ISSN 1403-2473 (Print)  
ISSN 1403-2465 (Online) 
 
 
 
 
 
 
 
 
 
Working Paper in Economics No. 687 
 
 
 
Prices versus Standards: Evidence from an 
Artefactual Field Experiment on Managerial 
Investment Behavior  
 
Magnus Hennlock, Åsa Löfgren, Conny Wollbrant 
 
Department of Economics, Rev May 2021 (Nov 2017, Jan 2017) 
 
 
                                                     
 
1  
P ri ces versus Standards : Evidence from an Art e fa ct ual  Fiel d Ex peri m ent on Mana geri al 
Inv es t m ent Beh avi or   
 
Magnus Hennl oc k a , Åsa Löf gr e n b , Conny Woll bra nt c  
a Pol ic y and Economi c s , IVL Swedi s h Envi r onme nt al Rese ar c h Inst it ute , P.O. Box 530 21 S - 400 14 
Got he nbur g, Swede n, tel . +46 31 708 65 08, e - ma i l : Magnus .He nnl oc k@i vl .s e . (cor r es pondi ng aut hor)  
b Depa r t me nt of Economi c s , Uni ve r s it y of Gothe nbur g, P.O. Box 640, S - 405 30 Got he nbur g, Swede n, 
tel: +46 31 786 1375, e - ma i l : Asa.Lof gr e n@ec onomi c s.gu.s e  
c * W ol l bra nt , Economi c s Divi s i on, Stir li ng Mana ge me nt School , Uni ve r s it y of Sti rli ng, Stir li ng, FK 9 
4LA, Scotl a nd UK , tel : +44 1786 467276, e - ma i l : Conny.W ol l br a nt @s t i r .a c .uk   
☆ Resea rc h funding from the Swedis h Founda ti on for Str at e gi c Envir onme nt a l Rese a r ch Mist r a thr ough 
the Mistr a Car bon Exi t resea r c h progr a mme , the Swedis h Rese ar c h Counci l For ma s thr ough the proj ect 
Pol ic y Desi gn for Corr e cti ng Mar ke t Fail ure s Pr eve nt ing a Cir c ula r Economy -  Anal ys i s and Evi de nce 
from Two Case Studie s , and from the Swedis h Envir onme nt a l Prot e ct ion Agenc y’ s envi r onme nt a l 
resea rc h fund thr ough the proj e c t Pol i ci es for Life - C yc l e s -  an Int e gr a t e d Asses s me nt , is grate f ul l y 
acknowl e dge d.  
Acknowledgments 
T he aut hors are grat e ful to Prof es s or O. J oha nss on- St e nma n, Depa r t me nt of Economi c s , Uni ve r si t y of 
Got he nbur g, Profe s sor M. Duf we nber g, Depa rt me nt of Economi c s , Uni ve r si t y of Ari zona , and Seni or 
Fel low D. Bur t r a w and K. McCor ma c k at the Resour c e s for the Futur e , Washingt on D.C. for their 
val ua bl e comme nt s and advi c e on a previ ous pil ot desi gn of the exper i me nt . The rese ar c h assis ta nce 
from Lovi s a Käl l ma r k, Fanny Isa ks s on Lantt o, Loui s e Hwar gå r d, Hanna h Doher ty, and J onas Lind at 
the Swed i s h Envi r onme nt al Rese ar c h Inst it ute is grate ful l y acknowl e dge d.  
 
Abs t ract  
A n art ef a ct ual fiel d exper ime nt is conduct e d usi ng  166 expe ri e nc e d mana ge r s and seni or advi s ors 
recr uit e d from the chemi c al , pul p and pape r , el ect ri ci ty, heat i ng , and steel indus tr i es  in Swede n . T he 
expe ri me nt pres e nt e d a str ongl y ince nt i vi ze d cost minimi za t i on task frame d as an abate me nt inve s t me nt 
decis i on in a hypot het i c al fir m . Subj e c ts were randomi ze d int o two treat me nt s , and they made  deci si on s  
in the prese nc e of an  emi s s ions tax or  an  emi s s i ons st anda r d . T reat me nt s wer e  cali br a te d to gene ra t e no 
trea t me nt eff ec t s if subj ec ts ’  deci s ions are cons is t ent wi t h the predic ti ons of rati ona l  choi c e theor y . The  
resul ts show that emi s si ons standa r ds reduc e mana ge r s ’  att e nt i ona l focus on cost mi ni mi z a t i on , causi ng 
them to choos e the most cost - e f f e ct i ve alte r na ti ve  less oft en  than is predi ct e d by  rat ional choi ce theor y . 
Alt hough the emi s si ons tax signi fi c a ntl y incre a se d mana ge r s ’  focus on mone t ar y inf or ma t i on, mana ge r s 
tended  to mini mi ze aver a ge abat e me nt costs , resul ti ng in abat e me nt leve l s lowe r than  opt i ma l leve l s and 
mar gi na l costs lower than  tax leve l s.  The result s d emo ns t ra te that the type of polic y inst r ume nts per se  
                                                          
 
2  
c a n induc e deci s ion - ma ki n g bias via att e nt i ona l focus , even when deci si on make r s are exper i enc ed  
mana ge r s and seni or advi s or s .  
Ke ywords : art e fact u al fiel d ex peri m ent , bounded rati onal i t y, attent i onal bias, poll ut i on cont rol , 
tax es , regul at i on  
J EL - codes : C93, H32, L20. 
 
1. Introduction 
A comm on and import ant regul at or y probl em is how to opti m al l y impl em en t  poll ut i on cont rol . 
Neocl as s i c al econom i c t heor y  posi t s that  p rofi t - m ax i m iz i ng firm s  shoul d equat e the econom i c 
values of mar gi nal produ ct s and mar gi nal cost s . This  impl i es choos i ng  the emiss i ons level at 
the point at which the ma rgi nal abat em ent cost of reduci n g emiss i ons equa l s  the emiss i ons tax  
level  in the pr esenc e of envi ronm ent al econom i c poli c y inst rum ent s , su ch as a Pigovi an 
emiss i ons tax , simul t aneous  with select i n g th e pr oduct i on level at which the pric e  of the goo d 
equal s  the ma r gi nal cos t of product i on. This re sul t , which also gua rant ees cost - eff ect i ve  
emiss i ons reduct i ons , is a corners t one in the econo m i cs literat ure that ar gues for using econom i c 
poli c y inst rum ent s (such as tax es , subsi di es , or tradabl e perm i t s ) over co mm and and cont rol 
inst rum ent s (such as emis s i ons standards ) to cor re c t for ex ternal i t i es . H owev er,  that the opti m al 
choi ce o f poli c y inst rum e nt  can b e affe ct ed b y oth er ci rcum s t anc es , incl udi ng cost unc ert ai nt y , 
is well establ i s hed  ( W ei t zm an  1974) . Hen ce, u nde r s t andi ng both the theor y  of poll ut i on cont rol  
and the potent i al limi t at ions  and ineffi ci enci es that might arise when movi ng from theoret i cal 
predi ct i ons  to the pract i c al impl em ent at i on of  spe ci fi c envi ronm ent al poli ci es  is import ant from 
a re gul at or ’ s persp ect i ve.  
E mpi ri cal eviden ce su gges t s  that de ci s i ons in resp ons e to poli c y devi at e in some circum s t an ces 
from predi ct i ons of neo c l as s i cal  econom i c theor y (for an ove rvi ew , se e , for ex am pl e, Sorrel l , 
Mall et t , and Nye ( 2011) ) . A  prom i nent ex am pl e  of such a devi at i on is the so - cal l ed  “ e ner g y 
e ffi ci en c y gap ,”  which refer s  to the observ at i on that re al i z ed ener g y effi ci enc y inv es t m ent s 
diverge from the cost - m i ni m iz i ng level . Allco t t and Gre ens t one (201 2)  compreh ens i ve l y  
revi ew ed  the  ex ist ence of the ener g y effi ci enc y ga p  and point ed to  subst ant i al hetero gen ei t y  in 
inves t m ent ineffi ci en ci e s a cros s dif fer ent t yp e s of agent s . Amon g th e caus es  of thes e 
inves t m ent ineffi ci enci es  are i mperf ect inform at i on and behavi oral const rai n t s ( e.g. , inat t ent i on 
to inform at i on ) , al though most of the discus s i on consi ders deci s i on  maki ng from a consum er 
3  
pers pect i ve. In thi s pape r , whet her t hese t ypes o f inves t m ent ineffi ci enci es are relevant for 
mana geri al de ci s i on  maki ng and whet he r the re gul at or ’ s choi ce of poli c y i nst rum ent (pri ce 
versus quant i t y measur es ) affe ct s the ex tent to which abat em en t inves t m ent deci s i ons are cost  
effe ct i ve ar e inves t i gat ed . 
Rega rdi n g firm s ’  deci s i on  maki ng , neocl as s i c al  econom i c theo r y  pr edi ct s that  compet i t i on 
force s  firm s with  long - ru n  inves t m ent ineffi ci enci es or low product i vi t y to ex it the market . Yet, 
well - docum ent ed and p ersi s t ent empi ri cal diffe rences in product i vi t y ex ist even within 
narrowl y defi ned busi nes s es and indus t ri es (for an overvi ew , see S yv ers on 2011). Ex pl ai ni ng 
empi ri cal produ ct i vi t y differen ces has attract e d strong  inter es t from  econom i s t s , and 
mana gem ent has rec ei ve d part i cul ar attent i on. Fo r inst ance, Bloom et al. (2013) show ed  that 
inform at i onal const rai nt s and rest ri ct i ons on compet i t i ve press ure can allo w “ badl y run ”  firm s 
to sta y in busi nes s .  
Armst rong and Hu ck (2 010) and C yert and Ma r ch (1963) ex pl ai ned devi at i ons from profi t 
max i m iz at i on in firm s by introduci n g th e con ce pt  of  satisficing , first de vel oped b y  Simon 
(1955). Sati s fi ci ng is an inst ance of boundedl y rati onal deci s i on maki ng and impl i es that , 
gen eral l y, a firm  mana ge r  does not compa re all univers al l y possi bl e i nves t m ent choi ces ; 
inst ead, he or sh e  choo s es the first opti on dee m ed  sati s fact or y . 1  Alternat i vel y, man a ge ri al 
deci s i on  maki ng  might be based on rules of thum b, such as imit ati ng strat e gi es of well -
perform i n g rival s or chan gi n g str at e gi es onl y whe n profi t s de cl i ne to lowe r than an ac cept abl e 
thres hol d rather than pe rform i ng ex pl i ci t calcul at i ons of opti m al strat egi es (Arm s t ron g and 
Huck 2010).  
The ps yc hol og y literat u r e has sugges t ed  that dec i s i on makers might fail to enga ge in cost -
benefi t an al ys es and , inst ead , rel y on deci s i on - b y - r ul es mechani s m s le arned  through ex peri enc e 
or soci al ex chan ge (Si m ons on 1989; Prel ec and Herrns t ei n 1991; Shafi r, Simons on, and 
Tvers k y 1993). Amir an d Ariel y (2007) show ed  in an ex peri m ent that the need for mon et ar y 
asses s m ent s might invo ke previ ous l y le arn ed rules , resul t i ng in incon s i s t enc i es  betwe en 
monet ar y - bas ed judgm en t s and judgm ent s based on other fact ors , such as effort and pleas ure. 
Simi l arl y , Sunst ei n (200 4) ar gu e d th at the d eci s i on - m aki ng cont ex t  coul d acti vat e swit ch i n g  
from pref eren ce max i m i zat i on to , inst ead , appl yi n g a cert ai n choi ce rule.  
                                                          
