Hacienda Pública Española / Re iew o Public Economics, 230-(3/2019): 95-124
© 2019, Ins i u o de Es udios Fiscales
h ps://doi.o g/10.7866/hpe- pe.19.3.4
Do Quasi-Hype bolic P e e ences Explain Academic
P oc as ina ion? An Empi ical E alua ion*
DAVID PATIÑO**
FRANCISCO GÓMEZ-GARCÍA***
Uni e sidad de Se illa
Recei ed: No embe , 2016
Accep ed: No embe , 2018
Abs ac
T adi ional neoclassical hough ails o explain ques ions such as p oblems o sel -con ol. Beha-
iou al economics ha e explained hese ma e s on he basis o he in e empo al p e e ences o indi-
iduals and, speci ically, he so-called (
β, δ
) model which emphasises p esen bias. This opens he way
o he analysis o new si ua ions in which people can adop inco ec indecisions ha make i necessa y
o he go e nmen o in e ene. The li e a u e which has de eloped he (
β, δ
) model and i s implica-
ions has gene a ed a ca ego isa ion o people ha is widely used bu which lacks a sys ema ic empi i-
cal e alua ion. I is impo an o alue he need o his public ac ion. In his a icle, we de elop a
me hod which makes i possible o e i y he main implica ions ha his model has o explain he
p oc as ina ion o uni e si y s uden s. Using an expe imen al ime discoun ask wi h eal mone a y
incen i es, we es ima e he s uden s’
β
and
δ
pa ame e s and we analyse hei co ela ion wi h hei
answe s o a se ies o ques ions conce ning how hey plan o s udy o an exam. The esul s a e am-
biguous gi en ha hey back some o he model’s conclusions bu ejec o he s, including a numbe o
he mos basic ones, such as he ela ion be ween p esen biases and some o he ca ego ies o people,
hese being essen ial o p edic hei beha iou .
Keywo ds: Beha iou al economics, p oblems o sel -con ol, wel a e analysis, expe imen al economics.
JEL Classi ica ion: I210, D90
1. In oduc ion
Economics has been b oadening i s subjec s o analysis o s udy ques ions which we e
p e iously ou side i s adi ional a ea o in e es . The ex ension has enabled he in oduc ion
* We hank Pablo B añas-Ga za and An onio M. Espín, he wo anonymous e e ees, and he execu i e edi o s o
hei help ul discussions and commen s.
** ORCID ID: 0000-0002-3313-561X.
*** ORCID ID: 0000-0002-6430-0331.
96
da id pa iño and ancisco gómez-ga cía
o a di e en pe spec i e o ha used in o he disciplines bu has highligh ed he di icul ies
o explaining a se ies o phenomena h ough he adi ional economic pe spec i e. This has
allowed o he econside a ion o nume ous ques ions o adi ional economic analysis and
has e en opened he doo o new p oposals o economic policies.
Beha iou al economics has been ounded on he s udy o phenomena ha con en ional
neoclassic economics does no ha e an explana ion o , o whose explana ions a e unsa is ac-
o y. I s analysis is based on he cogni i e limi a ions o people ha do no pe mi hem o
assimila e all he in o ma ion necessa y o adop complex decisions. This o en leads hem
o ollow simple ules which dis ega d a good pa o he salien in o ma ion. Fo example,
hey gi e mo e ele ance o e en s which ake place close o hem in space o in ime han
wha hey objec i ely ha e. Fu he mo e, i uses beha iou al biases ound by psychology o
unde s and beha iou al economics. Mad ian (2014) unde sco es he impo ance o his ques-
ion, as i allows o he loca ing o ma ke ailu es in addi ion o hose adi ionally consid-
e ed. Bu i also opens he way o p oposing new o mulas o economic policies o ca ying
ou a di e en alua ion o hose which a e usually applied. Congdon e al. (2011) de ine
h ee b oad ca ego ies o psychological biases ha can be he sou ce o ma ke ailu es:
impe ec op imisa ion, limi ed sel -con ol and non- adi ional p e e ences. Limi ed sel -
con ol, a phenomenon his wo k is cen ed on, is mani es ed in he disc epancy be ween
people’s in en ions and hei ac ual beha iou . People equen ly plan o beha e in a speci ic
way bu end up doing so ano he way. They likely p oc as ina e, o hey modi y hei choic-
es acco ding o hei emo ional s a e, o small ba ie s, which objec i ely a e no so, a e
signi ican impedimen s o hei ac ions. To dis ega d he e ec s o his issues can lead o
choosing mis aken ins umen s o economic policies. Fo ins ance, Campbell e al. (2011)
no e ha he e ec i eness o he supply o obliga o y in o ma ion as a way o esol ing
ma ke ailu es, such as he exis ence o ex e nali ies, is limi ed i he consume s do no
unde s and he in o ma ion, i hey belie e ha i is no ele an o he adop ing o hei deci-
sions, o i hey do no know how o access i o use i .
Sel -con ol p oblems can be unde s ood as he incapaci y o some subjec s o domina e
hei desi es o achie e hei aims. Among hem, p oc as ina ion s ands ou . This has con-
ce ned economis s since a leas S o z (1956). Acco ding o Ake lo (1991), p oc as ina ion
akes place when he cu en cos s a e unduly s essed in compa ison wi h hose o he u u e.
This leads people o pos pone asks wi hou ealising ha , when i is again he ime o do he
ask, hey will pu i o again. The mos widesp ead explana ion o his way o beha ing is
based on people’s in e empo al p e e ences and explains how beha iou is planned in ime,
and why such a plan is eneged when i implies ca ying ou asks ha a e cos ly in e ms o
e o . The decision o he p esen educes he u u e well-being and people la e eg e hei
choices. The phenomenon is analogous o an ex e nali y owa ds onesel and i is some imes
deno ed as in e nali y. The adi ional conclusion o neoclassical economics is ha people a e
he bes gua an o s o hei own in e es s and supposes ha hey a e he ones who bes know
how o choose wha will imp o e hei well-being. Ye his is no ensu ed when biases exis .
To measu e he in e nali ies equi es iden i ica ion o he impac o agen s’ choices unde
hei own expe ienced u ili y. This is simila o how a adi ional ex e nali y equi es iden i-
ica ion o an agen ’s impac on he u ili ies expe ienced by o he s.
97
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
The analysis o p oc as ina ion has been used o explain phenomena such as d ug addic-
ion and, in gene al, he adop ion o nume ous habi s conside ed ha m ul o unheal hy (Read
and Van Leeuwen, 1998). In he a ea o economics, i s consequences ha e been especially
s udied o decisions wi h espec o sa ings (Thale and She in, 1981). P oc as ina ion has
also been used o illus a e why he a ailabili y o pay wi h a c edi ca d g ows as pos poning
he paymen educes he cu en alue o he deb (P elec and Simes e , 2001); explain he
unc ioning o bu eauc acies in an al e na i e way o he agen -p incipal model (Ake lo ,
1991); and show ha spec a o s’ choices o hei ype o ilms gi e ise o biases owa ds
comme cial ilms (Read e al., 1999), o ci e only a ew ou s anding examples.
