac ions·
He e ogenei y
1 In oduc ion
Nume ous coun ies a e g appling wi h he challenge o expanding hei enewable
powe gene a ion capaci ies o mee clima e objec i es and sus ainabili y ambi-
ions, equi ing huge in es men s in gene a ion in as uc u e (McCollum e  al.
2013; Peake and Ekins 2017; Thacke e al. 2019). Financing cos s cons i u e a la ge
* Thomas Baldau
homas.baldau @dl .de
1 Ins i u e o Ne wo ked Ene gy Sys ems, Ge man Ae ospace Cen e (DLR), Cu ies . 4,
70563S u ga , Baden-Wü embe g, Ge many
760
T.Baldau , P.Jochem
ac ion o he o al cos s o enewable powe p ojec s, and a e o en a key igu e in
he isk assessmen o sus ainable echnologies (Egli e al. 2018; Hi h and S eckel
2016). The coo dina ion o ene gy inance is a key ac o o a success ul ansi-
ion (Polzin and Sande s 2020). In 2023, he o al in es men o inancing ene gy
wo ld-wide amoun ed o 1.7 illion US dolla , and i is impe a i e o sus ain high
in es men olumes in he upcoming yea s in o de o achie e a global wa ming
a ge below 2 °C (IEA 2023; McCollum e al. 2018). A signi ican sha e o hese
in es men e o s will be di ec ed owa d ans o ming he powe ma ke , enhancing
enewable elec ici y gene a ion. Mos powe ma ke s, e en i widely libe alized, a e
subjec o egula o y measu es suppo ing enewable powe plan s (Hu e al. 2018;
Nicolli and Vona 2019). The policy design o hese measu es plays a i al ole in
achie ing an op imal balancing be ween public subsidies and p i a e in es men s. In
his ega d, agen -based in eg a ed assessmen models (IAMs) a e a use ul analysis
ool. E ec i ely in eg a ing he ac o s d i ing enewable ene gy inance in o agen -
based IAM amewo ks is an essen ial p e equisi e. A main challenge in modeling
is o inco po a e he bo om-up ma ke aspec s o ene gy in es men s in su icien
de ail, while main aining he simplici y and gene ali y o a modeling amewo k.
When modeling ene gy in es men s, he ela ed ba ie s and isks ha e o be con-
side ed o a se o he e ogeneous in es o s (Mus a a e al. 2021; Semieniuk e al.
2021). Empi ically g ounded esea ch has de eloped an unde s anding abou he
composi ion o in es o s’ balance shee s and cos o capi al o enewable ene gy
inance (S e en 2020; Kempa e al. 2021), and has widely demys i ied eal-wo ld
in es o s’ isk pe cep ion and a i udes owa d powe ma ke suppo schemes as
well as he unde lying inancing condi ions (May and Neuho 2017; Polzin e al.
2019; Taghizadeh-Hesa y and Yoshino 2020).1
He e, agen -based models (ABMs) a e pa icula ly sui able, as hey na u ally
inco po a e bo om-up ansi ion dynamics and ac o s’ beha io unde he e ogenei y
(Vasileiadou and Sa a zynska 2010; Fa me e al. 2015; Dosi and Ro en ini 2019;
Nieddu e al. 2022). Con e sely, some o hese empi ically g ounded insigh s on
ene gy inance ha e no ye en e ed he domain o mac o-ABMs o he ene gy an-
si ion. Fo a eliable iden i ica ion o policy se ings which incen i ize he expan-
sion o enewable ene gy in es men s, mac o-ABMs and agen -based IAMs need
o be cha ged wi h mo e eal-wo ld e idence abou he ene gy in es men beha io .
Recen li e a u e has elabo a ed se e al open esea ch gaps in he ABM modeling
con ex . Cas o e al. (2020) poin ou a lack in agen -based modeling o clima e-
ene gy policy. Sande s e al. (2022) and Sa in e al. (2023) iden i y missing pa s
om o he disciplines such as elec ici y ma ke modeling. Howe e , only ew con-
ibu ions ha e been made in he di ec ion o conc e e model designs o (mac o)
ABMs linking inancial ma ke s and elec ici y ma ke s.2
The model o his pape is an a emp o imp o emen in his di ec ion. Two
aspec s o enewable ene gy inance a e highligh ed: he e ogenei y in in es o s and
echnologies, and he in es men isk ela ed o powe ma ke p emium schemes.
1 The isk aspec is especially impo an o ene gy p ojec s which ha e di icul ies o hedge agains isk
o he wise, o example h ough bila e al powe pu chase ag eemen s (Hollmén e al. 2022).
2 See appendixA o a mo e de ailed li e a u e e iew.
761
P ojec inance o co po a e inance o  enewable ene gy?…
Fi s , powe gene a ion echnologies a e he e ogeneous and he e o e need o be
conside ed as di e en asse classes. Typically, mac oeconomic models only sepa-
a e be ween “clean and di y” (o “g een and b own”) echnologies, blu ing ou
he he e ogeneous echnological speci ica ions o di e en ypes o powe plan s.
Fo example, sola pho o ol aic in es men s may o e ela i ely lowe up on cos s
and sho e p ojec imes compa ed o onsho e wind powe p ojec s, whe eas wind
powe plan s may un o e a highe numbe o hou s pe yea (Hi h e al. 2016).
The physical and echnological bounda ies o elec ici y gene a ion na u ally esul
in a ce ain deg ee o he e ogenei y among ene gy in es o s. In addi ion, he e can
be an in o ma ion asymme y on he c edi ma ke be ween he in es o s, who a e
amilia wi h hei echnological pa ame e s, and he c edi lende s, who apply bes -
p ac ice ules o assess p ojec isks in a mo e gene alized ashion. Wi hou u he
inancial in e media ies being engaged, his can lead o a signi ican misma ch (In
e al. 2020). Thus, he e ogenei y in ene gy in es o s and hei inancing s a egies
migh induce ins i u ional and ma ke - ela ed ba ie s o mobilizing p i a e inance
o low-ca bon echnologies (Ba azza and S achan 2020; Azhgaliye a e al. 2023;
Mazzuca o and Semieniuk 2018).3 As a esul o a ia ions in ope a ional and non-
ope a ional ( inancing) expendi u es among di e en in es o s and echnologies, he
cash low o powe plan p ojec s di e s om one p ojec o ano he .
Second, in es men isk is linked o powe ma ke suppo schemes and he occu -
ence o insecu e e enue s eams. Whe eas mos models assume an exogenous
powe p ice o a ixed eed-in p emium, a ypical ea u e employed in eal-wo ld
powe ma ke s is he sliding ma ke p emium.4 One-sided powe ma ke p emium
designs allow o wind all p o i s o enewable ene gy plan s whene e expensi e
ossil powe is p ice-se ing, while secu ing a lowe bound o e enue s eams i
elec ici y p ices all below a ce ain s ike p ice. Unde one-sided p emium
schemes, enewable ene gy in es o s mus he e o e accoun o he insecu i y o
acqui ing excess p o i s o no . Acco ding o he pecking o de heo y o in es men
(Mye s and Majlu 1984), when con on ed wi h insecu e u u e e enue s eams
in es o s will be es ic ed o equi y-based inancing and do no make use o ( yp-
ically cheape ) deb -based inancing o e ained pas ea nings. This aspec is also
connec ed o he use o p ojec inance (PF) (Pollio 1998; S e en 2018; Ga i 2023).
The la e o e s an oppo uni y o de- isk an in es o ’s balance shee by spli ing o
pa s o i s unds o a subsidia y, a so-called special pu pose ehicle (SPV). SPVs
a e economic en i ies which exis as an isola ed inancial ame o he unde lying
ene gy p ojec , he e o e being e alua ed o hei indi idual pe o mance a he
han he in es o ’s inancial le e age o pas pe o mance. The e a e also some mo e
in insic easons o using PF, ela ed o he o ganiza ional s uc u e o an in es -
men . Ci izen ene gy p ojec s o wind and sola a e ypical example o SPV a chi-
ec u es and high equi y o deb sha es, whe eas mos u ili y-scale coal- and gas- i ed
3 Because di e en powe plan ypes can a ac di e en ypes o in es o s, hey also play a ole in he
decen aliza ion o powe and a shi owa d local and egional au ho i ies, see (Iskanda o a e al. 2021;
Schlindwein and Mon al o 2023; Rome o-Cas o e al. 2023).
4 Fo example, see (Gawel and Pu kus 2013; Pu kus e al. 2015; Klobasa e al. 2013) o a mo e de ailed
analysis o he ma ke p emium on he Ge man elec ici y ma ke .
762
T.Baldau , P.Jochem
powe plan s classically make use o co po a e inance (CF) (S e en 2018). In sum-
ma y, he de- isking o ene gy in es men s o en esul s in di e en combina ions o
inancing s a egies (CF/PF) and deb - o-equi y composi ions among he he e oge-
neous in es o s, depending on he powe ma ke layou .
In o de o add ess hese wo unde - esea ched aspec s, his pape conside s six
di e en powe gene a ion echnologies and a s ylized powe ma ke wi h a one-
sided sliding p emium mechanism. In es o s and banks a e ma ched bo om-up on
impe ec inancial ma ke s in o de o accoun o s ickiness and in o ma ion asym-
me y. Bo h he PF and he CF s a egy a e conside ed.
The emainde o his pape is s uc u ed as ollows: In Sec .2, we add ess ou
model s uc u e and equa ions, and he imeline o e en s is cons uc ed. Fu he ,
model esul s o a baseline se ing and wo al e na i e scena ios wi h less p ojec
inance and wi h an addi ional c edi easing scheme a e discussed (Sec .3). We sum-
ma ize ou esul s and discuss he po en ial o u he esea ch in Sec .4.
2 Model
The model o his pape adop s wo dis inc le els o de ail, employing s ylized, ep-
esen a i e agen s o he go e nmen , cen al bank and es o economy on he mac-
oeconomic le el, while u ilizing mic oeconomically in e ac ing agen s o ene gy
in es o s, SPVs and banks. Figu e 1 p o ides a comp ehensi e o e iew o he
model a chi ec u e, illus a ing he ela ions and inancial ansac ions be ween he
agen s, as well as he ma ke in e ac ions. This model uses a s ock- low consis en
amewo k, meaning ha he inancial balances o agen s a e cohe en , and liabili ies
o one agen a e consis en ly ea ed as asse s o ano he .5 In es o s ei he ollow he
P ojec Finance (PF) o he Co po a e Finance (CF) s a egy, condi ional on he gen-
e a ion echnology hey aim o in es in. CF in es o s can di ec ly in es in o new
powe plan s. PF In es o s ollow a mo e indi ec s a egy, as hey sea ch o p ojec
oppo uni ies (which a e ep esen ed by newly o ming SPVs) ia he ma ke o
equi y, and in es in new powe plan s using SPVs as in e media ies. SPV p ojec s
and CF in es o s in e ac wi h banks ia he c edi ma ke in o de o ob ain c edi s
o new p ojec s. The in e ac ion dynamics in luence he cos o equi y and cos o
deb o each indi idual powe plan . Banks pay and ecei e in e es and p inciple
paymen s. The e alua ion o new bank loans is subjec o he mac o-p uden ial egu-
la ion o he bank ese es, as well as o he isk pe cep ion o he banks abou he
p ojec s o be inanced. Once powe plan s ha e been buil up, hey a e added o
he supply side o he powe ma ke . When he powe ma ke is clea ed, powe is
consumed and e enues om sales a e dis ibu ed o he in es o s. I is impo an
o no e wo dynamic p ope ies o ou model ega ding he agen ypes and numbe
o agen s. Fi s , in es o s can swi ch be ween PF and CF s a egies o e ime, such
ha he e is no s ic sepa a ion be ween CF in es o s and PF in es o s. Second,
5 Ou model does no include a ully ledged mac oeconomy wi h all economy-wide agen s being explic-
i ly modeled. The e o e, some ansac ions a e open-ended ( hey ne e en e he sys em again), bu hese
a e acked and we ne e iola e s ock- low consis ency.

