Social contacts, unemployment, and experienced well-being: evidence from time-use data

Hoang, Thi T uong An; Knabe, And eas
A icle — Published Ve sion
Social con ac s, unemploymen , and expe ienced well-
being: e idence om ime-use da a
Jou nal o Popula ion Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Hoang, Thi T uong An; Knabe, And eas (2025) : Social con ac s, unemploymen ,
and expe ienced well-being: e idence om ime-use da a, Jou nal o Popula ion Economics, ISSN
1432-1475, Sp inge , Be lin, Heidelbe g, Vol. 38, Iss. 1,
h ps://doi.o g/10.1007/s00148-025-01083-6
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323179
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
Vol.:(0123456789)
Jou nal o Popula ion Economics (2025) 38:26
h ps://doi.o g/10.1007/s00148-025-01083-6
ORIGINAL PAPER
Social con ac s, unemploymen , andexpe ienced
well‑being: e idence om ime‑use da a
ThiT uongAnHoang1· And easKnabe1
Recei ed: 23 Ap il 2024 / Accep ed: 10 Janua y 2025 / Published online: 10 Feb ua y 2025
© The Au ho (s) 2025
Abs ac
We analyze how di e ences in he equency and in ensi y o social con ac s con-
ibu e o he expe ienced well-being o employed and unemployed indi iduals. In
he UK Time-Use Su ey, we obse e ha people gene ally enjoy being wi h o he s
mo e han being alone. The unemployed end o eel wo se han he employed when
engaging in he same kind o ac i i ies, pa ly because hey a e mo e o en alone.
The unemployed canno spend ime wi h o he people a wo k bu spend sligh ly
mo e ime wi h p i a e con ac s han he employed. In e ms o expe ienced well-
being, he sligh inc ease in ime spen wi h amily and iends (which people enjoy
a lo ) o se s he loss o wo k con ac s (which people gene ally enjoy only a li le).
Hence, we do no ind ha he di e ences in he social-con ac composi ion be ween
he employed and he unemployed a e associa ed wi h di e ences in hei expe i-
enced well-being.
Keywo ds Unemploymen · Happiness· Expe ienced well-being· Time use· Social
con ac
JEL Classi ica ion I31· J60· J22
Responsible edi o : Milena Nikolo a
* And eas Knabe
and eas.knabe@o gu.de
Thi T uong An Hoang
hi.hoang@o gu.de
1 Facul y o Economics andManagemen , O o-Von-Gue icke-Uni e si y Magdebu g, P.O.
Box4120, 39016Magdebu g, Ge many
T.T.A.Hoang, A.Knabe
26 Page 2 o 28
1 In oduc ion
Using da a om he UK Time-Use Su ey, ou s udy in es iga es he ela ionship
be ween ime use, social con ac s, and expe ienced well-being o employed and
unemployed pe sons. We aim o analyze o wha ex en di e ences in he quan i y
and quali y o ime spen wi h o he people a e e lec ed in he expe ienced well-
being o employed and unemployed pe sons.
I has been shown in he li e a u e ha unemploymen is de imen al o a ious
measu es o subjec i e well-being, in pa icula o li e sa is ac ion and gene al
men al heal h ( o a su ey, see Suppa 2021). S udies on emo ional well-being
ha e ound ha unemployed pe sons also expe ience less posi i e and mo e nega-
i e emo ions han employed pe sons ou side o wo k. The nega i e impac o
unemploymen on emo ional well-being du ing non-wo k ac i i ies has been
called he “saddening e ec ” o unemploymen (K uege and Muelle 2012;
Knabe e al. 2010).
Va ious easons could be esponsible o he saddening e ec . Fo example,
K uege and Muelle (2012) suppose ha he unemployed migh ha e less income
o spend on leisu e, ha he e migh be diminishing ma ginal u ili y o leisu e, o
ha he unemployed migh ha e mo e ime o hink abou hei mise y. Ano he
eason could be he loss o non-mone a y, la en bene i s o employmen . In he
well-known enume a ion, Jahoda (1981) lis s i e la en bene i s o wo k: i ex e -
nally imposes a ime s uc u e on he day; en o ces ac i i ies; links people o
goals and pu poses ha anscend hei own; de ines pe sonal iden i y and social
s a us; and es ablishes social con ac s ou side he amily. Empi ical s udies on he
ela ionship be ween unemploymen and subjec i e well-being suppo he exis -
ence o some o hese la en bene i s (c . Zechmann and Paul 2019). Fo exam-
ple, he e is s ong e idence ha unemploymen hu s because i is associa ed
wi h a loss in social s a us and a de ia ion om social no ms ( he ou h i em
on Jahoda’s lis ; Cla k 2003; Schöb 2013; He schko e al. 2014). The e is also
e idence ha employmen links people o anscenden al goals ( he hi d la en
bene i ). When people a e asked abou how “meaning ul” o “ ewa ding” hei
daily ac i i ies a e, wo king usually anks on op (Whi e and Dolan 2009; Wol
e al. 2022). McKee-Ryan e al. (2005) and Hoa e and Machin (2010) p o ide e i-
dence ha he unemployed su e when hey eel ha hey do no use hei ime in
a s uc u ed and pu pose ul way.
