1
POLICY BRIEF
Inno a ing Clima e-Economy Modelling:
Dealing wi h Unce ain y, Risk and
Complexi y
Au ho s:
Kos as F agkiadakis, S elios Tsia as, Giannis
Cha alampidis (E3M)
Jonas Haas, Sebas iano Bacca (Global Clima e Fo um)
Ugne Keliauskai e (B uegel)
Reinha d Mechle (IIASA)
Femke Nijsse, Ian Bu on (Uni e si y o Exe e )
Jamie Pi ie, Ch is Thoung (Camb idge Econome ics)
This policy b ie was de eloped wi hin he amewo k o he DECIPHER p ojec , unded by he Eu opean
Union’s Ho izon Eu ope esea ch and inno a ion p og amme (G an Ag eemen No. 101056898).
2
1. Mo i a ion
Designing e ec i e clima e, biodi e si y and ene gy policy equi es making decisions oday
abou unce ain u u es. Policymake s ace complex isks ha un old o e decades, shaped
by echnological inno a ion, e ol ing inancial condi ions, and he mac oeconomic
impac s o ex eme wea he . To na iga e hese challenges, models o he clima e–
economy sys em a e indispensable o explo ing al e na i e u u es and s ess- es ing
policy choices. The alue o hese models depends on how well hey cap u e unce ain y,
inco po a e beha iou al dynamics, and e lec eal-wo ld cons ain s such as inancing
cos s, sys emic shocks, and physical damages. Wi hou hese elemen s, e en sophis ica ed
analyses can miss c i ical ulne abili ies o oppo uni ies.
The DECIPHER (Decision-making amewo k and p ocesses o holis ic e alua ion o
en i onmen al and clima e policies) p ojec , unded unde Ho izon Eu ope, aimed o
o e come limi a ions in con en ional policy app aisal by c ea ing a new, holis ic decision-
making amewo k ha be e cap u es he complex nexus o clima e change, biodi e si y,
and he economy. The p ojec included an i e a i e knowledge co-c ea ion p ocess o
mo e anspa en , inclusi e and ep esen a i e policy design and e alua ion. Mo eo e , i
de eloped new gene a ion o s a e-o - he-a economic models ea u ing eedback
loops wi h physical sys em models and embedded sys emic isks and unce ain y and
cap u e beha iou al and knowledge dynamics allowing i o imp o e he ep esen a ion o
he economy-clima e-biodi e si y nexus.
Compa ed o mains eam economic models ha o en neglec unce ain y and esilience,
DECIPHER in eg a es ad anced economic and biophysical modelling, empi ical me hods,
and s akeholde co-c ea ion o assess he easibili y, esilience, isk, and oppo uni y
dimensions o clima e and en i onmen al policy op ions. The amewo k is designed o be
ope a ional unde eal-wo ld condi ions and is being applied o key EU policy domains such
as he “Fi o 55” package, LULUCF egula ion, and na ional eco e y plans. Th ough his
app oach, DECIPHER equips policymake s wi h ools ha do mo e han o ecas : hey e eal
ade-o s, highligh obus s a egies, and s eng hen he legi imacy and adap abili y o
decisions in an unce ain u u e.
DECIPHER esea ch d aws upon he comp ehensi e amewo k o IRGC (2006)
1
o iden i y
discou ses ha science, policy, and applica ion mus add ess in he ‘unce ain y’ space.
These discou ses a e based on di e en classes o isk and unce ain y, b oadly aligned
wi h he Knigh ian isk and unce ain y de ini ions o simple and complex, high-unce ain y,
and high-ambigui y isk (see Figu e 1).
• Ins umen al discou se is sui able o add essing clea ly de ined isk p oblems,
employing well- es ed decision suppo me hods like cos -bene i analysis and
ocusing on economic incen i es and echnical solu ions.
1
IRGC-In e na ional Risk Go e nance Council (2006). Risk go e nance: Towa ds an in eg a i e app oach.
Whi e pape no. 1. IRGC, Gene a.
3
• S ong pa icipa o y discou se is c ucial, pa icula ly (bu no exclusi ely) o add essing
issues o ambigui y whe e alues a e con es ed, aiming a con lic esolu ion and
b oad-based s akeholde engagemen .