763
P ojec inance o co po a e inance o  enewable ene gy?…
he numbe o SPVs is g owing o e ime, i.e., whene e an SPV is success ully
launched, he old SPV exi s he ac i e ma ke , and a new SPV en e s he ma ke .6
The old SPV, howe e , emains in he simula ion un il he end o he p ojec li e-
ime. The numbe o he banks is kep ixed o e he en i e simula ion pe iod, and
de aul ed banks a e eplaced by new en an s i necessa y.
Ou model is se up using Py hon and he (xxx) open-sou ce package o agen -
based modeling (ci a ion blinded). Each agen ep esen s an isola ed inancial en i y,
possessing i s own balance shee and cash low accoun ing. The balance shee
ma ix (Table7 in appendixB.1) de ines he asse and liabili y en ies o each agen
class. S ock- low consis ency is acked by only making use o p e-de ined, con-
sis en ansac ions h oughou he simula ion (see Table8 in appendixB.2 o an
o e iew).
The ollowing subsec ions add ess he beha io al ules o each agen class in
de ail.
2.1 In es o s
The ene gy in es o s, indexed wi h j, a e he co e agen s o he model, aking pa
in mos o he ansac ions. In he beginning o he simula ion, he ini ial deposi s
and c edi s a e egis e ed a andomly selec ed banks. To accoun o a he e ogene-
ous s a ing dis ibu ion in size o he in es o s, he ini ial size o in es o agen s is
sampled om a Pa e o dis ibu ion (see appendixF o a mo e de ailed desc ip ion).
The inancial s uc u e o in es o s changes as a esul o di e en ac i i ies. Besides
looking o new in es men oppo uni ies, in es o s ha e o p o ide in e es and ax
paymen s, and conduc ope a ional e o s on exis ing powe plan p ojec s. In es-
o s ac on he deposi ma ke , he c edi ma ke and he p ojec equi y ma ke . In he
beginning, each agen is assigned a a o i e bank o deposi s and a a o i e bank o
c edi s. I deposi s all i s a ailable liquidi y
Dj,k(j)
on he a o i e bank o deposi s,
k(j), and will s a asking o c edi s (accoun ed as inancial liabili ies
Lj,k′
) om he
a o i e bank
k�(j)
o c edi s. Howe e , as ime e ol es, he in es o s can selec new
a o i e banks, and he e o e he c edi s o in es o j will be s o ed a a ini e se o
banks
Bj
. This is because powe plan p ojec - ela ed loans a e no allowed o be
eloca ed om hei o iginal bank: Re- inancing does no occu and in e es a es a e
ixed pe p ojec . In es o s which a e equi y holde s o an SPV ecei e a e u n on
hei in es men in o m o di idend paymen s. SPVs a e egis e ed as equi y on he
asse side o he in es o s’ balance shee s. When ope a ing powe plan s, a physical
quan i y
Yp,
equi alen o he plan p’s nominal powe yield, is gene a ed. The plan
owne pays he ope a ional cos s o he plan s. Finally, a ac ion
𝜑p,
o he plan ’s
physical yield is sold a he p ice p
el
a he powe ma ke . The ope a ional cos s sum
up o
Cp, (𝜑p, ,Yp, )
.7 The in es o s pay a ixed pa
𝜃
o hei posi i e a e - ax p o i s
Πj,
o he es o economy agen as di idends. The emainde o p o i s is held on
he deposi accoun as e ained ea nings.
6 Fo simplici y, he e is only one powe plan p ojec pe SPV.
7 The cos s o a powe plan a e spli in o ixed and a iable cos s, see also sec ion2.4.
764
T.Baldau , P.Jochem
In summa y, he ne income o an in es o is composed o inancing, powe plan
ope a ion and axes:
He e,
CFj,k
deno es he se o co po a e- inanced powe plan s o in es o j, wi h
loans held a bank k, and
PFj
i s se o SPVs (p ojec - inanced powe plan p ojec s).
The a es
𝛿
k(j)
and
𝓁
k,p
a e he bank-speci ic a es on deposi s and c edi s, espec i ely.
I can be seen om Eq. (1) ha he p ojec - inanced cash lows a e encapsula ed in
he SPV s uc u e. Co po a e axes
Tj,
a e cha ged on posi i e ne p o i s ( inancing
and ope a ion cash lows) a a ixed a e
𝜏0
. In es o di idends
Di j=𝜃max{0, Πj, }
a e paid o he es o economy. SPV di idends a e paid ou o in es o j om he
p o i s
Πi,
o an SPV i, depending on he quan i y in es ed
Ei,j
weigh ed by he o al
equi y
Ei
con ained in i.
Using a nes ed andom choice expe imen , in es o s ake he decision o in es in
ce ain echnologies, choose a inancing s a egy and compu e a sui able balancing
Π
j, = 𝛿
k(j), Dj,k(j), −1−
∑
k∈Bj
∑
p∈CFj,k
𝓁
k,p, Lj,p,k, −1
�������������������������������������������������������
inancing
+∑
p∈CFj
(𝜑p, Yp, pel
−Cp, (𝜑p, ,Yp, )) + ∑
i∈PFj
(Ei,j∕Ei)Πi,
�������������������������������������������������������������������������������
powe plan ope a ion
−Tj,
���
axes
Fig. 1 O e iew o he model s uc u e. Boxes ep esen agen classes, blue o als ep esen ma ke in e -
ac ions
765
P ojec inance o co po a e inance o  enewable ene gy?…
o deb and equi y. Figu e2a shows an o e iew o he nes ed decision p ocess o
in es o s. In his pape , i is assumed ha decisions o in es o s a e no ully a ional
and he e o e ollow immu able habi s in hei choice o he echnology and inanc-
ing s uc u e,8 de e mined using a ixed exogenous choice p obabili y. Howe e , he
po ion o deb s. equi y (D/E a io) is endogenously linked o he powe ma ke
isk. The decisions o echnology and inancing s a egy a e disc e e, and he D/E
a io a ies con inuously. In es o s plan new powe plan s acco ding o he ollow-
ing h ee-s ep sequence. Fi s , he e is an ini ial decision abou he echnology o
in es in, d awn andomly om he se o a ailable echnologies using echnology-
speci ic weigh ing ac o s. Fo simplici y, he ela i e equency o each echnol-
ogy is se equal o he exogenous echnology mix
T
, such ha he poli ical capac-
i y expansion a ge is p ac ically in e nalized o he in es o s’ echnology choice
𝜏
:
P(𝜏)=p𝜏 o 𝜏∈T={
sola , wind onsho e, wind o sho e, ligni e, ha d coal,
na u al gas
}
,
∑𝜏∈T
p
𝜏
=
1
whe e
P(𝜏)
is he p obabili y o choosing echnology
𝜏
.
The echnology decision is only e ised a e an in es o has succeeded in build-
ing a new powe plan , o i i has no succeeded o a ixed numbe o
N†
pe i-
ods. Second, he in es o s’ inancing s a egy (CF o PF) is de e mined. The s a egy
choice is condi ional on he echnology choice, and he p obabili y o a o PF o e
CF is speci ied acco ding o a simple lookup able o each o he echnologies (see
Table1), e lec ing he ela i e equencies o p ojec inance o a gi en echnol-
ogy9:
P(PF|𝜏)=p0,PF|𝜏,P(CF|𝜏)=1−p0,PF|𝜏.
Thi d, he sha e o equi y s. deb is
endogenously de e mined, depending on he expec ed e enues o he echnologies
a he powe ma ke . This s ep also akes he ma ke p emium mechanism o elec-
ici y p ices in o conside a ion, e-e alua ed in each simula ion pe iod. A de ailed
ea men o his decision s ep is gi en in sec ion2.5.1.
A e his mul i-s ep p ocedu e, in es o s wi h he CF s a egy will di ec ly ask
o loans a hei cu en a o i e bank o c edi s ( he c edi mechanism is desc ibed
in Sec .2.5.3). Once a bank accep s o issue he equi ed c edi s o he in es o ,
he powe plan p ojec is s a ed and he equi ed inancial deposi s a e con e ed
in o ixed asse s ia a ansac ion wi h he es o economy (whe e he machine y
and cons uc ion se ice sec o can be hough o ). In es o s wi h PF s a egy mus
sea ch o a newly opening SPV i ing hei echnology, ans e he equi ed equi y
sha e and wai o he SPV o ob ain a bank loan (o possibly addi ional equi y
unds), and o s a he p ojec independen ly ( o mo e de ails see Sec .2.5.2).
In es o s ope a e in hei no mal business un il hei deposi s u n ze o: A any
inancial ansac ion demanding mo e deposi s han a ailable, an illiquidi y bank-
up cy e en is igge ed, in which he in es o ies o es o e i s balance shee . I
8 This assump ion is backed by empi ical e idence abou ene gy in es o s, concluding ha in es men
decisions can be mo e in insic and i a ional, see e.g., (Gamel e  al. 2016; Hols enkamp and Kahla
2016) on communi y ene gy and (F ei e al. 2018) on la ge-scale u ili ies. Fu he decision ac o s could
be egional di e ences, social esponsibili y and desi e o local ene gy supply. The exogenous choice o
echnology and inancing s uc u e allows o an a chi ec u al design o he ansi ion condi ions by he
modele (Ande sen e al. 2023).
9 Mind ha he alues p o ided jus se e as an illus a i e se ing and only oughly e lec he eal-
wo ld si ua ion. We he e o e e ain om p o iding any pa icula policy ad ice and ocus on he heo-
e ical ea u es o he model.
766
T.Baldau , P.Jochem
inancing expendi u es canno be co e ed, he co esponding amoun is egis e ed as
a bad deb loss a he bank.10 Any nega i e liquidi y is assumed o be cushioned by
he es o he economy. Insol ency bank up cy occu s a nega i e ne wo h. A e
a bank up cy e en (ei he illiquidi y o insol ency), he in es o econside s i s un-
ning powe plan p ojec s. F om he se o unning p ojec s, in es o s dump exis ing
powe plan s which a e cu en ly p oducing nega i e cash lows, s a ing a he ossil
CF plan s, hen mo ing on o he enewable CF plan s, and inally o he PF plan s.
P ojec s wi h nega i e cash low a e p ema u ely e mina ed. Such uncon en ionally
e mina ed p ojec s a e s anded capaci ies and no longe con ibu e o he powe
supply. Rela ed loans u n in o non-pe o ming loans.
2.2 Banks
Banks (indexed wi h k) se e as lende s o c edi s and a e demande s o depos-
i s. They hold bo h deposi s o he es o economy (subsc ip R, in ol ing all non-
ene gy inance) and he ene gy-speci ic deposi s and loans. Ini ially, each in es o is
assigned a andom bank o deposi s and a andom bank o loans, shaping he ini ial
pools o deposi o s and deb o s and de e mining he ini ial le e age ac o o he
banks (mo e de ails abou he ini ializa ion sequence a e gi en in appendixF).
Banks compe e o liquidi y a he deposi ma ke , as hey a e obliged o hold
su icien liquidi y ese es. Besides deposi s, banks can ob ain addi ional liquidi y
ese es in o m o cen al bank loans
Ck
, such ha ese es can be exp essed as
R
k,
=
Dk,R,
+∑j∈
D
k
Dj,k,
+
Ck, , whe e
Dk
deno es he se o deposi o s (mic oeco-
nomic clien s) o bank k. The equi ed cen al bank c edi s a e subjec o he ese e
equi emen , such ha ideally
Rk, ≥𝜅minLk,
:
A global and exogenous in e es
⋆
is cha ged on cen al bank c edi s. In o de o
make p o i s, banks ake he cen al bank a e as he lowe bound o he in e es
(1)
Ck, =max{0, 𝜅minLk, −Dk, }.
Fig. 2 Illus a ion o he h ee-s ep decision p ocess o in es o s (a) and he c edi g an ing mechanism
o banks (b)
10 Fo simplici y, loans, e en i non-pe o ming, emain a he bank hey o igina e om and a e no
esold.
773
P ojec inance o co po a e inance o  enewable ene gy?…
ha d coal, na u al gas.27 This yields he ac ual, disc e e supply cu e (g ay a eas
in Fig.4a, b).
2. I e a e o e all plan s. The po en ial elec ici y supply o a plan p is gi en by
Y
⋆
p
=