Faced wi h he design o public ac ions, he de ailed knowledge o hese aspec s is
impo an as he di e ence be ween people’s in en ions and hei ac ions can a y as a
esponse o e y small changes in he con ex o hei choice (Mad ian and Shea, 2001).
Bu u he mo e, he deg ee o sel -con ol depends on he cu en s a e o he decide s and
hei emo ions. Elemen s such as s ess, an o e loading o in o ma ion o ea can se o
impa ience and mo i a e adical changes in beha iou . Acco ding o beha iou al econom-
ics, in he cases in which many people show cogni i e biases o a lack o sel -con ol, he
ole o he go e nmen should no be limi ed o a minimum, gi en ha people canno ee
hemsel es om he mis akes o hei decisions. I is indispensable o know he mechanism
which p oduces hese disc epancies o disce n when a nudge is necessa y (Thale and
Suns ein, 2008).
Che y (2015) poin s ou ha he decision o include beha iou al elemen s in economic
models mus be conside ed as being mo e a p agma ic han a philosophical choice. Ne e he-
less, gi en he mul i ude o biases which dis ance people om he beha iou p edic ed by
con en ional models, i is necessa y o de e mine which a e decisi e and in oduce hem. To
iden i y he op imal policy equi es e alua ing he ex en o which he u ili ies expe ienced
by people di e om he decisions ha hey eally adop . Ye his opens he doo o a bi-
a iness. This is why i is impe a i e o empi ically measu e he deg ee o which u ili y and
decisions a e de ached om each o he , which explains he me hods ha we p opose o use
in his wo k. Speci ically, he li e a u e has sugges ed measu ing expe ienced u ili y using
da a on sel - epo ed happiness. This is an analogous app oach o ha employed in he con-
ingen e alua ion me hods which assess ex e nali ies (Diamond and Hausman 1994). Like-
wise, he idea has a isen in o he a icles o calib a ing he s uc u al pa ame e s o a model
ha includes beha iou al biases. We employ his no ion in he cen al pa o his wo k.
In his line, his a icle p oposes me hods o measu e people’s deg ee o e o in hei
decisions and when hey do so, as well as o analyse hei consequences in e ms o well-
being. I s aim is o analyse a speci ic eali y – he daily ac i i y o uni e si y s uden s in
p epa ing a subjec – and measu e he deg ee o which his p ocess i s wha he heo y p e-
dic s. To do so, an empi ical me hodology is in oduced which enables his e i ica ion o be
ca ied ou .
The so-called quasi-hype bolic discoun ing allows o modelling o he beha iou o
people who pos pone hei decisions o p oc as ina e. I s use has been gene alised and has
98
da id pa iño and ancisco gómez-ga cía
os e ed he de elopmen o a ypology o people wi h di e en beha iou s ega ding sel -
con ol p oblems. Howe e , he ela ion be ween quasi-hype bolic discoun ing and di e en
ime p e e ences has no been he objec o a sys ema ic empi ical e alua ion. The main aim
o his a icle consis s in e i ying whe he he model explains he p oc as ina ion o a sam-
ple o uni e si y s uden s when pe o ming hei academic ac i i ies. To do so, we ha e ca -
ied ou wo su eys which pe mi us o ind ou hei s udy habi s and he s uden s’ cha ac-
e is ics. One o hem includes a habi ual discoun ask ha has enabled o us o in e he
s uden s’ ime p e e ences and o cha ac e ise hem acco ding o he deg ee o consis ency
ha hey p esen . This in o ma ion allows o e i ica ion o mos o he implica ions o he
explana ion o sel -con ol p oblems based on quasi-hype bolic p e e ences. Speci ically, he
ela ion be ween p esen biases, he ype o ime p e e ences people ha e, and hei beha -
iou and he cos s o hei sel -con ol p oblems in e ms o well-being and poo academic
pe o mances a e e i ied.
The a icle’s main conclusion pe mi s he es ablishmen o an in e se empi ical ela ion
be ween he size o he p esen bias and main aining beha iou s consis en wi h he s uden s’
ime p e e ences, as he model ha we aim o e i y p edic s. None heless, we ha e no
ound a ela ion wi h he es o he ca ego ies o people o wi h he es o he heo y’s im-
plica ions. We belie e ha ou esul s a e impo an in ha hey shed ligh on he almos
non-exis en empi ical basis o he (
β, δ
) model and i s conclusions. On he o he hand, ou
e alua ion also gi es keys o aluing he ex en o which i is necessa y o design new ac ion
ins umen s in he educa ional a ea.
The a icle is s uc u ed in 6 sec ions, including his in oduc ion. In he second, we e-
iew he economic li e a u e ha analyses sel -con ol p oblems, hei implica ions o pub-
lic policies and hei ela ionship wi h ime p e e ences. The hi d indica es he empi ical and
expe imen al me hodology ollowed o analyse he ques ions p oposed. The ou h desc ibes
how he da abase was buil and ca ies ou a b ie analysis o i s desc ip i e s a is ics. The
i h shows he models which ha e been used o empi ically e i y he aspec s analysed and
ex ensi ely analyses he esul s ob ained. The a icle ends wi h a conclusions sec ion.
2. The p oblem o sel -con ol and in e empo al p e e ences
The mains eam economic analysis ha assumes ha a ional people adop esul s con-
sis en wi h hei p e e ences has g ea di icul ies in explaining sel -des uc i e beha iou s,
o ins ance d ug addic ion o compulsi e ood consump ion. Dissa is ac ion wi h he ap-
p oach, in spi e o i s a emp s1 o explain such phenomena, has os e ed he sea ch o al e -
na i es based mainly on concep s common in psychology and amed in he a ea o beha -
iou al economics. These explana ions o lack o sel -con ol ha e e ol ed a ound ime
p e e ences and possible sho sigh ed calcula ions o he bene i s and cos s o ac ions2. I -
che and Za ghamee (2011) indica e ha he psychological amewo k o lack sel -con ol
o e laps wi h he economic concep o ime p e e ence.
99
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
The analysis o he p oblems o sel -con ol lies wi hin he s udy o pa hological di e -
gences be ween he choices o people and hei p e e ences. The mos accep ed explana ions
a e based on he p oposal ha he e a e wo ypes o hough : one which gi es swi , au-
oma ised and unconscious answe s; he o he is slow hough ha is logical and is done
consciously (Kanheman, 2001). Fo example, Be nheim and Rangel (2004) use his ame-
wo k o analyse d ug addic ion. Fo hese au ho s, he mechanisms o semi-au oma ic an-
swe s a e bene icial, especially in s able en i onmen s, because hey gene a e quick answe s
in mul iple ci cums ances. No wi hs anding, hey can lead o sys ema ic mis akes ha can be
se ious. In hei model, people can make decisions “coldly”, imposing cogni i e con ol.