• Re lec i e discou se becomes essen ial when dealing wi h high le els o unce ain y,
emphasizing p ecau iona y p inciples and he need o ca e ul conside a ion o 'dange '
and 'adap a ion limi s.’
• Epis emological discou se is pa icula ly impo an o cha ac e izing he a ailable
e idence o unde s anding isk ac oss he en i e isk spec um, especially in he ace o
complexi y and igno ance.
Figu e 1: The isk and unce ain y space.
Sou ce: Mechle e al., 2025.
2
Based on Knigh , 1921
This policy b ie p esen s i e ecen modelling ad ances ha s eng hen he e idence base
o clima e and biodi e si y ac ion and economic decision-making amids unce ain y
based on DECIPHER esea ch:
1. Emula ion o Unce ain y Quan i ica ion – enabling housands o simula ions o
iden i y which policies a e obus ac oss many u u es.
2. Ra ional Expec a ions in Compu able Gene al Equilib ium (CGE) Models – showing
how o wa d-looking in es men beha io can lowe ansi ion cos s and smoo h
shocks.
3. Cos o Financing in Technology Di usion – e lec ing how in e es a es and capi al
cos s a ec he up ake o clean echnologies.
4. Flood and Coas al Damage Assessmen – linking de ailed coas al isk assessmen
wi h mac oeconomic models o e eal he economic alue o adap a ion.
5. Mul iple Resilience Di idend – cap u ing a oided losses, de elopmen co-bene i s
and inequali y educ ions om isk-managemen in es men s, s eng hening he
case o sus ained adap a ion and mi iga ion.
2
Mechle , R., Żeb owski, P., Cle cq-Roques, R., Pa il, P., S e an Hoch aine -S igle (2025).The ole o
ex eme e en and sys emic isk - assessmen and guidance. DECIPHER p ojec , Deli e able 5.3
Ins umen al
Risk Ambigui y
Unce ain y Complexi y
Epis emological
Re lec i e
Ins umen al Pa icipa o y
4
These inno a ions sha e a common goal: o assis decision-make s in mo ing beyond
s a ic o ecas s owa d mo e esilien policy de elopmen . They iden i y a ge ed measu es,
such as in es ing in adap a ion, imp o ing access o a o dable inance o enewable
ene gy, o conside ing in es o expec a ions, which migh help con ain cos s and deli e
long- e m economic and social bene i s.
Inco po a ing hese app oaches in o clima e and ene gy planning and impac assessmen
ools can help policymake s make c edible choices ac oss a ious u u e scena ios,
inc easing con idence ha policies will sa egua d ci izens and economies while suppo ing
a well-managed ansi ion.
The ollowing sec ions p esen each modelling inno a ion by se ing ou he p oblem i
add esses and why i ma e s o policy, he solu ion wi h key modelling ea u es,
illus a i e igu es showing main esul s, guidance on applica ion, including when he
me hod is mos sui able and i s limi s, and e e ences o academic wo k by p ojec pa ne s
o eade s seeking u he de ail.
2. Inco po a ing unce ain y in impac assessmen ools
2.1. Emula ion o Unce ain y Quan i ica ion
P oblem
T adi ional clima e and mac oeconomic scena io models o en ail o cap u e unce ain y
in policy inpu s and assump ions sys ema ically, making i challenging o policymake s o
assess policy obus ness, de ined as he abili y o achie e policy goals unde unce ain y
and shocks, ac oss a ious possible u u es. Robus ness di e s om esilience, which
emphasises ‘ e u ning o a s able equilib ium poin a e a shock’.
Modelling solu ion
A machine-lea ning-based emula o was de eloped o se e as a compu a ionally
inexpensi e su oga e o complex simula ion models, such as FTT:Powe . The emula o
allows o :
• housands o scena io e alua ions a minimal cos ,
• simul aneous a ia ion in unce ain inpu s ( o example, echno-economic
pa ame e s such as lea ning a es
3
, as well as policy ambi ion).
The emula o sys ema ically explo es unce ain y in 15 echno-economic pa ame e s, such
as build and connec ion speed, lea ning a es, ene gy demand g ow h, cos o capi al, and
echnology li e imes. In addi ion, he emula o explo es he e ec o di e en policy
ins umen s wi hin he coun y ha implemen s hem and c oss-bounda y.