Yp⋅

K
p
. The maximum needed gene a ion capaci y
Yp
=max(L−

Y
p
,0
)
is
compu ed o each plan , whe e L deno es he exogenous powe demand (load)
and

Yp
he cumula i e elec ici y p oduc ion o all plan s wi h highe ank han p.
The load ac o esul s in
𝜑
,p=Yp∕Y
⋆
p
∈[0, 1
]
.
3. Repea he p ocess un il he demand is me o no plan s a e le .28 The esul ing
clea ing p ice in each pe iod is app oxima ed as a linea unc ion
whe e
𝜓
=(
∑RE
𝜑 ,pY
⋆
p
)∕L=Y
RE
∕
L
is he o e all sha e o enewable ene gy
p o ided. He e,
𝛽0
co esponds o he ma ginal cos o he mos expensi e ossil
powe plan in he sys em, and
𝛽1
de e mines he s eng h o he long- un me i
o de e ec .
To e ine he powe p ice dynamics, sho - e m p ice luc ua ions a e added by a
mean- e e ing s ochas ic p ocess (see Sec .E.1 o mo e de ails). All plan s wi h
𝜑 ,p>0
a e assigned he ma ke alue p
el
o his pe iod ( he emaining plan s ha e
ze o ma ke alue because hey a e no equi ed).
Ma ke P emium and Ra io o Equi y We expec om he long- e m me i o de
e ec ha he in es men s in enewable gene a ion echnology will lowe he powe
p ice le el. This, howe e , can be a possible bu den o in es o s as he expec ed
p o i s om powe sales g adually dec ease o e ime. In o de o suppo he enew-
able ene gy in es o s, a sliding p emium is employed. This sliding p emium can be
hough o as a inancial op ion con ac , g an ing a leas a minimum s ike p ice
S𝜏,
, such ha he e ec i e selling p ice o enewable plan s is ne e below his
bound. Fossil gene a ion is no included in his p emium scheme. The e ec i e p ice
esul s in
whe e
S𝜏,
is a alue which depends on he espec i e echnology
𝜏
and p
el
deno es
he (wholesale) clea ing p ice wi hou ma ke p emium. Figu e 4c schema ically
depic s he ela ion be ween he p ojec alue o enewable powe plan s and he
composi ion o deb and equi y. In con as o he ossil coun e pa s, enewable
elec ici y is gene a ed a e y low ma ginal cos s, such ha hei ma ke alue ( hei
selling p ice a he powe ma ke ) ends o be low in imes o high sha es o enew-
able ene gy and high whene e expensi e ossil plan s a e in he ma ke alongside. I
he e is no p emium, enewable powe plan s make high e enues i powe p ices a e
(8)
p
el
=𝛽
0
−𝛽
1
𝜓
(9)
pel =
{
pel
i 𝜏∈{ligni e, ha d coal, na u al gas}
max(S
𝜏,
,pel
)i 𝜏∈{sola , wind o sho e, wind onsho e
}
27 The o de in his model is ixed. We so plan s by echnologies in analogy o ma ginal gene a ion
cos s. Wi hin g oups o powe plan s possessing he same ank, plan s a e andomly shu led.
28 In case he e a e no plan s le (which does no happen in he scena ios analyzed in his pape ), i can
be assumed ha ene gy impo s a e used o co e he emaining load.

774
T.Baldau , P.Jochem
high and close o ze o e enues i powe p ices a e low. Wi h he p emium ac i e,
plan s can heo e ically ecei e e enues wi h no uppe bound, bu he minimum
e enues a e secu ed. Any posi i e di e ence be ween s ike p ice and ma ke alue
is paid by he consume (he e: es o economy). This ensu es he p o i abili y o
enewable powe plan s, e en unde pe manen ly high sha es o enewable ene gy
and hus low powe p ice le els. I he expec ed e enue o a powe plan is highe
han he le el o he s ike p ice S, excess e enues occu . Unde he assump ion ha
banks a e ully isk-a e se wi h espec o he inancing o enewable ene gy p o-
jec s, such addi ional luc ua ions in e enues ha e o be backed using equi y ins ead
o con en ional bank deb (due o he pecking o de heo y o capi al, see Mye s and
Majlu (1984)). The equi y equi ed o a p ojec unde a sliding ma ke p emium
is he e o e equal o he unce ain addi ional e enues beyond S, which can be also
in e p e ed as a minimal isk le el. The equi ed equi y is ma hema ically gi en by
he expec a ion alue o e enues highe han he s ike p ice.29 He e, i is assumed
o simplici y ha all powe plan s ecei e he ma ke p emium, and he s ike p ice
is se equal o an equilib ium alue, which is ob ained by equi ing ha he p emium
can jus co e he echnology cos s.30 The e o e, S is chosen op imally by he go -
e nmen such ha he in es men I can be on a e age co e ed om he sum o deb
and equi y:
Table 2 O e iew o he
ma ke s in he model and
mapping o supply and demand
Ma ke Supply Demand
Deposi s In es o s, SPVs Banks
Loans Banks In es o s, SPVs
P ojec Equi y In es o s SPVs
Powe Powe Plan s Res o Economy ( ixed)
Fig. 4 Le : Powe p ices when he enewable ene gy supply is low (a) and high (b). The g ay a eas
indica e he supply cu e. Righ : Composi ion o deb and equi y as a unc ion o he ma ke alue o a
p ojec ela i e o he s ike p ice (c)
29 Fo a mo e igo ous ma hema ical ea men see appendixE.2.
30 The eal-wo ld s ike p ice is auc ioned be ween a s a e ins i u ion and he in es o s. The implemen a-
ion o he ac ual coo dina ion mechanism o de e mine S i le o u u e esea ch.
775
P ojec inance o co po a e inance o  enewable ene gy?…
whe e he expec ed le el o deb D and equi y E depends on he dis ibu ion o
expec ed ma ke alues o a echnology
𝜏
. Fo sol ing his op imiza ion p oblem,
he me hod de eloped by Neuho e al. (2022) is used.31 The in ui ion is ha he
s ike p ice is chosen such ha enewable powe plan s can aise su icien deb
o co e hei ope a ional cos s, whe e a echnology-speci ic op imal ou pu

Y𝜏
is assumed, de i ed om he annual load hou s o he echnologies included (see
Table11 in he appendix).32 The cons ain s and g inco po a e he a e age inanc-
ing condi ions in he economy, i.e., he a e age equi y a e
⟨ e⟩
and c edi a e
⟨ 𝓁⟩
o all in es o , SPV and bank agen s (see appendixE.2). Fu he , he dis ibu ion o
expec ed ma ke alues
𝜙𝜏
beyond he s ike p ice S is included in he calcula ion o
he ac ion o equi y (cons ain g). No e ha
E
ises as ma ke alues g ow and
D
inc eases wi h inc easing s ike p ice. The e o e, a highe ma ke alues ela i e o
he s ike p ice, he expec ed sha e o equi y o in es men s ises, as schema ically
depic ed in Fig.4c.
Once he s ike p ice is se by he go e nmen , in es o s compu e hei indi idual
le els o deb and equi y using hei idiosync a ic inancing condi ions
e,j
, as well as
𝓁
j
and
S𝜏
such ha
D
j, (S𝜏, , e,j, ,
𝓁
j, )
and
Ej, (S𝜏, , e,j, )
luc ua e a ound he equilib-
ium assump ion in Eqs. (11) and (12) o each in es o j as a consequence o he
inancial ma ke impe ec ions. SPVs ollow he same sys ema ic, using he a e age
equi y a e o hei en isaged sha eholde s.33 The equi y a e is ob ained om he
expec a ion alue o med by he in es o s, i.e.,
e,j, =
𝔼
[ e(j,𝜏)]
, as will be u he
elabo a ed in Sec .2.5.2. Fo hei pe cep ion o
𝓁
j
, in es o s ake only hei cu en
a o i e bank in o accoun .
2.5.2 P ojec equi y ma ke
In o de o collec equi y, SPVs ha e o ind sui able in es o s on he ma ke o
p ojec equi y. PF in es o s compe e o SPVs by s a ing hei desi ed sha eholde
e u n ( e u n on equi y), based on hei expec a ions. The expec ed e u n a e on
(10)
S
𝜏, =a gminS�
𝜏
|
|
I𝜏−(D (S
�
𝜏
)+E (S
�
𝜏
))
|
|
(11)
s. .
D
= (