1  Decisio n - ma ke r relia nc e o n si mp l ifie d decis io n r ule s fo r co gniti ve l y de ma nd i ng tas ks has rece i ve d muc h 
atte ntio n i n t he lite r a t ur e on b o und e d ratio na lit y (Si mo n 19 5 5 , 197 9 ) . T he ter m re fe r s to the li mite d cap a c it y fo r 
ratio na lit y t hat can arise wh e n so lvi n g co mp le x pro b l e ms, pro c e ssing lar ge a mo u nt s of i nfo r ma tio n, or  mak in g 
decisio ns i n the prese nc e of unce r ta in t y a nd inco mp le te i n fo r ma tio n.  
4  
Thi s paper cont ri but es to th ese are as of resea rch b y condu ct i ng an art e fa ct ual fiel d ex peri m ent 
on inves t m ent choi ces among indus t ri al mana ge r s and seni or advi s ors in respons e to poli c y 
intervent i ons  (see Ha rri s on and List ( 2004)  for a  tax onom y o f ex peri m ent al approa ches ) . T he 
overal l aim is to test wh e t her  mana gers ’  and seni o r advi s ors ’  inves t m ent ch oi ces ar e consi s t ent 
with predi ct i ons of neocl as s i cal econom i c theo r y  and whet her dif fer e nt t yp es of poli c y 
inst rum ent s induce s yst em at i c diff eren ces in inves t m ent choi ces .  This resea rch  buil ds on 
empi ri cal eviden ce that docum ent ed beh avi oral  anom al i es and the liter at ure that emph as i z ed 
the need to consi der  findi ngs from behavi or al  econ om i cs and bounded rati onal i t y in the desi gn 
and an al ys i s o f envi ron m ent al poli c y (se e , for ex am pl e , Shogren ( 2002) ; Shogr en and Ta yl or 
( 2008) ; Gowd y ( 2008) ; Gsot t bauer and van den Ber gh ( 2 010) ;  Poll i t t and Shaors hadz e ( 2011) ; 
Carl s s on and J ohans s on- S t enm an ( 2012) ;  and Gs ot t bauer ( 2013) ) .  
To this end, 166 firm m ana gers and seni or advi s ors from Sw edi s h indus t r y we re rec rui t ed to 
part i ci pat e in an incent i vi z ed ex peri m ent . Subj ect s  were random l y assi gned t o one of two poli c y 
inst rum ent treat m ent s —e i t her a price ( f ram ed as an emiss i ons tax ) or a qua nt i t y ( f ram ed as  an  
emiss i ons standard)  inst rum ent . In the ex peri m ent , m ana gers and s eni or ad vi s ors  were asked to 
make an ab at em ent inv es t m ent choi ce and we r e inst ruct ed to mini m i ze total cost s  (wit h 
ident i cal cost struct ure s  and incent i ves at the margi n in both treat m ent s ). The  ex peri m ent al 
desi gn allow s  for t est i ng of  whet he r  the poli c y ins t rum ent  t ype  can i) incr e as e the consi s t enc y 
of mana ge rs ’  and seni or advi s ors ’  inves t m ent ch oi ces throu gh the opti m al i t y pr edi ct i on of  
neocl as s i cal  econom i c theor y, ii) s yst em at i cal l y shift inves t m ent choi ces toward other choi c e 
rules 2  than the opti m al i t y  rule, and iii) induce atten t i onal biases in inform at i on use for de ci s i on  
maki ng.  
The literat ure on theore t i cal and ex peri m ent al behavi oral  econom i cs h as reve al ed sev eral 
judgm ent and attent i onal biases that  depend on the inform at i on attri but e  tha t is more sali ent to 
deci s i on make rs . T h e diss oci at i on betwe en monet ar y asses s m ent s and p redi ct ed util i t y is a well -
known judgm ent anom a l y (se e , for ex am pl e, Thal er ( 1985) ; Hsee et al . ( 2003) ;  and Amir , 
Ariel y, and Carm on ( 200 8) ). Addi t i onal l y, eviden ce sugges t s  that monet ary asses s m ent s make 
deci s i on makers fo cus more on inform at i on va ri abl es relat ed to trans a ct i ons (for inst ance ,  
prices , cost s , and ma rket norm s )  and  less on vari ab l es relat ed to the ple as ure or util i t y o f owni n g 
or consum i ng.  
                                                          
2 A  c ho ic e rule is refe r r e d to as a map p i ng fro m a  set o f avail a b le info r ma tio n to a  set of alt e r na tive s .  
5  
Am i r , Ariel y, an d Car m on (2008) su gges t ed  that this  phenom enon ca n be ex pl ai ned b y 
focal i s m , meani n g that different t ypes of info r m at i on and feat ures o f the eval uat ed stim ul i 
inform diffe rent asses s m ent tasks . Th e y found ex peri m ent al support for this conj ect ure b y 
test i ng the impact of diff erent t ypes of inform at i o n on subj ect s ’  value asses s m ent s . T he y find a 
dispari t y betw een subj e ct s ’  will i ngnes s to pa y and predi ct ed util i t y  b y drawi ng subj e ct s ’ 
attent i on to either a monet ar y va ri abl e (product i o n cost ) or oth er attri but e vari abl es . Simi l arl y, 
Schkade  and Kahnem an (1998) show ed  that d e ci s i on - m akers ’  eval uat i o ns of chan ges are 
affe ct ed b y the  inform at i on emphas i z ed. Final l y, prefe renc e mi ght be reve rs ed  when choi c es 
becom e more inform ed b y the most prom i nent attr i but e of the eval u at ed opt i ons , known as the 
prom i nence effe ct  (Tve rs k y , Satt at h, and Slovi c  1 988). 3  
In the cont ex t of our  stu d y, th e resul t s show that mana ge rs ’  and seni or advi s ors ’  inves t m ent 
deci s i ons differ from pr edi ct i ons of neocl as s i c al theor y o f profi t max i m iz at i on. The resul t s 
sugges t that econom i c poli c y  inst rum ent s and emiss i ons standards nat ural l y invok e two 
different asses s m ent approaches —m on et ar y - b a s ed and pe rform ance - b as ed jud gm ent s —
l eadi ng to diff erent attent i onal and judgm ent biase s . Managers ’  attent i onal focus on abat em ent 
level s when fa ci n g emiss i ons standards s eem s  to cause them to und erwei ght  cost inform at i on  
and to fail to s elect the alt ernat i ve corr es pondi ng to less abat em ent , even when doing so is more 
cost effective. In cont ras t , when faci n g emiss i ons tax es , managers ’  focus on cost inform at i on 
increas es toward  consi s t e nc y with the opti m al rule of equal i z i ng mar gi nal abat em ent cost s  and 
the tax level  but with an attent i onal bias towar d  mini m iz i ng aver a ge abat em ent cost s . This 
ex peri m ent hence highl i ght s import ant aspect s of poll ut i on cont rol and poli c y choi c e, namel y ,  
that different t yp es of poli c y inst rum ent s , ceteris paribus , induce differe nces in inves t m ent 
behavi or.  
The rem ai nder o f this p aper is or gani z ed as fol l ows . In se ct i on 2, the surve y desi gn and 
ex peri m ent al treat m ent s  are des cri bed . In se ct i on 3, the resul t s  are pres ent ed , and the findi ngs 
and thei r welf are and pol i c y impl i cat i ons  ar e disc us s ed in sect i on 4. Sect i on 5 concl udes .  
 
                                                          
3  Reve r sa l o f a prefe r e nc e occur s wh en a sub j e c t prefe r s one alte r na tive i n one resp o n se mo d e (e. g. , cho ic e ) but  
exhib it s  t he op p o site prefe r e nc e ord e r in ano the r resp o nse mo d e (e. g. ,  a ratin g over an at t r ib ute ) .  
6  
2. Survey Design and Experimental Treatments  
2.1 Popul at i on and subj ect s  
E x peri enced mana gers and seni or  advi s ors  of lar ge and medi um - s i z e d fir m s in the Swedi s h 
indus t r y  we re recrui t ed for the ex peri m ent . Thei r  norm al work duti es invol ve deci s i on  maki ng,  
anal y s i s , pr epar at i on of  techni cal and econom i c back ground inform at i on for inves t m ent 
deci s i ons , and  d eli veri n g deci s i on recom m endat i o ns to  the firm ’ s CEO or board of direct o rs . 
The subj ect pool w as  ident i fi ed from Bisnode ’ s  databas e PAR AD and the Swedi s h regul at or y 
re gi s t er of plant s class i fi ed as en ga gi ng in envi ro nm ent al l y haz ardous acti vi t i es (EHA)  b y the 
Swedi s h Envi ronm ent al Code. 4  O btai ned  from Bisnode’s databas e PAR AD wer e person al 
cont act inform at i on (phone number and e - mai l addres s ) of mana ge rs and board members of 
compani es in the  ch em i cal , pulp and pap er, and steel sect ors , as well as of firm s oper at i ng 
combus t i on plant s to produce ele ct ri ci t y and heat in Sweden. These sea rch resul t s  incl uded a 
total of 2,010 firm s , each with more than 200 emp l o ye es . T he foll owi n g pr ofi l es we re us ed to 
search for subj ect s to be recrui t ed from these firm s to  the ex peri m ent : CEO , financi al man a ger ,  
envi ronm ent al man a ger ,  operat i ve mana ger , re gi onal mana ge r , boa rd m ember , and seni or 
a dvi s ors , leavi n g us with a sampl e of 629 indi vi dual s belongi n g to a t least one of these 
cate gori es .  
Both the Bisnode and EHA cate gor i es  of t he Swedi s h regul at or y re gi s t er of plant s repres ent 
large and medi um - s i z e fi rm s , incl udi ng the lar ges t firm s in Swed en . Moreo ver, ea ch re gul at ed 
plant in the EH A cate gor y is  gene ral l y  re gul at ed b y vari ous \ pol i c y inst rum ent s , incl udi ng the 
comm and and cont rol ac t, Swedi s h Envi ronm ent al Code , and econom i c poli c y  inst rum ent s , 
such as th e Swedi s h NO x char ge, the Sw edi s h carb on tax , and the EU  emiss i ons tradi n g s yst em 
( EU ETS ) . Hence, this stud y’s subj ect s gen eral l y have  bro ad ex peri en ce with vari ous t yp es of 
econom i c poli c y inst rum ent s and standards .  
A pilot stud y was cond uct ed  usin g 38 pa rt i ci pa nt s from the consul t i n g indus t r y who have 
previ ous ex peri ence as m ana ge rs o r seni or a dvi s or s in the chem i cal , pulp and paper, and steel 
sect ors in Sweden . Afte r anal yz i n g thei r respons es , 10 subj ect s part i ci p at ed in two focus groups  
in which the y  we re ask e d to  ex pl ai n  how the y u nders t ood the task and  how the y used the 
inform at i on avai l abl e w h en maki ng thei r choi ces .  S ubj ect s ’  report ed unde r s t andi ng of the task 
                                                          
4 T he S wed i sh re gu la to r y reg is te r is Sven s ka milj ö r a p po r te r ing sp o r ta le n (SM P ) . T he decla r a tio n data used are 
revie we d by t he S wed is h Env i r o n me nta l E mi ssio ns Data (Sv e ns ka Milj ö Emis sio n s Data , SME D) on beha lf of the 
S wed ish E n vir o n me nta l P ro tec tio n Age nc y.  
7  
and thei r respons es requ i red onl y minor adjus t m ent s to the scenari o tex t  used in the final 
ex peri m ent . 
Duri ng the re crui t i n g proces s to the final ex peri ment , the 629 manage rs and seni or a dvi s ors in 
the sampl e w ere  sea rched up to five times b y  phon e . P hone cont a ct was fina l l y establ i s hed with 
374 (59.5%) of the 62 9 subj ect s . D uri n g the phone call , the subj ect s recei v ed a  short 
introduct i on to the survey . 5  S ubj ect s were then as ked whet her the y were will i ng to compl et e  
the surve y . Out of the 37 4 subj ect s  cont act ed , 268 (71.7%) were will i n g to compl et e  the surve y  
and we re sent t he  surv e y link b y  e - mai l dire ct l y aft er the phon e call . Afte r a max i m um of three 
remi nders b y phon e and then b y e - m ai l , 166 subject s (110 mana ge rs and 56 seni or advi s ors ) 
compl et ed the ex peri m en t , resul t i ng in a 61.9%  re spons e rate.  
Part i ci pant s rec ei ved the onli ne surve y with the stat ed aim  of “ bet t e r under s t andi ng the effe ct s 
of envi ronm ent al re gul at i ons on firm s ’  inves t m ents in cleaner technol ogi es . ”  Part i ci pant s were 
never inform ed in phone call s , e - m ai l s , or throu gh the onli ne surv e y that the y wer e p art i ci pat i n g  
in an ex peri m ent or that different tr eat m ent s  ex ist ed . 
2.2 Information and choice sets 
The purpos e of this study is to inves t i gat e whet h er the t ype of poli c y inst rum ent (the poli c y 
cont ex t ) can:  
i)  caus e m an a ge rs ’  and seni or advi s ors ’  inves t m ent choi ces to be com e mor e consi s t ent 
with predi ct i ons of  opti m al i t y of neo cl as s i cal econ om i c theor y ;  
ii)  s hi ft thei r inves t m ent ch oi ces  toward  oth er choi c e rules than the opti m al respons e 
rule ; and ,  
iii)  caus e dif fer ent attent i on al biases in the reve al ed use of info rm at i on for deci s i on 
maki ng .  
A novel ex peri m ent al d es i gn was dev el oped to answer thes e  quest i ons . In an incent i vi z ed  onli ne  
ex peri m ent , subj ect s wer e present ed with a sc ena r i o in which a hypot het i ca l firm was about to 
inves t in a new abat em ent technol og y to redu ce its carbon dioxi de emiss ions  (see Appendi x 
A.1 for the full surv e y) . Duri ng inst a l l m ent of t he technol og y ,  the firm ha s  to choos e the 
abat em ent l evel . Subj ect s wer e inst ruct ed to min i m i z e compl i ance cost s  t o the h ypot het i c al 
                                                          
