Decentralized finance (DeFi) markets for startups: search frictions, intermediation, and the efficiency of the ICO market

Mom az, Paul P.
A icle — Published Ve sion
Decen alized inance (DeFi) ma ke s o s a ups: sea ch
ic ions, in e media ion, and he e iciency o he ICO
ma ke
Small Business Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Mom az, Paul P. (2024) : Decen alized inance (DeFi) ma ke s o s a ups: sea ch
ic ions, in e media ion, and he e iciency o he ICO ma ke , Small Business Economics, ISSN
1573-0913, Sp inge US, New Yo k, NY, Vol. 63, Iss. 4, pp. 1415-1447,
h ps://doi.o g/10.1007/s11187-024-00886-3
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/315614
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
h ps://doi.o g/10.1007/s11187-024-00886-3
RESEARCH ARTICLE
Decen alized inance (DeFi) ma ke s o s a ups: sea ch
ic ions, in e media ion, and he e iciency o he ICO
ma ke
Paul P. Mom az
Accep ed: 22 Janua y 2024
© The Au ho (s) 2024
Abs ac This pape examines he e iciency o he
Ini ial Coin O e ing (ICO) ma ke h ough a sea ch-
heo e ical lens. Sea ch in ensi y associa ed wi h he
p ocess o iden i ying aluable s a ups is inc easing
in ma ke g anula i y. DLT inc eases ma ke g an-
ula i y because asse okeniza ion lowe s en y ba -
ie s. Lowe -end en an s, howe e , inc ease agg e-
ga e sea ch in ensi y bu may lack sea ch skills. The
esul ing sea ch- ela ed ine iciency c ea es a niche o
in e media ies o ins i u ional in es o s ha special-
ize on sea ch. Consis en wi h he heo y, specialized
c yp o unds inc ease ICO ma ke e iciency by educ-
ing sea ch ic ions, in e alia, by sho ening he ime-
o- unding and inc easing he unding amoun . A he
same ime, c yp o unds ex ac sizable economic en s
o hei in e media ion se ices. O e all, he s udy
ela es o he gene al ade-o be ween cen aliza ion
and decen aliza ion in en ep eneu ial inance. I sug-
ges s ha ma ke ic ionsspeci ic oea ly-s agec owd-
unding o en ep eneu ship may p e en “pe ec ly”
Decen alized Finance (DeFi) ma ke s om unc ion-
ing e icien ly.
Plain English Summa y Decen alized Finance
(DeFi)ma ke smay equi e a subs an ial deg eeo cen-
aliza ion o unc ion e icien ly. We show ha cen-
aliza ion in he o m o ins i u ional in es o s ha
P.P. Mom az (B)
TUM School o Managemen , Technical Uni e si y o
Munich, A ciss . 21, 80333 Munich, Ge many
e-mail: [email p o ec ed]
in e media e Ini ial Coin O e ings (ICOs) lead o, i s ,
sho e imepe iods o each und aisinggoalsand, sec-
ond, highe alua ions. In a sea ch- heo e ical model,
we quan i y he ex en o which cen aliza ion mi i-
ga es ic ions in a decen alized ma ke . Cen aliza ion
educes adingdelaysandimp o esdecen alizedma -
ke e iciency especially in imes o ma ke down u ns
and when he e is unce ain y abou he eam o p od-
uc quali y. Thus, he p incipal implica ion o ou s udy
is ha decen alized ma ke s o s a ups may no be
op imal o socie y. Cen aliza ion is aluable because
i imp o es he speed wi h which en ep eneu s and
in es o s mee , and because i mi iga es ma ke ic-
ions a ising om asymme ic in o ma ion.
Keywo ds En ep eneu ial inance ·Blockchain-
based c owd unding ·Ini ial Coin O e ing (ICO) ·
Tokeniza ion ·C yp o unds ·Decen alized Finance
(DeFi)
JEL Classi ica ion G23 ·G24 ·L26
1 In oduc ion
New echnologiesa e con inuouslychanging hena u e
o en ep eneu ial inance. The end goes owa d
a dec easing deg ee o in e media ion. The a io-
nale is, in e alia, ha disin e media ion inc eases
he economic ansac ion su plus ha en ep eneu s
and in es o s ge o enjoy. Fo example, equi y-based
123
/ Published online: 23 Feb ua y 2024
Small Bus Econ (2024) 63:1415–1447
c owd unding and i s ela ed o ms ( o e iews, see
Mo i z and Block, 2016; Mochkabadi and Volkmann,
2020; Block e al., 2021) ha e pa ially elimina ed
Ven u e Capi alis s (VCs) om mo e adi ional en-
u e inancing, which has subs an ially inc eased he
po en ial e u n on in es men ha in es o s ecei e,1
and i has also expanded he supply-side ma ke o
en u e inancing o p e iously unde se ed indi idual
in es o s. The c owd unding e olu ion con inues o
ha e a d ama ic impac on how en u es aise inanc-
ing, and i also challenges classic schola ly pa adigms
in he en ep eneu ial inance li e a u e, which esul ed
in some o he mos impac ul esea ch in economics
and managemen o he las decade (e.g., Ahle s e al.,
2015; Belle lamme e al., 2014; Mollick, 2014).
Decen alized Finance (DeFi) ma ke s o s a ups
a e he nex s age in he e olu ion o en ep eneu ial
inance (Bella i is e al., 2021; Block e al., 2021; Chen
& Bella i is, 2020; Khe e al., 2021). DeFi ma ke s
o s a ups e e o Dis ibu ed Ledge Technology-
(DLT-)based c owd unding (mo e commonly known
as oken o e ings o Ini ial Coin O e ings, ICOs, see
Fisch (2019), o seminal wo k), which u he econo-
mize on in e media ion cos s by eplacing c owd und-
ing pla o ms, such as Kicks a e , wi h sma con ac s
(Adhami e al., 2018; Bella i is e al., 2020,2021;
Benede i & Kos o e sky, 2021; Bo eiko & Ris eski,
2021; Campino e al., 2022; Chalme s e al., 2022;
Fisch, 2019; Fisch e al., 2021; Howell e al., 2020;
Mom az, 2020a,2021c). The e o e, we ollow Schu-
e el (2021) in de ining DeFi as a pa adigm o pee -
o-pee inancial se ice p o ision wi hou a cen al-
ized in e media y. Sma con ac s a e compu e p o-
ocols ha au oma e he exchange o in es o s’ inan-
cial con ibu ions o ICOs and okens ha o en ep e-
sen claims on en u es’ u u e asse s a a p ede ined
exchange a e. The only ee incu ing in he execu ion
o a sma con ac is he ee o ope a e he blockchain
ne wo k. Fo example, he a e age ansac ion cos on
he mos popula blockchain, E he eum, was less han
$2 du ing Janua y 2022, which educes ansac ion
cos s o c owd unding o a minimum.2
This pape a gues bo h heo e ically and empi ically
ha , despi e i s ansac ional e iciency, he ICO ma -
1VCs ypically cha ge pe o mance ees o 20% and annual
managemen ees be ween 1 and 2%.
2In compa ison, equi y-based c owd unding pla o ms ypically
cha ge ees a ound 7%.
ke is ela i ely ine icien wi h espec o “sea ch.”
Sea ch b oadly e e s o he p ocess o inding a ma ch-
ing ansac ion coun e pa y. The in ui ion is s aigh -
o wa d: DeFi ma ke s o s a ups a e e y g anu-
la ; ha is, hey ha e high le els o ma ke pa ici-
pa ion (anyone wi h in e ne connec i i y may pa ici-
pa e) and ma ke comple eness (e e y hing can be ok-
enized). Sma con ac s enable ha anyone can ade
any hing wi h anybody a almos no cos ( ansac ional
e iciency).The lipsideis, howe e , ha ma ke g anu-
la i y is p opo iona e o he equi ed sea ch e o (see,
o a ecen su ey among indi idual in es o s, An e
e al., 2022). Mo e indi idual agen s and aded p od-
uc s and se ices mean ha agen s wishing o ans-
ac ha e o sc een deepe ma ke s, which akes mo e
ime, in o de o a oid esou ce misalloca ions h ough
subop imal ansac ions (sea ch- ela ed ine iciency).
Thep oblemisplausiblypa icula ly p onouncedin he
ICO ma ke because DLT and sma con ac s p omo e
ma ke g anula i yand ma ke segmen a ion(Benede i
&Nikbakh ,2021), while heydono o e a echnolog-
ical solu ion o acili a e sea ch. Fo his eason, c i -
ics showcase he ICO ma ke o a gue ha pe ec ly
decen alized und aising is u opian gi en he pe a-
si e sea ch ic ions, and ha en ep eneu ial inance
may e e back o in e media ed ma ke s (e.g., Ze -
zsche e al., 2020).
Consequen ly, his pape aims o ad ance he li e a-
u e on sea ch in en ep eneu ial inance by add essing
he ollowing, o e a ching esea ch ques ion:
How e icien is he ICO ma ke and do sea ch
ic ions educe agg ega e ma ke e iciency?
The ques ion is impo an because he cu en s a e
o he li e a u e on ICOs is ipe wi h e iciency losses
due o ma ke design p oblems (e.g., Bella i is e al.,
2021; Ho nu e al., 2021;Mom az,2021c), bu ails
in la ge pa o p o ide an explana ion o why no el,
specialized in e media ies, so-called “c yp o unds,”
a e apidly en e ing he ICO ma ke (Fisch & Mom-
az, 2020). C yp o unds a e a blend o en u e-s yle
hedge unds ha pool e ail in es o s’ unds and chan-
nel hem h oughsophis ica ed adings a egies o ok-
enized s a ups in liquid seconda y ma ke s o okens.
C yp o unds plausibly ha e eme ged as a esponse o
sea ch ic ions in ma ke s o okens, which a e e y
p onounced due o he high le els o asymme ic in o -
ma ion in ICOs (Block e al., 2021; Bo eiko & Vidusso,
2019; Ze zsche e al., 2020). This esona es wi h an
123
1416 P.P. Mom az
es ablished li e a u e ha in e media ies ex ac en s
om educing sea ch ic ions in decen alized ma ke s
(Demse z, 1968; Rubins ein & Wolinsky, 1987).
O cou se, sea ch- ela ed a gumen s a e implici in
many exis ing wo ks in en ep eneu ial inance, and
no an inno a ion o his s udy. Howe e , o ou bes
knowledge, sea ch has ne e been explici ly modeled
in ex an heo y no es ed empi ically in he con ex o
en ep eneu ial inance, which is he p incipal con i-
bu ion we claim o his s udy. Fo ins ance, a as li -
e a u e examines signaling ( o a e iew, see Colombo,
2021), e.g., in IPOs (A hu s e al., 2009; Colombo
e al., 2019), c owd unding (Ahle s e al., 2015;Vis-
ma a,2018b,2016),andICOs (An e al., 2019; Beli ski
& Bo eiko, 2021; Bella i is e al., 2020;Fisch,2019;
Giudici & Adhami, 2019; Lee e al., 2022), as well
as adjacen a gumen s, such as in o ma ion cascades
(e.g., Visma a, 2018a). While all hese s udies implic-
i lyassumesea ch ic ions o be animpo an easonas
o why signaling is he p ime de e minan o success in
he compe i ion o en ep eneu ial inance, hey ne e
make sea ch ic ions explici ; in ac , none o hese
s udies men ions “sea ch” a all. Ano he example is
he li e a u e on ins i u ional in es men s in s a ups,
wi h a ocus on en u e capi al (Be oni e al., 2011;
Colombo e al., 2010) and ICOs (Fisch & Mom az,
2020). These s udies es whe he he e is a selec ion
e ec in he o m ha ins i u ional in es o s a e able
o pick s a ups wi h mo e a o able g ow h p ospec s
han non-ins i u ional in es o s. Again, a he oo o
he selec ion e ec is sea ch (i.e., ins i u ional in es o s
possess be e skills and mo e esou ces o sc een he
ma ke and nego ia e deals), albei he p ecise na u e
o sea ch in hese ma ke s is ne e made explici .
Ou eclec ic heo y d aws upon mul idisciplina y
sea ch heo y in decen alized ma ke s, inancial in e -
media ion, as well as asymme ic in o ma ion and lim-
i s o signaling li e a u es in en ep eneu ship o p o-
pose wo o e a ching hypo heses. The i s hypo he-
sis, he Decen alized Ine iciency Hypo hesis (DIH),
posi s ha sea ch- ela ed ic ions ende he ICO ma -
ke ela i ely ine icien . Speci ically, excessi e sea ch
in he ICO ma ke educes he ma ke ’s agg ega e e i-
ciency in a leas h ee dis inc ways. Fi s , sea ch ic-
ions imply ha he ICO ma ke in ol es wo-sided
ma ching: S a ups conduc ICO campaigns o a ac
in es o s and in es o s, in u n, sc een he ma ke o
iden i y a ac i e s a ups o in es men pu poses.