This ype o decisions esul s in he choice o he al e na i e p e e ed. Bu he e also exis s
a “ho ” mode in which decisions and p e e ences can di e .
Thale and She in (1981) con empla e a double pe sonal plan in he adop ion o deci-
sions o explain sel -con ol p oblems. Each pe son has a a sigh ed-planne and sho sigh -
ed-doe na u e which main ains a kind o agen -p incipal ela ion wi h di e gen in e es s.
The planne ob ains u ili y uniquely h ough he ac ions ha he execu o ca ies ou . The
model p edic s ha people will es ablish es ic ions o hei own beha iou mainly in he
ac ions whose bene i s and cos s a e p oduced a di e en momen s. The ac ions o he plan-
ne s can consis o modi ying he p e e ences o he execu o , ac ing on hei incen i es o
limi ing hei se o possibili ies o choice. Gul and Pesendo e (2001) show, in a simila
amewo k, how emp a ions can be comba ed by es ablishing limi a ions o he se o
choices. Likewise, Fudenbe g and Le in (2006) indica e ha his iew is compa ible wi h
much e idence o magne ic esonance images, as many decision p oblems can be explained
as a game be ween a sequence o impulsi e sho - e m sel es and pa ien long- e m sel es.
Models based on an agen -p incipal p oblem cen e hei explana ion on ime p e e ence
biases. O’Donoghue and Rabin (1999) explain how people p oc as ina e. The “long- e m
sel ” es ablishes he plan, bu wha is commonly called losses o sel -con ol, caused by
p esen biased ime p e e ences, a ises. Thei e ec is ha immedia e g a i ica ions a e al-
ued o a g ea e ex en han i he ac ions had been ca ied ou a a la e momen . This same
idea has os e ed la e e sions ha ha e modelled a b oad ange o decisions, such as sa ing
and d ug consump ion.
This has p opaga ed he need o e hink he explana ion o how decisions in ime a e adop ed,
which has been domina ed by he heo y o discoun ed u ili y. This heo y was de eloped by
Samuelson (1937), who ex ended I ing Fishe ’s p e ious idea o mul iple pe iods. Discoun ed
u ili y educes all mo i es which lead people o alue he u u e in ela ion wi h he p esen o a
unique pa ame e known as he discoun a e. The discoun ac o enables people o in e change
he u u e u ili y wi h ha o he p esen . Koopman (1960) la e demons a ed ha he model
could be ob ained om a se ies o plausible axioms and his model gained in ele ance.
Mo e o mally, he s anda d model o empo al p e e ence designed by Samuelson
(1937) is based on he exis ence o an exponen ial empo al discoun ing a e which is cons-
an o e ime. Fo all , he u ili y o an indi idual would be:

100
da id pa iño and ancisco gómez-ga cía
(1)
Whe e
δ
€ (0, 1] is he discoun ing ac o .
I indi iduals ha e a bias owa ds immediacy, i is necessa y o weigh he emo eness o
nea ness o he e en . This can be in oduced by employing a quasi-hype bolic empo al
discoun ing model; see S o z (1956), Phelps and Pollak (1968) and Laibson (1997)3 Con-
c e ely, he bias o he p e e ences is inse ed ia a unc ion designed by Phelps and Pollak
(1968) in he con ex o in e gene a ional al uism. This unc ion adds an addi ional ac o o
Samuelson’s in e empo al p e e ences which weigh he u ili ies ob ained in pe iods ollow-
ing ha which is aken as a e e ence. In his way he model in oduces he p esen bias by
o e discoun ing he u ili y ob ained in pe iods subsequen o he e e ence. We can ew i e
he u ili y unc ion o include such biases as:
(2)
Whe e 0 ≤
β
,
δ
≤ 1,
β
measu es he p esen bias. I i is close o 1 i ha dly exis s, ha is
o say, he now is no especially alued wi h espec o he a e wa ds. On he con a y, a
β
close o 0 indica es an impa ience o excessi e eage ness o achie e an immedia e ewa d.
The model explains he decision o unde ake ac ions whose bene i s and cos s a e gen-
e a ed a di e en momen s. The p oblem o sel -con ol a ises when he discoun a e ises
a he ime o pe o ming he ac ion, gene a ing a ecalcula ion o he o al balance o ben-
e i s and cos s s emming om i . The esul may be di e en o ha p o ided by he long-
e m discoun ing a e and cause a change o decision. People do no change hei p e e -
ences, o a leas hey do no change hem pe manen ly o s ably. Once he momen has
passed, hey e u n o a s able o e lexi e si ua ion. To e alua e he cos o he lack o sel -
con ol, he e e ence is he decisions ha a pe son wi h ime consis en p e e ences would
adop and which a e hose ha would be chosen in he long un, gi en hei ime p e e ence.
O’Donoghue and Rabin (1999) p opose a classi ica ion o people acco ding o hei ime
p e e ences. People wi h a p esen bias ha e ime p e e ences consis en o e ime and do
no su e om sel -con ol p oblems. We can dis inguish wo ypes among hose who ha e
a p esen bias. On he one hand, sophis ica ed people a e awa e o hei bias and o he sel -
con ol p oblems ha his will cause hem. To a oid hem hey adop measu es which, in
gene al, consis in ca ying ou he ac ion be o e. The esul is subop imal bu be e han no
ac ion4. Naï e people do no o esee ha hey will su e sel -con ol p oblems. They ha e
p esen biases he same as sophis ica ed people bu , unlike hem, hey plan he u u e igno -
ing hei p esen biases. As hey do no adop any kind o cau iona y measu e, i is likely ha
hey will suppo he o ali y o he cos s o well-being. These a e due o no adjus ing o he
planned beha iou which, a pos e io i, hey would ha e liked o ca y ou .
O’Donoghue and Rabin (2008) la e in oduced he ca ego y o he pa ially sophis i-
ca ed o de ine people who a e awa e o hei p esen bias bu unde es ima e i s deg ee. The
101
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
condi ion can be in oduced using a pa ame e ha we can deno e by ^
β
, which measu es he
agen s’ es ima ions o he size o hei own biases. In he case o a pe son wi h consis en
ime p e e ences, ^
β
=
β
=1. I a bias exis s,
β
< 1. Naï e people belie e ha hei beha iou
will be consis en wi h hei p e e ences, bu ac ually hey ha e a p esen bias, he e o e,
^
β
=1>
β
. I he agen s a e sophis ica ed, hey co ec ly p edic hei p esen bias, and he e o e
hei sel -con ol p oblems, so ^
β
=
β
<1 will occu . Finally, pa ially sophis ica ed agen s will
ha e ^
β
<
β
<1 as hey a e awa e o hei sel -con ol p oblems bu unde es ima e hei magni-
ude5.