Y
𝜏
(S
�
𝜏
−C
a ,𝜏
)−C
ix,𝜏
,
⟨
𝓁⟩
,
⟨
e⟩ )
(12)
s. .
E
=g(

Y
𝜏
,S
�
𝜏
,
𝜙𝜏
,
⟨
e⟩ )
31 A mo e in-dep h ea men o his algo i hm is gi en in appendixE.2.
32 The pa ame e

Y
has a cons an alue, ela ed o he op imal ope a ion modes o each echnology,
a he han he ac ually ob ained ma ke sha es a he powe ma ke . I he e o e di e s om
𝜑Y
men-
ioned ea lie in his sec ion. I is ob ained by di iding he a e age annual ull-load hou s o a echnology
by 8760 (numbe o hou s pe yea ), see Table11.
33 I is ensu ed a all imes ha he op imiza ion objec i e is always close o ze o, such ha
D(S𝜏)+E(S𝜏)=I𝜏
holds wi hin e y small nume ical e o s. The luc ua ions in deb and equi y used o
inancing can he e o e be solely a ibu ed o he impe ec ions on he inancial ma ke .
776
T.Baldau , P.Jochem
equi y
𝔼[ e(j,𝜏)]
o in es o j o a echnology
𝜏
is compu ed using a capi al asse
p icing model (CAPM, c . Ga i (2023), Ki zing and Webe (2014)), such ha in es-
o s can gauge he ela ionship be ween isk and e u n:
whe e
= b
deno es a isk- ee a e.34 The expec ed ma ke e u n a e
𝔼
[ m
j,𝜏]
a he
equi y ma ke o echnology
𝜏
is app oxima ed by he ne cumula i e cash low

V
p�, =
∑
�=
0
(p�)Vp�,
�
di ided by he in es ed capi al
E′
p
, a e aged o e all exis ing
powe plan s
p�∈P𝜏
o his echnology,
weigh ing he powe plan s acco ding o hei age
ap′
. Hence,
𝔼
[ m
j,𝜏]
a ies o e ime
as he alues o
V′
p
change o e ime. In es o s a e conse a i e in hei expec a ion
o ma ion, weigh ing he cu en ly pe cei ed e u n a es on equi y (Eq.13) wi h
hei in insic belie s
𝔼[ e(𝜏)]0
, as included in Eq. (14). The (le e ed) equi y be a
𝛽e
j
is linked o he asse be a
𝛽a
j
ia he ollowing equa ion:
whe e
𝛽a
j
is he asse be a ac o assigned o in es o j.35 This pa ame e e lec s he -
e ogenei y in asse classes ( echnical designs, egional e ec s e c.). Equa ion (16)
conside s ha in e es paymen s can be deduc ed om ax, and he e o e he a ge ed
deb - o-equi y a io
(Dj∕Ej)∗
and he co po a e ax a e
𝜏0
mus be aken in o
accoun .36 The ma ke ma ching p o ocol b ings oge he supply and demand in he
ollowing sequence o e en s:
(13)
𝔼
[ e(j,𝜏)]
�
= +𝛽
e
j
⋅max{0, 𝔼 [
m
j,𝜏
]−
}
(14)
𝔼 [ e(j,𝜏)] = 𝛼e
𝔼
[ e(𝜏)]0+(1−𝛼e)
𝔼
[ e(j,𝜏)]�
(15)
𝔼
[ m
j,𝜏]=
1
|
P𝜏
|∑
p
�
∈
P
𝜏

Vp�, e−𝛿map�∕Ep�
,
(16)
𝛽
e
j=𝛽a
j
(
1+(1−𝜏0)
(
Dj
E
j)∗)
34 The 10-yea go e nmen bond yield is ypically used in p ac ice. In ou model, we assume a cons an
alue.
35
𝛽a
j
is e-assigned om a uni o m dis ibu ion
u∼(𝛽a
min,𝛽a
max)
each ime an in es o edecides he a ge
echnology.
36 Fo simplici y, in es o s a e conse a i e in hei expec a ions and assign
(D∕E)∗=(
1
−
𝛼
)∕𝛼
as a
ixed p oxy ac o ,
𝛼
being he minimum equi ed equi y sha e o a p ojec . This also a oids a ci cula
dependence be ween he le el o equi y used and he expec ed e u n on equi y. Al hough in ou model,
he assump ions unde lying he capi al asse p icing model a e no en i ely ul illed— o example he e
is no ully e icien ma ke and no ull a ionali y—in es o s ha e his o mula a hand when i comes
o es ima ing he e u n on equi y hey expec om hei powe plan in es men s. The e o e, ou model
e lec s a ce ain limi ed a ionali y o he ma ke ac o s who do no ully ake ad an age o hei indi-
idual ma ke si ua ion, bu a he end o apply common-p ac ice decision making.
777
P ojec inance o co po a e inance o  enewable ene gy?…
1. I e a e o e a subse o SPVs wi h ela i e size
0≤𝜒E≤1
, andomly selec ed
om he demand pool. Fo each SPV, a new andom subse o ela i e size
𝜒E
o
in es o s (equi y supplie s) is ma ched.
2. Ou o he andom subse o in es o s seen by an SPV, he bes in es o candida es
a e chosen using he ollowing sys ema ic:
• In es o s wi h ma ching echnology choice a e selec ed as sui able candida es.
• Second, hose in es o s a e il e ed who p o ide su icien inancial means o
ully mee he en i e demand, gi en by he SPV’s a ge equi y, conside ing
p eexis ing unds o he SPV al eady collec ed ea lie . I he e a e sui able
candida es, he bes ma ch is chosen acco ding o he lowes expec ed e u n
on equi y o he in es o .
• I no single in es o mee s he c i e ia o suppo unds on i s own, he ma ke
algo i hm a emp s o ill up he SPV’s demand by a emp ing o c ea e a join
en u e om smalle in es o s, accumula ing hei po en ial equi y supply
un il he demand is me , o un il he pool o supply agen s is exhaus ed. These
smalle in es o s a e selec ed a andom om he a ailable pool.
I is wo h no ing ha he SPVs canno di ec ly choose among in es o s, bu can
only send applica ions o hei a o ed ma ches. In es o s will hen go h ough hei
o e s in ch onological o de and pu hei unds in o he join en u e (o possi-
bly single en u e).37 In o de o a oid o e ly lumpy in es men s, in es o s a e only
allowed o pu a maximum a sha e o
𝜉E
o hei a ailable deposi s in o one SPV a a
ime. Fu he , an SPV can only eques up o

N
in es o s pe ma ke clea ing ound.
A e ob aining su icien unding in e ms o sha eholde equi y, he SPV will apply
o a bank c edi in o de o ob ain he equi ed emaining unding om deb . As
soon as he SPV ecei es a c edi ia he c edi ma ke , he SPV pu chases a powe
plan a he es -o -economy agen and s a s ope a ing he powe plan (which is
assumed o be a ailable immedia ely a e , i.e., a ze o cons uc ion ime).38 I he
sea ching o he SPV is no success ul o a pe iod o
N†
pe iods, he SPV edecides
i s in es men choice and a ge echnology. This decision is in ull analogy o he
andom choice expe imen o he co po a e in es o s.
2.5.3 Deposi andc edi ma ke
The main idea o he deposi and c edi ma ke is ha banks al e hei o e ed in e -
es a es in o de o expand hei business and egula e hei a ac ion owa d po en-
ial clien s.39
37 This mechanism is simila o he “b ochu e mechanism” o he Schumpe e Mee ing Keynes model,
see Dosi e al. (2010).
38 The es o economy agen can be hough o as p o iding machine pa s o cons uc he powe plan .
Also, planning and cons uc ion imes can signi ican ly delay he enewable capaci y expansion. This
aspec will be included in u u e e sions o he model.
39 The deposi and c edi ma ke a e simila , bu no ully iden ical o he ma ching p o ocols o Ricce i
e al. (2015).
778
T.Baldau , P.Jochem
Deposi Ma ke On he deposi ma ke , banks a e demande s, and in es o s and
SPVs a e ma ched om he supply side. As a esul o he ma ching, deposi o s
ind new a o i e banks o deposi s. Banks a e ini ially assigned a andom alue
𝛿
k,0
∈u(
𝛿
min,0
,
𝛿
max,0)
as hei cu en o e o he in e es a e on deposi s. When e-
ma ching he deposi ma ke , he sequence o e en s is as ollows:
1. A andom subse o size
0≤𝜒D≤1
ela i e o he o al numbe o demand agen s
(banks) is d awn. The selec ed banks upda e hei o e ed deposi a es. Banks
p e e a ac ing deposi s o e cen al bank loans, as he in e es a e on deposi s
is by design lowe han he in e es a e on cen al bank loans. As a consequence
o compe i ion o o he banks, each bank ies o se he in e es a e on deposi s
as low as possible and as high as necessa y o a ac deposi o s. Thus, i liquidi y
equi emen s a e no me , i.e.,
Rk, <𝜅
minLk, −1
, o i he e is no mo e han one
clien , he bank aims a a ac ing mo e deposi o s. This mechanism wo ks also in
e e se: lowe ing he in e es a e will dis-incen i ize deposi o s om choosing
his bank. Also, i he deposi s jus ma ch he ese e equi emen s, he banks
inc eases he in e es a e o be able o expand i s liquidi y a ailable o new loans
in he nex pe iod:
The in e es a es on deposi s mo e be ween he legs (
 𝛿
, ⋆
) wi h
𝛿
k,
=min(
⋆
,max(
𝛿
,
∗𝛿
k, ))
.
𝛼↑
d
,𝛼
↓
d
a e lea ning speed pa ame e s o he upwa d
and downwa d adjus men o deposi a es, and u(0, x) is a uni o m andom
alue be ween 0 and x.
2. We loop o e supplie s in a andomly d awn subse o ela i e size
0≤
𝜒
′
D≤1
.
Each deposi o in his subse is andomly assigned a pool o banks wi h ela i e
size
𝜒D
. Banks in his subse a e so ed acco ding o hei in e es a e on deposi s.
The bank wi h he highes a e wins. The supplie will ans e all i s deposi s o
he bank wi h he highes o e ed a e i he a e is a leas as high as he cu en
a e. O he wise, he old bank is kep .
The p o ocol jus ou lined is p ocessed sepa a ely in a ma ke o in es o s and
a ma ke o SPVs. Despi e his sepa a ion, he deposi ma ke does no u he
dis inguish be ween SPVs and in es o s. A e he ma ke clea ing o he mic o-
economically modeled agen s, he deposi s in he es o economy a e shi ed.
Each bank is compa ed wi h a numbe
Nm
o he banks a andom. A sha e
𝜂R
D
o
he economy’s deposi s is ans e ed om he cu en bank o he bank o e ing
he bes condi ions in he subse (i di e en om he cu en bank). The numbe
o clien s o each bank is equal o he numbe o deposi o s in he cu en pe iod.
C edi Ma ke The c edi ma ke mechanism is simila o he deposi ma ke .
Banks a e ini ially assigned a andom alue
𝛿
k,0
∈u(
𝓁
min,0
,
𝓁
max,0)
as hei cu en
o e o he in e es a e on c edi s. The sequence o e en s o he ma ke p o o-
col is as ollows:
(17)
∗𝛿
k, =
{
𝛿
k, −1(1+u(0, 𝛼↑
d)) i Rk, −1
≤
𝜅minLk, −1o Nclien s <=
1
𝛿
k, −1(1−u(0, 𝛼↓
d)) i Rk, −1>𝜅
minLk, −1and Nclien s >
1