5  Call scr ip ts are avai la b le fro m the a ut ho r s on req ue st.   
8  
fi rm 6  and  were then pr ese nt ed with a choi c e betw e en  two diff erent abat em e nt level s —A and B 
( hence fort h ref err ed to as a choice set ) —wi t h dif fere nt  mar gi nal and av er age cost  level s .  The 
subj ect s were furt her in fo rm ed that alternat i ve A had a smal l er ab at em ent le vel than alternat i ve 
B in all choi ce s ets  and  t hat abat em ent  cost s we re “ continuously increasing” . This inform at i on 
was foll owed b y th e cl ari fi cat i on that each unit of reduct i on in carbon diox i de emiss i ons  makes 
it more ex pens i ve to re duce  emiss i ons b y  anot her  ton of carbon diox ide. As a resul t , the 
margi n al cost of alternat i ve B was alw a ys high er t han that of alternat i ve A. 7  Each subj ect w as  
present ed with nine choi ce sets and , h enc e, made in total nine abat em ent deci s i ons . S ubj ect s 
were inform ed that succe edi ng in mini m i z i ng compl i ance cost s for all nine inves t m ent deci s i ons 
woul d earn them approx i m at el y USD 350 (aft er t ax es ), paid as a direct ca sh trans fer to thei r 
bank ac count . 
Tabl e 1 present s an ex am pl e of a choi c e set fo r the tax tre at m ent . Fo r a choi ce s et, cost 
inform at i on for each alte rnat i ve—A and B —is  pr esent ed as ma r gi nal an d avera ge cost s  of one 
addi t i onal  ton of carbon emiss i ons reduct i on. 
Tabl e 1. Ex am pl e of choi ce set in the tax treat m en t (choi ce set 1 in App end i x A.1)  
 In form at i on in the tax treat m ent  A  B  
Tax per ton of carbon dio x i de (SEK/ t on)  1 000  1 000  
Mar gi nal cost per ton of an addi t i onal unit of redu ct i on (SEK/ t on)  
 
200  500  
Avera ge cost per ton o f an addi t i onal unit of reduc t i on (SEK/ t on)  1 000  610  
 
Bec aus e ma r gi nal cost s are cont i nuous l y in cre as i ng , av era ge cost  inform a t i on and numeri cal 
abat em ent level  values  co rres pondi n g to each alter nat i ve are redund ant for ident i f yi n g  the cost -
m i ni m iz i ng alternat i ve . The cost - m i ni m i z i ng alternat i ve is alwa ys that for which the mar gi nal 
cost equal s  the tax level  ( or  is closes t  to the tax  level for a corn er solut i on). For a corn er solut i on, 
                                                          
6 A b a te me nt co st s are incl ud e d as a par t of co mp lia nc e c o st s . In the c ho ic e exp e r i me nt tas k, sub j e c ts wer e 
instr uc te d to ma xi miz e the co st savi n gs, and that the se co sts incl ud e d bo th ab ate me nt co s ts and an y tax co st s (in 
the tax trea t me nt) o r sanc tio n fees (in the e mis sio n sta nd a r d trea tme nt) wa s exp lic itl y stat e d . See Ap p e nd ix A.1 
fo r deta ils ab o ut the sur ve y de sig n.  
7 T hus,  A and B are lo ca te d along t he sa me ab ate me nt co st fu nc tio n  wh ic h li mit s the op ti ma l prob le m to a prob le m 
co nsis tin g on l y o f co mp a r i ng mar gi na l ab ate me nt co st le ve l s bet we e n t he t wo co r ne r so lut io n alte r na t ive s — A a nd 
B —and the P igo uv ia n tax lev e l (o r the P igo u via n sa nc tio n fee in t he stand a r d s trea t me n t) , wher e a s t he fi xe d co st 
is red u nd a nt fo r maki n g t he r atio na l c ho ic e . Mo r e o ve r , to si mp li f y the cho ic e set desi g n ther e is no rela tio ns hip 
bet we e n the pro d uc tio n a nd e mis sio n s le ve ls, i mp l yin g t hat the cha n ge in ab ate me nt do es no t affe c t t he op ti ma l 
pro d uc tio n leve l . T he op tima l pro d uc tio n leve l is t her e fo r e c o nsta nt dur in g t he decisio n o n ab ate me nt le ve ls.  
9  
as in T abl e 1 , a subj ect  first  ident i fi es the mar gi n al cost as th e relevant in form at i on  and th en  
ident i fi es the altern at i ves for which thei r mar gi n a l cost equal s (o r i s close s t to) the mar gi nal 
cost of 1,000 SEK/ t on (alt ernat i ve B  in Tabl e 1 ).  (See A ppendi x A.1  for the full surve y an d 
detai l s on the choi ce s et desi gn .)  
The refo re,  th e ex peri m en t provi des s uffi ci ent  infor m at i on for subj e ct s to us e form al re asoni ng 
to ident i f y the cost - m i ni m i z i ng alternat i ve. Nonet hel es s , ident i f yi n g the cost - m i ni m i zi ng 
abat em ent opti on is stil l not a trivi al task , which i s a cruci al fe at ure o f t he desi gn sinc e  the 
ex peri m ent aim s  to mimic the more compl ex deci s i ons that mana gers and s eni or advi s ors must 
make. Be caus e m ar gi n al cost s are rar el y av ai l abl e for inves t m ent deci s i ons , indus t r y man a gers 
might need to rel y on ch oi ce rules d ependi n g on other inform at i on than t hat requi r ed b y  the 
opti m al i t y condi t i on of neocl as s i cal theor y. Fo r inst ance, a  subst i t ut e for mar gi nal cost s c oul d 
be aver a ge cost s , which also account for p art of the fix ed cost s  in the deci s i on. Hence, a possi bl e 
choi ce rule c oul d be “ s et avera ge cost equ al to the tax rate,”  which woul d resul t in reveal i n g  
choi ces with av era ge cost s  close to the tax rate. A not her possi bl e ch oi ce rule coul d be 
“ mini m i z e avera ge cost ,”  which  woul d  resul t  in re veal ed choi ces with av er age cost s  tend i n g  to 
be the lowes t amon g alte rnat i ves .  
As descri bed above, e ac h subj ect was pres ent ed with nine binar y choi ce sets  to ident i f y the 
choi ce rules to which the subj ect s adhered . This de si gn allow s  us to make stat i s t i cal inferenc es 
about the choi ce rules re veal ed  withi n subj ect s and, import ant l y, whet h er a subj ect ’s reveal ed 
choi ce rul e is consi s t ent with the ( incent i vi z ed )  o pti m al rule  of mini m i z i ng compl i ance cost s . 
For inst an ce, as illus t rat ed in Fi gure 1, c hoi ce set 5 cont ai n s  the bin ar y choi ce betw een 
alternat i ves A and B, wh ich impl i es that alternat i ve B is opti m al (B being a corner solut i on) , 
given the lo cat i on on the margi n al cost curv e in relat i on to the tax level . In choi ce set 7 , the 
opti m al choi ce be twe en A and B woul d be A (A also the unique interi or s olut i on) . 8  
Figu re 1. Ex am pl e s  of choi ce set s  and thei r loca t i on s  on the average and margi nal cost s of 
abat em ent  cu rves  ( choi ce  sets 5 and 7 in Appendi x  A.1 ) .  
                                                          
8 T o keep the mar gi na l reaso ni ng at t he si mp le s t leve l wi th mar gina l co st s bet we e n al te r na ti ve s A a nd B being 
co mp a r e d wit h the ta x le ve l, cho ic e sets co mp a r i n g alte r na ti ve s lo ca te d at dif fe r e nt sid e s of the i nte r io r so lut io n 
wer e exc lud e d . Othe r wi se , ad d itio na l in fo r ma t io n ab o ut nu me r ic a l va lue s of t he ab ate me nt leve ls wo u ld have 
been req uir e d to id entif y the o p ti ma l alte r na tive s.  
10  
 
T hi s ex peri m ent al desi gn therefore allows for test i ng the consi s t enc y b etwe en subj ect s ’  choi ces 
and the foll owi n g fiv e ch oi ce rules : (i ) average cost equals tax/sanction fee , (ii) marginal cost 
equals tax/sanction fee, (iii ) minimize average cost , (iv) minimize marginal cost , and (v) 
minimize emissions (maximize abatement). 
Tabl e 2 illus t rat es  the cho i ce sequ ences of the nine choi ce s ets cor res pondi n g to ea ch o f the five 
choi ce rules  (i ) – ( v) . For ex am pl e, the opti m al choi ce rule, MC  =  tax , impl i es choos i ng  the 
abat em ent level for whic h the margi nal ab at em ent cost equal s  (or is closes t to) the tax level . I n 
Tabl e 2, this impl i es  cho os i ng  alternat i ve B in ch oi ce sets 1 –6 and altern at i ve A in choi ce sets 
7 to 9. In the ex peri m e nt , t he order of the nine choi ce sets present ed t o each subj ect w as  
random i z ed to rule out order ef fect s .  
Wilcox on matched - pai rs signed - ranks test s we re u sed to ve ri f y th at the choi ce s equenc es of the 
five choi ce rules in Tabl e 2 are signi fi cant l y dif fer e nt from random sequence s (W il cox on 1945).  
The ri ght - m os t colum n shows the test resul t s for t he choi ce sequen ce of ea ch choi ce rule ve rsus 
the sequenc e of ex pect ed values from a random i z e d sequenc e. T he likel i ho od is 0.195% that a 
subj ect ’ s random i z at i on among alte rnat i ves b y ch ance shoul d yi el d an y o f the choi ce rules in 
Tabl e 2.  
Tabl e 2. Choi ce sequ enc es of choi c e rules  
11  
 
T he inves t m ent alternat i v e with the lar ges t emiss i o n reduct i on is placed inte nt i onal l y as the B -
al t ernat i ve in all choi ce sets  to reduce the compl ex i t y of the ex peri m ent . I n form at i on on the 
poli c y  (and poli c y t reat m ent )  betwe en the choi c e sets  (th e poli c y tr eat m en t s ar e discus s ed in 
the nex t subsect i on)  was also kept const ant .  
S ubj ect s were info rm ed that the y woul d earn ap prox i m at el y USD 350 (aft er tax es ) if the y 
succeed ed in mini m i z i ng  the cost in  all nine abat em ent choi ces , that is , fo ll owi ng the choi c e 
sequence MC  =  tax in Ta bl e 2. Th e inc ent i ve sch e m e was test ed in the  pilot ex peri m ent and in 
focus groups . The pa ym ent was a si gni fi c ant inc ent i ve for the subj ect po ol to focus on the 
ex peri m ent al ta sk . No time limi t s were impos ed for compl et i n g the ex peri m ent and ex - pos t 
quest i onnai re, each subj e ct need ed on ave ra ge 1 5 minut es to compl et e the full surve y  in  both 
the pilot and  the full ex peri m ent . 9  
2.3 Policy treatments 
S ubj ect s were random i z ed to one of the two  poli c y  tre at m ent s . I n the tax treat m ent , the 
afore m ent i oned nine cho i ce sets  wer e pr esent ed as abat em ent de ci s i ons subj ect to a carbon 
emiss i ons tax . I n the stan dards tre at m ent , the choi ce sets wer e pr esent ed as deci s i ons subj ect  to 
                                                          
9  Exclud i ng o utlie r s i n resp o ns e ti me exce e d i n g 24 ho ur s.  
12  
an emiss i ons  standard  descri bed  in term s of an emiss i on limi t value.  T he two treat m ent s start ed 
with ident i cal des cri pt i on s of the inves t m ent de c i s i on and we re th en dist i ngui s hed b y a short 
inform at i on tex t of each poli c y as descri bed in Ta bl e 3 (see App endi x  A.1  for the full surve y) .  
Tabl e 3. Poli c y inst rum e nt treat m ent s (trans l at ed from Swedi s h)  
Investment decision (identical in both treatments) 
W e ask yo u to t hin k ab o ut a si tua tio n i n wh ic h yo u are i n vo lve d in an i nve st me nt decisio n in a fir m t hat 
wil l in ve st in a ne w ab ate me nt tech no lo g y to red uc e their car b o n dio xid e emi ssio ns. In ve st me nts i n 
techno lo g y i mp o se fixe d co sts and mar g ina l co sts t hat are continuously increasing ; that is, the mo r e the 
fir m red uc e s i ts car b o n dio xi d e e mis sio n s, the mo r e e xp e nsi ve it beco me s to red uc e a no the r to n o f 
car b o n dio xid e . With ne w tech no lo g y, t he fir m i s co n sid e r in g t wo dif fe r e nt ab ate me nt le ve ls : alte r na ti ve 
A, wh ic h i mp lie s a smalle r e mis sio n red uc tio n, and alte r n a tive B, whic h i mp lie s a lar g e r emi ssio n 
red uc tio n. Bo th op tio ns ha ve t he sa me eco no mic life .  
 