The empi ical li e a u e has concen a ed on es ing he ela ionship be ween discoun ing
a es and beha iou s which e eal a lack o sel -con ol o a emp o unde line he lack o
cohe ence o he adi ional ision in explaining compulsi e beha iou s. The mos common
p ac ice has been o exploi he e idence p o ided by labo a o y o ield expe imen s, which
a e g ounded on some me hod o in e ence o people’s empo al p e e ences. These expe i-
men s usually consis o asking he indi iduals o choose be ween sums o money, eal o
ic i ious, which a e smalle in a close momen in ime and g ea e la e 6, in o de o calib a e
when he u ili y o bo h is balanced7. Fo example, Meie and Sp enge (2012) ha e s udied
he ela ionship be ween p esen bias and he inancial sol ency o indi iduals. Reynolds
(2006) explains d ug consump ion and gambling, Ki by e al. (1999) he oin addic ion,
Bickel e al. (1999) smoking and Welle e al. (2008) obesi y.
Reuben e al. (2015) s and ou o ha ing a di ec ela ion wi h ou s udy aim. Thei wo k
analyses he ela ion be ween ime p e e ences and p oc as ina ion h ough a se ies o labo-
a o y expe imen s and ield wo k wi h a popula ion o s uden s. They es ima e he pa ame-
e s which de ine he ime p e e ences ia a se o asks o he ype indica ed in he p e ious
pa ag aph and he le el o us , cogni i e skills and gende a e among he con ols used.
Bu ks e al. (2012) compa e he goodness o di e en me hods o in e ing ime p e e -
ences, con as ing he ex en o which he discoun ac o s es ima ed by each one explain
di e en phenomenon. Speci ically, hey analyse he accumula ion o human capi al (Eckel
e al., 2007), sa ings (Ash a e al., 2006) and academic esul s (e.g., Shoda e al., 1990).
The di e en es ima ions use expe imen s ca ied ou on middle-aged wo ke s wi h low skill
le els. The unc ional o m which bes p edic s he decisions analysed is quasi-hype bolic
discoun ing, calcula ed om a se o choices o e sums o money a di e en momen s in
ime.
Ano he ou s anding wo k is Na do o’s (2011), which iden i ies he di e en ca ego ies
o people desc ibed abo e, along wi h hei cha ac e is ics. I uses a sample o equencies
o access and a i s con ac ed by use s o a uni e si y gymnasium along wi h hei aca-
demic quali ica ions. They build a subop imal index o “cos ” o no ul illing hei own plan
ha is explained by he people’s cha ac e is ics. This di ides he people in o consis en (o
a ional), naï e o sophis ica ed, compa ing he planned beha iou wi h wha is inally ca -
ied ou . To build his ixed classi ica ion, a h eshold o 25% o mis akes includes he unp e-
dic able mo i es ha p e en ul illing he plan. This wo k inds ha 40.6% o he people
102
da id pa iño and ancisco gómez-ga cía
p edic well and a e ca alogued as a ional. 51.1% ha e op imis ic p e e ences wi h espec
o hei o ecas s o a endance and a e ca alogued as naï e, and 4.3% a e ca alogued as so-
phis ica ed.
Las ly, Wong (2008) analyses he p epa a ion o he inal exam o a subjec by a g oup
o deg ee s uden s. He iden i ies consis en , sophis ica ed, ingenuous and pa ially ingenuous
indi iduals using wo ques ionnai es. The i s ques ionnai e is abou he amoun o s udy
ha s uden s conside ideal and which hey es ima e ha hey will ac ually do and is asked
hal way h ough he e m. The second ques ionnai e is done he day o he exam and in e s,
a pos e io i, he amoun o s udy eally done. This wo k inds a small pe cen age o consis-
en s uden s who us in ul illing hei ideal s udy plan and indeed do so. Among he in-
consis en , h ee beha iou pa e ns a e iden i ied. The naï e who p edic he ul ilmen o
hei ideal s udy plan bu do no ul il i . The sophis ica ed who ul il hei o ecas s bu
whose plans do no co espond wi h wha is ideal. The au ho in e p e s his beha iou wi h
an awa eness o u u e sel -con ol p oblems and he design o a plan o minimise hei con-
sequences. Las ly, he pa ially naï e a e awa e ha hei sel -con ol p oblems will lead
hem o no ul illing hei ideal plans and so hey also design plans which y o compensa e
o his, bu hey do no ul il hem. Wong (2008) employs he delay o eseen in he ideal plan
o measu e he deg ee o empo al inconsis ency and he delay o eseen in he chosen plan
as a measu e o he indi idual’s deg ee o sophis ica ion.
The a icle concludes ha any delay, o eseen o no , has nega i e e ec s on he aca-
demic esul s, e en con olling o he ime eally dedica ed o s udying, which egis e s
hose caused by easons o he han sel -con ol p oblems. I unde lines ha sophis ica ed
indi iduals do no manage o educe he nega i e e ec s o sel -con ol p oblems. The a icle
in e p e s his esul as being a consequence o hese indi iduals’ poo dis ibu ion o s udy
ime.
Ou wo k ollows Wong’s closely, bu addi ionally in oduces an elici ing o ime p e e -
ences. In his way, we can empi ically e alua e i he (
β, δ
) model p edic s he cha ac e isa-
ion o people acco ding o hei ime p e e ences om he p esen bias and he discoun
ac o , in he way a gued by O’Donaghue and Rabin (1999, 2008).
3. Me hodology and da abase
3.1. P ocedu e
Ou s udy is ounded on a da abase o deg ee s uden s ha we elabo a ed ou sel es and
which was ob ained ia su eys. I s design, desc ibed below, analyses hei beha iou in
planning, p epa ing and de eloping o hei academic ac i i ies, as well as hei esul s. This
beha iou di e s among hem in he planning o he p epa a ion o he subjec and in he
deg ee o which hey ul il his plan. Fu he mo e, wi hin he g oup o hose who do no
103
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
ul il hei plan, people can be dis inguished by hei deg ee o awa eness o hei u u e
ailu es. The s uden s likewise di e in hei in e empo al p e e ences, cha ac e ised mainly
by each one’s discoun ac o and p esen bias.
A ime discoun ask was implemen ed o in e he discoun a es and he possible p e-
sen biases. On he o he hand, in o de o classi y he people we had o ob ain in o ma ion
on he ex en o which hey ul il hei plan and i hey a e awa e, a p io i, o wha hey a e
going o do. The ques ionnai es ha include he expe imen al discoun ask ha e been de-
signed combining he me hodologies de eloped by Bu ks e al. (2012) and by Wong (2008).