779
P ojec inance o co po a e inance o  enewable ene gy?…
1. A andom subse o ela i e size
0≤𝜒C≤1
is selec ed om he banks, who a e
allowed o adap hei in e es a es, based on he pas expe ience. The in e es
a e on loans is adjus ed as ollows:
using he up- and downwa d adjus men speed
𝛼↑
𝓁
,𝛼
↓
𝓁
o he c edi a e. The
in e es a e o a bank k is aised i he ma ching success a e
s,k
o his bank is
la ge han he a e age ma ching success a e
⟨ s⟩
. I he success a e is ela i ely
low, o i he e is enough liquidi y p esen in he bank, he in e es a e is low-
e ed in o de o a ac new c edi ake s.
2. A subse o ela i e size
0≤
𝜒
′
C≤1
is andomly d awn om he demand se . We
loop o e each in es o /SPV in he demand se . Demande s which ha e nonze o
demand o c edi s a e il e ed ou . Fo all emaining demande s, a andom subse ,
o ela i e size
𝜒C
each, is d awn om he supply se . The bank o e ing he lowes
in e es a e in his subse is chosen as a po en ial candida e. The in es o /SPV
changes o he candida e bank as a new a o i e bank o c edi s i i s c edi a e
is below he in es o ’s cu en c edi a e, i.e., he a e i would ob ain o a new
c edi a he cu en a o i e bank.
I his is he case, i is a success o he new bank and a ailu e o all emaining
banks in he isible subse . O he wise, a ma ching ailu e o he en i e subse is
egis e ed.
The c edi ma ke ma ching success a e is de e mined by di iding he numbe o
successes o banks by he sum o successes and ailu es in he pas i e simula ion
pe iods. In equi alence o he deposi ma ke , he e is a sepa a e c edi ma ke o
in es o and SPVs. A e he ma ke clea ing o he mic oeconomically modeled
agen s is done, he c edi s o he es o economy a e upda ed in ull analogy o
he es -o -economy deposi ma ke . He e, a sha e
𝜂R
L
o c edi s is eloca ed i one
in he andomly selec ed
Nm
o he banks conside ed o e s a lowe o equal c edi
a e.
The ma ke ac i i ies o deposi s and c edi s lead o a he e ogeneous dis ibu ion
o deposi and c edi a es, and agen s a e no able o obse e he whole ma ke in a
simula ion pe iod.
2.6 Sequence o e en s
Be o e he ac ual simula ion, an ini ializa ion sequence akes place. This in ol es
he ollowing s eps40:
(18)
𝓁
k, =
�
𝓁
k, −1(1−u(0, 𝛼↓
𝓁)) i s,k<
⟨
s
⟩
o Rk, −1
≥
𝜅minLk, −1
𝓁
k, −1(1+u(0, 𝛼↑
𝓁)) i s,k>
⟨
s
⟩
and Rk, −1<𝜅
minLk, −
1
40 Mac oeconomic agen -based models ypically equi e a ’bu n-in’ phase, in which he quasi-s eady
s a e is eached beginning om a oughly posed pa ame e se . We y o keep his phase sho by expe i-
men ing wi h di e en heo e ical anges o pa ame e s.
780
T.Baldau , P.Jochem
1. Agen s and hei co esponding balance shee s a e se up in a consis en manne .
2. An ini ial powe gene a ion mix is se up.41
3. Ma ke links o c edi s and deposi s a e ini ialized: In he beginning, a o i e
banks o in es o s and SPVs a e comple ely andom.
A e he ini ializa ion has aken place, he main model loop is i e a ed:
1. Each in es o (SPV) ecei es e enues om las pe iod’s business and pays di i-
dends o he sha eholde s ( es o economy o he pa en in es o )
2. The powe ma ke clea s and s ike p ices a e upda ed. Each powe plan upda es
i s p oduc ion yield o his pe iod.
3. Powe plan s upda e hei cash lows om powe gene a ion ( e enues ne gene a-
ion cos s). New expec a ions abou he ma ke e u n a e o med.
4. The ma ke s o deposi s, c edi s and p ojec equi y clea .
5. PF and CF In es o s y o in es in new powe plan s, SPVs apply o loans a
he banks.
6. Cash lows om axes and in e es paymen s a e upda ed.
7. Banks a emp o epai hei ese es by buying o selling cen al bank loans,
bonds o loans om he es o economy.
8. Bank up in es o s and banks a e es o ed.
Each model un consis s o a o al o 1000 simula ion pe iods.
3 Resul s
The pu pose o his sec ion is o s udy he simula ion ou pu and o check some qual-
i a i e s ylized ac s. Subsec ion3.1 s a s by showcasing wo ep esen a i e base-
line model uns. Addi ionally, o illus a e he s a is ical p ope ies o ou esul s,
we p esen he ou comes o 100 Mon e Ca lo simula ions. Subsec ion3.2 con in-
ues wi h he analysis o wo al e na i e scena ios, in which he impac o a lowe ed
p obabili y o p ojec inance, as well as an enhanced g een c edi g an ing p obabil-
i y (g een c edi easing) a e in es iga ed.
3.1 Baseline model
In he baseline se ing, he economic sys em is se up such ha mos o he model
dynamics, like e.g., he c edi a e o he in es o ailu e a e, a e in a quasi-s eady
s a e.42 Two independen uns o he model a e obse ed in o de o gi e an in ui ion
41 These ini ial powe plan s a e no owned by he in es o s bu a e exogenous. Thei e enues and losses
a e o wa ded o he es -o -economy agen .
42 Albei he e a e empi ically easonable pa ame e s included, he model is no calib a ed o i any pa -
icula empi ical s a is ics o his o ical da a. We lea e his ask o u u e esea ch. See (Fagiolo e al.
2006, 2019; Lampe i 2018; Vandin e al. 2021) o a ea men o calib a ion me hods o agen -based
781
P ojec inance o co po a e inance o  enewable ene gy?…
abou he model ou pu and o ske ch he possible a ia ion among model uns. The
analysis o he esul s ollows a wo-s ep p ocedu e. Fi s , he unc ioning and s abil-
i y o he inancial ma ke s, as well as he esul ing inancing condi ions o powe
plan in es o s is in es iga ed. Second, we inspec he numbe o capaci y addi ions
o elec ici y gene a ion, he ansi ion o powe mix owa d enewable powe plan s
and he e olu ion o powe p ices.
Figu e5 shows some key inancial ma ke indica o s o he model. To ensu e a
p ope unc ioning o he deposi ma ke , banks need o ha e enough compe i i e-
ness such ha he ma ke concen a ion does no a o one bank a all imes, wi h
all he o he s ha ing ze o clien s.43 Figu e5a shows he inancial ma ke concen a-
ion measu ed by he He indahl-Hi schman index (HHI) o he dis ibu ion o bank
clien s. The ma ke concen a ion exhibi s endogenous cycles o bo h o he simula-
ions, anging be ween 0.05 (all banks ha e app oxima ely equal ma ke size) and
0.5 ( wo banks sha e he ma ke powe ). The model ou pu sugges s ha he cho-
sen baseline pa ame iza ion is able o sus ain ela i ely s able cycles o highe and
lowe concen a ion h oughou he simula ed pe iod.44 The ma ching success a e
(Fig.5c), measu ed as he a io be ween success ul in e ac ions on he c edi ma ke
and he o al numbe o in e ac ions aken place, equi es a bu n-in up o app oxi-
ma ely
=250
, om whe eon i emains ela i ely s able sligh ly abo e 20%. Bank
ailu es a e in equen e en s, sugges ing ha he e a e small-scale endogenous bank
c ises, bu he model does no yield colla e al damage o he banking sys em, wi h
mo e han six o he 20 banks ailing wi hin a ime window o en modeling pe iods.
These esul s o he inancial ma ke s a e in line wi h o he decen alized ma ch-
ing models in li e a u e, ob aining endogenous business cycles as well as occasional
i m and bank ailu es, see e.g., (Ricce i e  al. 2015, 2022). The mean le e age
o in es o s is shown in Fig.5b, emaining ela i ely s able be ween 46 and 49%
along he en i e ime span o he simula ion. This indica es ha he in es men isk
o he powe plan p ojec s is ela i ely cons an , e lec ed in he deb sha e o he
in es men s.
Fu he , he inancing condi ions o powe plan s a e analyzed (Fig.6). The e is
a sligh inc ease in c edi a es, as well as a s able de elopmen in expec ed e u n on
equi y almos no long- e m end componen . In e es a es on c edi s a e in luenced
by he igh ening ese es o banks in he second hal o he simula ion (
>500
).
The alling elec ici y p ice (Fig.6c), he main dynamic pa ame e o powe plan
e enue s eams, exhibi s clea ly isible sho - e m luc ua ions and a long- e m
downwa d end, as expec ed om Eq. (8). The s ike p ices o sola and wind
ene gy (Fig.6d) ise un il app oxima ely
=500
, exceeding he wholesale ma ke
p ice a a ound
=300
. In he mo e ma u e phase o he ansi ion, s ike p ices
43 Clien s a e de ined as agen s s o ing his bank as hei a o i e bank o deposi s.
44 Fo mo e de ailed s udies on he eme gence o ma ke concen a ion in agen -based models, we o -
wa d he in e es ed eade o (San os and Nakane 2021; Te ano a and Tu co 2022).
models. We cha ac e ize he quasi-s eady s a e by a isually s able de elopmen o he s ochas ic model
ou pu o he i s 200 ime s eps in wo ep esen a i e uns.
Foo no e 42 (con inued)
782
T.Baldau , P.Jochem
almos sa u a e owa d he echnology cos s. Whe eas he s ike p ice o sola and
wind onsho e ene gy a e in he same o de o magni ude, wind o sho e p ices each
signi ican ly highe le els, due o he high echnology cos s, hus lowe p o i abili y,
o he o sho e wind echnology.
In Fig.7a-b, he numbe o powe plan s and he esul ing enewable ene gy sha e
o o al gene a ion is plo ed o wo ep esen a i e uns. In he i s un (blue cu e),
he e is an in es men pla eau a a ound
=400
and ano he a
=600
. This pla eau
is no exis en in he second simula ion (o ange cu e); howe e , a sligh kink a e
=400
is isible. The second un eaches ull enewable powe gene a ion be o e
he end o he simula ion (
<1000
). Apa om he empo a y dis up ions, he
powe plan expansion is a he linea , and he g ow h a e o PF p ojec s (solid lines
in Fig.7 exceeds he one o he CF p ojec s (dashed lines). In Fig.7c, he numbe o
bank up in es o s pe simula ion pe iod is plo ed. Fo he blue cu e, i is clea ly
isible ha he bank up cies a e highly co ela ed wi h he phases o low powe
plan capaci y expansion. This bank up cy wa e is an icipa ed in he ejec ed c edi
eques s, which endogenously eme ge be ween
=200
and
=400
(Fig.7d, blue
cu e). As can be seen in he blue cu e in Fig.5b, he in es o le e age also expe-
iences a boom be ween
=200
and
=500
and a subsequen bus un il
=600
,
he poin whe e he in es men pla eau ends. This is no obse able o he o he
simula ion un, indica ing ha his ype o c isis is an eme gen phenomenon o he
model. In ac , his con agion dynamic is a ypical ea u e o bank-in es o ne wo k,
Fig. 5 Financial ma ke indica o s o wo dis inc uns o he baseline model (blue and o ange cu e). a
Deposi ma ke concen a ion (He indahl-Hi schman index o numbe o clien s pe bank), b in es o
le e age, c a e age c edi ma ke ma ching success a e, d a e age numbe o bank ailu es. The alues
shown o he bank ailu es a e en-pe iod a e ages o p o ide su icien s a is ics and o imp o e isual
cla i y
789
P ojec inance o co po a e inance o  enewable ene gy?…
he e is a la ge sp ead o de aul a es among he Mon e Ca lo uns. The la e is
simila o he baseline case.
G een C edi Easing Suppo Ins umen Due o he imma u i y o echnologies
and high indi idualism o enewable ene gy p ojec s, g een CF in es o s o en ace
di icul ies when sea ching o inancial capi al. As an al e na i e o PF, g een c edi
easing (GCE) migh guide g een CF in es o s owa d a success ul launch o a deb -
in ensi e p ojec (Lampe i e al. 2021; Alha bi e al. 2023) bu is also linked o
inc eased agili y and isk in he inancial sys em (DelGaudio e al. 2022). The
in oduc ion o G een C edi Easing (GCE) in he “Less P ojec Finance wi h GCE”
scena io aims o mi iga e he ad e se e ec s o educed PF by imp o ing access
o c edi o g een p ojec s. This policy in e en ion is designed o guide g een
CF in es o s owa d success ul deb -in ensi e p ojec launches, he eby sus aining
in es men momen um. We echnically do so by changing he pa ame e s
𝜌1
and
𝜌3
in he c edi mechanism (Eqs.3 and5), and alle ia ing he ese e equi emen s
𝜅min
, see Table5. We in oduce he modi ica ions
𝜌RE
1
,𝜌
RE
3
,𝜅
RE
min
o enewable ene gy
such ha he new bank pa ame e s depend on he echnology choice
𝜏
:
Fig. 9 Boxplo s o 100 Mon e Ca lo samples o he baseline scena io (blue), he al e na i e scena io wi h
less p ojec inance (o ange) and he al e na i e scena io wi h less p ojec inance and g een c edi eas-
ing (GCE) a
=500
(g ay) o di e en model a iables. Resul s a e shown o ou di e en simula-
ion ho izons: I (
=
0–249), II (250–499), III (500–749) and IV (750–999). a Sha e o enewable ene gy
(RE), b c edi ma ke success a e ( a io o success ul o o al numbe o eques s), c mean in es o le e -
age, d mean capi al adequacy a io o banks, e mean expec ed equi y a e o in es o s, mean c edi a e
o banks, g mean deposi a e o banks, h a e age numbe o bank de aul s, i a e age numbe o in es o
de aul s