Regulatory information in each treatment 
Emissions standard treatment Tax treatment 
T he prer e q uisite fo r yo ur decis io n is t hat t he fir m 
pro d uc e s in a co un tr y with an e mis sio n s stand a r d 
that li mit s ho w mu ch car b o n di o xid e e mis sio n s is 
allo we d . T he alte r na tive wit h the lo we r 
ab ate me nt le ve l (alte r na ti ve A) results in an 
e mis sio n s le ve l tha t exce e d s t he e mis s io ns l i mit 
value . I f th is alte r na ti ve is c ho se n, t he fir m need s 
to pay SEK   100 0 per to n in sanc tio n fee fo r ever y 
ad d itio na l to n that t he e mi ssi o ns leve l e xce e d s 
the e mi ssio ns li mit val ue .  a If the alte r na ti ve wit h 
a hig he r ab ate me n t is cho se n (alte r na ti ve B), the 
e mis sio n s le ve l do es no t e xc e e d the sta nd a r d s, 
and no sanc tio n fee need s to be paid .  
T he prere q uisite fo r yo ur deci sio n is t hat t he fir m 
pro d uc e s in a cou ntr y wi th a c ar b o n tax rate eq ua l 
to SEK 100 0 per to n car bo n dio xid e emiss io n eac h 
year . a  Bo th ab ate me nt a lte r na t ive s A and B red uc e 
e mis sio n s and , he nc e , the ta x payme nts.  
 
a  SEK = Swed ish kro na  
Although the re gul at or y i nform at i on vari ed acros s treat m ent s , it was ident i c al for al l choi c e sets 
withi n each treat m ent . The clari fi c at i on for the standards tre at m ent was that onl y alternat i ve B 
woul d fulfi l l the standard s . Choos i ng the lower abat em ent level , A, woul d lead to a Pigouvi an 
sanct i on fee id ent i cal to the tax rate in the tax t rea t m ent . 1 0  The subj ect s we re ex pl i ci t l y asked 
to choos e the abat em ent l evel , A or B ,  the y beli ev e gives the firm the great es t savi ngs , in cl udi ng 
                                                          
10 I n po lic y in prac tic e , fixe d or var ia b le sanc tio n fees are us ed . To ob ta in id entic a l eco no mic ince nti ve s at the 
mar gi n  bet we e n the t wo po lic y trea t me nt s, a “P igo uvia n sa n c tio n fee” paid per unit e mis si o ns was used .  
13  
bot h abat em ent inves t m e nt cost s and an y s anct i on fees  (standard treat m e nt ) or tax pa ym ent s 
(tax treat m ent ) . 
The inform at i on provi de d in the choi ce sets in th e two treat m ent s was ide nt i cal ex cept for the 
poli c y inform at i on in the second row (see Tabl e 3 ). The tax treat m ent repo rt ed the tax per ton , 
where as the stand ards tre at m ent report ed the sanction fee per ton. In ea ch choi ce set, the cost s 
and the t ax / s anct i on fee are ident i cal acros s tre at m ent s , as obse rved in the ex am pl es in Tabl es 
4a and 4b.  
Tabl e 4 a . Ex am pl e of  ch oi ce set  in the tax  treat m ent  
 In form at i on in the tax treat m ent  A  B  
Tax per ton  of carbon dio x i de (SEK/ t on )  1 000  1 000  
Mar gi nal cost per ton  at attai ned level of reduct i o n (SEK/ t on )     200     500  
Avera ge cost per ton  at attai ned level of re duct i on (SEK/ t on )  1 000     610  
 
Tabl e 4b. Ex am pl e of ch oi ce set in the standard t r eat m ent  
 In form at i on in the stand ard tre at m ent  A  B  
Sanct i on fee pe r ton  of carbon diox i de (SEK/ t on ) in ex ces s of the standard  1 000  1 000  
Mar gi nal cost per ton  at attai ned level of reduct i o n (SEK/ t on )  
 
   200     500  
Avera ge cost per ton  at attai ned level of re duct i on (SEK/ t on )  1 000     610  
 
Although the two treat m e nt s had ident i cal econom i c incent i ves at the margi n , possi bl e fram i ng 
effe ct s mi ght resul t  from pa yi n g a sanct i on fee for viol at i ng the re gul at i on ( e mi s s i ons  standard s ) 
rather  than pa yi n g a tax . Fram i n g effe ct s  might also ex ist becaus e of i nform at i onal bias es  
resul t i ng from the dif fe rences in  inform at i on cont ex t s between econo m i c and  re gul at or y 
inst rum ent s . This possi bil i t y is fu rt her discus s ed i n the resul t s sect i on.  
2.4 Descriptive statistics  
Of the 166 subj e ct s who  compl et ed the ex peri m en t , 90 we re assi gn ed to the tax treat m ent and 
76 to the stand ard s  tr eat m ent . The aver a ge subj ec t was 50 yea rs old , 70% were male, and 80 % 
had a univers i t y de gree . The subj ect s ’  educat i onal back ground was  p rim ari l y in te ch or 
14  
engi n eeri n g (60 %). Tw en t y - fi v e per cent had a ba ck ground in s cience or envi ronm ent al scienc e, 
and 10% had a busi nes s admi ni s t rat i on backgr ound, and onl y 2% of t he subj ect s had an 
econom i cs ba ck ground. The subj ect s had signi fi c ant ex peri ence part i ci pat i ng in envi ronm ent al 
inves t m ent deci s i ons at compani es : 68% of the subj ect s stat ed that the y ha d long (5–10 yea rs ) 
or ver y lon g (mor e than 1 0 ye ars ) ex peri ence .  Onl y 2.5% of the subj e ct s had no ex peri ence with 
mak i ng such d eci s i ons . 
 
3. Results 
Thi s sect i on discus s es t he an al ys i s of  the ex per i m ent al data. First , whet her the re gul at or ’ s 
choi ce o f poli c y inst ru m ent (pri c e ve rsus  emis s i ons standards ) aff ect s the  ex tent  to which  
abat em ent inves t m ent de ci s i ons resul t in cost - effect i ve choi ces  is test ed . More speci fi c al l y, 
whet her  or not the t ype o f poli c y inst rum ent cause s  man a ge rs ’  and seni or advi s ors ’  inves t m ent 
choi ces to becom e mo r e  or less consi s t ent wit h the predi ct ed opti m al i t y of  neocl as s i c al  
econom i c theo r y. Secon d, the choi c e rul es that s ubj ect s use in thei r ab at e m ent deci s i ons and 
whet her the y diff er  be ca us e the deci s i ons wer e made subj ect to a tax or  an emiss i ons standard  
are ex am i ned . Fin al l y, the effe ct s of poli c y t ype o n diffe rent attent i onal bia ses and rev eal ed use 
of inform at i on for de ci s i on maki ng  are an al yz ed.  
3.1 Rationality and treatment effects 
Tabl e 5 illus t rat es the fre quenci es and per cent a ges of the observed choi ces acros s the two poli c y 
treat m ent s .  
 
  
15  
Tabl e 5. C hoi ce b y tre at m ent  
 Tax treat m ent  Standards treat m ent  
Abat em ent level  A  
Low er 
abat em ent  
B  
Highe r 
abat em ent  
A  
Low er 
abat em ent  
B  
Highe r 
abat em ent  
Observed choi ces *  382  428  219  465  
Percent a ge observ ed 
choi ces *  
47.2  52.8  32.0  68.0  
Total number of choi c es  
made withi n each 
treat m ent *   
810  684  
*The lar ge r number o f ch oi ces in the tax treat m ent is the resul t of the lar ger number of subj ect s 
in the tax treat m ent (90) relat i ve to the standards treat m ent (76) . Each subj ect made nine 
choi ces .  
 
Table 5 provi des clear  ev idence of treat m ent eff ect s on abat em ent inves t m ent level s  ( recal l that 
abat em ent incr eas es fro m A to B ; that is , for all choi ces , inves t m ent choi ce A cor res ponds to 
the lowes t abat em ent level  and B  to  the highes t ) . A lternat i ve B  is  a more frequent choi ce fo r 
the standards treat m ent than for the tax  treat m ent : for the standard s  tre at m ent ,  68%  of the  
subj ect s chos e  alternat i v e B ( B compl i es  with the emiss i ons  standar ds ) , and 32%  select ed 
alternat i ve A . For the tax treat m ent , 47%  sele ct ed A  and 53%  sele ct ed B . Hence, althou gh t he 
most comm onl y sele ct ed alternat i ve in both tr eat m ent s was  alte rnat i ve B  ( 6 8%  and 53% in the 
standard and tax treat m en t s , respect i vel y) ,  a  la r ger share o f the subj e ct s chos e abat em ent choi c e 
A ( lowe r ab at em ent l evel ) in  the tax treat m ent rela t i ve to the stand ard t reat m ent . ( Tabl e A. 1 in 
Appendi x A.2 provi de s  a quant i t at i ve overvi ew o f frequen ci es pe r  choi c e set and  treat m ent .)  
Nex t ,  whet her signi fi can t treat m ent effect s ex ist on opti m al abat em ent choi ces  is test ed . For 
this purpos e, a dumm y v ari abl e is cre at ed that indi cat es  for each choi c e wh et her it is consi s t ent 
with the opti m al choi ce (dum m y = 1) or not (du m m y = 0). Wilcox on matched- pai rs sign ed -
ranks test s (W il cox on 1945) are p erfo rm ed on pair s of these dumm y v ari abl es from ea ch 
treat m ent . Tabl e 6  pres e nt s t he Wilcox on matched - pai rs si gned - ranks test s of equal i t y of the 
matched observ at i on  pairs . The null h ypot hes i s is that  no difference in  dist ri but i ons ex ists 
between the two tr eat m e nt s . The t otal number of observat i ons is 1,494 ch oi ces made b y 166 
subj ect s in nine choi ce s ets.  A  stat i s t i cal dif fer en ce in th e frequen c y o f ra ti onal choi ces m ade 
16  
bet ween tre at m ent grou ps  is found. The tax treat m ent seem s  to caus e  subj ect s to reveal a 
somewhat  highe r consi s t enc y with the opti m al rule that  equal i z es margi n al cost and the tax 
(sanct i on fee ) level . For the tax treat m ent , 36% of all  choi ces are consi s t e nt with the rati onal 
choi ce, whe reas 32% of t he choi ces  ar e for the sta ndards treat m ent . Th is  ef fect is signi fi cant at 
the 10%  level .  
Tabl e 6. Propo rt i on of op ti m al choi ces (m ean  of  dumm y va ri abl es indi c at i ng consi s t enc y with 
the opti m al rule )   
 Tax treat m ent  Standards treat m ent  Wilcox on matched -
pai rs sign ed - r ank test   
 Mean  Mean  Prob  
Proport i on of opti m al 
choi ces  
0.363  0.322  0.0565  
 
Overal l , these resul t s  su gges t that the p resen ce o f a tax improves subj ect s ’  consi s t enc y with the 
rati onal choi ce theor y  pr edi ct i ons  relat i ve to those faci n g an emiss i ons standard, even thou gh 
econom i c incent i ves and margi n al  cost s ar e ident i c al acros s the two t ype s  o f poli c y tre at m ent s .  
3.2 The effect of policy instrument on choices and attentional biases 
Thi s sect i on discus s es withi n - s ubj ect consi s t enc y with a speci fi c choi ce rule. Tabl e 7 shows 
that 25 subj ect s  (27.8%) i n the tax tre at m ent and 2 4 subj ect s (31.6 %) in the standards tr eat m ent 
are fully consi s t ent with some  speci fi c choi c e rule  for all nine choi c es . Notabl y, t he most 
frequent choi ce sequen c es among  subj e ct s are co nsi s t ent with the choi ce rules prescri bi n g to 
either minimize emissions or minimize average abatement cost . Onl y th e third most frequent 
rule consi s t ent with the choi ces made b y the sub j ect s is to equalize marginal abatement cost 
with the tax (sanction fee) level , that is , the  opti m al cost - m i ni m i z i ng choi ce rule . The prob abi l i t y 
that subj ect s  choos i ng random l y  amon g alternat i ves shoul d yiel d an y of the se  choi ce rules  
consi s t ent l y ove r the nin e choi ces  is a mer e 0.195 %. 1 1  Stat i s t i cal l y, the refo r e the clai m can be 
made that if a subj ect ’s  choi ce sequenc e is found to be perfect l y consi s t ent with a given rule , 
his choi ces  were bas ed on some s yst em at i c beli ef about how the inform at i on shoul d be used to 
mini m iz e cost s rather tha n  b y ch anc e.  
                                                          