The hypo hesis which we e i y is ha he likelihood o he s uden s ul illing hei plans
depends on he ype o in e empo al p e e ences ha hey ha e, con olling o hei di e -
en pe sonal and socio-economic cha ac e is ics. Mo eo e , we examine i he p esen biases
lead o sel -con ol p oblems and, whe e app op ia e, o suppo ing he cos s o well-being
which hey cause, as O’Donaghue and Rabin (1999, 2008) p edic . In his way we es he
empi ical basis o hei model.
The sample includes s uden s o he compulso y subjec o Mac oeconomics, co e-
sponding o he second cou se-yea o he Deg ee o Finance and Accoun ing o he Uni e -
si y o Se ille, Spain. The subjec is augh in 8 g oups o a simila size: hal in he mo ning
and he o he hal in he a e noon. The cen e de e mines he assigna ion o he s uden s o
he g oups by ex a-academic c i e ia, which es ablishes a simila p o ile in all o hem. The
con en o he cou se is he same and includes an iden ical exam o all he s uden s.
3.2. Ques ionnai es and classi ica ion o indi iduals
The da a was ob ained h ough wo ques ionnai es. The i s ques ionnai e is done mid-
e m and con ibu es mos o he in o ma ion, including he expe imen and he con ols. A
his poin in he cou se, he s uden s had in o ma ion on he con en o he subjec and i s
di icul y and could ca y ou a p ecise es ima ion o he equi emen s o he wo k needed o
p epa e i . The su ey was ca ied ou du ing he same week wi h all he g oups in he second
hou o a wo-hou session. Each s uden signed his/he au ho isa ion o pa icipa e in he
expe imen and ead he pape ’s ins uc ions. The ins uc ions cla i ied he olun a iness o
he ac i i y and ha i would no a ec he ma k o he subjec in any way. Likewise, hey
indica ed p ecisely how o do he ime discoun ask. The p ocess by which he da a would
be anonymised was also explained, placing special emphasis on he need o gi e since e
answe s. In his way, he s uden s had incen i es o espond o he discoun ask. This made
i possible o also since ely answe he es o he ques ions. The esea che s did no each
mos o he g oups and signi ican di e ences we e no ound in he answe s gi en by hei
s uden s and he es . In any case, in spi e o he insis ence on he since i y o he answe s
and ha he e was an anonymous handling o he ques ionnai es om he momen in which
hey we e handed ou , he e exis s he possibili y o an expe ienced demand e ec , as au ho s
such as Zizzo (2010) and De Quid e al. (2017) indica e. Ne e heless, hese same au ho s
110
da id pa iño and ancisco gómez-ga cía
include all he con ols. The esul shows ha he likelihood o being consis en dec eases he
g ea e he p esen bias is. This ag ees wi h ou ini ial hypo hesis, hough he ela ion is
ela i ely weak. The comple e model (4) also iden i ies o he cha ac e is ics o people wi h
consis en ime p e e ences. Speci ically, hey end o do a g ea e o al numbe o s udy
hou s. In con as , s uden s wi h lowe le els o mon hly allowances ha e a lowe endency
o ha e ime p e e ences o his ype. This esul sugges s ha s uden s wi h lowe le els o
income could be adop ing wo se decisions, which coincides wi h ecen indings in his
ein15.
Table 3
RELATIONSHIP BETWEEN CONSISTENT PEOPLE AND THEIR TIME PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a -7.107 -6.601 -8.901 -8.846
(7.250) (7.302) (7.058) (7.085)
Be a 0.238 0.232 0.273* 0.278*
(0.170) (0.172) (0.163) (0.163)
Mon hly allowance unde €200 -0.127* -0.115*
(0.067) (0.067)
Mon hly allowance be ween €200 and €300 -0.135* -0.129*
(0.073) (0.075)
To al s udy ime 0.002*** 0.002***
(0.000) (0.000)
Cons an 7.002 6.920 8.957 8.600
(7.122) (7.179) (6.902) (6.939)
Obse a ions 214 214 209 209
R-squa ed 0.009 0.019 0.235 0.250
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
The analysis has been limi ed o subg oups in o de o be e know he cha ac e is ics o
people wi h hese p e e ences. Speci ically, he ela ion be ween he
β
ac o and he condi-
ion o consis en is only ound when he sample is limi ed o women, o s uden s o he a -
e noon classes and o non- epea e s. In he case o he i s wo g oups, he signi icance ises
o 1%. A ela ion be ween he p esen biases and consis en p e e ences is no no ed in he
es o he g oups. This indica es ha he e may be di e en pa e ns which explain people’s
ype o ime p e e ences. The obus ness o he esul s has been es ed es ima ing logi and
p obi models. The es ima ions a e simila o hose ob ained wi h he LPM, hough he
s a is ic o he
β
ac o coe icien only a ains he alue o 1.32 and canno be conside ed
s a is ically signi ican .

111
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
Table 4 shows he alue o he coe icien s and he s anda d de ia ion o he
δ
discoun
a e and o he
β
ac o o he es ima ions o he g oups conside ed and o he disc ee choice
models o he model which includes all he con ols.
Table 4
RELATION BETWEEN CONSISTENT PEOPLE AND THEIR TIME PREFERENCES.
ANALYSIS OF THE ROBUSTNESS OF THE RESULTS
Va iables (1)
P obi
(2)
Logi
(3)
Men
(4)
Women
(5)
Repea -
e s
(6)
Non- e-
pea e s
(7)
Mo ning
classes
(8)
A e -
noon
classes
Del a -23.256 -39.977 -2.029 -12.542 -15.586 -9.509 -1.837 -11.850
(52.058) (95.744) (11.465) (9.352) (18.671) (8.247) (11.622) (10.031)
Be a 1.416 2.752 0.093 0.559*** 0.213 0.339* -0.235 0.770***
(1.133) (2.088) (0.290) (0.205) (0.331) (0.199) (0.242) (0.239)
Cons an 14.931 20.211 2.990 11.463 15.294 9.258 2.628 10.962
(453.614) (899.217) (11.199) (9.215) (18.687) (8.066) (11.414) (9.741)
Obse a ions 209 209 89 120 44 165 104 105
R-squa ed 0.303 0.422 0.436 0.281 0.280 0.384
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
The es o he ime p e e ence ypologies do no ha e a s a is ical ela ion wi h he p e-
sen biases. The speci ic case o he naï e is especially ou s anding as he heo y conside s
ha he disp opo iona e p e e ence o immedia eness is i s main de e minan . Tables 5 and
6 show he inexis ence o a s a is ical ela ion be ween discoun a es and p esen biases and
he conside a ion o people as naï e o sophis ica ed.