The ime o conduc ICO campaigns o en akes se -
e al mon hs (Mom az, 2020a), which is ime in which
s a up-in es o ma ches a e delayed. Thus, he i s
way in which sea ch ic ions impede ICO ma ke e i-
ciency is h ough a delay in oken alloca ions. Sec-
ond, sea ch is cos ly. I is cos ly o s a ups o ma -
ke ICO campaigns o in es o s, and i is cos ly o
in es o s (in e ms o bo h ime and inancial esou ces)
o pe o m a due diligence on po en ially in e es ing
in es men a ge s. These sea ch- ela ed cos s imply
ha some in es men s ha would be socially op imal in
a ic ionless economy do no ake place i sea ch cos s
exceed he an icipa ed ansac ion su plus. The e o e,
hesecondwayinwhichsea ch ic ionsimpedema ke
e iciency is in e ms o an agg ega e unde in es men
in high-quali y, okenized s a ups. Thi d, because he
ICOma ke ischa ac e izedbyhigh le els o asymme -
ic in o ma ion and he e a e limi s o signaling, he e
is subs an ial unce ain y in he ICO ma ke , which can
cause a misalloca ion o inancial esou ces o unde-
se ing okenized s a ups. One way o his o occu
is h ough ad e se selec ion (Ho nu e al., 2021)o
mo al haza d (Mom az, 2021c). Thus, he e iciency o
he ICO ma ke is also impeded by o e in es men s in
low-quali y, okenized s a ups.
The second hypo hesis, he In e media ed E i-
ciency Hypo hesis (IEH), posi s ha new DeFi in e -
media ies, in pa icula c yp o unds, inc ease ICO
ma ke e iciency by educing sea ch- ela ed ic ions.
In e media ies ha e long been known o ex ac ing
en s by educing sea ch ic ions in decen alized ma -
ke s (Demse z, 1968; Rubins ein & Wolinsky, 1987;
Schue el, 2021). C yp o unds de elop a compe i-
i e ad an age in sea ch h ough economies o scale
in c yp o-speci ic human capi al in es men s. C yp o
unds sc een he ma ke and in es in he bes s a ups,
signaling s a up quali y o he ma ke and ce i ying
p ojec legi imacy (Fisch & Mom az, 2020). They also
educe sea ch ic ions ela ed o pos -ICO in o ma ion
p oduc ion. Gi en he salien mani es a ions o mo al
haza d in he ICO ma ke (Ho nu e al., 2021;Mom-
az, 2021c), in es o s need o moni o s a ups pos -
unding and coo dina e collec i e ac ions agains shi k-
ing eams, which isp oblema ic o indi idualin es o s
because heymayno beable ode ec mani es a ionso
mo al haza d o coo dina e collec i e ac ions di ec ed
agains such beha io . C yp o unds no only ha e he
skills and esou ces o sea ch o indica o s o s a up
eams’ e o p o isions, he h ea o exi in liquid sec-
123
1417Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
onda y ma ke s o okens may p e en shi king and
o he o ms o de imen al beha io in he i s place.
Tes ing he DIH and IEH is challenging. The key
di icul y is ha bo h hypo heses a e ela ed o ICO
ma ke e iciency, which is a ela i e cons uc , and
a pe ec ly e icien ma ke is a coun e ac ual bench-
ma k ha is no obse ed in eali y. Fo his eason,
ou empi ical app oach is wo old. The i s empi ical
pa in ol es educed- o m eg ession analyses o wo
es able ela ions ha a e ela ed o ou o e all a gu-
men a ion ha he ICO ma ke has p onounced sea ch
ic ions, which in e media ion ia c yp o unds help
educe. Speci ically, we es (i) whe he in e media ed
ICOs a e mo e e icien in e ms o he ime i akes
o achie e he c owd unding goal, and (ii) whe he
en ep eneu s in non-in e media ed ICOs need o sell
hei okens a a discoun o a ac enough in es o s.
The wo empi ical ela ions a e no ee o endo-
genei y conce ns. Fo example, i is possible ha only
ICOs wi h s ong success p ospec s a e able o secu e
in e media ion se ices (selec ion e ec ), a he han i
is he in e media ion ha sho ens he ime- o- unding
o inc eases he oken alue ( ea men e ec ). To
his end, we employ se e al wo-s age and ins umen-
al a iable app oaches o disen angle he ue e ec s
o ICO in e media ion. The esul s sugges ha (i)
in e media ed ICOs achie e he c owd unding goal
25% as e and (ii) non-in e media ed ICOs ha e o
o e okens a a subs an ial discoun o 57%. These
esul s a e in line wi h anecdo al e idence, in pa -
icula ha non-in e media ed ICOs o e okens a
discoun s in he ange o 50 o 70%. The e o e, hese
esul s join ly sugges ha ICO in e media ion makes
he ma ke mo e e icien in e ms o ime- o- unding,
while ICO in e media ies plausibly a e able o ex ac
subs an ial en s o hei se ices.
In he second empi ical pa , gi en he challenging
na u e o he DIH and IEH, we jux apose he educed-
o m eg ession-based e idence wi h s uc u al es i-
ma es om a simple model o he ICO ma ke . The
model allows o es ima e he ma ke ’s agg ega e e i-
ciency, which is a no el y in he en ep eneu ial inance
li e a u e. In he model, he e a e indi idual in es o s
and in e media ies. Only indi idual in es o s enjoy a
u ili y om holding okens (because he oken owne -
ship enables hem, o ins ance, o pa ake in an online
gaming communi y), while in e media ies ha e no u il-
i y om holding okens, bu hey ex ac en s om
ading okens. S a up i ms a e he e ogeneous in he
model wi h espec o hei unde lying pla o m sizes.
In ui i ely, okens o la ge pla o ms a e mo e alu-
able han okens o smalle pla o ms. We calib a e
he model wi h ac ual ICO ma ke da a co e ing he
2017–20 pe iod. The model p edic s se e al agg ega e
quan i ies e y well. Consis en wi h ou educed- o m
es ima es and indings in ela ed s udies (e.g., Bella i-
is e al., 2021; Fisch and Mom az, 2020), he s uc-
u al es ima ion o he model shows ha in e media ies
help educe ading delays and ha sea ch cos s a e
p onounced in he ICO ma ke . Impo an ly, indi idual
selle s and buye s sha e he ansac ion su plus mo e
equally in non-in e media ed ICOs han in in e medi-
a ed ICOs, in which in e media ies pocke he ansac-
ion su plus almos exclusi ely.
O e all, he model-implied es ima es sugges ha
he ICO ma ke c ea es onlye one- hi d o one- i h o
hewel a ei couldpo en iallyc ea ei i we epe ec ly
e icien , wi h he loss s emming om sea ch- ela ed
ine iciency.
Theo e ical con ibu ions, p ac ical implica ions,
limi a ions and a enues o u u e esea ch a e dis-
cussedinSec .8.P eceding ha ,wep o idesomeins i-
u ional backg ound on DeFi, ICOs, and c yp o unds
in Sec .2, de i e o e a ching hypo heses in Sec .3,dis-
cuss da a and eg ession esul s in Sec s.4and 5, he
o mal model in Sec .6, and he s uc u al es ima ion
in Sec .7.
2 Ins i u ional backg ound: DeFi, ICOs, and
c yp o unds
2.1 Decen alized inance (DeFi) and he pu sui o
disin e media ion
DeFi ma ke s may ha e se e al ad an ages o e adi-
ional inance ma ke s. Fi s , DeFi may imp o e ma -
ke pa icipa ion. Mo e indi iduals and small en e -
p ises may gain (equi able) access o inance because
DeFi educes he en y ic ions, such as, o example,
h ough a mi iga ion o local bias in en u e inanc-
ing (So enson e al., 2016) o lending (Becke , 2007).
Second, DeFi may make ma ke s mo e comple e by
acili a ing inancial inno a ions. This could be spu ed
by he open-sou ce cha ac e o DeFi, pai ed wi h i s
lack o bo de s and ocus on in e ope abili y s anda ds
(Ha eye al.,2021).Thi d,DeFip omisesasigni ican
educ ion in ansac ion cos s s emming om mul iple
123
1418 P.P. Mom az

sou ces (Gao & Li, 2021). Fo example, disin e me-
dia ion inc eases he sha e o he ansac ion su plus
om which ansac ion pa ies can exclusi ely bene i ,
he anspa ency o public ledge s signi ican ly educes
audi ing cos s, and he de e minis ic and us less cha -
ac e o sma con ac s minimizes he execu ion isk.
A he same ime, DeFi has ye o add ess a numbe
o no el and idiosync a ic isks ha all b oadly in o
wo ca ego ies: in a-p o ocol and in e -p o ocol isks.
In a-p o ocol isks include consensus ailu es, such as
51% a acks on P oo -o -Wo k (PoW) blockchains and
alida o ca els on P oo -o -S ake (PoS) blockchains,
as well as in a-p o ocol a bi age on au oma ed ma ke
make (AMM) exchanges, known as mine ex ac ed
alue (MEV) (Daian e al., 2020). In e -p o ocol isks
include so-called o acle a acks, in which biased o
ake ou side in o ma ion is ed in o sma con ac s,
and “ lash loans” ha pa e he way o in e -con ac
a bi age (Wang e al., 2021).3Bo h in a- and in e -
p o ocol isks ha e a common a ibu e in ha hey
ep esen echnical ulne abili ies ha a e ex emely
di icul o indi idual pla o m use s o de ec o
e en unde s and. The e o e, hese isks dis inguish
DeFi om in e media ed inancial ma ke s. The conse-
quences o hese isks may be salien in c owd unding
ma ke s, such as he ICO ma ke , because indi idual
backe s may no possess he echnological knowledge
o adequa ely e alua e he no el p o ocol isks.
2.2 The DeFi ma ke o s a ups: ini ial coin
o e ings (ICOs)
Token o e ings (o ini ial coin o e ings, ICOs) a e an
en ep eneu ial inance mechanism ha sha es some
common ea u es wi h c owd unding, en u e capi al,
and ini ial public o e ings ( o an excellen ecen
e iew, see B ochado and T oilo, 2021). Speci ically,
ICOs ha e e ol ed om c owd unding by employing
DLT o bo h issue and exchange s akes in s a up i ms
(Bella i ise al., 2021;Fisch,2019;Howelle al.,2020;
Mom az, 2020a). ICOs a e pee - o-pee s a up inanc-
ing ansac ions ha ely on sma con ac s o au o-
ma e us less ansac ions be ween en ep eneu s and
in es o s (Fisch e al., 2022; Rawhouse e al., 2023).
In an ICO, an en ep eneu aises en u e inancing
3See Ca e and Jeng (2021) o an o e iew o DeFi p o ocol
isks.
by selling c yp og aphically p o ec ed digi al asse s,
known as okens o coins, o in es o s. Tokens can ep-
esen di e en ypes o alue and igh s. C yp ocu -
ency okens a e me e mediums o exchange, such as
Bi coin; secu i y okens may include o ing and con-
ol igh s; and u ili y okens a e paymen ins umen s
(Howelle al.,2020; Lambe e al., 2021). U ili y
okens a e he mos equen ly issued oken ype in
ICOs (Bella i is e al., 2020), hough de elopmen s in
ICO egula ion ha e ini ia ed a g adual shi o secu i y
oken o e ings (Lambe e al., 2021). U ili y okens
a e ouche -like asse s ha can be edeemed o one
uni o he en u e’s u u e p oduc o se ice. The
eliance on DLT means ha ICO in es o s equi e no
only business skills bu also a g ea deal o echnolog-
ical knowledge (Bella i is e al., 2021;Fisch,2019).
Unlike o he en ep eneu ial inancing mechanisms,
ICOs in eg a e he ull spec um o “ icke sizes”, ang-
ing om mic o-cap ICOs (<$100,000) o mega-cap
ICOs (>$1,000,000, such as he EOS campaign, wi h
mo e han $4 billion aised).
The i s ICO (Mas e Coin) ook place in July 2013,
and he ma ke has s eadily e ol ed since hen. Figu e1
shows he e olu ion o he ma ke o oken o e ings
o e he 2017–2020 pe iod. Du ing ha ime, oughly
5,500 oken o e ings we e comple ed, wi h he majo -
i y in 2018.
Bella i is e al. (2021) es ima e ha 2,598 oken
o e ings aised an agg ega e unding amoun o $12.3
billion in 2018 alone.
U ili y okeno e ings a eo en hough o bepe ec ly
disin e media ed pee - o-pee ansac ions and issued
okens a e ypically aded pos -ICO in liquid sec-
onda y ma ke s. Sma con ac s allow en ep eneu s
and in es o s o au oma e he ansac ion in a us -
less way, he eby edis ibu ing he ansac ion su plus
exclusi ely o en ep eneu s and in es o s; in con as ,
in e media ies in c owd unding o ini ial public o e -
ings ypically cha ge a ee o 5–7%. Disin e medi-
a ion migh also democ a ize en ep eneu ial inance
ma ke s by lowe ing bo h supply- and demand-side
en y ba ie s (Bu icé & Visma a, 2022; Fisch e al.,
2022; Meoli e al., 2022; Rawhouse e al., 2023), lead-
ing o mo e comple e ma ke s wi h highe pa icipa-
ion. Mo eo e , because okens can be aded a close-
o-ze o ansac ion cos s and limi ed ading delays
h ough DLT, ICO a e ma ke s a e highly liquid. Liq-
uid pos -ICO oken exchange ma ke s educe s a up
i m discoun s associa ed wi h illiquidi y (Ba g e al.,
123
1419Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
Fig. 1 E olu ion o he
ICO ma ke
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
2017 2018 2019 2020 2021
# ICOs
2021) and p o ide in es o s wi h apid exi oppo -
uni ies (Fisch & Mom az, 2020;Mom az,2020a).