11  1/( 2 9 ) = 0.00 1 95 3  
17  
Tabl e 7. Fr equen c y and p ercent a ge of subj e ct s per treat m ent consi s t ent with a choi ce rul e ac ros s 
all  nine choi ces  
 Tax treat m ent  S tandards treat m ent  Mann -
W hi t ne y 
U - tes t  
Choi ce rule  Frequ enc y  Percent a ge  Frequ enc y  Percent a ge  Prob  
AC = tax level  
AC = sanct i on fe e level  
0  0.0%  0  0.0%  n/a  
MC = tax  level  
MC = sanct i on fee l evel  
5  5.6%  1  1.3%  0. 1460  
Min AC  10  11.1%  9  11.8%  0 .8832  
Min MC  0  0.0%  1  1.3%  0. 2765  
Min emiss i ons  10  11.1%  13  17.1%  0. 2668  
 
It is note wort h y that a cl ear dif fer enc e se em s to ex ist between tr eat m ent s when ex am i ni ng the 
number of subj e ct s whos e choi c e sequen ce ex act l y foll ow ed the opti m al choi ce sequen ce (see 
Tabl e 7). For the tax treat m ent , 5.6% of the subject s were full y consi s t en t with the opti m al 
rule —M C  =  tax level —a cros s all of th ei r choi c es , where as 1 .3% o f the subj ect s we re consi s t ent 
with the opti m al rule —M C  = sanct i on fee l evel —i n the standard tre at m e nt .  
Howeve r, t he Mann - W hi t ne y U - tes t  resul t s in Tab l e 7 show no si gni fi cant t reat m ent eff ect s on 
the  number of subj e ct s full y consi s t ent with a sp eci fi c choi ce rul e, suppo rt i ng  the  conj ect ur e 
that the  subj ect s  in tabl e 7 made thei r choi c es on the basi s of s yst em at i c  b eli ef s  about ho w to 
use the inform at i on t hat were stron g enou gh to be unaffe ct ed b y the poli c y treat m ent s .  
Nex t , whet her th e t ype o f  poli c y affe ct s  a  subj ect ’ s total number of choi ce s consi s t ent with a 
given rule  is test ed . Speci fi cal l y, are ther e treat m ent effect s that resul t in higher consi s t enc y  
with a cert ai n rule  within subjects ?  For this purpose, Mann - W hi t ne y U test s were pe rform ed on 
the total number of choi c es  for each subj ect , consi s t ent with a n y o f the choi ce  rule s . Th e resul t s 
in Tabl e 8 indi cat e that the tax treat m ent signi fi c ant l y incr eas es subj e ct s ’  consi s t enc y  (hi gher 
18  
m ean of the number of choi ces consi s t ent with a given choi c e rule ) wit h choi ce rules that 
mini m iz e avera ge and m ar gi nal abat em ent cost s  compared with the standa r d treat m ent . 1 2  
Tabl e 8. Mean and standa rd devi at i on of total number of choi ces consi s t e nt with a given  choi ce  
rule for each subj e ct  
 Tax treat m ent  S tandards treat m ent  Mann -
W hi t ne y U -
tes t  
Choi ce rule  Mean  Std. Dev  Mean  Std. Dev  Prob  
AC = tax level  
AC = sanct i on fe e level  
1.63  2.10  1.74  2.43  0.0155  
MC = tax level  
MC = sanct i on fee l evel  
3.27  3.18  2.90  3.26  0.1087  
Min AC  3.25  3.34  2.38  3.01  0.0155  
Min MC  2.30  2.74  1.32  2.14  0.005  
Min emiss i ons  2.58  2.94  2.80  3.44  0.005  
 
Moreover, the tax treat m ent makes subj ect s ’  choi ces signi fi cant l y l ess con si s t ent with the rule 
of equal i z i ng av era ge co st and the tax (penal t y) level and the  rule o f mini m i z i ng emiss i ons  
relat i ve to the standards  treat m ent . A n indi cat i on also ex ists that the tax treat m ent causes 
subj ect s to be more consi s t ent with the opti m al rule that equal i z es the                                                                                                                                                                                                                   
margi n al cost and the  tax (penal t y) rate (si gni fi c an t at the 10.1% level ) .  
Final l y,  whet he r an y si gn i fi cant tre at m ent eff ect s o n singl e choi ces  is test ed . A dumm y v ari abl e 
is creat ed to indi cat e  the choi ce rule with which ea ch indi vi dual choi ce is consi s t ent . Wilcox on 
matched - pai r signed - ran ks test s  (W il cox on 1945) are perfo rm ed on the matched pair s  of these 
dumm y va ri abl es from  ea ch tre at m ent . The total nu mber of observ at i ons in each test is the 1 ,49 4  
choi ces mad e b y the 166 subj ect s in the nine choi ce sets. T abl e 9 shows th at  the tax treat m ent  
seem s to increas e subj ect s ’  consi s t en c y  with the o pti m al choi ce rule (treat m ent effect incr eas e 
from 32.2% to 36.3 % an d p- val ue = 0.0565) ,  the avera ge cost - m i ni m i z i ng rule  ( the tre at m ent 
                                                          
12 T a x ver sus  per fo r ma nc e stand a r d : Mann - W hit ne y U - te st, p = 0.0 15 5 ver sus p = 0.0 05 fo r mi ni miz i ng a ver a ge 
and mar g ina l ab ate me n t co sts, resp e c tive l y.  
19  
effe ct incre as es  from 26. 4% to 36.1%  and p- val ue <  0.001),  and the mar gi n al cost - m i ni m i z i ng 
rule  (tre at m ent eff ect inc reas es  from 14.7% to 25. 6%  and p- val u e <  0.001) .  
T he resul t s in Tabl es 8 and 9 indi cat e  that the tax treat m ent  (com pared with the standard s  
treat m ent )  si gni fi c ant l y i ncreas es subj e ct s ’  consi s t en c y  with choi ce rules th at minimize average 
abatement costs and marginal abatement costs, and equalizes the                                                                                                                                                                                                                   
marginal cost and the tax (penalty) rate. N o indi ca t i on exist s that the tax treat m en t signi fi cant l y 
affe ct s subj ect s ’  consi s t e nc y with the  equalize the average abatement cost and the tax level rul e 
or a rule that minimizes emissions . 
Tabl e 9. Me an of dumm y va ri abl es indi cat i n g pr oport i on consi s t ent with a giv en choi c e rule 
(dum m y  =  1 if consi s t ent with a choi ce rul e and d umm y  =  0 if not )  
 Tax treat m ent  Perform anc e standards 
treat m ent  
Wilcox on 
matched -
pai rs signed -
rank test  
Choi ce rule  Mean  Mean  Prob  
AC = tax level  
AC = sanct i on fe e 
level  
0.181  0.193  0.4216  
MC = tax level  
MC = sanct i on fee 
level  
0.363  0.322  0.0565  
Min AC  0.361  0.264  0.0000  
Min MC  0.256  0.147  0.0000  
Min emiss i ons  0.287  0.311  0.2157  
 
A condi t i onal lo gi t model is appl i ed to the full sampl e with the abat e m ent choi ce as the 
dependent vari abl e , cos t vari abl es as alternat i ve- s peci fi c vari abl es , 1 3  and treat m ent and 
back ground vari abl es as indi vi dual - s peci fi c cont r ol s  to furt her inves t i gat e whet her the use of 
                                                          
13B e c a use co st in fo r ma tio n i n th e cho ic e sets are ge ne r a te d alo ng a no n - li ne a r co st func tio n, t he aver a ge and 
mar gi na l co st le ve l s are no t li ne a r l y co r r e la te d . T he co rr e la tio n co eff ic ie n t  is 0.2 3 1 4 . A m ultic o lli ne a r it y a nal ys is 
by Belsle y, K uh, a nd Welsc h ( 1 9 8 0)  sho we d no e vid e nc e of mu ltic o lli ne a r it y bet we e n t he aver a ge a nd mar gi na l 
co st leve ls.  
20  
i nform at i on can ex pl ai n observed abat em ent choi ces (M cF adden and Trai n, 2000) . Tabl e A.2 
in the Appendi x  present s predi ct ed probabi l i t i es and margi nal effe ct s of the condi t i onal logi t 
model esti m at ed for the full sampl e.  
The condi t i onal logi t model shows that t he poli c y t reat m ent s  ha ve  the stro nges t effe ct on the 
choi ce prob abi l i t i es . T he tax treat m ent increas es th e probabi l i t y of an av era ge subj ect  cho os i ng 
A rather than B b y 1 2.9 percent a ge point s  (p <  0. 001) . A ver a ge and ma r gi nal cost inform at i on 
repres ent s alternat i ve - s p e ci fi c attri but es that are stat i s t i cal l y si gni fi cant at the 1%  level  for both 
inves t m ent choi ces  (p ≤ 0.001) . Overal l , ave ra ge cost inform at i on has a stronger ef fect on 
choi ces than mar gi nal cost inform at i on. A  one - uni t increas e in avera ge cost increas es the 
probabi l i t y o f choos i n g the lower abat em ent level A rather than the higher abat em ent level B 
by 0.060 per cent a ge poi nt s . A  one - uni t increas e in margi nal cost incr eas e s the probabi l i t y of 
choos i ng A b y 0.018 p er cent a ge point s .  
Among the ba ck ground vari abl es  in the surve y  o nl y educ at i on level  show s  signi fi cant effe ct s 
on the choi c e. 1 4  A unive rs i t y de gree incr eas es  the probabi l i t y o f choos i n g a lower abat em ent  
level (increas i n g the pro babi l i t y of inves t m ent level A  by 17.4% ) . F urt he r m ore , if the subj ect s  
stat e d that envi ronm ent - r el at ed R&D is a priori t i z ed are a withi n  the firm empl o y i n g  th em , the 
probabi l i t y that the y cho os e a highe r abat em ent level incre as es (incr eas i n g the probabi l i t y o f 
inves t m ent level B b y 9.2 % ).  The gender , age, t ype of de gre e , and ye ars o f work i ng  ex peri ence 
back ground v ari abl es do not have an y si gni fi cant effe ct on abat em ent choi ces .  
4. Discussion 
T he opti m al respons e to an envi ronm ent al Pigouv i an tax is to choos e an abat em ent level for 
which the mar gi nal abat e m ent cost equal s (or comes the closes t ) the tax  level . 1 5  The  ex peri m ent 
in this stud y rev eal s  that the number of choi c es  tha t coinci de s  with such an opti m al respons e  i s 
higher wh en de ci s i ons are mad e subj e ct to a tax than subj ect to a n emis s i ons  standard  with 
ident i cal econom i c in cen t i ves as a tax . Howev er, most import ant l y , oth er choi ce rules b ased on 
cost  inform at i on are found to be more comm onl y us ed for the tax treat m ent than for the 
standard s  treat m ent . T he choi ces that coinci de with choi ce rules that mini m i z e average and 
margi n al cost s are more frequent  in the tax treat m ent than the choi ces corres pondi n g to the 
                                                          