Table 5
RELATIONSHIP BETWEEN NAIVE PEOPLE AND THEIR TIME PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a -0.385 -0.643 0.145 -0.232
(8.519) (8.568) (9.059) (9.151)
Be a -0.008 -0.010 0.010 0.004
(0.200) (0.201) (0.209) (0.210)
Age-squa ed -0.001 -0.001* -0.001
(0.000) (0.001) (0.001)
Cons an 0.567 0.140 -0.621 -0.073
(8.369) (8.424) (8.858) (8.962)
Obse a ions 214 214 209 209
R-squa ed 0.000 0.013 0.058 0.065
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
112
da id pa iño and ancisco gómez-ga cía
Table 6
RELATIONSHIP BETWEEN SOPHISTICATED PEOPLE AND THEIR TIME
PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a -3.962 -4.195 -10.191 -11.262
(11.099) (10.997) (11.291) (11.400)
Be a -0.076 -0.051 0.028 0.039
(0.261) (0.258) (0.261) (0.262)
Age -0.088** -0.084** -0.081*
(0.040) (0.041) (0.042)
Age-squa ed 0.001* 0.001* 0.001*
(0.001) (0.001) (0.001)
Gende -0.107 -0.152** -0.167**
(0.069) (0.072) (0.074)
Li ing wi h hei pa en s -0.195*** -0.196***
(0.072) (0.072)
College-educa ed a he 0.239** 0.260**
(0.101) (0.103)
College-educa ed mo he -0.225** -0.225**
(0.106) (0.107)
Cons an 4.464 6.078 12.078 13.023
(10.903) (10.812) (11.041) (11.166)
Obse a ions 214 214 209 209
R-squa ed 0.003 0.045 0.144 0.151
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
As is no ed in Table 6, he s uden s o he sample who a e women, hose who do no li e
wi h hei pa en s and he younges ha e a g ea e endency o beha e as sophis ica ed. No
cha ac e is ic is ound which a ou s he conside a ion o people as naï e.
The obus ness checks ca ied ou con i m hese esul s wi h he unique excep ion o
men wi h naï e p e e ences. This g oup has a endency o p esen lowe discoun a es. Tha
is o say, he mos impa ien men a e mo e likely o ha e naï e p e e ences bu , con a y o
wha he heo y sugges s, his p obabili y dec eases wi h he p esen bias (i g ows wi h he
β
ac o ) in a ma ginally signi ican manne . Table 7 shows he es ima ed alues o naï e
s uden s in di e en g oups o he sample.
As we indica e, he endency o p oc as ina e is also measu ed h ough a subjec i e index o
pos ponemen o academic asks. Table 8 es ima es he ela ion be ween his and ime p e e -
113
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
ences h ough he LPM. As we see, he deg ee o which people p oc as ina e is no explained by
he pa ame e s which de e mine hei ime p e e ences. The o al numbe o hou s o s udy o he
exam and he s uden s’ sa is ac ion wi h he ca ee s udied a e he only de e minan s o hei
p opensi y o p oc as ina e. Tha is o say, he mo e mo i a ed and mo e s udious s uden s a e he
ones who end o pos pone hei asks less. In spi e o he ela ions no being s a is ically signi i-
can , he impo an quan i a i e e ec s ha he discoun a es and he p esen biases ha e also
s and ou , conside ing he small a ia ions which a e p oduced in hese a iables.
Table 7
RELATION BETWEEN NAÏVE PEOPLE AND THEIR TIME PREFERENCES.
ANALYSIS OF THE ROBUSTNESS OF THE RESULTS
Va iables (1)
P obi
(2)
Logi
(3)
Men
(4)
Women
(5)
Repea -
e s
(6)
Non- e-
pea e s
(7)
Mo ning
classes
(8)
A e -
noon
classes
Del a 0.210 -10.522 -30.654** 17.081 -0.209 -8.907 -18.235 5.429
(36.970) (64.152) (14.299) (12.981) (38.391) (9.547) (14.387) (14.058)
Be a 0.042 0.273 0.678* -0.301 -0.303 0.295 0.464 -0.328
(0.834) (1.480) (0.362) (0.284) (0.680) (0.231) (0.300) (0.335)
Cons an -8.409 -4.889 29.145** -16.063 7.031 7.801 16.807 -4.791
(35.700) (61.869) (13.967) (12.791) (38.423) (9.338) (14.130) (13.652)
Obse a ions 196 196 89 120 44 165 104 105
R-squa ed 0.240 0.122 0.411 0.125 0.156 0.115
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
The analysis o he obus ness o his a iable does no include he es ima ion o disc e e
choice models gi en ha we ha e ea ed i as con inuous. Table 9 shows he es o he
g oups conside ed be o e. As we can see, a s a is ically signi ican ela ion be ween he dis-
coun a es and he p oc as ina ion index has been ound in he case o he men, he non- e-
pea e s and he s uden s in he mo ning g oups. In hese g oups, he people wi h g ea e
discoun a es end o pos pone hei academic ac i i ies o a g ea e ex en han he es . This
e eals a ela ion be ween impa ience and p oc as ina ion.
The explana ion o p oc as ina ion h ough he quasi-hype bolic discoun also p edic s ha
people wi h high p esen biases end o ha e sel -con ol p oblems as hey pos pone he asks ha
hey plan o ca y ou . In ou case his can lead o he s uden s no ul illing he plan o p epa a ion
o he exam and inishing up wi h bad esul s. The e o e, he e should exis a ela ion be ween
he academic esul s and he p esen biases. To check his, we ha e es ima ed a model simila o
he p e ious ones, bu which con ols by he g oup in which he s uden a ends class in o de o
g oup oge he he di e ences in he eache , he companions and he class hou . Addi ionally,
a iables ela ed o he ac i i y and he educa ional aining o he pa en s ha e been in oduced.
The hi d g oup o a iables has been main ained as in he es o he es ima ions.