790
T.Baldau , P.Jochem
Table 6 S a is ics (mean wi h s anda d de ia ion in b acke s) o he Mon e Ca lo simula ion in he al e na i e scena ios wi h less PF, and less PF wi h g een c edi easing
(GCE)
Range Less PF Less PF + GCE
I II III IV I II II IV
( =0−249)
II
( =250 −499)
(
=500 −749)
( =750 −999)
( =0−249)
( =250 −499)
(
=500 −749)
( =750 −999)
Sha e PF 0.26 (0.10) 0.40 (0.02) 0.43 (0.00) 0.44 (0.00) 0.24 (0.09) 0.36 (0.02) 0.42 (0.02) 0.46 (0.01)
RE Sha e 0.28 (0.04) 0.42 (0.04) 0.54 (0.04) 0.68 (0.04) 0.27 (0.04) 0.40 (0.04) 0.68 (0.11) 0.95 (0.04)
In es o Le e age 0.47 (0.00) 0.47 (0.00) 0.47 (0.00) 0.47 (0.00) 0.47 (0.00) 0.47 (0.00) 0.49 (0.00) 0.49 (0.00)
C edi Ma ke Success Ra e 0.19 (0.04) 0.19 (0.00) 0.20 (0.00) 0.21 (0.00) 0.19 (0.04) 0.19 (0.00) 0.19 (0.00) 0.18 (0.00)
Capi al Adequacy Ra io (%) 7.90 (1.96) 8.59 (1.82) 8.11 (0.68) 8.05 (0.90) 7.91 (2.50) 8.72 (1.91) 8.08 (0.67) 8.19 (1.15)
C edi Ra e (%) 2.25 (0.09) 2.13 (0.02) 2.23 (0.03) 2.35 (0.03) 2.29 (0.07) 2.17 (0.01) 2.27 (0.06) 2.40 (0.02)
Equi y Ra e (%) 8.54 (1.18) 9.30 (0.08) 8.83 (0.12) 8.65 (0.05) 8.65 (1.23) 9.27 (0.10) 7.74 (0.27) 8.20 (0.07)
Bank De aul s pe Pe iod 0.60 (0.32) 0.22 (0.23) 0.10 (0.13) 0.13 (0.14) 0.55 (0.27) 0.26 (0.23) 0.11 (0.14) 0.12 (0.13)
In es o De aul s pe Pe iod 0.63 (0.23) 0.97 (0.12) 0.83 (0.11) 0.73 (0.12) 0.53 (0.20) 0.87 (0.13) 1.34 (0.38) 1.45 (0.14)
Elec ici y P ice (1/phys. uni s) 49.95 (1.60) 44.44 (1.46) 39.36 (1.46) 33.90 (1.72) 50.22 (1.44) 44.96 (1.48) 33.77 (4.28) 23.00 (1.50)
S ike P ice (Sola ) 0.95 (2.92) 35.05 (8.12) 46.80 (1.22) 50.35 (0.91) 0.95 (2.92) 35.10 (8.19) 46.68 (1.14) 50.16 (0.89)
S ike P ice (Wind Onsho e) 0.29 (0.88) 23.66 (8.46) 39.32 (2.03) 44.26 (1.07) 0.30 (0.91) 23.76 (8.47) 39.30 (2.00) 44.06 (1.03)
S ike P ice (Wind O sho e) 5.41 (11.39) 62.55 (8.14) 75.15 (1.79) 80.91 (1.55) 5.45 (11.47) 62.97 (8.19) 74.97 (1.64) 80.58 (1.52)
P emium (Sola ) % 0.00 (0.00) 0.00 (0.00) 3.12 (2.17) 13.65 (3.89) 0.00 (0.00) 0.00 (0.00) 14.66 (9.40) 43.23 (5.04)
P emium (Wind Onsho e) % 0.00 (0.00) 0.00 (0.00) 0.10 (0.26) 7.30 (4.01) 0.00 (0.00) 0.00 (0.00) 6.02 (6.73) 33.95 (5.96)
P emium (Wind O sho e) % 0.00 (0.00) 0.28 (0.31) 2.56 (0.94) 8.90 (3.21) 0.00 (0.00) 0.29 (0.29) 9.34 (6.27) 31.95 (4.25)
791
P ojec inance o co po a e inance o  enewable ene gy?…
This s uc u al modi ica ion s a s a
=500
, when he go e nmen likely no ices
ha he phase-ou is no happening as as as expec ed om he baseline scena io,
and he e o e in oduces sho -handed measu es o eco e he ansi ion pa hway.
The GCE esul s in a mo e op imis ic e alua ion o de aul p obabili ies, as well
as a lowe isk ma k-up o le e ed in es o s. Howe e , hese imp o ed condi ions
a e only g an ed o enewable in es o s. Fu he , we allow banks o sligh ly “o e -
s ain” hei capi al adequacy a ios when lending o g een in es men s (
Δ
𝜅
RE
min
).
The emaining mechanisms o banks and in es o s s ay unchanged. As depic ed in
Fig.8c and , he numbe o enewable ene gy p ojec s and he powe mix show a
much mo e inclined g ow h ajec o y as soon as he GCE is in oduced, app oach-
ing he le els o enewable sha e and powe p ice obse ed in he baseline sce-
na io by he end o he simula ion pe iod. The e o e, GCE e ec i ely b idges he
inancing gap c ea ed by educed PF, enabling con inued in es men in enewable
ene gy p ojec s. In e es ingly, he c edi ma ke success a e s ay app oxima ely he
same wi h GCE, indica ing imp o ed a ailabili y and accessibili y o inancing o
g een p ojec s, bu an app oxima ely equal o e all success a e in c edi s. The a e
o in es o de aul s is sligh ly highe bu mo e s abilized wi h espec o he o he
scena ios, sugges ing a balanced isk en i onmen o in es o s acili a ed by GCE.
E en hough he a e age alue o bank de aul s sligh ly educes in pe iods I and II,
he occu ence o bank de aul c ises pe sis s (see Fig.9). Howe e , he banking
sys em seems o e ec i ely s ee i s c edi g an ing owa d g een in es o s, shielding
he ossil in es men s, while main aining a low- isk le el o ese es.51 In summa y,
he GCE policy e ec i ely mi iga es he nega i e impac o educed PF, p omo ing
a s eady g ow h in enewable ene gy in es men s. This scena io sugges s ha com-
bining PF wi h suppo i e inancial policies like GCE migh enhance he esilience
and e ec i eness o he inancial sys em in d i ing he sus ainable ene gy ansi ion.
GCE helps main ain a high in es men a e while managing isk, unde sco ing he
impo ance o policy in e en ions in sus ainable inance.
(19)
𝜌�
1=
{
𝜌1,𝜏+Δ𝜌
RE
1i 𝜏is RE
𝜌1else
,
𝜌�
2,𝜏
={𝜌2+Δ𝜌RE
2i 𝜏is RE
𝜌2else ,
𝜅
�
min,𝜏
=
{
𝜅min +Δ𝜅RE
min i 𝜏
is RE
𝜅
min
else.
51 The eason behind his shielding e ec could po en ially lie in he design o he capi al ese e bu e .
We lea e he explo a ion o he in luence o he beha io al pa ame e s o he banks, like
𝜅RE
min
o u u e
esea ch.
792
T.Baldau , P.Jochem
4 Conclusions andou look
In summa y, his pape has de eloped and employed an agen -based model o
s udy he e ec s o di e en inancial s uc u es on enewable ene gy in es men s.
The model’s de ailed ep esen a ion o in es o beha io , inancial ma ke in e -
ac ions, and powe ma ke isk allows o an analysis o he oles o PF and CF in
he enewable ene gy ansi ion. By in eg a ing elemen s such as p ojec -speci ic
isks, in es o he e ogenei y, and a ma ke p emium design, he model p o ides a
comp ehensi e amewo k o unde s and he inancial mechanisms d i ing sus-
ainable ene gy in es men s, and is able o ep oduce basic s ylized ac s abou
ene gy in es men s. The indings sugges ha p ojec inance is essen ial o apid
enewable ene gy deploymen , as i enables a second inancing channel o powe
plan s. Wi hou PF, he in es men pace slows down signi ican ly, and achie -
ing sus ainabili y a ge s becomes mo e challenging. Howe e , also GCE can
e ec i ely suppo enewable ene gy in es men s by imp o ing access o g een
c edi and educing inancial isks, c ea ing a obus inancing en i onmen ha
can accele a e he ansi ion o enewable ene gy. These esul s sugges ha PF
is sui able o os e ing enewable ene gy g ow h due o i s abili y o manage and
dis ibu e inancial isks h ough special pu pose ehicles (SPVs).
Fu u e esea ch should u he explo e he model’s pa ame e space in o de o
ind an op imal mix o inancial mechanisms and policy in e en ions o enhance he
esilience and e ec i eness o he inancial sys em in suppo ing enewable ene gy
in es men s. This includes in es iga ing he in e play be ween a ious o ms o
inancing, egula o y amewo ks, and ma ke condi ions. Addi ionally, expanding
he model o include o he o ms o enewable ene gy echnologies such as s o -
age and g id echnology, and di e en powe ma ke policies will p o ide deepe
insigh s in o mo e complica ed ene gy sys ems. Mo eo e , add essing he obse ed
endogenous c ises si ua ions in inancial indica o s will be c ucial o de eloping
obus s a egies o mi iga e inancial isks. By iden i ying he condi ions ha lead o
ex eme alues and inancial ins abili y, policymake s and inancial ins i u ions can
be e design in e en ions o ensu e a smoo h and con inuous ansi ion o a sus ain-
able ene gy u u e. Fu he expansions could include he combina ion o his model
wi h an in eg a ed assessmen model, o a u he dis-agg ega ion o he es -o -
economy agen . Finally, he ole o ene gy ade has emained unexplo ed, o which
he analysis o inancial s abili y in he con ex o geopoli ical isks is le o u u e
esea ch. A u he expansion o he model could be he in oduc ion o al e na i e
banking mechanisms, such as he e alua ion o in es o s acco ding o he expec ed
ope a ing cash low ins ead o he inancial le e age.
In conclusion, his s udy un eils he ole o PF in d i ing he enewable ene gy
ansi ion and he ole o policy in e en ions like GCE in suppo ing his p o-
cess. This model sugges s a balanced app oach ha le e ages he s eng hs o PF,
supplemen ed by a ge ed inancial policies, o e ec i ely add ess he challenges
o inancing enewable ene gy p ojec s and con ibu e o achie ing an economy’s
sus ainabili y goals. The model ep esen s a good basis o u he analyses o
usage as an ene gy module in a la ge in eg a ed assessmen model.
793
P ojec inance o co po a e inance o  enewable ene gy?…
Appendix
Appendix A: Expanded li e a u e e iew
The linkage be ween powe ma ke aspec s and agen -based in eg a ed assess-
men modeling is elabo a ed in his pa ag aph, ex ending he li e a u e e iew o
he main a icle.
In models ocusing on inance, bank lending plays a supe io ole o e echno-
economic aspec s. In es men mechanisms a e closely linked o changes in he
balance shee o in es o s and banks. This esea ch s eam add esses he ques ion
o possible ex e nal inance o ca bon- iendly i ms om a pe spec i e o com-
me cial banks, h ough he channels o con en ional bank lending, ma ke deb
(i.e., g een bonds) and ma ke equi y in es men s (Campiglio 2016; Campiglio
e al. 2018). Fu he , D’O azio and Popoyan (2019) p opose he deca boniza ion
o banks’ balance shee s in a mac o-p uden ial policy con ex . The au ho s d aw
conclusions o cen al banking and ela ed ese e equi emen s.
In agen -based elec ici y ma ke models, he expansion planning o powe
gene a ion capaci ies is ypically de e mined ia he ne p esen alue o powe
plan p ojec s, based on he elec ici y p ice. De elopmen s in ma ke p ice le els
can be modeled subjec o empo al supply sca ci y o as a esponse o speci ic
ma ke mechanisms (K aan e al. 2018; Chappin e al. 2017; Jimenez e al. 2024).
Thus, he eac ion o he e ogeneous powe supplie s o ma ke incen i es can be
analyzed (Pu kus e al. 2015; Jonson e al. 2020; Ba azza and S achan 2020), bu
mino modeling ocus is a ibu ed o he ac ual inancing o m and balance shee
composi ion esul ing om hese ma ke incen i es.
Rep esen ing ye ano he s eam o li e a u e, ull-sized mac o-ABMs aim a a
mo e holis ic unde s anding o he economy in he con ex o sus ainable ansi-
ions. Fo example, in (Nieddu e al. 2022; Pon a e al. 2018; Rabe o e al. 2019)
and (Lampe i e al. 2020, 2021), g een inancial policies in ABMs wi h he e oge-
neous i ms and banks a e in es iga ed. Howe e , hese ypes o ABM a e able o
add ess he ene gy-economy-clima e nexus in a highly in eg a ed ashion, mino
a en ion is a ibu ed o he elec ici y ma ke clea ing mechanisms and associ-
a ed in es men isks, nei he a e di e en gene a ion echnologies add essed.
Simila ly, Ciola e  al. (2023) conside he in oduc ion o wind all p o i s and
changes in ossil uel p ices, bu do no mi o long- e m me i o de e ec s.
Gene ally, in ABMs, he b idge be ween powe ma ke p emia and al e na i e
inancing s uc u es o enewable ene gy (in pa icula p ojec inance) is sel-
dom buil . Solely A i and Koc (2019) ha e p esen ed a i s agen -based model
o sus ainable inancing o Qa a sola a ms. The model includes a ep esen a-
i e bank and a ini e numbe o uni o mly dis ibu ed agen s, who can make use
o equi y-based inancial in e media ies o con en ional bank loans. In a ollow-
up wo k, he au ho s also conside a model wi h a philan h opic c owd unding
o ganiza ion (A i and Koc 2021). Whe eas he au ho s aim a desc ibing weal h
de elopmen s, hey exogenize assump ions abou go e nmen incen i es and isk
accommoda ion.
794
T.Baldau , P.Jochem
In summa y, models conside ing sus ainable inance in agen -based mac oeco-
nomic modeling ha e been iden i ied in li e a u e, as well as models inco po a ing
capaci y in es men s in he con ex o powe ma ke s. Ne e heless, a esea ch gap
lies in he combina ion o ene gy inance, powe ma ke incen i es and echnologi-
cal he e ogenei y. This pape aims a p o iding a con ibu ion owa d illing his gap.
Appendix B: Model de ails
Appendix B.1: Balance shee ma ix
Table7shows he balance shee ma ix o he model.
Appendix B.2: T ansac ions‑ low ma ix
Table8 shows he ansac ions- low ma ix o he model.
Table 7 Balance shee ma ix showing he balance shee en ies o banks, in es o s, special pu pose
ehicles (SPVs), he cen al bank (CBank), he go e nmen and he es o economy (RoE)
Bank In es o SPV CBank Go e n-
men
RoE
A L A L A L A L A L A L
Deposi s D
DI
DS
DR
Loans L
LI
LS
LR
Cen al Bank Loans C C
Cu ency in ci cula ion M
Bonds B B
P ojec Equi y
Ep
Ep
(Powe Plan Asse s)
KI
KS
Ne Wo h
NWB
NWI
NWS
NWC
NWG
NWR