Facili a ed ades o s akes in s a ups migh make
en ep eneu ial inance ma ke s mo e e icien (e.g.,
by means o ( ai ) oken alua ions ob ained om
equilib ium p ices in liquid oken exchange ma ke s
ha a e in o ma i e o he ma ke ; see Mom az,
2021c), he eby imp o ing capi al alloca ion o he
bes en ep eneu ial p ojec s and po en ially p omo ing
long- e m economic g ow h (Acs & Sze b, 2007;
Aud e sch, 2018; Bella i is e al., 2020).
Empi ical wo ks a e mos ly conce ned wi h suc-
cess de e minan s o oken o e ings (e.g., Adhami e
al., 2018; Beli ski and Bo eiko, 2021; Bella i is e al.,
2020; Bella i is e al., 2021;Fisch,2019; Giudici and
Adhami, 2019; Ho nu e al., 2021;Howelle al.,2020;
Mom az, 2020a). The oles o sea ch, in e media ion,
and agg ega e oken ma ke e iciency ep esen a oid
in he li e a u e.
Fig. 2 E olu ion o
in e media ion in he ICO
ma ke
0
50
100
150
200
250
300
350
2017 2018 2019 2020
# In e media ed ICO
123
1420 P.P. Mom az
2.3 New in e media ies en e ing he ICO ma ke :
c yp o unds
S uc u al p oblems in he ICO ma ke , such as he
sys ema ic mani es a ion o mo al haza d (Ho nu e
al., 2021;Mom az,2021c) and egula o y (Cumming
e al., 2019) and in o ma ional ic ions (Bou eau e
al., 2022), ha e led o he eme gence o a no el, spe-
cialized in e media y: he “c yp o und.” Mos c yp o
unds esemble en u e capi al o hedge unds, wi h he
impo an di e ence ha hey ade in “non-secu i ies.”
The numbe o c yp o unds is apidly g owing. Mo e
han 800 c yp o unds a e ac i e and ha e agg ega e
asse s unde managemen o he amoun o $57 bil-
lion in he i s qua e o 2021. The a e age c yp o
unds g oss e u n in he i s qua e o 2021 was 98%,
sligh ly below Bi coin’s 103%. Like mos hedge unds,
a la ge po ion o c yp o unds a e domiciled in he
B i ish Vi gin Islands o Cayman Islands o ax, legal,
o o he egula o y easons, al hough hal o hem hold
p ima y o ices in he U.S.4Figu e2illus a es he pen-
e a ion o he ma ke o oken o e ings by c yp o
unds o e he 2017–2020 pe iod.
C yp o unds a e an in iguing asse class because
hey di e om adi ional en u e capi al unds in se -
e al impo an ways. Fi s , c yp o unds mos ly ade
in non-secu i ies, a oiding much o he egula ion a-
di ional unds ace. T ading in non-secu i ies la gely
exemp s hem om heIn es men Company Ac ,which
enables hem o ca e o a new ma ke o small and indi-
idual in es o s, who a e no acc edi ed o quali ied in
helegalsense(Mokh a ian&Lindg en,2018).Indeed,
c yp o unds a ac small in es o s wi h signi ican ly
lowe minimum in es men equi emen s. Acco ding
o C yp o Fund Resea ch (2021), he median mini-
mum und in es men amoun s o $100,000. Mo eo e ,
c yp o unds a e la gely exemp ed om he Ad ise s
Ac . This li s limi s on pe o mance ees ha can be
cha ged o small in es o s, making c yp o unds mo e
inanciallya ac i e(albei aisingconce nsabou mis-
alignmen o incen i es). Second, DLT sa es c yp o
unds ime and ees ha would o he wise be incu ed
o hi d-pa y cus odians pu suan o he Ad ise s Ac .
Thi d, wi h some excep ions, okens a e axable only
in he case o “ ecogni ion e en s,” i.e., i hey a e
exchanged o ia money. This allows in es o s o op i-
4See h ps://c yp o und esea ch.com/q1-2021-c yp o- und-
epo /.
mize bo h he iming and he amoun o hei pe sonal
ax liabili ies in coo dina ion wi h hei o e all po o-
lios (Mokh a ian & Lindg en, 2018). Finally, he liq-
uidi y o okens li s en u e capi al unds’ bu den o
iden i y and in es in “unico ns” o compensa e o he
ela i ely la ge numbe o ailed p ojec s, because liq-
uid oken ma ke s allow c yp o unds o exi a any ime
(Kas elein, 2017).
3 Theo y and hypo heses
3.1 In ui ion
Be o ede eloping ou concep ual amewo kmo e o -
mally, we p e ace he heo e ical discussion by s a ing
ou wo o e a ching hypo heses and p o iding some
in ui ion behind hem. The i s hypo hesis, he Decen-
alized Ine iciency Hypo hesis (DIH), posi s ha
sea ch- ela ed ic ions ende he ICO ma ke ela-
i ely ine icien . The second hypo hesis, he In e -
media ed E iciency Hypo hesis (IEH), posi s ha
new DeFi in e media ies, in pa icula c yp o unds,
inc ease ICO ma ke e iciency by educing sea ch-
ela ed ic ions. The in ui ion is simple: he ICO ma -
ke is e y g anula ; ha is, i has high le els o ma -
ke pa icipa ion (anyone wi h in e ne connec i i y
may pa icipa e) and ma ke comple eness (e e y hing
can be okenized).5Sma con ac s enable anyone o
ade any hing wi h anybody a almos no cos ( ans-
ac ional e iciency). Howe e , sma con ac s do no
p o ide echnological solu ions o acili a e sea ch-
ing o ma ching ansac ion coun e pa ies. The e o e,
because he numbe o ading agen s and aded claims
po en ially eaches a maximum in he ICO ma ke and
agen s bea he bu den o inding he agen wi h he pe -
ec lyma chingclaim o ade, heICOma ke mayno
achie e i s wel a e po en ial when many socially op i-
mal ades do no occu i he expec ed ansac ion su -
plus does no compensa e o he expec ed sea ch cos s
(sea ch- ela ed ine iciency). In he ICO ma ke , hese
sea ch ic ions a e plausibly e en mo e p onounced
due in la ge pa o he highly asymme ic in o ma ion
and helimi s oe ec i esignaling(Ho nu e al.,2021;
Mom az, 2021c). The e o e, a pe ec ly decen alized
5O cou se, he e a e limi a ions o okenized ma ke pa icipa-
ion and okeniza ion, such as he echnical sophis ica ion ha is
equi ed om indi iduals.
123
1421Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
ICOma ke maybe ela i elyine icien (i.e., heDIH),
and ein oducing a ce ain deg ee o in e media ion
imp o es hema ke ’so e alle iciency (i.e., he IEH).
The ollowing sec ion in oduces he building blocks
o ou heo y and o mally de i es he hypo heses.
3.2 Sea ch- ela ed ic ions and ICO ma ke e iciency
Sea ch- ela ed ic ions e e o economic cos s s em-
ming om ma ke impe ec ions ha impede he e i-
cien ma ching o ansac ion coun e pa ies in decen-
alized ma ke s (e.g., Du ie e al., 2005). As such,
sea ch ic ions a e p opo iona e o he deg ee o ma -
ke decen aliza ion. In p inciple, ma ke ailu e may
occu i sea ch cos s exceed he wel a e a ising om
he exchange o asse s (Weill, 2020). The e o e, he
p obabili y o ma ke ailu e inc eases in he deg ee
o ma ke decen aliza ion. As we discuss in Sec .3.3
below, decen alized ma ke s ha ace salien sea ch-
ela ed ine iciency o en e e back o in e media ed
ma ke mic os uc u es, in which in e media ies o e
se ices ha educe sea ch ic ions (Ga azza, 2016;
Rubins ein & Wolinsky, 1987). The ollowing explains
why sea ch ic ions in he ICO ma ke a e plausibly
e yp onounced.ICO-speci icsea ch ic ionsinclude,
in e alia, (1) p o ocol-in e ace isks, (2) p o ocol-
immanen isks, (3) sma -con ac isks, (4) o acle
isks, and (5) go e nance- ela ed isks (Ha ey e al.,
2021).
The ICO ma ke is p one o sea ch ic ions by
design, la gely because i imp o es on bo h ma ke pa -
icipa ion and ma ke comple eness. Ma ke pa icipa-
ion e e s o henumbe o agen s ha canaccessama -
ke . DLT has signi ican ly lowe ed he en y ba ie s o
en ep eneu ial inance ma ke s, in e alia, h ough a
d ama ic educ ion o he ansac ion cos s o c owd-
unding campaigns (demand-side en y ba ie ) and a
educ ion in he minimum in es men amoun hanks
o ac ional oken owne ship (supply-side en y ba -
ie ) (Bella i is e al., 2021;Fisch,2019; Huang e
al., 2020; Lambe e al., 2021; Ze zsche e al., 2020).
Indeed, Fisch e al. (2022) epo ha he ICO ma ke
has democ a ized en ep eneu ial inance, e idenced,
e.g., by he inc eased numbe o in es o s om e hnic
mino i ies (see, also, Bu icé and Visma a, 2022; Meoli
e al., 2022; Rawhouse e al., 2023). Ma ke comple e-
ness e e s o he a ie y o asse s in a ma ke . Asse
he e ogenei yalsoc ea es sea ch p oblems becausei is
p opo iona e o he in es o s’ e o equi ed o de e -
mine he ela i e i o a ocal asse in he ligh o an
in es o ’s subjec i e p e e ences (Rubins ein & Wolin-
sky, 1987). Sma con ac s ha e inc eased asse he -
e ogenei y subs an ially because hey allow he ok-
eniza ion o any claim. Fo example, Fisch and Mom-
az (2020) epo ha he ICO ma ke ’s demand side
is e y compe i i e, wi h o en mo e han 1,000 com-
pe ing oken o e ings p esen a he same ime. Gi en
his high in ensi y, i is e iden ha ma ke comple ion
exace ba es sea ch ic ions.
The high le el o asymme ic in o ma ion in he
ICO ma ke u he agg a a es he sea ch p oblem
o in es o s (Bella i is e al., 2020; Block e al.,
2021;Fisch,2019; Ho nu e al., 2021). Asymme ic
in o ma ion is a pe asi e p oblem in en ep eneu ial
inance (Colombo e al., 2019;Visma a,2018b). A
i s co e, he p oblem wi h asymme ic in o ma ion is
ha inancial in es o s lack he in o ma ion o gauge
he ue quali y o an in es men , esul ing in equi-
lib ium p ices ha a e based on he popula ion a e -
age ins ead o a mo e disc imina o y p icing mecha-
nism based on he unde lying in es men alue (Jensen
& Meckling, 1976; Leland & Pyle, 1977). Conse-
quen ly, high-quali y in es men s could sell a a dis-
coun , de e ing issue s om pu ing hose in es men
oppo uni ies on he ma ke en i ely, which may c e-
a e a ma ke o lemons (Ake lo , 1978). In o ma ional
asymme ies a e salien in he ICO ma ke , in e alia,
because blockchain-sa y en ep eneu s a e ypically
young and lack a ack eco d (An e al., 2019;Fisch,
2019); he okens sold a e o ye unde eloped, u u e
p oduc s (Fisch, 2019;Howelle al.,2020;Mom az,
2020a); he e a e li le manda o y disclosu e laws (see
Bella i is e al., 2021; Bo eiko e al., 2019); and oken
issue s a e known o embellish he in o ma ion dis-
closed in ICO whi epape s (Mom az, 2021c). These
p oblems inc ease agg ega e unce ain y in he ICO
ma ke , which acco dingly exace ba es sea ch- ela ed
ine iciency.
Finally, sea ch- ela ed ine iciency is also pa ly
d i en by he limi s o signaling in he ICO ma ke .
Se e al s udies a gue ha he absence o an ins i u-
ional amewo k o ICOs may c ea e a mo al haza d
in signaling (Ho nu e al., 2021;Mom az,2021c). Fo
example, Mom az (2021c, p. 2) a gues ha “issue s
123
1422 P.P. Mom az
in Sec .5.2, and o he impac o non-in e media ion
on oken alue discoun s in Sec .5.3.
5.1 Econome ic speci ica ion: wo-s age app oaches
o mi iga e endogenei y
Se e al wo-s age app oaches help mi iga e conce ns
abou po en ial endogenei y pe aining o sample selec-
i i y,9In pa icula , ou app oach o debias he ea -
men e ec s o in e media ion mi iga es selec ion
based on unobse ables, which a e po en ially p o-
nounced con ounde s in en ep eneu ial inance.10
The gene al app oach is o es ima e a i s -s age
model ha p edic s he p obabili y ha an ICO is
in e media ed o non-in e media ed. These “selec ion
p obabili ies” a e hen ans o med and included in
he second-s age models, which es ima e he ea men
e ec s. Speci ically, we a e in e es ed in wo dis inc
ea men e ec s; namely, he ea men e ec o in e -
media ionon he ime i akesICO i ms oachie e hei
c owd unding goals and he ea men e ec o he lack
o in e media ion on oken alua ion. Equa ions1and 2
ep esen he po en ially con ounded OLS models:
ICO du a ioni=β×1In e media edi+Ωiγ+εi(1)
Token alua ioni=β×1Non-in e media edi+Ωiγ+εi(2)
whe e iindexes ICOs, 1In e media ediand 1Non-in e me-
dia edi ep esen indica o a iables o whe he o no
ICO iis in e media ed, and i ep esen s a ec o o
con ol a iables.