114
da id pa iño and ancisco gómez-ga cía
Table 8
RELATIONSHIP BETWEEN THE SUBJECTIVE INDEX OF POSTPONEMENT
AND TIME PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a 52.510 67.667 66.617 54.050
(70.122) (69.194) (71.373) (70.154)
Be a 0.047 -0.364 -0.690 -0.545
(1.642) (1.621) (1.642) (1.610)
Age 0.396 0.511** 0.335
(0.248) (0.257) (0.257)
Age-squa ed -0.008** -0.009** -0.006
(0.004) (0.004) (0.004)
To al s udy ime -0.009*** -0.009***
(0.003) (0.003)
Sa is ac ion wi h hei s udies -0.453***
(0.156)
Cons an -44.760 -64.660 -61.378 -43.958
(68.916) (68.012) (69.806) (68.743)
Obse a ions 211 211 206 206
R-squa ed 0.005 0.053 0.134 0.184
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
Table 9
RELATIONSHIP BETWEEN THE SUBJECTIVE INDEX OF POSTPONEMENT AND TIME
PREFERENCES. ANALYSIS OF THE ROBUSTNESS OF THE RESULTS
Va iables (1)
Men
(2)
Women
(3)
Repea e s
(4)
Non-
epea e s
(5)
Mo ning
classes
(6)
A e noon
classes
Del a 213.095* 14.602 -86.065 136.019* 189.116* 78.862
(119.312) (91.503) (209.734) (75.653) (102.543) (106.190)
Be a -5.760* 0.403 -0.637 -1.972 -0.479 -2.013
(2.956) (2.030) (3.714) (1.832) (2.111) (2.529)
Cons an -191.504 -11.644 66.884 -120.056 -180.452* -55.253
(116.759) (90.124) (209.740) (74.029) (100.721) (103.124)
Obse a ions 87 119 43 163 101 105
R-squa ed 0.379 0.177 0.537 0.285 0.295 0.304
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
115
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
As can be seen in Table 10, he e is a posi i e ela ion be ween he
β
ac o , he in e se
o he p esen bias and he ma ks. Addi ionally, he nega i e sign o he discoun ac o indi-
ca es ha as i g ows ( he lowe he discoun a e), he ma ks dec ease o , in o he wo ds,
people wi h a lowe p e e ence o immedia eness o wi h a g ea e endency o wai ob ain
be e sco es. Bo h esul s a e cohe en wi h he in ui i e idea and e i y he hypo hesis p e-
dic ed by he heo y. Howe e , no s a is ical ela ion has been ound be ween he di e en
ypes o s uden s and hei ma ks.
Table 10
RELATIONSHIP BETWEEN MARKS AND TIME PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a 10.216 -26.666 -93.501** -91.110**
(40.998) (38.622) (40.161) (40.140)
Be a 0.123 0.945 1.956** 1.967**
(0.965) (0.916) (0.950) (0.943)
Gende 0.353 0.542** 0.444*
(0.243) (0.260) (0.268)
G oup 2 1.995*** 1.385** 1.245*
(0.623) (0.667) (0.679)
G oup 3 2.667*** 2.063*** 1.886***
(0.597) (0.654) (0.655)
G oup 4 1.579** 0.892 0.923
(0.612) (0.660) (0.658)
G oup 5 2.738*** 1.735** 1.679**
(0.624) (0.680) (0.683)
G oup 6 1.416** 0.539 0.394
(0.623) (0.681) (0.682)
G oup 7 0.456 -0.338 -0.376
(0.624) (0.681) (0.678)
G oup 8 1.732*** 0.652 0.502
(0.665) (0.724) (0.723)
Mon hly allowance be ween €200 and €300 -0.706 -0.725*
(0.430) (0.430)
Li ing wi h hei pa en s -0.610** -0.578**
(0.259) (0.257)
Deg ee o isk 1 2.360** 1.967*
(1.117) (1.123)
Deg ee o isk 2 2.826*** 2.417**
(1.056) (1.065)

116
da id pa iño and ancisco gómez-ga cía
(Con inued)
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Deg ee o isk 3 2.433** 1.903*
(1.024) (1.045)
Fa he en ep eneu +10 employees 0.756 0.876*
(0.520) (0.523)
Mo he pensione -1.405** -1.507**
(0.630) (0.633)
Mo he wi hou educa ion -0.878* -0.760*
(0.451) (0.452)
Mo he sel -employed wo ke -1.793*** -1.677***
(0.478) (0.477)
Sa is ac ion wi h hei s udies 0.214**
(0.095)
Cons an -3.558 30.611 95.599** 91.168**
(40.277) (37.920) (39.208) (39.292)
Obse a ions 206 206 187 187
R-squa ed 0.001 0.227 0.395 0.416
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
The con ols indica e ha he men ob ain be e ma ks. The e exis s a g oup e ec – ha is
o say, ei he he eache , he companions o he class ime– which gene a es consequences in
he ma ks. The s uden s who li e wi h hei pa en s ha e wo se esul s. I may be ha hei
p oduc i i y has less s imuli as hey ha e less di ec cos s. The mo i a ion which s udying a
sa is ac o y ca ee causes is also ans o med in o be e esul s. Las ly, he child en o pen-
sione mo he s, hose o sel -employed wo ke s o , hose wi hou s udies, ge wo se ma ks.
The las implica ion o he model ha we a e going o es is he p edic ion ha sel -
con ol p oblems gene a e well-being cos s in he people who su e om hem. Acco ding
o his, no ul illing he objec i es planned causes dissa is ac ion wi h one’s own ac ions.
The e o e, we measu e he deg ee o which u ili y and decision dis ance hemsel es om
each o he and i he di e gence is b ough abou by ime p e e ences. To es his implica-
ion, he ela ion be ween he ac o s ha de e mine
β
and
δ
is es ima ed and he le el o
subjec i e well-being is measu ed h ough sa is ac ion wi h li e in gene al. Table 11 shows
he esul s.
As can be seen, he discoun ing ac o has a posi i e e ec upon he deg ee o sa is ac-
ion wi h li e, ha ing a s a is ical signi icance a he 5% le el. Acco ding o his, mo e impa-
ien people would be, emo ing he es o he elemen s conside ed, hose who epo
117
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
g ea e le els o sa is ac ion. The p esen biases do no ha e a s a is ically signi ican ela ion
wi h he a es o happiness16. In his sense, and conside ing ha his ac o is he main de e -
minan o he sel -con ol p oblems, he e idence ound in he sample o s uden s analysed
does no co obo a e his implica ion o he heo y. The o he ac o s which a ou sa is ac-
ion wi h li e a e sel -con idence, sa is ac ion wi h s udies and, us in o he s. Ha ing a u al
o igin and isk a e sion also ha e a posi i e and signi ican e ec .
Table 11
RELATION BETWEEN THE LEVEL OF SUBJECTIVE WELL-BEING
AND TIME PREFERENCES
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Del a 52.267 59.246* 53.580 53.074**
(35.979) (35.215) (34.540) (26.809)
Be a -0.499 -0.574 -0.512 -0.336
(0.835) (0.817) (0.798) (0.617)
Age -0.236* -0.195 -0.021
(0.125) (0.126) (0.099)
Mon hly allowance unde €200 -0.886*** -0.622**
(0.326) (0.255)
Mon hly allowance be ween €200 and €300 -1.034*** -0.862***
(0.359) (0.283)
Mon hly allowance be ween €300 and €500 -1.014** -0.900***
(0.443) (0.342)
Deg ee o isk 1 1.962** 0.943
(0.973) (0.764)
Deg ee o isk 2 2.520*** 1.726**
(0.911) (0.717)
Deg ee o isk 2 2.687*** 1.612**
(0.896) (0.714)
To al s udy ime 0.002* 0.002
(0.001) (0.001)
T us in o he s 0.111**
(0.053)
Sa is ac ion wi h hei s udies 0.224***
(0.060)
Sel -con idence 0.405***
(0.047)
Cons an -44.104 -46.909 -43.634 -49.765*
(35.349) (34.605) (33.776) (26.258)
118
da id pa iño and ancisco gómez-ga cía
(Con inued)
Va iables (1)
model 1
(2)
model 2
(3)
model 3
(4)
model 4
Obse a ions 213 213 209 209
R-squa ed 0.011 0.078 0.213 0.539
S anda d e o s in pa en heses.