Illus a i e example
Key unce ain ies in he speed o ansi ion a e iden i ied in Figu e 2. O e he las decade,
he pace o building ou sola and wind ene gy has accele a ed, and g id expansion has
s uggled o keep pace wi h he g ow h o enewables in some coun ies. These wo
3
A measu e o educ ion in cos s o ene gy echnologies o each doubling o cumula i e p oduc ion o
capaci y.
5
unce ain ies lead o he highes a ia ion in FTT:Powe . Lea ning a es o onsho e wind
come nex ; i s mean lea ning a e is much lowe , so ha sligh a ia ions in he lea ning a e
can ha e a la ge e ec on cos -compe i i eness. Finally, Chinese policy plays a key ole in
he global cos o ce ain echnologies, as hei la ge ma ke has he s onges abili y o
induce inno a ion. US subsidies, and hei po en ial ollback, ha e a limi ed di ec e ec ia
induced inno a ion. Sola PV is no longe sensi i e o hese dynamics o egional policy.
Figu e 2: One-a -a- ime analysis (oaa ), o he op 19 a iables in he analysis, (a) in e ms
o global emissions, capaci y o onsho e wind and capaci y o o sho e wind. Panel (b)
shows he a e age sensi i i ies ac oss all h ee ou pu s.
No e, he igu e was p oduced using code adap ed om McNeall e al. (2024).
Figu e 3 shows he obus ness o policy combina ion agains key unce ain ies. None o he
policies a e e y obus agains g id delays and wo sening o ela i e build imes o sola
and wind (bo om ow). In e ms o o he unce ain ies, such as he cos o inance and
high demand g ow h, he combina ion o subsidies and phase-ou s is he mos obus .
Figu e 3. Sha e o emula o uns mee ing India’s 2030 a ge s. The bo om ow shows he
policy combina ions es ed (cu en policy, and combina ions o up on subsidies (Sub),
ca bon p icing (CP), and phase-ou s (Phase).
6
Applica ion
Use when: a high numbe o model uns is equi ed o assess he sensi i i y o obus ness
o policy ou comes.
A oid when: he a ge model has apidly shi ing dynamics ha a e oo complex o be
accu a ely cap u ed by emula o s wi hou a la ge amoun o aining da a.
2.2. Ra ional Expec a ions in Compu able Gene al Equilib ium (CGE)
Models
P oblem
Expec a ions shape economic decisions as hey e lec how agen s belie e key economic
undamen als will e ol e. In economic modelling, he way expec a ions a e o med s ongly
a ec s model esul s and policy implica ions, de e mining no only he e en ual equilib ium
bu also he pa h he economy akes o each he equilib ium. Typically, in economic
models, expec a ions e e o he ajec o y o p ices o cos s, and he e a e wo mains eam
app oaches:
• Myopic expec a ions, whe e agen s lack o esigh and base hei decisions solely
on cu en condi ions,
• Ra ional expec a ions, whe e agen s ha e pe ec o esigh , meaning hey possess
comple e in o ma ion on how, o example, p ices will change in he u u e, and hei
beha iou is in luenced no only by cu en condi ions bu also by an icipa ed u u e
condi ions.
Myopic expec a ions, when coupled wi h es ic ions on capi al mobili y, may lead o an
o e es ima ion o cos s ela ed o he clean ene gy ansi ion. The li e a u e has highligh ed
how myopic expec a ions can lead o s onge esponses o shocks and how a ional
expec a ions can lead o milde esponses. Howe e , since hese a e wo ex eme cases o
expec a ion o ma ion, esea che s ha e also explo ed al e na i e me hods, such as
inco po a ing sa ings in o he in e empo al p oblem.
7
Modelling solu ion
A sho e sion o CGE model, he GEM-E3 model, was de eloped ha in oduces wo
enhancemen s:
• Ra ional expec a ions, allowing in es o s o o m o wa d-looking iews abou
u u e e u ns,
• Capi al mobili y cons ain s, which can be se as pa ial o ull, o cap u e limi s on
how easily capi al lows be ween sec o s o egions.
These ea u es enable a mo e ealis ic ea men o policy impac s and capi al ealloca ion
dynamics. Addi ionally, a ional expec a ions we e in oduced o he ull e sion o he GEM-
E3 model, and he economic implica ions o he g een ene gy ansi ion we e examined o
Ge many and I aly.