Fo he models in Eqs. 1and 2 o es ima e an unbi-
ased ea men e ec o ICOs’ in e media ion s a us, i
is necessa y o assume ha he independen a iables
a e o hogonal o he e o e m, i.e., E[i,ε
i]=0.
This condi ion is iola ed i selec i i y is p esen ; o
example, in he case ha only high-quali y ICOs a e
able o secu e in e media ion se ices. The e o e, he
9The echniques used in ou s udy ha e been employed be o e
in simila con ex s (e.g., Fisch and Mom az 2020; Be oni e al.,
2011; Colombo e al., 2010; Colombo e al., 2021; Cumming e
al., 2023).
10 Fo example, unobse ed he e ogenei y in s a ups’ ime- o-
unding by en u e capi alis s can be so p onounced ha i biases
common ime- o-e en models (Mom az, 2021b)
i s s age explici ly models he selec i e ma ching
be ween ICO i ms and in e media ies. Speci ically, we
p edic ha ICO iis in e media ed, 1In e media edi,by
a ec o o exogenous con ol a iables ha possibly
in luence he selec ion mechanism, Ω(s)
i:
1In e media edi=Ω(s)
iδ+ξi(3)
Es ima es om Eq. 3help con ol o unobse ed
he e ogenei y as ollows. We use Gene alized Residu-
als (GRs) bo h as explici con ols o selec ion-based
endogenei y, and as ins umen al a iables o ICOs’
in e media ion s a us (Gou ie oux e al., 1987), which
con ols o unobse ed he e ogenei y by explici ly
modeling any endogenei y in he e o e m, de ined
as ollows:
GRi=1In e media edi×
φ−Ω(s)
iδ
1−−Ω(s)
iδ
+(1−1In e media edi)
×
−φΩ(s)
iδ
−Ω(s)
iδ(4)
whe e φ(.)and (.) deno e he p obabili y densi y
and he cumula i e densi y unc ions o he s anda d
no mal dis ibu ion, espec i ely. We es ic he s an-
da d de ia ion o he e o e m o in e media ed
(σε, 1In e media edi) obeequal o ha o non-in e media ed
ICOs (σε,1Non-in e media edi). The es ic ion ensu es ha
GRican be added as an ins umen al a iable o Eq. 1.
Fo Eq. 2, he adjus men s a e iden ical, wi h he
excep ion ha “selec ion p obabili ies” a e es ima ed
o he case ha ICO iis no in e media ed, i.e.,
1Non-in e media edi.
5.2 Reg ession esul s: in e media ion and ICO
du a ion
Table 3shows he eg ession esul s o how ICO in e -
media ion impac s ICO du a ion. All models include
qua e -yea and coun y ixed e ec s o abso b bo h
ime- ela ed and geog aphical a ia ion. All epo ed
s anda d e o s a e obus . The selec ion model o
Eq. 3is in column (1), wi h an indica o a iable o
123
1429Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...

Table 3 Two-s age analysis o sea ch ic ions in in e media ed s. non-in e media ed ICOs
S age: 1s 1s 2nd 2nd
Model: Selec ion Con ol GR IV
Dependen a iable: 1In e media ed Du a ion Du a ion Du a ion
Key a iable:
1In e media ed −0.273*** −0.273*** −0.287***
(0.076) (0.077) (0.079)
ICO- ela ed con ols:
E he eum 0.030 −0.032 −0.032 −0.032
(0.041) (0.069) (0.069) (0.069)
P e-sale −0.013 0.090* 0.090* 0.090*
(0.030) (0.050) (0.050) (0.050)
# compe ing o e ings (log.) −0.098** 2.616*** 2.616*** 2.615***
(0.039) (0.066) (0.066) (0.066)
So cap, in $ mil. (log.) −0.001 −0.268*** −0.268*** −0.268***
(0.041) (0.068) (0.068) (0.068)
Ha d cap, in $ mil. (log.) −0.006 0.007 0.007 0.007
(0.009) (0.015) (0.015) (0.015)
Whi elis 0.036 −0.038 −0.038 −0.038
(0.026) (0.044) (0.044) (0.044)
KYC −0.007 −0.027 −0.027 −0.027
(0.033) (0.054) (0.054) (0.054)
Team- ela ed con ols:
Expe a ing 0.023 −0.015 −0.015 −0.014
(0.026) (0.043) (0.043) (0.043)
# eam membe s 0.001 −0.001 −0.001 −0.001
(0.002) (0.003) (0.003) (0.003)
% eam wi h echnical deg ee 0.207*** −0.154 −0.154 −0.151
(0.078) (0.130) (0.131) (0.130)
% eam wi h yp o expe ience 0.008 0.042 0.042 0.042
(0.064) (0.106) (0.106) (0.106)
% eam wi h Ph.D. 0.348* −0.784*** −0.784*** −0.779***
(0.177) (0.295) (0.297) (0.296)
Popula ion weigh s ✓✓✓✓
Gene alized Residuals ✗✗✓✗
Ins umen al Va iable ✗✗✗✓
Coun y ixed e ec s ✓✓✓✓
Qua e -yea ixed e ec s ✓✓✓✓
# Obs. 567 567 567 566
Adjus ed R20.018 0.823 0.823 0.823
No e: *, **, and *** deno e s a is ical signi icance a he 10%, 5%, and 1% le els, espec i ely
whe he an ICO is in e media ed as he dependen a i-
able. Ou baseline eg ession esul s o Eq. 1a e p e-
sen ed in column (2), wi h he na u al loga i hm o he
imebe weenICOs a andendindaysas hedependen
a iable. Thesecond-s age eg ession esul sa eshown
in columns (3) and (4), wi h he gene alized esidual,
as measu ed in Eq. 4, as an added con ol and as an
ins umen al a iable o he in e media ion indica o ,
123
1430 P.P. Mom az
espec i ely.Inpa icula , heseapp oachesou pe o m
ma ching-on-obse ables in ou con ex because hey
do no equi e ICO in e media ion o be independen
o unobse ed ac o s and he ma ginal p obabili y o
ICO in e media ion does no need o equal he a e -
age p obabili y. Finally, gi en he sample popula ion
di e ences epo ed in Table 1, we weigh all eg es-
sion coe icien s by he in e se o he (absolu e alue)
sum o ela i e de ia ions om popula ion means o
each oken o e ing o mo e ou es ima es o he local
ea men e ec in ou sample close o he a e age
ea men e ec in he popula ion o ICOs.
The esul s sugges ha he e ec o ICO in e medi-
a ion on ICO du a ion is signi ican ly nega i e h ough-
ou all models in columns (2) o (4). The coe icien s
ange om −0.273 o −0.287, s a is ically highly sig-
ni ican wi h p- alues consis en ly below 1%. The sim-
ila i y o hese coe icien s may imply ha selec i -
i y does no signi ican ly bias he causal e ec o ICO
in e media ion on ICO du a ion. Indeed, he adjus ed
R2 o he selec ion model in column (1) is ela i ely
low (1.8%). O e all, he ma ginal e ec o ICO in e -
media ion on ICO du a ion is a ela i e dec ease in he
ime i akes o in e media ed ICO i ms o achie e
hei c owd unding goals o −24.9% (exp(−0.287)–
1). This s ongly suppo s he o e a ching hypo hesis
ha in e media ies help educe sea ch ic ions in he
ICO ma ke .
Fo he selec ion model, he coe icien s o he con-
ol a iables la gely show plausible e ec s, sugges ing
ha (i) he numbe o compe ing o e ings is nega i ely
associa ed wi h he p obabili y o ICO in e media ion,
while (ii) he ela i e amoun o eam membe s wi h a
echnical backg ound o (iii) a Ph.D. deg ee ha e pos-
i i e e ec s.
Fo he key models in columns (2) o (4), he coe -
icien s o he con ol a iables also appea plausible,
sugges ing ha (i) p e-sales and (ii) he numbe o com-
pe ing o e s inc ease he ime i akes ICO i ms o
achie e he c owd unding goal, while (iii) he size o
he so cap and (i ) he pe cen age o eam membe s
wi h a Ph.D. deg ee dec ease he du a ion. I is no e-
wo hy ha he coe icien s o he con ol a iables a e
consis en in e ms o bo h hesignsand he magni udes
ac oss columns (1) o (3).
O e all, he esul s in Table 3sugges ha ICO i ms
ha e an economic mo i e o secu e in e media ion o
hei o e ings o educe hei ime- o- unding.
5.3 Reg ession esul s: lack o in e media ion
and discoun on oken alue
Table 4p esen s eg ession esul s o he e ec o
ICO in e media ion on oken alua ions. In pa icula ,
we es whe he okens o e ed in non-in e media ed
ICOs ade a a ela i e discoun . The es s a e sim-
ila o hose in Table 3, wi h he di e ence ha he
dependen a iables a e eplaced o an indica o a i-
able o non-in e media ed ICOs in column (1) and he
na u al loga i hm o oken alua ions in $ million in
columns (2) o (4). The models also include qua e -
yea and coun y ixed e ec s, and he s anda d e o s
a e obus . Columns (3) and (4) show second-s age
eg essions ha con ol o unobse able he e ogene-
i y by con olling o he gene alized esiduals in col-
umn (3) and o ins umen ing he indica o a iable o
non-in e media ed ICOs wi h he gene alized esiduals
in column (4).
The esul s sugges ha okens o e ed in non-
in e media ed ICOs ade a a signi ican discoun .
Speci ically, he a e age non-in e media ed ICO o e s
okens a a discoun o up o −57.7% (exp(−0.861)–1)
in he IVmodel incolumn (4).Thecoe icien es ima es
o ou key a iable anges om −0.793 o −0.861,
which a e highly s a is ically signi ican , wi h p- alues
consis en ly below 1%. I is no ewo hy ha he key
coe icien o he con ol model (−0.793) is clea ly
di e en om ha in he GR and IV models (−0.850
and −0.861, espec i ely), indica ing ha he selec ion
o in e media ed s. non-in e media ed ICOs would
unde es ima e he oken alua ion e ec in he absence
o he adjus men s in columns (3) and (4).
No e ha he adjus ed R2is simila ly high as in ela ed
s udies (Bella i is e al., 2020;Fisch,2019), and he
con ol a iables also seem o be consis en . Speci i-
cally, (i) expe a ings, (ii) so cap amoun s, (iii) ha d
cap amoun s, (i ) whi elis s, ( ) he numbe o ICO
eam membe s, and ( i) he ela i e amoun o eam
membe swi hp io c yp o indus y expe iencea e pos-
i i ely ela ed o ICO oken alua ions. In con as , ( ii)
only he pe cen age o echnical eam membe s has a
nega i e e ec . These es ima es a e consis en ac oss
all model speci ica ions in columns (2) o (4).
O e all, he non-in e media ed ICOs o e okens a
d ama ic ce e is pa ibus discoun s o up o −57.7%,
123
1431Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
Table 4 Two-s age analysis o alua ion discoun in non-in e media ed ICOs
S age: 1s 1s 2nd 2nd
Model: Selec ion Con ol GR IV
Dependen a iable: 1No in e media ed Valua ion Valua ion Valua ion
Key a iable:
1No in e media ed −0.793*** −0.850*** −0.861***
(0.262) (0.262) (0.274)
ICO- ela ed con ols:
E he eum −0.030 −0.031 −0.011 −0.033
(0.041) (0.236) (0.236) (0.237)
P e-sale 0.013 −0.207 −0.250 −0.206
(0.030) (0.172) (0.173) (0.172)
# compe ing o e ings (log.) 0.098 −0.293 −0.278 −0.287
(0.039) (0.226) (0.225) (0.227)
So cap, in $ mil. (log.) 0.001 0.561** 0.571** 0.562**
(0.041) (0.233) (0.232) (0.233)
Ha d cap, in $ mil. (log.) 0.006 0.431*** 0.439*** 0.432***
(0.009) (0.051) (0.051) (0.051)
Whi elis −0.036 0.253* 0.268* 0.250*
(0.026) (0.150) (0.150) (0.151)
KYC 0.007 −0.083 −0.068 −0.082
(0.033) (0.187) (0.187) (0.188)
Team- ela ed con ols:
Expe a ing −0.023 0.347** 0.348** 0.345**
(0.026) (0.149) (0.148) (0.149)
# eam membe s −0.001 0.038*** 0.036*** 0.038***
(0.002) (0.010) (0.010) (0.010)
% eam wi h echnical deg ee −0.207*** −1.533*** −1.663*** −1.547***
(0.078) (0.448) (0.450) (0.448)
% eam wi h c yp o expe ience −0.008 0.970*** 0.993*** 0.970***
(0.064) (0.365) (0.364) (0.366)
% eam wi h Ph.D. −0.348* 1.242 1.420 1.218
(0.177) (1.018) (1.018) (1.019)
Popula ion weigh s ✓ ✓✓✓
Gene alized Residuals ✗✗✓(-**) ✗
Ins umen al Va iable ✗ ✗✗✓
Coun y ixed e ec s ✓ ✓✓✓
Qua e -yea ixed e ec s ✓ ✓✓✓
# Obs. 567 567 567 566
Adjus ed R20.018 0.264 0.269 0.262
No e: *, **, and *** deno e s a is ical signi icance a he 10%, 5%, and 1% le els, espec i ely
123
1432 P.P. Mom az
whichis consis en wi hanecdo ale idence ha epo s
discoun s in he ange o 50 o 70%. This implies ha he
a e age non-in e media ed ICO issue “lea es money
on he able” o he amoun o $5.34 million. By impli-
ca ion, he causal na u e o hese esul s sugges s ha ,
in compe i i e DeFi ma ke s, in e media ies can cha ge
e y high ees commensu a e wi h he es ima ed dis-
coun s o hei in e media ion se ices, an in e p e a-
ion ha we will use o iden i y ou heo e ical model
below.