795
P ojec inance o co po a e inance o  enewable ene gy?…
Table 8 T ansac ions low ma ix including he cu en accoun (CA) and capi al accoun (KA) o banks, in es o s, special pu pose ehicles (SPVs), he cen al bank
(CBank), he go e nmen and he es o economy (RoE)
T ansac ion Bank CBank Go e nmen In es o RoE Sp To al
CA KA CA KA CA KA CA KA CA KA CA KA
Di idends
−Di b
−Di I+Di S
Di R
−Di S
0
Elec ici y Consump ion
+EleI
−Ele
+EleS
0
Powe Plan Ope a ion
−C(Y)I
+C(Y)
−C(Y)S
0
In e es Bonds
+ bB
− bB
0
In e es Cb C eds
− cC
+ cC
0
In e es C edi s
+In 𝓁
−
𝓁
,ILI
−
𝓁
,RLR
−
𝓁
,SLS
0
In e es Deposi s
−In d
+ dDI
+ dDR
+ dDS
0
In es men
−II
II
−IS
IS
0
Taxes T
−TI
−TS
0
Bailou
+bo −boB
−boG
0
Re ained Ea nings
−(Re B)
+(Re B)
−(Re C)
+(Re C)
−(Re G)
+(Re G)
−(Re I)
+(Re I)
−(Re R)
+(Re R)
−(Re S)
+(Re S)
0
Δ
Deposi s
−ΔDB
ΔDI
ΔD
ΔDS
0
Δ
Loans
ΔLB
−ΔLI
−ΔLS
0
Δ
P ojec Equi y
ΔE
−ΔE
0
Δ
Cen al Bank Loans
−ΔC
+ΔC
0
Δ
Bonds
ΔB
−ΔB
0
To al 0 0 0 0 0 0 0 0 0 0 0 0 0
796
T.Baldau , P.Jochem
Appendix C: S uc u al model pa ame e s
Table9 summa izes he alues and desc ip ions o he s uc u al pa ame e s o he
model. Table10 shows he ac ions o ese es sold o bough by banks o op imize
hei composi ion o ese es wi h espec o he mac o-p uden ial equi emen s.
Table 9 Pa ame iza ion o he powe ma ke model. Sou ce: own alues
Va iable Desc ip ion Value
𝜃
F ac ion o in es o p o i s paid ou as di idends 0.6
𝜏0
Tax a e 0.3
⋆
Cen al bank a e 0.0
b
Go e nmen bond yield 0.0
𝜅min
Rese e equi emen 0.08
𝜒E
Subse sampling size on he equi y ma ke 0.10
𝜒C
,𝜒
′
C
Subse sampling size on he c edi ma ke 0.20, 0.05
𝜒D
,𝜒
′
D
Subse sampling size on he deposi ma ke 0.15, 0.05
N†
Numbe o a emp s be o e e hinking an in es men decision 4