14 A  mu ltic o lli ne a r it y a nal ysi s on alte r na t ive s -  a nd case - sp e c i f ic var ia b le s in T ab le 3 was per fo r me d acco r d ing to 
Belsle y, K uh, a nd Welsc h (1 9 8 0 ) ,  resulting i n a fa ir l y mo d e r a te co nd itio n nu mb e r = 39 . 6 2 .  
15  T his ho ld s as lo n g as t he tax l eve l is hi ghe r t han t he min i mu m po int at t he aver a ge ab ate me nt co st  cur ve , 
i mp l yin g t hat the sav in gs i n e nvir o n me nta l ta x pa yme nt s  al s o co ve r the fi xe d po r tio n of ab ate me nt in ve st me nt 
co sts.  
21  
opt i m al respons e o f se tt i ng the tax equ al  to  the mar gi nal abat em ent cost . Hen ce, the 
ex peri m ent al resul t s  indicat e that m ana gers ’  and seni or advi s ors ’  use of i nform at i on for thei r 
inves t m ent choi ces was  signi fi c ant l y infl u enced b y th e t ype of poli c y inst rum ent . In th e 
standard tre at m ent , the total number of choi ces wit hi n subj ect s  tend s  to be more consi s t ent with 
a rule to mini m i z e emiss ions  level s . 
The  differ ent poli c y tr eat m ent s seem to direct  subj ect s ’  attent i on to either a monet ar y v ari abl e 
(the tax treat m ent ) or the emiss i ons limi t (the standard tr eat m ent ). Th e ex peri m ent al resul t s  
show that these attent i onal biases induce  diff eren c es between inv es t m ent choi ces in the tax and 
standards treat m ent s . In the standards treat m ent , subj ect s ’ focus on abat e m ent seem s  to have 
caused them to under wei ght  cost inform at i on, fail i ng to choos e alternat i ve A , whic h  
corres ponds to less ab at e m ent , even when this is more cost  ef fect i ve.  
Howeve r, re ga rdi ng the  tax inst rum ent , managers ’  and seni or advi s or s ’  reveal ed choi ce 
sequences ar e more con si s t en t with the cost - ef fe ct i ve choi c es that const i t ut e the opti m al i t y  
condi t i on of a mar gi nal cost that equal s the tax . Stil l , the stron ges t tr eat m ent eff ect s wer e found 
for choi c e rules that mini m i z e avera ge o r mar gi nal abat em ent cost s rather th an the total cost to 
the firm (the sum of abat em ent and compl i an c e cost s ). Consequent l y, subj ect s tended to 
mini m iz e abat em ent cost s  acros s alternat i ves rather than the sum  of the abat em ent and 
compl i ance cost s . ( In the ex peri m ent , the subj ec t s were ex pl i ci t l y inst ru ct ed to choos e the 
inves t m ent that resul t ed in the lar ges t cost savi ngs consi d eri n g compl i a nce cost s incl udi n g 
abat em ent cost s ; see th e i nst ruct i ons in the full sur ve y  in Appendi x A.1.) Thus, subj ect s in the 
tax treat m ent overw ei gh ed the impact o f thei r o wn abat em ent cost s  and fail ed to consi de r th e 
full compl i ance cost s .  Th e resul t s show that the att ent i onal bias towa rd  mini m i z i ng aver a ge and 
margi n al abat em ent cost s  resul t s  in mar gi nal abat e m ent cost s in the tax treat m ent that are low er 
than the margi nal cost for the opti m al choi ce sequence ; 1 6  853.7 SEK/ kg compared with the 
ex pect ed means of the op ti m al choi ce of 900 SEK / kg.  
The literat ur e on behavi o ral econom i cs su gges t s a number of ex pl anat i ons and resul t s showi n g 
simi l ar biases as in our findi ngs . Schkade and Kahnem an (1998) found that subj ect s 
overes t i m at e the impact of event s that ar e more s a li ent to them than other event s . Amir, Ariel y, 
and Carm on (2008) fou nd an addi t i onal ex pl anat i on—t hat price asses s m ent s focus on the 
feat ur es of the trans act i o n cues (pri c es , cost s , ma rket nor m s , and so on). Hsee et al. (200 3) 
sugges t ed  that attent i ona l bias with respect to the t yp e of vari abl e can l ea d to incons i s t enci es 
                                                          
16 T he op tima l cho ic e seq ue nc e i mp lie s usi n g co r ne r so lut io ns wh e n inte r io r so lu tio ns are no t po ssib le .  
22  
bet ween choi ces and pr e feren ces . Furt he r, the p r om i nence effe ct might resul t in pref eren ce 
revers al s  becaus e choi c es becom e more inform ed b y the most prom i nent attri but e of the 
eval uat ed alternat i ves (T vers k y, Satt at h, and Slovi c 1988). Inat t ent i on to inf orm at i on is amon g 
the caus es of the ex ist ence o f the ene r g y effi ci en c y gap (All cot t and Gree ns t one, 2012) . E v en 
if this liter at ure has so far fo cus ed most l y on co nsum ers , the resul t s indi cat e that man a ge ri al 
deci s i on  maki n g is also subj ect to bias in thei r at tent i on when respondi ng to poli ci es , whic h 
coul d cause inves t m ent s in , for ex am pl e, en er g y eff i ci enc y to diver ge from th e cost - m i ni m i z i ng 
level .  If the probl em is underi nves t m ent , then th e resul t s indi cat e that inf orm at i on about the 
emiss i ons or ene r g y pe rf orm ance might n eed to be made more sali ent to t he de ci s i on maker.  
The se  resul t s furt her su gges t that appropri at el y b a lanc i ng  th e import ance of consi derat i ons that 
are the fo cus of attent i on with the import ance of consi derat i ons that are cu rrent l y in the 
back ground mi ght b e di ffi cul t . O ur resul t s sho w that this holds true even for ex peri enc ed 
deci s i on - m akers .  
4.1 Welfare implications 
W hat gener al welfar e con cl us i ons can be reached from the ex peri m ent al resul t s? Theor y shows 
that tax es resul t in an effi ci ent outcom e when a fir m subj ect to a soci al l y ef fi ci ent tax bas es its 
inves t m ent deci s i ons on the opti m al choi ce rule. Howeve r, mini m iz i ng  the avera ge abat em ent 
cost  was the most frequent choi ce rule used b y mana gers and seni or ad vi s ors . This  t ype o f 
mana geri al inves t m ent behavi or gene ral l y resul t s  i n abat em ent level s lower than the soci al l y 
effi ci ent ab at em ent level  becaus e the mini m um point of the average abat e m ent is associ at ed 
with a lower  ab at em ent l evel than that of the so ci al l y effi ci ent abat em ent l evel , resul t i n g in a 
welfar e los s . 1 7  
T he effi ci en c y of tax es v ersus standa rds dep ends on the choi ce rule us ed b y firm s . Tax es are 
weakl y mor e effi ci ent than standards when a fi rm ’s  inves t m ent deci s i ons are bas e d on the 
opti m al choi ce rul e.  How ever, as our ex peri m ent al resul t shows , mini m i z i ng avera ge abat em ent 
cost was the  most fr equ ent choi ce rule , which  i ntroduces ambi gui t y in t erm s of th e opti m al 
poli c y choi ce fo r which s tandards can be more effi ci ent than tax es .  
                                                          
17 Fo r any ta x leve l t hat is hig he r than the mi ni mu m po int of t he fir m’ s aver a ge co st cur ve , t he op ti ma l cho ic e o f 
the fir m is a po sitive ab ate me n t leve l fo r whic h the mar gi na l c o st eq ua ls the ta x leve l. Beca us e the po sitive l y slo p e d 
mar gi na l co st cur ve al wa ys cu ts the mi ni mu m po int of the  U - s ha p e d  aver a ge ab ate me n t co st cur ve , an y so cia ll y 
eff ic ie n t ab ate me nt leve l t hat e xce e d s the mi ni mu m po int of t he aver a ge ab ate me n t co st i mp lie s a hi ghe r ab ate me nt 
leve l t han t hat at t he mini mu m po int o f the a ver a ge ab ate me nt co st .  
23  
Th i s  ambi gui t y consi s t s of the net welfar e eff ect o f two oppos i n g beh avi oral anom al i es . On the 
one hand, mini m i z i ng av era ge abat em ent cost s  un der tax es leads to welfar e losses bec aus e of 
under - com pl i an ce. On th e other hand, ov er - com pl i ance unde r standa rds lea ds to welfar e losses 
from ex ces s abat em ent highe r than the soci al l y effi ci ent abat em ent lev el .  
Hence, t he compa ri s on between tax es  and standards  depends  on the diff erenc e betwe en the 
abat em ent l evel at  the mi ni m um point of the aver age abat em ent cost  curve  and the ab at em ent 
level associ at ed with the soci al l y ef fi ci ent tax level  relat i ve to the differenc e betwe en the 
abat em ent level associ at e d with over - com pl i anc e and the soci al l y effi ci ent abat em ent level .  
The inst rum ent that brin gs the abat em ent level s  clos e r  to those cor res pon di ng to the soci al l y 
effi ci ent t ax level also re sul t s in lower welf are los s es . T hat t he relat i ve slo pes of the m ar gi nal 
abat em ent cost and mar gi nal bene fi t funct i on s  do not  aff ect whet her tax es or standards resul t in 
the lowes t welfa re  losses is strai ght for ward  to sho w . 
5. Conclusions 
Thi s paper f inds  ex peri en ced man a gers ’  ab at em ent  deci s i ons  ex hi bi t  s yst em at i c diff eren ces that 
depend on whet her  a  deci s i on is made subj ect to emiss i ons tax es  or emiss i ons standard s . 
R ati onal choi ce theor y pr edi ct i ons sugges t that  no differen ce shoul d ex ist between the outcom e 
of  rati onal de ci s i ons in these two cir cum s t anc es . Howeve r, the ex peri m en t conduct ed in this 
stud y su gges t s that ex per i enced man a gers are sub j ect to attent i onal biases . D iffer ent t yp es of 
poli c y inst rum ent s  seem to draw mana gers ’  atte nt i on to either monet ar y vari abl e s  (the tax 
treat m ent ) o r emiss i ons l imi t value s  (the stand ard s  treat m ent ). Mana gers ’  focus on abat em ent 
level s in the standards treat m ent seem s  to cause them to have underwei gh t  cost inform at i on, 
and the y fail to select th e alternat i ve corr es pondi n g to less abat em ent —eve n when doin g so is 
more cost  effe ct i ve. Acc ordi ngl y, the tax treat m ent seem s to increas e managers ’  consi s t en c y 
with the opti m al choi ce rule relat i ve to the emiss i ons standard treat m ent . This resul t highl i ght s 
import ant aspect s of poll ut i on cont rol and poli c y choi ce, namel y , th at diffe r ent t yp es of poli c i es , 
despi t e ident i cal econom i c incent i ves , induce di ffe rent inves t m ent behavi or.  
That mana gers us e simpl i fi ed rules o f thum b re ga rdl es s of th e t ype of poli c y inst rum ent the y 
face is part i cul arl y clea r from the resul t s of this stud y . Part i ci pant s wer e p resent ed  with nine 
binar y choi ce s ets  to ide nt i f y th e choi ce rules to which the y adhe re . This desi gn allowed us to 
draw stat i s t i cal infer enc es withi n subj ect s  about the choi ce rules used. The tax treat m ent 
signi fi cant l y inc reas e d  mana ger s ’  consi s t enc y with choi ce rules invol vi ng monet ar y 
inform at i on. The choi ce rules incl uded the opti mal rule of equal i z i ng the margi n al abat em ent 
24  
cos t with the t ax level an d mini m i z i ng  mar gi nal and ave ra ge ab at em ent co st s . The latt er choi ce 
rule was the most signi fi c ant . T hat  is , managers ten d to mini m iz e avera ge ab at em ent cost s rathe r 
t han the  sum of the total abat em ent and re gul at or y cost s , resul t i n g  in abat em ent level s lower 
than opti m al and ma r gi nal cost s lowe r than t ax level s . In this stud y’s inc e nt i vi z ed ex peri m ent , 
if mana ge rs mad e de ci s i ons consi s t ent with rati onal choi c e theor y, the re sul t shoul d not have 
reveal ed an y tre at m ent effect s . The resul t s  of thi s ex peri m ent  ar e in line with the resea rch 
docum ent i ng beh avi oral anom al i es  (see, fo r ex am pl e, Thal er 1985; Hse e et al. 2003; Amir, 
Ariel y, and Carm on 2008 ) and de ci s i on- b y - rul es m echani s m s lea rned eithe r through ex peri ence 
or soci al ex change rather than ca rri ed out b y full y rati onal deci s i ons mak ers (see,  for ex am pl e, 
Simons on 1989; Prel ec and Herrns t ei n 1991; Shaf i r, Simons on, and Tvers ky 1993).  
Rega rdi n g welfa re eff ect s , an y poli c y inst rum ent that brings the abat em ent level close r  to the 
opti m al level also resul t s in lowe r  welf ar e loss es . When firm s are rati onal and use the opti m al 
rule, tax es are alwa ys we akl y mo re ef fi ci ent than s tandards . Ho weve r, in the ex peri m ent in this 
stud y, the most frequ ent choi ce rule used b y man a ge rs and seni or advi s ors was to mini m i z e the 
avera ge ab at em ent cost . This t ype of m ana ge ri al inves t m ent behavi or gen eral l y resul t s in 
abat em ent lev el s lower t han the soci al l y effi ci ent abat em ent lev el , introduc i ng an ambi gui t y in 
term s of opti m al poli c y choi ce for whi ch standard s can be mor e ef fi ci ent th an tax es .  
  