*** p<0.01, ** p<0.05, * p<0.1.
5. Conclusions
The consequences o biases in beha iou and p oblems o sel -con ol ha e become
impo an in economic analysis and a e one o he bases o beha iou al economics. Among
he mos accep ed explana ions o sel -con ol p oblems i is no ed ha people’s ime p e e-
ences ollow a quasi-hype bolic o m ins ead o he adi ional discoun . Acco ding o his,
he absence o sel -con ol is he consequence o he e ec ha p esen biases o a endency
o sho sigh edly alue he e en s which ake place in he p esen ha e, o e es ima ing hem
compa ed o hose ha will happen in he u u e. The basic in e p e a ion o his p oposal
means ha he people who ha e a disp opo iona e p e e ence owa ds he p esen adop di -
e en decisions when he momen o making a pos poned decision app oaches – and ha his
signi ies assuming cos s – om hose ha hey would ha e made i hey had decided in ad-
ance. The bias leads hem o modi y hei cos -bene i analysis and o adop decisions a
emo ed om hei long- e m p e e ences. Unde hese p emises, O’Donaghue and Rabin
de ined in a ious a icles a classi ica ion o people which makes hem mo e o less inclined
o su e sel -con ol p oblems. These ca ego ies a e mainly explained by he exis ence o
p esen biases. Some o hese ypes o people, speci ically he naï e, do no manage o plan
hei beha iou o , seen om ano he pe spec i e, epen , a pos e io i, abou decisions
adop ed in he pas .
The p oposi ion is a ac i e and a icula es a logical explana ion o his phenomenon bu
opens he doo o new concep ual di icul ies. Fo example, he economic analysis is based
on people wi h s able p e e ences, a leas du ing he ime ha he analysis las s. This s abi-
li y is he logical consequence o supposing a ional people who do no andomly change
hei beha iou and who can he e o e be he objec o p edic ion. Admi ing he possibili y
o p esen bias implies ecognising ha indi iduals can change hei p e e ences o ques ion
he a ionali y o hei beha iou . This has a - eaching implica ions o he ways o eason-
ing o economis s.
The quasi-hype bolic o mula ion has been conside ed in he li e a u e o beha iou al
economics, which uses i ex ensi ely o model ime p e e ences ha unde es ima e he u u e
(o po en ial) u ili y in ela ion o he cu en u ili y due o p esen biases. In b ie , i cons i-
u es a o m o measu ing he deg ee o which he decisions o he agen s dis ance hemsel es
om ideals and, he e o e, opens he way o measu ing he need o a public ac ion. In gen-
e al, o calcula e he deg ee o which he eal beha iou s o people a e emo ed om hei
119
Do Quasi-Hype bolic P e e ences Explain Academic P oc as ina ion? An Empi ical E alua ion
p edic ions h ough con en ional models is a way o de e mining i policymake s should
in luence hei ac ions.
Ne e heless, he main sho coming o he app oach is ha he empi ical e i ica ion
is limi ed and, in any e en , a good pa o i s logical de elopmen s is pa ial. The p esen
a icle p esen s an unp eceden ed empi ical e i ica ion o di e en aspec s o he explana-
ion and o he ela ionship be ween discoun ing a es, p esen -bias and sel -con ol p o-
blems, ca ied ou wi h a sample o uni e si y s uden s du ing hei ac i i y. Speci ically,
i empi ically checks he de e minan s o he di e en ypes o empo al p e e ences as
well as hei consequences. The p epa a ion o he inal exam o a subjec is an example
o a cos ly ac i i y ha is ca ied ou in he u u e, suscep ible o being planned o and
he e o e a candida e is likely o be a ec ed by sel -con ol p oblems. And hough we ha e
cen ed on he e ec s o p esen bias on he ac ions o uni e si y s uden s, he app oach
can be ex ended o o he opics. Fo example, Pase man (2008) es ima es an employmen
sea ch model wi h simila p emises and uses i o p edic he e ec o he policies o un-
employmen bene i s.
Ou esul s show ela i ely weak e idence o he app oach. Speci ically, we ha e
ound empi ical e idence ha he lack o p esen biases leads o consis en beha iou s, o ,
in ano he wo ds, people who end o ul il he ime plan ha hey had imposed on hem-
sel es. Fu he mo e, he ela ion only occu s o some g oups o he sample, no o all,
al hough in hese g oups i is su icien ly s ong o be applied o all he sample. Howe e ,
we ha e no ound a clea empi ical gua an ee o he es o he explana ion’s implica ions.
We ha e especially no encoun e ed a ela ion be ween he p esen biases and he ime
p e e ences o he people who beha e as naï e o sophis ica ed. We ha e only iden i ied a
ela ion be ween discoun a es and naï e p e e ences in men. No do ou calcula ions en-
able co obo a ion o he majo i y o he consequences ha he (
β
,
δ
) model p edic s o
sel -con ol p oblems. Though we ha e no ed ha he absence o p esen biases imp o es
academic esul s, a s a is ical ela ion has no been es ablished wi h he le els o sa is ac-
ion wi h li e ha we ha e used as a measu e o he sel -con ol p oblems’ cos s o well-
being.
Ou esul s mus be e alua ed e y cau iously. To ake due accoun o hem, i is neces-
sa y o ecognise ha he mechanism ha unde lies inancial beha iou is no , necessa ily,
he same as ha o ano he ype o decision and o be awa e ha he beha iou in p epa ing
an exam has been in e ed h ough a hypo he ical ques ionnai e. All his implies he need
o addi ional esea ch o be done o explo e i he p edic ions o he model ail o be con-
i med because o a ailu e o he heo e ical model i sel o o al e na i e easons. Ne-
e heless, hese limi a ions do no in alida e he main conclusion o ou wo k, which
shows ha e en ecognising ha he (
β
,
δ
)
model ep esen s a logical solu ion o explain
sel -con ol p oblems, i is necessa y o analyse i s empi ical basis o be su e ha i is an
app op ia e explana ion o he beha iou o people, which may se e as a guide o pos-
sible p oposals o public ac ion.