To summa ize he educed- o m e idence in Sec s.5.2
and 5.3, ICO in e media ion is causally ela ed o he
ime i akes ICO i ms o achie e hei c owd und-
ing goals and oken alua ion discoun s, espec i ely.
The e o e, he esul s o e ini ial suppo ing e idence
o he o e a ching conjec u es ha in e media ion
helps educe sea ch ic ions in ICO ma ke s, and in e -
media ies can cha ge subs an ial ees o hei se ices.
Table 5 Pos hoc analyses: second-s age eg ession esul s o di e en subsamples
Dependen a iable: Funding amoun , in USD (log.) ICO du a ion, in days (log.)
Panel A: Ma ke phases
Bull ma ke Bea ma ke Bull ma ke Bea ma ke
1No in e media ed −1.637∗∗ −0.713∗∗ 0.056 0.282∗∗∗
(0.720) (0.287) (0.391) (0.064)
Panel B: Team quali y
Abo e-median Below-median Abo e-median Below-median
eam a ing eam a ing eam a ing eam a ing
1No in e media ed −0.771∗∗ −0.537 0.435∗∗∗ 0.277∗∗
(0.318) (0.472) (0.130) (0.127)
Panel C: P oduc quali y
Abo e-median Below-median Abo e-median Below-median
p oduc a ing p oduc a ing p oduc a ing p oduc a ing
1No in e media ed −0.767∗∗ −0.427 0.448∗∗∗ 0.167
(0.329) (0.489) (0.134) (0.129)
Panel D: Ven u e’s ision quali y
Abo e-median Below-median Abo e-median Below-median
ision a ing ision a ing ision a ing ision a ing
1No in e media ed −0.712∗∗ −0.763∗0.494∗∗∗ 0.034
(0.318) (0.418) (0.123) (0.128)
Panel E: En ep eneu ial incen i iza ion ( oken e en ion)
Abo e-median Below-median Abo e-median Below-median
oken e en ion oken e en ion oken e en ion oken e en ion
1No in e media ed −0.833 −0.838∗∗∗ 0.043 0.351∗∗∗
(0.547) (0.299) (0.108) (0.126)
Model in o ma ion:
O he con ols ✓✓✓✓
Popula ion weigh s ✓✓✓✓
Gene alized Residuals ✓✓✓✓
Coun y ixed e ec s ✓✓✓✓
Qua e -yea ixed e ec s ✓✓✓✓
No e: *, **, and *** deno e s a is ical signi icance a he 10%, 5%, and 1% le els, espec i ely
123
1433Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
5.4 Pos hoc analyses and obus ness checks
The iden i ied e ec s o ICO in e media ion migh di -
e in a ious subsamples, and in es iga ing he sensi-
i i y o he main e ec s may yield addi ional insigh s
(Newbe e al., 2022). Table 5shows e-es ima ed
second-s age ea men e ec s based on he GR model
o subsamples based on i e di e en ca ego ies, wi h
anindica o a iable o non-in e media edICOs.Panel
A shows he main e ec s o bull o s able e sus bea
phases in he ICO ma ke . The esul s sugges ha he
alua ion e ec is mo e p onounced in bullma ke s and
he du a ion e ec is mo e p onounced in bea ma ke s.
Panel B, C, and D show he main e ec s o abo e- and
below-median subsamples based on ICObench’s eam,
p oduc ,and en u e ision a ings.The esul s indica e
ha he alua ion and du a ion e ec s a e mo e p o-
nounced o abo e-medianquali y i ms.PanelCshows
he main esul s o abo e- and below-median i ms
based on hei oken e en ion a ios. Re aining mo e
okens is o en ega ded as a cos ly signal o i m qual-
i y and associa ed wi h s onge en ep eneu ial incen-
i iza ion (e.g., Leland and Pyle, 1977). The indings
sugges he alua ion and du a ion e ec s a e mo e p o-
nounced o he below-median oken e aining i ms.
Fu he , se e al addi ional ad hoc checks con i m
he esul s’ obus ness. In pa icula , he esul s a e
obus o (i) excluding so cap and ha d cap con ols
om he eg essions, (ii) adding hype- ela ed a iables
as con ols, such as mon hly GoogleT ends o ICO,
blockchain, and c yp ocu ency sea ch e ms, (iii) con-
olling o p ojec size in absolu e (i.e., he o al oken
supply mul iplied by he ICO oken p ice) and ela-
i e (i.e., he p ojec size as he numbe o okens sold
di ided by he o al okensupply) e ms,and(i ) adjus -
ing he so cap and ha d cap a iables by hypo he i-
cal in es men amoun s by c yp o unds in he amoun
o 10%, 25%, and 50%. Mo eo e , we also disagg e-
ga e he ICO in e media ion a iable in o whe he he
c yp o und’s in es men s a egy esembles mo e en-
u e capi al o hedge unds and whe he he c yp o
und’s in es men s a egy ollows a specialized o a
di e si ied s a egy. While hese esul s lack s a is i-
cal powe due o small sample sizes gi en he g an-
ula subca ego ies, hey a e ne e heless quali a i ely
consis en wi h he main esul s and show ha c yp o
hedge unds and di e si ied unds ha e a s onge e ec
on alua ion, while c yp o en u e unds and special-
ized unds ha e a s onge e ec on ICO du a ion.
Finally, we eplace he unding amoun as he p oxy
o ICO i m alua ion wi h a ela i e measu e (ac ual
unding di ided by ha d cap), which leads o a s a-
is ically highly signi ican coe icien on he indica o
o non-in e media ed ICOs o −0.179, con i ming he
main esul s on he ICO i m alua ion discoun in he
absence o ICO in e media ion.
6 Model
6.1 Se up: he ICO ma ke
Any model is an abs ac ion om eali y. Ou model
aces he addi ional cons ain ha we wan o s uc-
u ally es ima e i . This equi es u he abs ac ion
and may come a he cos o igno ing in e es ing ea-
u es o he ICO ma ke ha can be conside ed in a
pu ely heo e ical model, such as ha by Wang e al.
(2022). We model he ICO ma ke as a decen alized
sea ch-and-ba gaining ma ke popula ed by indi idual
in es o s and in e media ies who discoun he u u e
a a e ρ>0. The key di e ence be ween he wo
in es o ypes is ha only indi idual in es o s de i e
u ili y om holding okens (e.g., he oken may be used
as a membe ship ee o a ideo gaming online com-
muni y, o o pu chase cloud s o age o sa e elec onic
iles, o o any o he pu pose), while in e media ies
ac as ins i u ional in es o s ha ex ac economic en s
om exploi ing ma ke impe ec ions. The model is se
in con inuous ime wi h an in ini e ho izon.11
Indi idual in es o s (hence o h, indi iduals)a e
assumed o be isk neu al. A e e y ins an , a mass
μo high- alua ion indi iduals wi h oken alua ion
zh>0 en e s he ICO ma ke . Thei alua ion d ops
o zl<zhand hey become low- alua ion indi iduals
wi h in ensi y λ, obeying a con inuous- ime Ma ko
chain. Because indi iduals’ alua ions a e indepen-
den , he e is a mass o μ
λhigh- alua ion indi iduals
in s eady s a e.
In e media ies (indexed by d o “middlemen”) en e
he ICO ma ke as an (endogenous) mass μd, and we
assume ha ma ke en y is ee. μdo(a)and μdn(a)
deno e he endogenous masses o in e media ed and
11 Weill (2020) o e s an excellen o e iew o ela ed wo k on
sea ch heo y. Ou analysis ollows Ga azza (2016), combining
elemen s om Du ie e al. (2005), Rubins ein and Wolinsky
(1987).
123
1434 P.P. Mom az

non-in e media ed ICOs o a pla o m o size a, espec-
i ely. We discuss he ole o pla o m size below.
Because in e media ies do no enjoy u ili y om hold-
ing u ili y okens, hey se e as po en ial ansac ion
pa ies independen o any oken- alua ion pa ame-
e .12 In e media ies ha e ope a ing cos s k.
Tokens We assume ha en ep eneu s launch ICOs
such ha okens en e he ma ke a e e y ins an as
amasso x<μand ansac s a endogenous p ice
p∗. Tokens a e he e ogeneous wi h espec o hei
unde lying pla o m size and a e gene ically anked,
such ha a=1...100 (1 =la ges pla o m pe -
cen ile, 100 =smalles pla o m pe cen ile). Token
alue dec eases in aand has a sal age alue s≥0.
A wo hless oken can also be sc apped ee-o -cha ge
a any poin in ime. Pla o m size ma e s because
i ela es o ne wo k e ec s, which ul ima ely de e -
mine oken alue (Chen & Bella i is, 2020;Fisch,
2019). We equi e μ
λ>A, whe e A ep esen s he
o al amoun o pla o m okens in he ma ke , so ha ,
in a Wal asian ma ke , he ma ginal oken-holde is
o he high- alua ion ype. Ins an aneous low u ili y
π(z,a)is a unc ion o a oken’s pla o m size aand i s
holde ’s alua ion z. U ili y is inc easing in alua ions
(π(zh,a)>π(zl,a)), dec easing in oken’s pla o m
size anking (∂π(z,a)
∂a<0) (i.e. smalle pla o m, lowe
oken alue), and i has nega i e z-acomplemen a i y
such ha ∂π(zh,a)
∂a<∂π(zl,a)
∂a.
Sea ch Indi iduals pay a sea ch- ela ed cos s cs o
ind oken ansac ion pa ies a pai wise independen
Poisson a i al a es γ>0. The e o e, an indi id-
ual wishing o sell a oken mee s a po en ial buye a
a e γs=γμ
b, and an indi idual looking o buy a
pla o m-size-a oken mee s a po en ial selle a a e
γb(a)=γμ
s(a). The masses μband μs(a)a e de e -
mined in equilib ium. Simila ly, indi iduals mee in e -
media iesa pai wiseindependen Poissona i al imes
wi h in ensi y γ>0. An indi idual wishing o pu -
chase a pla o m-size-a oken mee s a pla o m-size-a
oken-holding ins i u ional a a e γbd(a)=γμdo(a),
and in e media ies mee indi iduals o sell hei okens
a a e αds =γμb.Thesumo γbd (a)and αds ep-
12 In e media ies do no in es in mo e han one pla o ms a
once. This assump ion la gely simpli ies he model, and co e-
sponds closely o s a is ics abou ins i u ional in es o in ol e-
men in he ICO ma ke , e.g., Fisch and Mom az (2020).
esen s he combined sea ch in ensi y a which buy-
ing indi iduals and selling in e media ies mee . Sim-
ila ly, γsd and αdb(a)exp ess he sea ch in ensi ies
wi h which selling indi iduals mee in e media ies and
buying in e media ies mee indi iduals wishing o sell
pla o m-size-a okens.
Ba gaining Mee ings be ween ma ching buying and
selling indi iduals o wi h in e media ies lead o p ice
nego ia ions acco ding o a gene alized Nash ba gain-
ing amewo k. The pa ame e s θs∈{0,1}and θd∈
{0,1}deno e he ela i e ba gaining powe o he selle
in a non-in e media ed ICO and o he ins i u ional in
an in e media ed ICO, espec i ely. Thus, we assume
symme ic in o ma ion abou oken quali y, which is
why we es ima e he model based on esidual p ices
a e con olling o de e minan s o oken alue o he
han pla o m size in Sec . 7.
6.2 Value unc ions
6.2.1 Indi idual in es o s
We dis inguish ou ypes o indi idual in es o s in ou
model.Indi idualscanha eahigh(zh)o low(zl) oken
alua ion, and own o do no own he espec i e oken.
Fu he , indi idual in es o s who own a oken decide
be weenkeepingo sellingi ,whileindi idualin es o s
who do no hold a oken decide be ween ac i ely ying
o pu chase okens in he ma ke o no .
An indi idual wi h alua ion zholding a pla o m-
size-a oken con inuously chooses be ween keeping o
selling he asse . I she is no seeking o sell he oken,
she enjoys u ili y π(z,a) om holding a high- alued
oken. He alua ion swi ches o zl<zha a e λ,a
which poin she aces he decision p oblem o choos-
ing be ween he u ili y om a low- alued oken and he
p oceeds omselling i (max{Ulo(a), Slo(a)}).Finally,
shemayincu a capi al change in heamoun o U
ho(a),
which de ines he alue unc ion as: ρUho(a)=
π(zh,a)+λVlo(a)−Uho(a)+U
ho(a). Al e na i ely,
shemayac i ely y osell he oken.Shealsoenjoys he
cu en low u ili y π(zh,a), aces he decision p ob-
lem max{Ulo(a), Slo(a)}when he alua ion d ops,
and sus ains capi al change due o ne wo k e ec s
U
ho(a). Mo eo e , she aces he low sea ch cos cs.