25  
References 
Al l cot t , H., and M. Gre ens t one. 2012. “Is Th er e an Ene r g y Effi ci enc y Gap? .” Journal of 
Economic Perspectives 26 (1): 3 –28. 
Amir, O., and D. Ari el y. 2007. “Deci s i ons b y Ru les : The Case of Un wi l l i ngnes s to Pa y fo r 
Bene fi ci al Dela ys .” Journal of Marketing Research 44 (1): 142–52. 
Amir, O., D. Ariel y, and Z. Carm on. 2008. “The Dissoci at i on between M onet ar y Asses s m ent 
and Predi ct ed  Util i t y. ” Marketing Science  27 (6): 1055–64. 
Armst rong, M., and S. Huck. 2010. “Beh avi oral Econom i cs as Appl i ed to Firm s : A Prim er. ” 
CES i fo Worki ng Paper No. 2937. 
Bloom , N., B. Eifert , A. Mahaj an, D. McKenz i e, and J . Robert s . 2013. “Does Mana gem ent 
Matt e r? Evidence from Indi a.” Quarterly Journal of Economics  128 (1): 1– 51. 
Camerer, C. F., and U. Malm endi er. 2007. “Behavi oral Econom i cs of Organi z at i ons ” In 
Behavi o ral Econom i cs and Its Appl i c at i ons , edit e d b y P. Diam ond and H. Vart i ai nen, 235–90. 
Princet on, NJ : Princet on Univers i t y Pr ess . 
Carl s s on, F., and O. J ohans s on- S t enm an. 2012. “Beh avi oral Econom i cs and Envi ronm ent al 
Poli c y.” Annual Review of Resource Economics  4 (1): 75 –99. 
C yert , R., and J . March. 1963. A Behavi or al Theor y of the Firm . Engl ewood Clif fs , NJ: Prent i ce 
Hall .  
Elli s on, G. 2006. “Bounded Rati onal i t y in Indus t ri al Organi z at i on” In Adv a nces in Econom i cs 
and Econom et ri cs : Theor y and Appl i c at i ons , Nint h Worl d Congres s , Volu m e 2, edit ed b y R. 
Blundel l , W. Newe y, and T. Perss on, 142–74. Ne w York:  Cambri d ge Uni vers i t y Pr ess .  
Goldber g, J ., and R. Vo n Nitz s ch. 2001. Behavi oral Finance. Chiches t e r, UK: J ohn Wile y & 
Sons. 
Gowd y, J . M. 2008. “Beh avi oral Econom i cs and Clim at e Change Poli c y.” Journal of Economic 
Behavior & Organization 68 (3–4): 632–44. 
G reen, D., and I. Shapi ro. 1994. Pathol ogi es of Rati onal Choi ce Theor y. Ne w Haven, CT: Yale 
Univers i t y Pr ess .  
Gsot t bauer, E. 2013. “Beh avi oral Econom i cs and Envi ronm ent al P oli c y: Theo r y and 
Ex peri m ent s .” Doct oral t hesi s . Spai n: Univers i t at Autónom a de Barc e l ona.  
Gsot t bauer, E., and J . C. J . M. van den Ber gh. 2 011. “Envi ronm ent al Poli c y Th eor y Given 
Bounded Rati onal i t y and Other - R e ga rdi ng P refe renc es .” Environmental & Resource 
Economics 49 (2): 263–3 04. 
Harri s on, G. W., and J . A. List . 2004. “Fiel d Ex per i m ent s .” Journal of Economic Literature  42 
(4): 1009–55. 
26  
Hs ee, C. K. 1999. “Val u e Seeki n g and Predi ct i on–Deci s i on Incons i s t enc y: Wh y Don’t Peopl e 
Take What The y Predi ct The y’ l l Like the Most ? .” Psychonomic Bulletin & Review 6 (4): 555–
61. 
Hsee, C. K., J . Zh an g, F. Yu, and Y. Xi. 2003. “La y R ati onal i s m and Inc ons i s t enc y b etwe en 
Predi ct ed Ex peri ence and Deci s i on.” Journal of Behavioral Decision Making 16 (4): 257–72. 
J ones , B. D. 1999a. “Bounded Rati onal i t y. ”  Annual Review of Political Science 2 (1): 297–321.  
———. 1999b. “Bound e d Rati onal i t y, Poli t i cal In st i t ut i ons , and the Anal ys i s of Outcom es ” In 
Compet i t i on and Cooperat i on: Convers at i ons with Nobel i s t s about Econom i cs and Poli t i cal 
Science, edit ed b y J . Alt , M. Levi , and E. Ostrom , 85–111. New York: Russel l Sage Fo undat i on.  
Kahnem an, D., and A. Tv ers k y. 1979. “Pros pect Theor y: An Anal ys i s of De ci s i on under Risk.” 
Econometrica 47 (2 ): 263–91. 
Kocher, M. G., P. Mart i n s s on, and M. Visser. 200 8. “Does Stake Siz e  Matt er fo r Cooper at i on 
and Puni s hm ent? .” Economics Letters  99 (3): 508 –11. 
Muni er, B., R. Selt en, D. Bou ys s ou, P. Bour gi ne, R. Da y, N. Ha rve y, D. Hi lt on, M. J . Machi na, 
P. Parker, J . Sterm an, E. Weber, B. Wern erf el t , an d R. Wensl e y. 1999. “Bo unded Rati ona l i t y 
Model i ng.” Marketing Letters 10 (3): 233–48.  
McFadd en, D., and K. Trai n. 2000. Mix ed MN L Model s for Discret e Re spons e. Journal of 
Applied Econometrics 15 (5): 447–70. 
Poll i t t , M. G., and I. Shaors hadz e. 2011. “The Role of Behavi our al Econo m i cs in Ener g y and 
Clim at e Poli c y. ” EPR G Worki ng Paper 1130. Cambri dge Worki n g Pa per in Econom i cs , 
Volum e 1165. 
Prel ec, D., and R. Herr ns t ei n. 1991. “Pref eren c es or Princi pl es : Alte rna t i ve Guidel i nes for 
Choi ce” In Strat e g y and Choi ce, edit ed b y R.J . Ze ckhaus er, 319–40. Ca mbri dge, MA: M IT 
Press .  
Salant , Y. 2011. “Procedural Anal ys i s of Choi ce Rules with Appl i cat i ons to Bounded 
Rati onal i t y. ” American Economic Review  101 (2): 724–48. 
Schkade, D. A., and D. Kahnem an. 1998. “Do es Livi n g in Cali forni a Mak e Peopl e Happ y? A 
F ocus i n g Illus i on in J udgm ent s of Lif e Sati s fact i o n.” Psychological Science  9 (5): 340–46.  
Shafi r, E., I. Simons on, and A. Tvers k y. 1993. “Reas on- B as ed Choi ce. ” Cognition 49 (1–2): 
11–36. 
Shogren, J . F. 2002. “A Beh avi oral Minds et on Envi ronm ent al Poli c y. ” Journal of Socio-
Economics 31 (4): 355–6 9. 
Shogren, J . F., and L. O. Ta yl o r. 2008. “On Behav i oral - Envi ronm ent al Eco nom i cs .” Review of 
Environmental Economics & Policy 2 (1 ): 26 –44.  
27  
S i m on, H. A. 1955. “A Behavi oral Model o f Rati onal Choi ce.” Quarterly Journal of Economics 
69 (1): 99–118. 
———. 1979. “Rat i onal Deci s i on M aki ng in Bu si nes s Or gani z at i ons .” American Economic 
Review 69 (4): 493–513.  
Simons on, I. 1989. “Choi ce Based on Reasons : The Case of Attra ct i on and Comprom i s e 
Effect s . ” Journal of Consumer  Research 16 (2): 15 8–74. 
Simons on, I., and A. Tv ers k y. 1992. “Choi ce in Cont ex t: Tradeoff Cont r as t and Ex trem enes s 
Avers i on.” Journal of Marketing Research 29 (3) : 281–95. 
l, S., A. Mall et t , and S. Nye . 2011. Ba rri ers to Indus t ri al En er g y Effi c i enc y: A Lite rat ur e 
Revi ew. Dev el opm ent Poli c y, Stat i s t i cs and Res ear ch Branch, Worki ng Pap e r 10/2011. Vienna: 
Unit ed Nati ons Indus t ri al Devel opm ent Or gani z at i on. 
Spiegl er, R. 2011. Bou nded Rati onal i t y and In dus t ri al Or gani z at i on. Ox ford, UK: Ox ford 
Univers i t y Pr es s .  
Sunst ei n, C. R. 2004. “Moral Heuri s t i cs and Mo ral Fram i n g.” Minnesota Law Review  88 (6 ): 
1556–97. 
S yvers on, C. 2011. “W h at Determ i nes Produ ct i vi t y? .” Journal of Economic Literature  49 (2): 
326–65. 
Thal er, R. 1985. “Ment al Account i ng and Consum er Choi ce.” Marketing Science 4 (3 ): 199–
214. 
Tvers k y, A., and D. Ka hnem an. 1991. “Loss Avers i on in Riskl es s Choi ce: A Referen ce -
Dependent Mod el .” Quarterly Journal of Economics  106 (4): 1039–61.  
Tvers k y, A., S. Satt at h, and P. Slovi c. 1988. “Cont i ngent Wei ght i ng in J udgm ent and Choi ce. ” 
Psychological Review 95 (3): 371–84. 
Tvers k y, A., and I. Simons on. 1993. “Cont ex t - Dependent Pref eren ces .” Management Science  
39 (10): 1179–89. 
Weit z m an, M. L. 1974. “Pri ces vs. Quant i t i es .” Review of Economic Studies  41 (4): 477–91. 
  
28  
Appendix  
Appendix A.1 Survey (translated from Swedish) 
Introduction 
Thank you for taki n g the time to part i ci pat e in this rese arch stud y.  Th e purp os e of this stud y is 
to bett er unders t and how firm s ’  envi ronm ent al inves t m ent s are aff ect ed b y va ri ous 
envi ronm ent al poli c y inst rum ent s . The rese ar ch stud y is ca rri ed out b y a rese ar ch proj ect 
invol vi ng rese ar chers fro m IV L Sw edi s h Envi ron m ent al Resear ch Inst i t ut e and the School of 
Busi nes s , Econom i cs , an d  La w at the Unive rs i t y of Gothenbur g. The p roj ect is funded b y the 
Swedi s h foundat i on for s trat e gi c envi ronm ent al re search  (M ist ra ) ,  which ai ms to cre at e stron g 
resea rch envi ronm ent s , cont ri but e with solut i ons to envi ronm ent al probl em s , and strengt hen 
compe t i t i venes s .  
In the stud y , we ask you to make an inves t m ent deci s i on for a repr es ent at i ve (fict i t i ous ) firm . 
Howeve r, when answe ri ng the qu est i ons , we woul d like you to use th e pra ct i cal ex peri ence that 
yo u have  that also migh t be relevant to the deci s i ons . Your task is to make a total of nine 
indi vi dual inves t m ent deci s i ons for the fict i t i ous firm . Note that the inform at i on tex t in each 
select i on is ident i cal . After the nine inves t m ent deci s i ons , we ask you to answer a short 
quest i onnai re with b ack ground quest i ons . The t as k and the su rve y to get her take approx i m at el y 
15–20 minut es to compl et e. 
If you choos e th e opti o n that gives the firm th e lar ges t cost savi n gs in all nine inves t m ent 
deci s i ons , we will pa y you SEK 4,200 (app rox i mat el y  USD 485) in comp ens at i on. Tax es and 
soci al securi t y fees wil l be deduct ed from the compens at i on, which means that your  
compens at i on afte r tax will be approx i m at el y SEK 3 ,000 ( approx i m at el y  USD 35 0).  
You r ident i t y remai ns  an on ym ous , and the final re sul t s will onl y be pr esent ed at a summ ari z ed 
level without the possi bi li t y of linki ng your answ e rs to you or the firm for which you work. The 
final surve y resul t s will also be made avai l abl e to interes t ed pa rt i ci pant s .  
 