N
Maximum numbe o equi y ma ke eques s pe SPV 2
𝜉E
Maximum sha e o equi y unds pu in o one SPV 0.2
𝜂R
D
,𝜂
R
L
Sha e o es o economy c edi s/deposi s shi ed 0.005
𝛼e
Weigh ing ac o o expec ed e u n on equi y 0.9
Nm
Random numbe o banks in es o economy inancial ma ke s 2
𝛽a
min,𝛽a
max
Asse be a dis ibu ion 0.8,1.0
𝛼↑
d
,𝛼
↓
d
Lea ning speeds (deposi ma ke ) 0.057,0.050
𝛼↑
𝓁
,𝛼
↓
𝓁
Lea ning speeds (c edi ma ke ) 0.002,0.14
𝛽0,𝛽1
Powe p ice s a ing le el and me i o de slope 60.0,40.0
𝛿
min,0
,
𝛿
max,0
Ini ial in e es a e pa ame e s (deposi s) (0.004,0.006)
𝓁
min,0
,
𝓁
max,0
Ini ial in e es a e pa ame e s (c edi s) (0.015,0.025)
𝜚1
Bank pa ame e (c edi e alua ion) 0.5
𝜚2
Bank pa ame e (c edi e alua ion) 0.001
𝜚3
Bank pa ame e (c edi e alua ion) 0.03
𝜆
Bank pa ame e (c edi e alua ion) 4
p0
SPV c edi accep ance p obabili y 0.85
𝜁d,𝜁c
F ac ion o c edi s and deposi s o he RoE agen shi ed on he inan-
cial ma ke s
0.005
LPowe load 150
Table 10 O e iew o he ac ions o ese es sold and bough by banks o each hei ese e equi e-
men goals. Sou ce: own assump ions
Pa ame e
𝜑L
𝜑buy
B
𝜑sell
B
𝜑buy
CB
𝜑sell
CB
Value 0.9 0.6 0.6 0.8 0.98
797
P ojec inance o co po a e inance o  enewable ene gy?…
Appendix D: Technology pa ame e s
Appendix E: Powe ma ke de ails
Appendix E.1: Sho ‑ e m powe ma ke dynamics
The clea ing p ice o he ma ke is app oxima ed as a linea unc ion o he enew-
able sha e in he powe plan pool, such ha
P0=𝜂1−𝜂2YRE∕(YRE +Y ossil)
, whe e
pa ame e s
𝜂1,𝜂2
de e mine he s eng h o he me i o de e ec . A second p ice
e m is added o accoun o luc ua ions in wea he and egional he e ogenei y no
explici ly modeled. This sho - e m powe ma ke model is adap ed om li e a u e
(Ki zing and Webe 2014), employing a wo- ac o s ochas ic model, whe e he log-
a i hm o he powe p ice is di ided in o a long- e m and a sho - e m componen ,
i.e.,
whe e
d

P
is a long- e m B ownian mo ion and
d𝜒
is a mean- e e ing sho - e m
p ocess, in pa icula an O ns ein–Uhlenbeck p ocess.52 In ou model, we use he
(20)
ln P =𝜉 +𝜒
(21)
d

P =
P
(
𝜇𝜉+
1
2
𝜎2
𝜉
)
d +
P 𝜎𝜉dz
𝜉
(22)
d𝜒 =−𝜅𝜒 d +𝜎𝜒dz𝜒
Table 11 Technology da a used in his pape : Typical sys em size (MW), annual ull-load hou s (h),
p obabili y o p ojec inance (p(PF)) in he baseline scena io (BL) and he al e na i e scena ios wi h
less p ojec inance (PF), p obabili y o echnology choice p(T), ixed ope a ional cos s (kEUR/kW), a i-
able ope a ional cos s (kEUR/kWh) and capi al expendi u es (kEUR/kW). Sou ces: own assump ions
Technology Size

K
(MW) Load
Hou s (h/
yea )
p(PF) BL /
less PF %
p(T) %
C ix
(Eu /kW)
C a
(Eu /
kWh)
Capex
Sola PV 50 1130 28.8/0.96 29 20 0.0 700
Wind
Onsho e
40 2570 24.0/0.8 56 19 0.002 1800
Wind O -
sho e
100 2820 15.0/0.5 5 70 0.004 3000
Ligni e 75 6970 0.0/0.0 0 40 0.0009 2300
Ha d Coal 100 6550 6.6/0.22 0 42 0.018 1900
Na u al Gas 50 2600 1.8/0.06 10 25 0.050 800
52 As he Phelix u u e ma ke is ba ely used in ecen yea s (2021–2022), a smoo hed spo p ice as a
p oxy o he base p ice. The calib a ion p ocess o his model is desc ibed in (Ki zing and Webe 2014).
In his pape , we only make heo e ical assump ions abou he sho - e m pa ame e s.
798
T.Baldau , P.Jochem
a bi a y pa ame e s
𝜉=3.7, 𝜒=0.5, 𝜇𝜉=0.001, 𝜅=0.5, 𝜎xi=0.1, 𝜎𝜒=0.1
and o e lay he esul ing sho - e m luc ua ions wi h he long- e m clea ing p ice
o accoun o mo e s ochas ic (and hus close - o- eali y) e enue s eams o he
in es o s.
Appendix E.2: Nume ical de e mina ion o  inancing condi ions
Fo compu ing he s ike p ice endogenously, we make use o he algo i hm de i ed
by Neuho e al. (2022).53 This me hod sums o e he dis ibu ion
𝜙
o all expec ed
ma ke alues o powe plan s o yield he equi ed amoun o equi y (E) in he p o-
jec , gi en he s ike p ice S:
Equa ion (23) s a es ha he equi y-se ing cash low equals he expec ed e enue
s eams om a powe plan beyond he s ike p ice S, yielding i s physical ou pu

Y
o a gi en pe iod. He e,
ae
is he equi y-se ing ac o ( he expec ed ac ion o
equi y o be paid o sha eholde s in each inancing pe iod). This equa ion e lec s
ha insecu e e enue s eams a e backed by equi y. The dis ibu ion
𝜙
in luences
how s ic his ule applies. I expec a ions a e ela i ely s able,
𝜙
migh be a unc-
ion wi h a sha p peak a ound S, and equi y will only accoun o a sho igh
ail. I expec a ions a e uzzy,
𝜙
exhibi a b oade ail beyond S and he amoun o
equi y equi ed inc eases. In his pape ,
𝜙(pi)∼N(𝜇𝜙,𝜎𝜙)
is a no mal dis ibu ion
o expec ed ma ke alues ac oss powe plan s o he same echnology,54 whe e
𝜇𝜙
is gi en by las pe iod’s powe p ice le el and he s anda d de ia ion
𝜎𝜙(= 2.0)
is
an indica o o he e ogenei y in expec a ions. Mind ha his is a conse a i e and
myopic s a egy as ma ke alues end o all wi h inc easing enewable amp-up
in he long un. The po ion o deb D in he in es men co e s a leas he cos s o
elec ici y gene a ion:
He e, he cos s
C a
and
C ix
o a iable and ixed ope a ional cos s a e included.
The deb -se ing ac o
ad
and equi y-se ing ac o
ae
ead55
(23)
a
eE(S)!
=
Y
∫∞
S
𝜙(p�)dp�
.
(24)
ad
D
(
S
)!
=

Y
(
S
−
C a
)−
C ix
.
53 We o wa d he in e es ed eade o his a icle, which ep esen s a ho ough de i a ion and in es iga-
ion o he me hod summa ized he e. The only di e ence in ou model is ha also ixed expendi u es a e
co e ed by deb .
54 We app oxima e he in eg al in equa ion23 o e a disc e e dis ibu ion o expec ed ma ke alues.
55 The deb and equi y-se ing ac o s esul om he geome ic sum o e discoun ed cash lows in he
en i e p ojec li e ime T, i.e.,
D
=∑
T
p
=1
a
d
D∕(1+ 𝓁) ,E=∑
T
p
=1
a
e
E∕(1+
e
)
. A mo e ho ough de i a-
ion can be ound o example in Gö ze e al. (2008).
805
P ojec inance o co po a e inance o  enewable ene gy?…
Wang J, Qiang H, Liang Y, Huang X, Zhong W (2024) How ca bon isk a ec s co po a e deb de aul s:
e idence om Pa is ag eemen . Ene gy Econ 129:107275
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps
and ins i u ional a ilia ions.