She mee s po en ial buye s a a e γs, a which poin
123
1435Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
she has o decide be ween selling he oken a p ice
p(a)and becoming a high- alua ion non-holde wi h
alue Vhn, o keeping i . Selling i esul s in he cap-
i al gain p(a)+Vhn −Sho(a). Simila ly, she mee s
in es o s a a e γsd, a which poin she has o decide
whe he o accep he in es o ’s bid p ice pB(a). Thus,
he alue unc ionsa is ies he ollowingBellmanequa-
ion: ρSho(a)=π(zh,a)−cs+λVlo(a)−Sho(a)+
γsmax{p(a)+Vhn −Sho(a), 0}+γsd max{pB(a)+
Vhn −Sho(a), 0}+S
ho(a).
Low- alua ion oken-holding indi iduals can also
choose be ween keeping and selling he asse . In he
o me case, because zlis an abso bing s a e and alu-
a ions canno swi ch anymo e, he alue unc ion sa is-
ies: ρUlo(a)=π(zl,a)+U
lo(a). In he la e case, he
condi ions o he alue unc ion ollow om he ac
ha zlis an abso bing s a e: ρSlo(a)=π(zl,a)−cs+
γsmax{p(a)+Vln −Slo(a), 0}+γsd maxpB(a)+
Vln −Slo(a), 0+S
lo(a).
Indi iduals who do no own okens can ei he pas-
si ely emain in hei owne ship s a us o ac i ely y
o pu chase okens in he ICO ma ke . High- alua ion
indi iduals who nei he hold a oken no a e ac i ely
ying o pu chase one in he ma ke ha e ze o u il-
i y: ρUhn =0. High- alua ion indi iduals who do no
hold a oken bu a e ac i ely looking o pu chase one
in he ICO ma ke pay a sea ch cos csand incu a
capi al loss o Vln −Shn when hei alua ion d ops.
Because hey canno be ex-an e su e o he ul ima e
pla o m size a ha is up o sale when he indi idual
encoun e s a po en ial ansac ion pa ne , bu which is
highly alue- ele an , hey ake he expec a ions o e
all pla o m sizes. They mee a selling indi idual a a e
γb(a)and a selling in es o a a e γbd(a)and expe i-
encesacapi al gain o max{Vho(a)−p(a)−Shn,0}and
max{Vho(a)−pA(a)−Shn,0}, espec i ely. This leads
o:ρShn =−cs+λ(Vln−Shn)+γb(a)max{Vho(a)−
p(a)−Shn,0}da +γbd(a)max{Vho(a)−pA(a)−
Shn,0}da.
Simila ly o ype-ln indi iduals wi hou oken own-
e ship, low- alua ion indi iduals who nei he hold a
oken no a e ac i ely ying o pu chase one in he ma -
ke ha e ze o u ili y: ρUln =0. Low- alua ion indi id-
uals who do no hold a oken bu a e ac i ely looking o
pu chase one in he ma ke pay he de e minis ic sea ch
cos cs. Like he high- alua ion non-holde s, he low-
alua ionindi idual akes heexpec a ionso e allpla -
o m sizes. She mee s a selling indi idual a a e γb(a)
and a selling in e media y a a e γbd(a)and expe i-
ences a capi al gain o max{Vlo(a)−p(a)−Sln,0}
and max{Vlo(a)−pA(a)−Sln,0}, espec i ely. This
yields: ρSln =−cs+γb(a)max{Vlo(a)−p(a)−
Sln,0}da+γbd(a)max{Vlo(a)−pA(a)−Sln,0}da.
6.2.2 In e media ies
Because in e media ies do no enjoy low u ili y om
holding okens, he ea eonly wo ypes: in e media ies
ha cu en ly hold okens and hose ha cu en ly do
no hold okens. A pla o m-size-a oken-holding in e -
media y pays he low ope a ing cos k, and decides
be ween selling he oken o ask p ice pA(a)and eal-
izing capi al gain pA(a)+Jdn −Jdo(a)ne o any
ne wo k- ela ed e ec s on capi al o ealizing he sal-
age alue (e.g., e u n okens in case a ha d-cap goal
was no me in he o e ing). The oken-holding in e -
media y’s alue unc ion sa is ies he ollowing Bell-
man equa ion: ρJdo(a)=max{−k+αdspA(a)+
Jdn −Jdo(a)+J
do(a), ρ Jdn. In con as , he alue
unc ion o he oken-nonholding in e media y is cha -
ac e ized by he low ope a ing cos k ha he in e -
media y incu s while ac i ely sea ching o a po en-
ial ansac ion and he expec a ion o e he pla o m
sizes when he in e media y mee s a selle a a e
αdb(a)and ealizes a capi al gain in he amoun o
max{Jdo(a)−pB(a)−Jdn,0}. I sa is ies: ρJdn =
−k+αdb(a)max{Jdo(a)−pB(a)−Jdn,0}da.No e
ha he ee-en y condi ion implies ha in e medi-
a ies’ expec ed capi al gains is exac ly o se by hei
ope a ing cos in he la e equa ion.
6.3 P ices
Following he li e a u e (e.g., Ga azza, 2016; Weill,
2020), we sol e o he endogenous p ice in he
non-in e media ed ICO ma ke segmen ia gene -
alized Nash ba gaining: maxp(a)[Uho(a)−p(a)−
Shn]1−θs[p(a)+Vln −Slo(a)]θssubjec o Uho(a)−
p(a)−Shn ≥0 and p(a)+Vln −Slo(a)≥0. This
yields he endogenous p ice: p(a)=(1−θs)Slo(a)−
Vln+θsUho(a)−Shn. The ask p ice pA(a)and
bid p ice pB(a)a e de e mined in a simila ashion:
pA(a)=(1−θd)Jdo(a)−Jdn+θdUho(a)−Shn
and pB(a)=(1−θd)Jdo(a)−Jdn+θdSlo(a)−
Vln).
123
1436 P.P. Mom az
6.4 Policies
6.4.1 Indi idual in es o s
Su plus en s om adea isewhen hn- ypeindi iduals
wi hou oken owne ship mee lo- ype oken-holde s.
Those su plus en s om ade a e highe o okens
wi h ela i ely la ge unde lying pla o ms due o he
nega i e complemen a i y be ween he inpu ac o s o
low u ili y π(z,a). By implica ion, no all okens a e
ealloca ed in equilib ium, which helps simpli y he
analyses. We can e o mula e he alue unc ions o ho-
ype oken-holding indi iduals. Vho(a)equals Uho(a)
o a<a∗
ho and Shn o a≥a∗
ho. These cases ollow
om U
ho(a)<0. Because he u ili y ho- ype indi-
iduals de i e om holding pla o m-size-a okens is
dec easing in a, he e mus be a cu o a∗
ho a which
oken-holde s decide o ealize he sal age alue Shn.
The e o e, Uho(a∗
ho)=Shn. Likewise, such a cu o
also exis s o hn- ype oken-nonholding indi iduals
a which hey decide o pu chase a pla o m-size-a
oken. The e o e, we ha e a∗
hn ≤a∗
ho, wi h a po en ial
wedge om ading ic ions. Thus, oken-nonholding
hn- ypeindi idualsha e he alue unc ion Vhn =Shn.
The alue unc ion o oken-holding lo- ype indi idu-
als can be simpli ied as ollows: Vlo(a)equals Slo(a)
o a<a∗
l,Ulo(a) o a≤a∗
l<T, and Vln o a=T.
Because U
lo(a)<0 and csis cons an , he e exis s a
cu o a∗
lsuch ha a lo- ype indi idual decides o sell
he okeni i is below he cu o andkeepi i i isabo e
he cu o , espec i ely. I she keeps he oken, hen she
holds i un il she can ealize he sal age alue a T.
Fu he , equilib ium conside a ions dic a e ha oken-
holdinglo- ypeindi idualssell o pu chase-willinghn-
ype indi iduals, implying a∗
l≤a∗
hn. Finally, because
zlis an abso bing s a e, he alue unc ion o oken-
nonholding ln- ype indi iduals is Vln =ywhe e y
ep esen s he alue o he smalles pla o m.
6.4.2 In e media ies
In e media ies p e e okens wi h la ge unde lying
pla o ms because hey ade a g ea e ma gins, bu
in e media ies’ ope a ing cos s ka e cons an . The e-
o e, in e media ies pu chase pla o m-size-a okens
such ha a≤a∗
dn and ealize hesal age aluea a∗
dn <
a∗
do.Equilib ium conside a ions alsodic a e a∗
do ≤a∗
hn,
ha is, in e media ies always sell o pu chase-willing
indi iduals.
6.5 Dis ibu ions o indi idual in es o s and
in e media ies
6.5.1 Laws o mo ion o indi idual in es o s
The mass o pla o m-size-a oken-holding ho- ype
indi iduals e ol es o e ime as non-holding hn- ype
indi iduals mee selling pla o m-size-a oken-holding
indi iduals (γb(a)μhn) o pla o m-size-a oken-
holdingin e media ies(γbd(a)μhn),andpla o m-size-
a oken-holding indi iduals swi ch alua ions o he
abso bing s a e zl(λμho(a)); o mally: ˙μho(a)=
(γb(a)μhn +γbd(a)μhn)−λμho(a) o a<a∗
ho.
Simila ly, he e olu ion o pla o m-size-a oken-
holding lo- ype indi iduals o e ime is a ec ed by
in lows om oken-holding indi iduals wi h unde -
lying oken-pla o m sizes a<a∗
ho (because hese
indi iduals would jus ealize he sal age alue i
hei okens we e based on a smalle pla o m)
whose alua ions jus swi ched o he abso bing zl-
s a e (λ1(a<a∗
ho)μho(a)), and he ou low o lo-
ype holde s o okens wi h ne wo ks g ea e han
a∗
l(holde s o okens wi h hose pla o m sizes
would p e e o sell a he hen keep i ) who mee
pu chase-willing indi iduals (γs1a<a∗
lμlo(a))
o in e media ies (γsd1a<min{a∗
l,a∗
dnμlo(a)).
Technically, o low- alua ion owne s: ˙μlo(a)=
λ1a<a∗
hoμho(a)−γs1a<a∗
lμlo(a)−
γsd1a<min{a∗
l,a∗
dnμlo(a) o a<T; o high-
alua ion non-owne s: ˙μhn =(μ −x)+μho(a∗
ho)−
λμhn −μhn a∗
l
0γb(a)da −μhn a∗
dn
0γbd(a)da; and
o low- alua ion non-owne s: μln =λμhn +
γsa∗
l
0μlo(a)da+γsd min{a∗
l,a∗
dn}
0μlo(a)da+μlo(T).
6.5.2 Laws o mo ion o in e media ies
Simila logic gi es he laws o mo ion o in e -
media ies in he model: ˙μdo(a)=αdb(a)1a<
a∗
dnμdn −αdsμdo(a) o a<a∗
do and ˙μdn =
αds a∗
do
0μdo(a)da −μdn min{a∗
l,a∗
dn}
0αdb(a)da +
μdo(a∗∗
dn).
7 S uc u al es ima ion
Thissec iondesc ibeshow we es ima e and iden i y he
model o he ICO ma ke desc ibed in Sec . 6.I also
epo skeypa ame e es ima es,includingsea chcos s,
123
1437Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...
ansac ion su plus, and en sha ing be ween indi id-
ual in es o s and in e media ies. Finally, we bench-
ma k he ICO ma ke o he pe ec ly e icien Wal-
asian equilib ium o quan i y he e ec o ma ke ic-
ions on agg ega e wel a e c ea ed in he ICO ma ke .
7.1 Es ima ion
We he model as ollows: The discoun a e is ρ=
0.5%, he o al mass o ICOs is se o equal he sample
median, and he a e age numbe o ac i e in e medi-
a ies, μd, is se o he sample mean. ICOs di e in hei
unde lying ne wo k size. All ICOs’ ne wo k sizes a e
anked by pe cen iles om 1 (la ges ) o 100 (smalles ).
The uni o ime, a which we es ima e he model, is se
o mon hs, co e ing 48 mon hs du ing he 2017–2020
pe iod. The sal age p ice is S=$0, s ipula ing ha
pla o ms wi hou use s a e wo hless.
Tokenholde s’ low payo equals π(z,a)=
ze−δ2a. We es ima e he ec o ψ=
{λ,γs,γ
sd,α
ds,zh,zl,δ
2,cs,θ
s,θ
d}. The endogenous
con ac a es {γs,γ
sd,α
ds}can be in e ed om he
da a and help iden i y o he pa ame e s o he model,
which in u n join ly de e mine indi idual in es o ’
and in e media ies’ policy unc ions and dis ibu ions.