Investment decision  
W e ask  you to think abo ut a  situat i on in which yo u are invol ved in an inve s t m ent deci s i on in a 
firm that will inves t in a new abat em ent technol o g y to reduce thei r carbon diox i de emiss i ons . 
29  
I nv es t m ent s  in technol o g y impos e  fix ed cost s and  mar gi nal cost s that are continuously  
increasing; that is,  the more the firm redu ces its carbon diox i de emiss i ons , the more ex pens i ve 
it becom es to reduce anot her ton of carbon d iox i de. With new technol og y, the firm  is 
consi deri n g two diffe ren t abat em ent level s : alter nat i ve A, which i mpl i es  a smal l er emiss i on 
reduct i on, and alternat i ve B, which i mpl i es  a large r emiss i on reduct i on. Both  opti ons have the 
same econom i c lif e.  
Treatment 1: Environmental Tax 
The prer equi s i t e for your deci s i on is that the firm produces in a count r y wit h a carbon tax rate 
equal to SEK 1000 p er t on carbon diox i de emiss ion each yea r. Both abat e m ent alternat i ves A 
and B redu ce emiss i ons and, hence,  the tax pa ym en t s . Both altern at i ves A an d B entai l incr eas ed 
margi n al cost s to reduce the firm ’ s  carbon dio x i de emissi ons . The cost s are shown in the 
foll owi ng tabl e.  
We ask you to choos e ab at em ent level A or B wit h the new t echnol o g y. Ch oos e the opti on that 
yo u b eli eve giv es the firm  the gr eat es t savi n gs in SEK , incl udi ng both inves t m ent and tax 
pa y m ent s . 
Investment decision 1 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
200 500 
Average cost per ton at attained reduction level (SEK/ton) 1 000 610 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions. 
  
30  
Investment decisions 2–9 (The investment decision and environmental tax treatment texts 
above repeated with tables as follows) 
Investment decision 2 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
200 1 000 
Average cost per ton at attained reduction level (SEK/ton) 1 000 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 3 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
300 500 
Average cost per ton at attained reduction level (SEK/ton) 750 610 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
  
31  
Investment decision 4 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
500 1 000 
Average cost per ton at attained reduction level (SEK/ton) 610 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 5 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
600 900 
Average cost per ton at attained reduction level (SEK/ton) 600 650 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
  
32  
Investment decision 6 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
900 1 000 
Average cost per ton at attained reduction level (SEK/ton) 650 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 7 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 000 1 200 
Average cost per ton at attained reduction level (SEK/ton) 680 750 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
  
33  
Investment decision 8 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 000 1 800 
Average cost per ton at attained reduction level (SEK/ton) 680 1000 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 9 
 
A B 
Tax per ton of carbon dioxide (SEK/ton) 1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 200 1 800 
Average cost per ton at attained reduction level (SEK/ton) 750 1000 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
 
34  
Treatment 2: Emissions Standard 
The prer equi s i t e for you r deci s i on is that the firm produces in a count r y with an emiss i ons 
standard that limi t s how much carbon diox i de em iss i ons is allowed. The alternat i ve with the 
lower ab at em ent lev el (a lt ernat i v e A) resul t s in an emiss i ons level that ex ceeds th e emiss i ons 
limi t value. If this alterna t i ve is chosen, the firm need s  to pa y SEK 1000 per ton in sanct i on fee 
for eve r y addi t i onal ton that the emiss i ons level ex ceeds the emiss i ons limi t value. If the 
alternat i ve with a hi gher abat em ent is chosen (al t ernat i ve B), the emiss i ons level does not 
ex ceed the standards , and no sanct i on fee needs to be paid. B oth alternat i ves A and B entai l 
increas ed ma r gi n al cost s to reduc e the firm ’ s  ca rb on diox i de emiss i ons . The cost s are shown in 
the tabl e  below .  
We ask you to choos e ab at em ent l evel A or B wit h the new t echnol o g y. Ch oos e the opti on that 
yo u beli ev e gives the fi rm the gre at es t savi n gs , incl udi ng both inves t ment cost s and an y 
sanct i on  fee p a ym ent s .  
Investment decision 1 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
200 500 
Average cost per ton at attained reduction level (SEK/ton) 1 000 610 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions. 
 
  
35  
Investment decisions 2–9 (The investment decision and emissions standard treatment 
texts above repeated with tables as follows) 
Investment decision 2 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
200 1 000 
Average cost per ton at attained reduction level (SEK/ton) 1 000 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 3 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
300 500 
Average cost per ton at attained reduction level (SEK/ton) 750 610 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
36  
 
Investment decision 4 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
500 1 000 
Average cost per ton at attained reduction level (SEK/ton) 610 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 5 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
600 900 
Average cost per ton at attained reduction level (SEK/ton) 600 650 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
  
37  
Investment decision 6 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
900 1 000 
Average cost per ton at attained reduction level (SEK/ton) 650 680 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 7 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 000 1 200 
Average cost per ton at attained reduction level (SEK/ton) 680 750 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
  
38  
Investment decision 8 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 000 1 800 
Average cost per ton at attained reduction level (SEK/ton) 680 1000 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
Investment decision 9 
 
A B 
Sanction fee per ton of carbon dioxide (SEK/ton) in excess of the 
standard 
1 000 1 000 
Marginal cost per ton at attained reduction level (SEK/ton) 
 
1 200 1 800 
Average cost per ton at attained reduction level (SEK/ton) 750 1000 
 
I make the following decisions: 
□ Invest in the new technology and abatement level A, which results in a smaller 
reduction in emissions. 
□ Invest in the new technology and abatement level B, which results in a greater 
reduction in emissions 
 
  
39  
Background questions 
1. Enter your gend er:  
□ Woman 
□ Man 
 
2. Enter the ye ar when you were bo rn: _________ 
 
3. What is you r high es t compl et ed educat i on:  
□ Primary school 
□ High school 
□ University 
□ Other, specify which__________________________________________________ 
 
4. Stat e your main educat i onal back ground  
□ Business Administration 
□ Law 
□ Economics 
□ Natural Science 
□ Environmental Science 
□ Psychology or Sociology 
□ Engineering 
□ Other, indicate which: 
 
40  
5. How man y yea rs of work ex peri ence do you ha ve in you r cur rent main tasks ? 
Answer  year     Not e tha t onl y numbers can be wr it t en in this fiel d. 
 
6. Have you pa rt i ci pat ed in a deci s i on with the aim of redu ci n g a firm ’s envi ronm ent al impact 
at an y time duri ng th e las t 10 ye ars ?  
□ Yes 
□ No. 
 
7. (If yes to B6): What ro le have you mainl y pla ye d in such deci s i ons ? 
□ M ana ger  
□ Technical advice 
□ Financial advice 
□ Other, specify which__ ________________________________________________ 
 
8. (I f ye s on 6): Indi cat e how much ex peri ence yo u hav e of part i ci pat i n g in envi ronm ent al 
inves t m ent deci s i ons withi n firm s 
□ Very extensive experience (>10 ye a rs )  
□ Extensive experience (5–10 ye ars )  
□ Little experience (<5 years) 
□ No experience 
 
9. Do you consi der th at resear ch and d evel opm ent in envi ronm ent al issues is a priori t y are a at 
the firm  wher e you ar e empl o yed ? 
□ High priority 
41  
□ Quite a priority 
□ Less priority 
 
  
42  
Appendix A.2 Tables 
Tabl e A.1 Obse rved choi ces per choi ce s et  
Obse r ve d cho ic e s per cho ic e set  
 
T ax trea tme nt  P er fo r ma nc e sta nd a r d s trea t me nt  
Cho ic e set  A  B  A  B  
1  9  81  5  72  
2  28  62  13  64  
3  19  71  11  66  
4  55  35  33  44  
5  42  48  23  54  
6  37  53  18  59  
7  52  38  26  51  
8  72  18  51  26  
9  68  22  48  29  
T o ta l  38 2  42 8  22 8  46 5  
 
Tabl e A.2. Predi ct ed prob abi l i t i es and mar gi n al eff ect s on choi ces  in the con di t i onal logi t model  
Pr( c ho ic e = A |1 sele c te d ) = 0 .1 5 5  
Var ia b le  dp /d x *  Std . Err .  z  P>|z| [    95% C.I .    ]  X  
AC         
A  - 0 .0 0 0 5 99    0 .0 0 01 3 6    - 4.4 0    0.0 0 0   - 0 .0 0 0 8 66   - 0 .0 0 0 3 33     74 7  
B  0 .0 0 05 9 9    0 .0 0 01 3 6     4.4 0    0.0 0 0    0 .0 0 03 3 3    0 .0 0 08 6 6        740  
MC         
A  - 0 .0 0 0 1 77    0 .0 0 00 5 2    - 3.4 1    0.0 0 1    - 0 .0 0 0 2 8   - 0 .0 0 0 0 75     65 6  
B  0 .0 0 01 7 7    0 .0 0 00 5 2    3.4 1    - 0.0 0 1    0 .0 0 00 7 5     0 .0 0 02 8   107 8  
C ase  var ia b le s         
1.T r e a tme nt  0 .1 2 9     0. 0 3 47  3.7 1    0.0 0 0    0 .0 6 06 2 2    0 .1 9 7           0  
        1 .B1  0 .0 2 40    0 .0 2 05     1.1 7    0.2 4 2   - 0 .0 1 6 2    0 .0 6 42          1  
B2  - 0 . 0 0 1 3 9     0 .0 0 12 3    - 1.1 3    0.2 5 9   - 0 .0 0 3 8 0    0 .0 0 10 2     19 7 0  
3.B 3  0 .1 7 4      0. 0 3 5 9  4.8 6    0.0 0 0    0 .1 0 4    0 .2 4 5           2  
    1.B 4  - 0 . 0 2 8 0      0 .0 2 4 6     - 1.1 4    0.2 5 5   - 0 .0 7 6 2     0 .0 2 02          0  
B5  - 0 . 0 0 0 5 94    0 .0 0 11 4    - 0.5 2    0.6 0 3   - 0 .0 0 2 8 4     0 .0 0 16 5     16. 2  
        2.B 6  0 .0 2 73    0 .0 2 46     1.1 1    0.2 6 7   - 0 .0 2 0 9      0 .0 7 5 6          1  
B 7  0 .0 9 2 2     0 .0 2 29     4.0 2    0.0 0 0    0 .0 4 7 3     0 .1 3 7    1.7 6  
 
  
43  
Fa c to r v ar ia b le        B ase  
T rea tme nt       0  
B1  1  
B3  2  
B4  0  
B6  1  
*dp /d x fo r facto r leve l s is t he disc r e te chan ge fro m the base leve l  
 
 
Tabl e A. 3. Des cri pt i on of back ground v ari abl es  
Var ia b le  Desc r ip tio n  Ob se r va tio n s  
B1  Gend e r  Fe ma le  (0 ) , Male (1 )  
B2  Bir th  year  Cale nd a r Year  
B3  Hig he st co mp le te d ed uc a tio n  P rima r y scho o l  (1 ) , High scho o l  (2 ) , Unive r sit y  
(3 ) , Othe r  (4 )  
B4  Main ed uc a tio na l back gr o und  So cia l scie nc e s  (1 ) , Natur a l sc ie nc e  or 
Engi ne e r i ng (2 ) Othe r  (3 )  
B5  Year s of wo r k exp e r ie nc e i n c ur r e nt main 
task s  
Year  
B6  Year s of exp e r ie nc e par tic ip a t ing i n 
envir o n me nta l i n ve st me nt decisio n s  
Very extensive experience (>10 years) (1) , 
Exte n si ve exp e r ie nc e (5 – 1 0 year s ) (2 ) , Little 
exp e r ie nc e (<5 year s)  (3 ) , No exp e r ie nc e (4 ) 
B7  Rese a r c h a nd deve lo p me nt i n envir o n me nta l 
iss ue s i s a prio r ity area at the fir m wher e yo u 
are emp lo ye d  
Les s prio r i t y  (1 ) , Quite a prio rit y  (2 ) , High 
prio r ity  (3 ) 
 
 
Tabl e A. 4. Des cri pt i ve st at i s t i cs  
Var ia b le  T yp e  Mean  Std . Dev  Min  Max  
AC  Nu me r ic a l  74 0  14 4 . 57 8         610  10 0 0  
MC  Nu me r ic a l  10 7 7 .7 7 8     444 . 31 5         500  18 0 0  
B1  Facto r  0.6 9 9     0.4 5 9           0  1  
B2  Nu me r ic a l  19 7 0     8.6 4 7        195 1  19 8 8  
B3  Facto r  2.9 3 4     0.4 4 1           2  4  
B4  Facto r  0.8 4 9     0.3 5 8           0  1  
B5  Nu me r ic a l  15.7 05     9.6 4 5           1  40  
B6  Facto r   2.0 1 3      0.8 6 1           1  4  
B7  Nu me r ic a l  1.7 4 7     0.7 0 9           1  3