Illus a i e example
To illus a e he in luence o expec a ion o ma ion and capi al mobili y, a pe manen
demand shock o pho o ol aic (PV) equipmen is simula ed in he sho e sion o he GEM-
E3 model wi hin he h ee- egion model. Regions R1 and R2 ac as ne impo e s o PV
equipmen , while Region R3 se es as he expo e . The shock o igina es in R1. Scena ios
compa e ull capi al mobili y (ALL) wi h pa ial mobili y (PART).
Figu e 4: Uni cos o capi al unde myopic s a ional expec a ions.
Myopic expec a ions lead o no an icipa o y adjus men , while a ional expec a ions igge
ea lie in es men , mode a ing cos spikes. The assump ions on capi al mobili y g ea ly
in luence he magni ude o he impac s, as capi al cos s change unde pa ial mobili y,
peaking a a ange o 10% o 18% compa ed o 0.7% unde he assump ion o ull capi al
mobili y.
Figu e 5: In es men pa e ns unde myopic s a ional expec a ions.
8
Unde a ional expec a ions, in es men begins ea lie (2025) in an icipa ion o 2030
demand. Myopic agen s eac only a he ime o he shock, esul ing in highe capi al cos s
and ine icien alloca ion. These implica ions a e clea e unde he assump ion o limi ed
capi al mobili y, as he shock in capi al p ices is signi ican ly highe , hence he adjus men
o in es men s unde pe ec o esigh begins much ea lie .
Then, a ional expec a ions we e inco po a ed in o he ully- ledged GEM-E3 model e sion.
To achie e his, a i s -o de app oxima ion was pe o med due o he model's la ge scale
and complexi y.
Figu e 6: Change in he uni cos o capi al (A) and in es men s (B), (C) in % om he
e e ence.
-0.40%
-0.30%
-0.20%
-0.10%
0.00%
0.10%
0.20%
2025 2026 2027 2028 2029 2030 2035 2040 2045 2050
(A)
Ge many I aly
-2.0%
-1.0%
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
8.0%
2025 2026 2027 2028 2029 2030 2035 2040 2045 2050
(B)
Ge many ( a ional) Ge many (myopic)
-4.0%
-3.0%
-2.0%
-1.0%
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
2025 2026 2027 2028 2029 2030 2035 2040 2045 2050
(C)
I aly ( a ional) I aly (myopic)
9
Changing he way expec a ions a e o med may also a ec he sec o al s uc u e o
in es men , shi ing i away om adi ional manu ac u ing sec o s and owa ds he
p oduc ion o clean ene gy and he manu ac u ing o clean ene gy equipmen .
Fu he mo e, his shi may pu p essu e on he cu en accoun balance due o highe
in es men s in he sho o medium e m, which leads o an inc ease in impo s o
in es men goods.
O e all, he a ional expec a ions assump ions lessen he impac s o he ansi ion on he
economy and lead o a highe GDP compa ed o myopic expec a ions. This, along wi h
o he s uc u al changes and shi s in mac o d i e s du ing he ansi ion, highligh s he
need o ca e ully conside he o ma ion o expec a ions when assessing he impac s o he
clean ene gy ansi ion.
Applica ion
Use when ying o assess he empo al e ec s o shocks o policy announcemen s wi h
long lead imes.
A oid when ocusing on e y sho - e m impac s o o si ua ions whe e eliable da a on
expec a ions a e lacking.
2.3. Cos o Financing in Technology Di usion
P oblem
The ansi ion equi es high le els o in es men in low-ca bon echnologies. These
echnologies a e mo e capi al-in ensi e han ossil- uel echnologies, making he ques ion
o inancing mo e impo an , bo h in i sel and when compa ing he ela i e a ac i eness
o he echnology op ions. An unde s anding o he in es men en i onmen and he
ela i e me i s o di e en echnology op ions is i al o in o m e ec i e policy. Con e sely,
ailing o accoun o such speci ici ies isks unde mining he eliabili y o p ojec ed
echnology ansi ions, especially in scena ios in which wide mac oeconomic o policy
condi ions migh change.
Unde con en ional modelling ea men s, a ixed discoun a e ( o example, 10%) is mo e
common, ac oss all echnologies ( ailing o iden i y echnology-speci ic ea u es) and
egions ( ailing o igno e mo e local inancing condi ions). This simpli ica ion hus igno es
di e en ials ha migh be consequen ial in de e mining he pace and global dis ibu ion o
he ansi ion, o he ole o di e en ial changes in such condi ions e.g. policy-induced
unce ain y and ising in e es a es.