The ollowing ICO momen s, m1(ψ), a e compu ed
based on he model’s solu ion: (1) The ac ion o
okens o sale, m1[1]=a∗
l
0μlo(a)da+a∗
do
0μdo(a)da
A;
(2) he cumula i e numbe o in e media ed ICOs
ela i e o all ICOs, m1[2]=a∗
do
0μdo(a)da
A;(3) he
ac ion o ac i e non-in e media ed ICOs, m1[3]=
γSa∗
l
0μlo(a)da
A; (4) he ac ion o ac i e in e media ed
ICOs, m1[4]=αds a∗
do
0μdo(a)da
A; and (5) he a e -
age anking o ICO pla o m’s unde lying ne wo k
size, m1[5]=a∗
l
0aμlo(a)da+a∗
do
0aμdo(a)da
a∗
l
0μlo(a)da+a∗
do
0μdo(a)da
. Fu he , six
p ice momen s, m2(ψ) ={β0,β
1,β
2,β
3,β
4,β
5},a e
ob ained om nonlinea leas squa es om he ollow-
ing wo auxilia y eg essions: p(a)=β0+β1e−β2a
and pB(a)=β3+β4e−β5a. Then, we es ima e
ψ ia he wo-s ep es ima o om Hansen (1982):
ˆ
ψ=a g minψ∈m(ψ) −mS( ˜
ψ)m(ψ) −mS
whe e mSdeno es he simula ed oken o e ing and
p ice momen s, and ( ˜
ψ) is he consis en es ima e
o he asymp o ic a iance-co a iance ma ix o he
momen s, and we use m(ψ)−mS
mS o ha e a simila scale.
7.2 Iden i ica ion
Iden i ica ion ollows la gely he mul idisciplina y
sea ch li e a u e, including labo (Ecks ein & Van
den Be g, 2007) and decen alized eal asse ma ke s
(Ga azza, 2016). ICO momen s iden i y he ansi ion
a es be ween s a es. The ac ion o non-in e media ed
ICOs iden i ies he a e a which indi iduals mee ,
γs, while he ac ion o in e media ed ICOs iden i-
ies in e media ies’ con ac a es, αds. The cumula i e
numbe o in e media ed ICOs iden i ies he a e a
which indi iduals mee in e media ies, γsd. Fu he -
mo e, he Ma ko -chain pa ame e ha go e ns in en-
o s’ alua ion changes, λ, is iden i ied by he o al ac-
ion o ac i e ICOs in any gi en ins an .
The alua ion pa ame e and he endogenous
momen s oge he wi h he ixed pa ame e s Aand μd
and he s eady s a e condi ion μ
λ=μhn +T
0μho(a)da
hen allow o in e he e iciency pa ame e s o he
ma ching unc ions, γand γ, and he mass o new
ma ke en an s, μ. Speci ically, we sol e o μho(a),
μlo(a), and μdo(a)in he law-o -mo ions equa ions in
Sec .6.5.
To iden i y he emaining pa ame e s
{δ2,zh,zl,θ
s,θ
d,cs}, we use he p ice momen s and
he i h ansac ion momen ha cha ac e izes he
a e age pla o m’s unde lying ne wo k sizes o he
ma ke ed ICOs. The p ice equa ions iden i y he alue
dep ecia ion pa ame e , δ2, om p ice a ia ions ac oss
okens o di e en unde lying ne wo k size.
We ollow Ga azza (2016) o iden i y ba gaining
and alua ion pa ame e s who sugges s o exploi di -
e ences be ween he p ices in in e media ed and non-
in e media ed ICOs, as well as he e ical he e ogene-
i y o he oken ne wo k sizes. β0and β3in he p ice
eg essionsiden i yhn non- okenholde s’ alueo con-
inuing o sea ch, Shn, which is hei ou side op ion
in Nash ba gaining wi h equilib ium p ices. Because
non- okenholde s do no know he oken’s unde lying
ne wo k size o he oken o he okenholde hey mee
nex , Shn does no depend on he oken’s unde lying
ne wo k,and hence hein e cep s om he p ice eg es-
sions a e su icien o iden i ica ion. Shn depends on
alua ions {zl,zh}and ba gaining pa ame e s {θs,θ
d}.
Shn is nega i e i he ansac ion su plus is exclusi ely
app op ia ed by selling indi idual and in e media ies.
Shn inc eases in he combined buying indi iduals’ ba -
gaining powe , (1−θs)+(1−θd). This also implies
123
1438 P.P. Mom az
An, J., Duan, T., Hou, W., & Xu, X. (2019). Ini ial coin o e ings
and en ep eneu ial inance: The ole o ounde s’ cha -
ac e is ics. The Jou nal o Al e na i e In es men s, 21(4),
26–40. h ps://doi.o g/10.3905/jai.2019.1.068
An e, L., Fiedle , I., Meduna, M ., & S einme z, F. (2022).
Indi idual c yp ocu ency in es o s: E idence om a
popula ion su ey. In e na ional Jou nal o Inno a ion
and Technology Managemen , 2250008. h ps://doi.o g/10.
1142/S0219877022500080
A hu s, J. D., Buseni z, L. W., Hoskisson, R. E., & Johnson, R.
A. (2009). Signaling and ini ial public o e ings: The use
and impac o he lockup pe iod. Jou nal o Business Ven-
u ing, 24(4), 360–372. h ps://doi.o g/10.1016/j.jbus en .
2008.02.004
Aud e sch, D. B. (2018). En ep eneu ship, economic g ow h,
and geog aphy. Ox o d Re iew o Economic Policy, 34(4),
637–651. h ps://doi.o g/10.1093/ox ep/g y011
Ba g, J. A., D obe z, W., & Mom az, P. P. (2021). Valuing
s a -up i ms: A e e se-enginee ing app oach o ai -
alue mul iples om en u e capi al ansac ions. Finance
Resea ch Le e s, 43, 102008. h ps://doi.o g/10.1016/j. l.
2021.102008
Bech , M., Bol on, P., & Röell, A. (2003). Co po a e go -
e nance and con ol. In Handbook o he economics o
inance (pp. 1–109). Else ie . h ps://doi.o g/10.1016/
S1574-0102(03)01005-7
Becke , B. (2007). Geog aphical segmen a ion o us capi al
ma ke s. Jou nal o Financial economics, 85(1), 151–178.
h ps://doi.o g/10.1016/j.j ineco.2006.07.001
Beli ski, M., & Bo eiko, D. (2021). Success ac o s o ini ial
coin o e ings. The Jou nal o Technology T ans e , 1–17.
h ps://doi.o g/10.1007/s10961-021-09894-x
Bella i is, C., Cumming, D., & Vanacke , T. (2020). Ban, boom,
and echo! en ep eneu ship and ini ial coin o e ings.
En ep eneu ship Theo y and P ac ice.h ps://doi.o g/10.
1177/1042258720940114
Bella i is, C., Fisch, C., & Wiklund, J. (2021). A comp ehensi e
e iew o he global de elopmen o ini ial coin o e ings
(icos) and hei egula ion. Jou nal o Business Ven u ing
Insigh s, 15, e00213. h ps://doi.o g/10.1016/j.jb i.2020.
e00213
Belle lamme, P., Lambe , T., & Schwienbache , A. (2014).
C owd unding: Tapping he igh c owd. Jou nal o Busi-
ness Ven u ing, 29(5), 585–609. h ps://doi.o g/10.1016/j.
jbus en .2013.07.003
Benede i, H., & Kos o e sky, L. (2021). Digi al ulips? e u ns
o in es o s in ini ial coin o e ings. Jou nal o Co po a e
Finance, 66, 101786. h ps://doi.o g/10.1016/j.jco p in.
2020.101786
Benede i, H., & Nikbakh , E. (2021). Re u ns and ne wo k
g ow h o digi al okens a e c oss-lis ings. Jou nal o
Co po a e Finance, 66, 101853. h ps://doi.o g/10.1016/j.
jco p in.2020.101853
Ben enis e, L. M., & Spind , P. A. (1989). How in es men
banke s de e mine he o e p ice and alloca ion o new
issues. Jou nal o Financial Economics, 24(2), 343–361.
h ps://doi.o g/10.1016/0304-405X(89)90051-2
Be oni, F., Colombo, M. G., & G illi, L. (2011). Ven u e
capi al inancing and he g ow h o high- ech s a ups:
Disen angling ea men om selec ion e ec s. Resea ch
Policy, 40(7), 1028–1043. h ps://doi.o g/10.1016/j. espol.
2011.03.008
Block, J. H., G oh, A., Ho nu , L., Vanacke , T., & Visma a, S.
(2021). The en ep eneu ial inance ma ke s o he u u e:
A compa ison o c owd unding and ini ial coin o e ings.
Small Business Economics, 57(2), 865–882. h ps://doi.
o g/10.1007/s11187-020-00330-2
Bo eiko, D., Fe a ini, G., & Giudici, P. (2019). Blockchain
s a ups and p ospec us egula ion. Eu opean Business
O ganiza ion Law Re iew, 20(4), 665–694. h ps://doi.o g/
10.1007/s40804-019-00168-6
Bo eiko, D., & Ris eski, D. (2021). Se ial and la ge in es o s
in ini ial coin o e ings. Small Business Economics, 57(2),
1053–1071. h ps://doi.o g/10.1007/s11187-020-00338-8
Bo eiko, D., & Vidusso, G. (2019). New blockchain in e medi-
a ies: Do ico a ing websi es do hei job well? The Jou nal
o Al e na i e In es men s, 21(4), 67–79. h ps://doi.o g/
10.3905/jai.2019.21.4.067
Bou eau, T., De Geo ge, E. T., Ellahie, A., & Macciocchi, D.
(2022). The ole o disclosu e and in o ma ion in e me-
dia ies in an un egula ed capi al ma ke : E idence om
ini ial coin o e ings. Jou nal o Accoun ing Resea ch,
60(1), 129–167. h ps://doi.o g/10.1111/1475-679X.12404
B ochado, A., & T oilo, M. L. (2021). Ini ial coin o e ings: An
eme gen esea ch a ea. Digi al Policy, Regula ion and Go -
e nance. h ps://doi.o g/10.1108/DPRG-06-2020-0084
Bu icé, V., & Visma a, S. (2022). Inclusi e digi al inance:
The indus y o equi y c owd unding. The Jou nal o
Technology T ans e , 47(4), 1224–1241. h ps://doi.o g/10.
1007/s10961-021-09875-0
Campino, J., B ochado, A., & Rosa, Á. (2021). Ini ial coin
o e ings (icos): The impo ance o human capi al. Jou nal
o Business Economics, 91(8), 1225–1262. h ps://doi.o g/
10.1007/s11573-021-01037-w
Campino, J., B ochado, A., & Rosa, Á. (2022). Ini ial coin o e -
ings (icos): Why do hey succeed? Financial Inno a ion,
8(1), 1–35. h ps://doi.o g/10.1186/s40854-021-00317-2
Ca e , N., & Jeng, L. (2021). De i p o ocol isks: The pa adox
o de i. Reg ech, Sup ech and Beyond: Inno a ion and
Technology in Financial Se ices RiskBooks-Fo hcoming
Q, 3.h ps://doi.o g/10.2139/ss n.3866699
Chalme s, D., Fisch, C., Ma hews, R., Quinn, W., & Recke ,
J. (2022). Beyond he bubble: Will n s and digi al p oo
o owne ship empowe c ea i e indus y en ep eneu s?
Jou nal o Business Ven u ing Insigh s, 17, e00309. h ps://
doi.o g/10.1016/j.jb i.2022.e00309
Chen, Y., & Bella i is, C. (2020). Blockchain dis up ion and
decen alized inance: The ise o decen alized business
models. Jou nal o Business Ven u ing Insigh s, 13, e00151.
h ps://doi.o g/10.1016/j.jb i.2019.e00151
Colombo, M. G., Fisch, C., Mom az, P. P., & Visma a, S.
(2021). The ceo beau y p emium: Founde ceo a ac i e-
ness and i m alua ion in ini ial coin o e ings. S a egic
En ep eneu ship Jou nal.h ps://doi.o g/10.1002/sej.1417
Colombo, O. (2021). The use o signals in new- en u e
inancing: A e iew and esea ch agenda. Jou nal o
Managemen , 47(1), 237–259. h ps://doi.o g/10.1177/
0149206320911090
Colombo, M. G., & G illi, L. (2010). On g ow h d i e s o high-
echs a -ups:Explo ing he ole o ounde s’humancapi al
123
1445Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...

and en u e capi al. Jou nal o Business Ven u ing, 25(6),
610–626. h ps://doi.o g/10.1016/j.jbus en .2009.01.005
Colombo, M. G., Meoli, M., & Visma a, S. (2019). Signaling
in science-based ipos: The combined e ec o a ilia ion
wi h p es igious uni e si ies, unde w i e s, and en u e
capi alis s. Jou nal o Business Ven u ing, 34(1), 141–177.
h ps://doi.o g/10.1016/j.jbus en .2018.04.009
C yp o Fund Resea ch. (2021). C yp o und qua e ly
epo – 2021 q2. h ps://c yp o und esea ch.com/
q2-2021-c yp o- und- epo /
Cumming, D., Domb owski, N., D obe z, W., & Mom az, P.
P. (2023). Decen alized inance, c yp o unds, and alue
c ea ion in okenized s a up i ms. A ailable a SSRN,
4102295. h ps://doi.o g/10.2139/ss n.4102295
Cumming, D., Ho nu , L., Ka ami, M., & Schweize , D.
(2021). Disen angling c owd unding om aud unding.
Jou nal o Business E hics, 1–26. h ps://doi.o g/10.1007/
s10551-021-04942-w
Cumming, D. J., Johan, S., & Pan , A. (2019). Regula ion
o he c yp o-economy: Managing isks, challenges, and
egula o y unce ain y. Jou nal o Risk and Financial Man-
agemen , 12(3),126.h ps://doi.o g/10.3390/j m12030126
Cumming, D. J., & Visma a, S. (2017). De-segmen ing esea ch
in en ep eneu ial inance. Ven u e Capi al, 19(1–2), 17–27.
h ps://doi.o g/10.1080/13691066.2016.1225910
Daian, P., Gold ede , S., Kell, T., Li, Y., Zhao, X., Ben o ,
I., B eidenbach, L., & Juels, A. (2020). Flash boys 2.0:
F on unning in decen alized exchanges, mine ex ac able
alue, and consensus ins abili y. 2020 IEEE Symposium
on Secu i y and P i acy (SP), 910–927. h ps://doi.o g/10.
1109/SP40000.2020.00040
Demse z, H. (1968). The cos o ansac ing. The Qua e ly
Jou nal o Economics, 82(1), 33–53. h ps://doi.o g/10.
2307/1882244
Du ie, D., Gâ leanu, N., & Pede sen, L. H. (2005). O e - he-
coun e ma ke s. Econome ica, 73(6), 1815–1847. h ps://
doi.o g/10.1111/j.1468-0262.2005.00639.x
Ecks ein, Z., & Van den Be g, G. J. (2007). Empi ical labo
sea ch: A su ey. Jou nal o Econome ics, 136(2), 531–
564. h ps://doi.o g/10.1016/j.jeconom.2005.11.006
Fisch, C., Meoli, M., & Visma a, S. (2022). Does blockchain
echnology democ a ize en ep eneu ial inance? an empi i-
calcompa isono icos, en u ecapi al, and ei s.Economics
o Inno a ion and New Technology, 1–20. h ps://doi.o g/
10.1080/10438599.2020.1843991
Fisch, C. (2019). Ini ial coin o e ings (ICOs) o inance new
en u es. Jou nal o Business Ven u ing, 34(1), 1–22.
h ps://doi.o g/10.1016/j.jbus en .2018.09.007
Fisch, C., Masiak, C., Visma a, S., & Block, J. (2021). Mo i es
and p o iles o ico in es o s. Jou nal o Business Resea ch,
125, 564–576. h ps://doi.o g/10.1016/j.jbus es.2019.07.
036
Fisch,C.,&Mom az,P.P.(2020).Ins i u ionalin es o sandpos -
icope o mance:Anempi icalanalysiso in es o e u nsin
ini ial coin o e ings (icos). Jou nal o Co po a e Finance,
64,101679.h ps://doi.o g/10.1016/j.jco p in.2020.101679
Gao, S., & Li, Y. (2021). An empi ical s udy on he adop ion
o blockchain-based games om use s’ pe spec i es. The
Elec onic Lib a y, 39(4), 596–614. h ps://doi.o g/10.
1108/EL-01-2021-0009
Ga azza, A. (2016). An empi ical equilib ium model o a decen-
alized asse ma ke . Econome ica, 84(5), 1755–1798.
h ps://doi.o g/10.3982/ECTA10847
Giudici, G., & Adhami, S. (2019). The impac o go e nance
signals on ico und aising success. Jou nal o Indus ial
and Business Economics, 46(2), 283–312. h ps://doi.o g/
10.1007/s40812-019-00118-w
Giudici, G., Moncayo, G. G., & Ma inazzi, S. (2020). The ole
o ad iso s’ cen ali y in he success o ini ial coin o e ings.
Jou nal o Economics and Business, 112, 105932. h ps://
doi.o g/10.1016/j.jeconbus.2020.105932
Gou ie oux, C., Mon o , A., Renaul , E., & T ognon, A. (1987).
Gene alised esiduals. Jou nal o Econome ics, 34(1–2),
5–32. h ps://doi.o g/10.1016/0304-4076(87)90065-0
G een, R. C., Holli ield, B., & Schü ho , N. (2007). Deale
in e media ion and p ice beha io in he a e ma ke o
new bond issues. Jou nal o Financial Economics, 86(3),
643–682. h ps://doi.o g/10.1016/j.j ineco.2006.10.004
G een, R. C., Holli ield, B., & Schü ho , N. (2007). Financial
in e media ion and he cos s o ading in an opaque ma ke .
The Re iew o Financial S udies, 20(2), 275–314. h ps://
doi.o g/10.1093/ s/hhl012
Hansen, L. P. (1982). La ge sample p ope ies o gene alized
me hod o momen s es ima o s. Econome ica: Jou nal
o he econome ic socie y, 1029–1054. h ps://doi.o g/10.
2307/1912775
Ha ey, C. R., Ramachand an, A., & San o o, J. (2021). De i
and he u u e o inance. John Wiley & Sons.
Ho nu , L., Kück, T., & Schwienbache , A. (2021). Ini-
ial coin o e ings, in o ma ion disclosu e, and aud.
Small Business Economics, 1–19. h ps://doi.o g/10.1007/
s11187-021-00471-y
Howell, S. T., Niessne , M., & Ye mack, D. (2020). Ini ial coin
o e ings: Financing g ow h wi h c yp ocu ency oken
sales. The Re iew o Financial S udies, 33(9), 3925–3974.
h ps://doi.o g/10.1093/ s/hhz131
Huang, W., Meoli, M., & Visma a, S. (2020). The geog aphy
o ini ial coin o e ings. Small Business Economics, 55(1),
77–102. h ps://doi.o g/10.1007/s11187-019-00135-y
Jensen, M. C., & Meckling, W. H. (1976). Theo y o he i m:
Manage ial beha io , agency cos s and owne ship s uc u e.
Jou nal o Financial Economics, 3(4), 305–360. h ps://
doi.o g/10.2139/ss n.94043
Kas elein, R. (2017). Wha ini ial coin o e ings a e, and why c
i ms ca e. Ha a d Business Re iew,24.
Khe , R., Te jesen, S., & Liu, C. (2021). Blockchain, bi coin,
and icos: A e iew and esea ch agenda. Small Business
Economics, 56(4), 1699–1720. h ps://doi.o g/10.1007/
s11187-019-00286-y
K eppmeie , J., & Laschinge , R. (2023). Signaling in he ma ke
o secu i y okens. Jou nal o Business Economics, 1–38.
h ps://doi.o g/10.1007/s11573-023-01175-3
Lambe , T., Liebau, D., & Roosenboom, P. (2021). Secu i y
oken o e ings. Small Business Economics, 1–27. h ps://
doi.o g/10.1007/s11187-021-00539-9
Lee, J., Li, T., & Shin, D. (2022). The wisdom o c owds in
in ech: E idence om ini ial coin o e ings. The Re iew
o Co po a e Finance S udies, 11(1), 1–46. h ps://doi.o g/
10.1093/ c s/c ab014
Leland, H. E., & Pyle, D. H. (1977). In o ma ional asymme ies,
inancial s uc u e, and inancial in e media ion. The Jou -
123
1446 P.P. Mom az
nal o Finance, 32(2), 371–387. h ps://doi.o g/10.2307/
2326770
Mansou i, S., & Mom az, P. P. (2022). Financing sus ainable
en ep eneu ship: Esg measu emen , alua ion, and pe o -
mance. Jou nal o Business Ven u ing.h ps://doi.o g/10.
1016/j.jbus en .2022.106258
Meoli, M., Rossi, A., & Visma a, S. (2022). Financial li e acy
and secu i y-based c owd unding. Co po a e Go e nance:
An In e na ional Re iew, 30(1), 27–54. h ps://doi.o g/10.
1111/co g.12355
Meoli, M., & Visma a, S. (2022). Machine-lea ning o ecas ing
o success ul icos. Jou nal o Economics and Business, 121,
106071. h ps://doi.o g/10.1016/j.jeconbus.2022.106071
Mochkabadi, K., & Volkmann, C. K. (2020). Equi y c owd-
unding: A sys ema ic e iew o he li e a u e. Small
Business Economics, 54(1), 75–118. h ps://doi.o g/10.
1007/s11187-018-0081-x
Mokh a ian, E., & Lindg en, A. (2018). Rise o he c yp o hedge
und: Ope a ional issues and bes p ac ices o an eme gen
in es men indus y. S an. JL Bus. & Fin., 23, 112.
Mollick, E. (2014). The dynamics o c owd unding: An
explo a o y s udy. Jou nal o Business Ven u ing, 29(1),
1–16. h ps://doi.o g/10.1016/j.jbus en .2013.06.005
Mom az, P. P. (2020b). Ini ial Coin O e ings, Asymme ic
In o ma ion, and CEO Loyal y. Small Business Economics,
in p ess.
Mom az, P. P. (2020). Ini ial coin o e ings. Plos one, 15(5),
e0233018. h ps://doi.o g/10.1371/jou nal.pone.0233018
Mom az, P. P. (2021). Ceo emo ions and i m alua ion in
ini ial coin o e ings: An a i icial emo ional in elligence
app oach. S a egic Managemen Jou nal, 42(3), 558–578.
h ps://doi.o g/10.1002/smj.3235
Mom az, P. P. (2021). Econome ic models o du a ion da a in
en ep eneu ship wi h an applica ion o s a ups’ ime- o-
unding by en u e capi alis s ( cs). Jou nal o Applied
S a is ics, 48(13–15), 2673–2694. h ps://doi.o g/10.1080/
02664763.2021.1896686
Mom az, P. P. (2021). En ep eneu ial inance and mo al haza d:
E idence om oken o e ings. Jou nal o Business Ven-
u ing, 36(5), 106001. h ps://doi.o g/10.1016/j.jbus en .
2020.106001
Mo i z, A., & Block, J. H. (2016). C owd unding: A li e a u e
e iew and esea ch di ec ions. C owd unding in Eu ope,
25–53. h ps://doi.o g/10.1016/j.jbus en .2020.106001
Newbe , S. L., Khe , R., & Yang, S. (2022). Now ha ’s in e es -
ing and impo an ! mo ing beyond a e ages o inc ease he
in e en ial alue o empi ical indings in en ep eneu ship
esea ch. Jou nal o Business Ven u ing, 37(2), 106185.
h ps://doi.o g/10.1016/j.jbus en .2021.106185
Rawhouse , H., Visma a, S., & Kshe i, N. (2023). Blockchain
and ulne able en ep eneu ial ecosys ems. En ep eneu -
ship & Regional De elopmen , 1–26. h ps://doi.o g/10.
1080/08985626.2022.2162979
Rubins ein, A., & Wolinsky, A. (1987). Middlemen. The Qua -
e ly Jou nal o Economics, 102(3), 581–593. h ps://doi.
o g/10.2307/1884218
Schue el, P. (2021). De i: Decen alized inance-an in oduc ion
and o e iew. Jou nal o Inno a ion Managemen ,9(3),
I–XI. h ps://doi.o g/10.24840/2183-0606009.0030001
So enson, O., Asseno a, V., Li, G.-C., Boada, J., & Fleming,
L. (2016). Expand inno a ion inance ia c owd unding.
Science, 354(6319), 1526–1528. h ps://doi.o g/10.1126/
science.aa 6989
Ti ole, J. (2001). Co po a e go e nance. Econome ica, 69(1),
1–35.
Visma a, S. (2018b). Signaling o o e come ine iciencies
in c owd unding ma ke s. In The economics o c owd-
unding (pp. 29–56). Sp inge . h ps://doi.o g/10.1007/
978-3-319-66119-33
Visma a, S. (2016). Equi y e en ion and social ne wo k heo y
in equi y c owd unding. Small Business Economics, 46(4),
579–590. h ps://doi.o g/10.1007/s11187-016-9710-4
Visma a, S. (2018). In o ma ion cascades among in es o s in
equi y c owd unding. En ep eneu ship Theo y and P ac-
ice, 42(3), 467–497. h ps://doi.o g/10.1111/e ap.12261
Wang, D., Wu, S., Lin, Z., Wu, L., Yuan, X., Zhou, Y., Wang,
H., & Ren, K. (2021). Towa ds a i s s ep o unde s and
lash loan and i s applica ions in de i ecosys em. P oceed-
ings o he Nin h In e na ional Wo kshop on Secu i y in
Blockchain and Cloud Compu ing, 23–28. h ps://doi.o g/
10.1145/3457977.3460301
Wang, S., Cai, X., Guan, L., & Zhang, L. (2022). Wha do
ins i u ional in es o s b ing o ini ial coin o e ings (icos)?
T anspo a ion Resea ch Pa E: Logis ics and T anspo a-
ion Re iew, 167, 102876. h ps://doi.o g/10.1016/j. e.
2022.102876
Weill, P.-O. (2020). The sea ch heo y o o e - he-coun e ma -
ke s. Annual Re iew o Economics, 12, 747–773. h ps://
doi.o g/10.1146/annu e -economics-091819-124829
Ze zsche, D. A., A ne , D. W., & Buckley, R. P. (2020). Decen-
alized inance. Jou nal o Financial. Regula ion, 6(2),
172–203.
Zhao, R., Hou, W., Kuma , V., & Kuma , A. (2023). Signalling
e ec s o doing good in global ico ma ke s. Jou nal o
Business Resea ch, 168, 114224. h ps://doi.o g/10.1016/j.
jbus es.2023.114224
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d
o ju isdic ional claims in published maps and ins i u ional
a ilia ions.
123
1447Decen alized inance (DeFi) ma ke s o s a ups: sea ch ic ions...