Modelling solu ion
The FTT: Powe model was enhanced o inco po a e weigh ed a e age cos o capi al
(WACC), which akes in o accoun he cos o deb and cos o equi y ha a y by:
• Technology ype ( o example, sola PV, onsho e/o sho e wind, gas, coal),
• Region.
The implemen a ion o he WACC a e is done a a g anula le el o allow o he modelling
o how inancing cos s could a y unde di e en mac oeconomic and policy condi ions
( o example, in e es a e hikes, g een inancing op ions). This en iches he FTT-Powe ’s
16
Applica ion
Use when: e alua ing de elopmen —o ien ed disas e isk educ ion and adap a ion
policies ac oss scales.
A oid when: doing echnically—minded disas e isk educ ion and adap a ion policies
ac oss scales.
Re e ence
Mechle , R. , Żeb owski, P. , Cle cq-Roques, R., Pa il, P. & Hoch aine -S igle , S. (2025). Posi i e
Ex e nali ies in he Polyc isis: E ec i ely Add essing Disas e and Clima e Risks o
Gene a ing Mul iple Resilience Di idends. In e na ional Jou nal o Disas e Risk Science
10.1007/s13753-025-00661-2.
3. Conclusion
The DECIPHER p ojec shows ha ad ances in clima e–economy modelling can p o ide
mo e policy- ele an insigh s when hey explici ly add ess unce ain y, beha iou al
dynamics, and eal-wo ld cons ain s. Ac oss he me hods p esen ed, anging om
machine-lea ning emula o s o high- esolu ion damage assessmen s and inancial isk
models, he common heme is a s onge ounda ion o assessing policy impac agains a
wide ange o po en ial u u es.
These inno a ions do no eplace adi ional modelling; ins ead, hey imp o e i by
iden i ying whe e policy ou comes a e mos sensi i e o ex e nal shocks, whe e a ge ed
in e en ions can educe cos s, and how adap a ion and mi iga ion measu es can p oduce
co-bene i s beyond emissions educ ion. Inco po a ing hese app oaches in o na ional and
17
egional planning can assis go e nmen s in de ising s a egies ha s ay c edible and
e ec i e e en as economic and clima ic condi ions e ol e.
Key Policy Takeaways
• In eg a e unce ain y analysis in o policy o mula ion. Employ emula ion and
o he me hods o assess whe he policies mee hei objec i es ac oss a ange o
p obable scena ios, no jus a cen al o ecas . This is pa icula ly ele an o he
2025–2026 NECP e isions.
• Conside expec a ions and inancing condi ions. Fo wa d-looking beha iou and
he cos o capi al g ea ly in luence in es men pa e ns and he speed o
echnology di usion. Policies should ake hese dynamics in o accoun o p e en
unde es ima ing ansi ion cos s. Policies should accoun o hese dynamics o
a oid unde es ima ing ansi ion cos s ha is a key conside a ion o he G een Deal
Indus ial Plan and moni o ing in es men lows unde Nex Gene a ionEU.
• In eg a e looding and coas al damage isk in o mac oeconomic planning. High-
esolu ion lood and coas al damage assessmen s, as well as inancial isk models
ha link clima e impac s o asse alues, p o ide i al e idence o adap a ion and
inancial s abili y s a egies. These insigh s can in o m he EU S a egy on
Adap a ion o Clima e Change and he design o esilience componen s in Cohesion
Policy unds.
• Recognise po en ial b oade bene i s o clima e ac ion. Policies ha lowe isk can
also p omo e economic g ow h, social ai ness, and de elopmen ad an ages.
Valuing hese “mul iple esilience di idends” ein o ces he economic a gumen o
ea ly and sus ained in es men , in line wi h he Eu opean G een Deal’s Jus
T ansi ion Mechanism, which combines emissions educ ion wi h social ai ness.
• Encou age collabo a ion be ween modelle s and policymake s. Ongoing dialogue
helps gua an ee ha modelling inno a ions mee policy needs and ha esul s a e
in e p e ed and used app op ia ely.
By applying hese lessons, policymake s can design clima e and ene gy s a egies ha a e
mo e esilien , economically sound, and capable o p o ec ing ci izens and economies in
an unce ain u u e.