In his pape , we ocus on he i h o Jahoda’s la en bene i s and examine
he hypo hesis ha he unemployed ha e less con ac wi h o he people han he
employed and ha his con ac loss ela es, ce e is pa ibus, o a well-being loss
expe ienced by he unemployed. The li e a u e has shown ha social connec ed-
ness and spending ime wi h o he s gene ally con ibu e o subjec i e well-being
(Kahneman e al. 2004). The empi ical e idence on he e ec o unemploymen
on social connec edness and isola ion appea s less clea . While some s udies ind
ha he unemployed eel mo e socially isola ed (e.g., Pohlan 2019), o he s udies
show ha unemploymen does no weaken he social ne wo k o a ec ed pe sons
(K ug and P echsl 2022; Röze e al. 2020). All hese s udies analyze su eys in
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 3 o 28 26
which esponden s a e asked o epo on he gene al size o hei social ne wo k
o ypical con ac equencies. To he bes o ou knowledge, no s udies analyze
hese issues using dia y da a. Dia y s udies allow us o examine how much ime
employed and unemployed pe sons spend wi h o he s and hei emo ional well-
being when engaging in speci ic ac i i ies alone o wi h o he s.
We use da a om he UK Time-Use Su ey, which con ains ich in o ma ion
on how people spend hei ime, how much ime hey spend wi h di e en ypes
o social con ac s, and how hey eel du ing hese episodes. We ocus speci ically
on he di e ences in ime use and expe ienced well-being be ween employed and
unemployed pe sons. We measu e he in ensi y o a pe son’s con ac s by calcula ing
he sha e o he o al waking ime spen wi h o he people and being alone and he
quali y by he le el o enjoymen a pe son pe cei es while spending ime wi h hese
con ac s. In line wi h he li e a u e, we ind ha he unemployed eel wo se han
he employed du ing many non-wo k ac i i ies. Howe e , he employed epo pa -
icula ly li le enjoymen du ing wo k episodes. On a e age, he unemployed do no
enjoy hei days less han he employed.
The no el con ibu ion o his s udy is o quan i y he ex en o which con ac s
wi h amily membe s and o he s ou side he household con ibu e o he expe ienced
well-being o employed and unemployed pe sons. We de elop a decomposi ion
echnique o iden i y i e sepa a e componen s ha illumina e he ole o di e ences
in he composi ion o ac i i ies and social con ac s as well as di e ences in enjoy-
men a ings in pa icula ac i i y-con ac combina ions. The oppo uni ies o spend
ime wi h o he s gene ally di e depending on a pe son’s ma i al/pa ne ship s a-
us. Thus, we conduc sepa a e analyses o pe sons who a e ma ied o cohabi ing,
hence o h “pa ne ed,” and pe sons who ha e ne e been ma ied o a e di o ced o
widowed, hence o h “single.”1 We also e i y ou indings using en opy balancing,
which ma ches he employed and he unemployed based on obse able, po en ially
con ounding ac o s.
Ou analysis shows ha he employed and he unemployed enjoy being wi h o h-
e s mo e han being alone. Unemploymen dep i es people o he oppo uni y o
mee o he s a wo k. Hence, he unemployed spend, on a e age, o e wo mo e wak-
ing hou s alone han he employed. In ha sense, ou indings suppo he claim by
Jahoda (1981) ha access o social con ac s ou side he amily is one o he la en
bene i s o wo k.
Howe e , ou esul s also sugges ha he obse ed di e ences in ac i i ies and
social con ac s a e no associa ed wi h lowe le els o expe ienced well-being among
he unemployed. We gi e a b ie p e iew o ou main esul s he e. Conce ning he
ac i i y composi ion, he unemployed bene i subs an ially om being able o shi
ime om wo k o o he , mo e enjoyable ac i i ies. This shi pa ly cap u es he
ime-composi ion e ec ha has al eady been desc ibed in p e ious s udies (Knabe
e al. 2010). In his s udy, we no only look a di e ences in he ype o ac i i ies bu
1 In his pape , we use he e ms “pa ne ed,” “ma ied/cohabi ing,” and “wi h pa ne ” in e changeably
o he subg oup o indi iduals who decla e o be ma ied o cohabi ing, and a he ime o he su ey a e
li ing wi h hei pa ne .
T.T.A.Hoang, A.Knabe
26 Page 4 o 28
also a di e ences in he p esence o o he pe sons du ing hese ac i i ies. We ind
di e en esul s o single and pa ne ed pe sons. Mee ing o he s a wo k does no
seem o p o ide addi ional enjoymen o people wi h pa ne s compa ed o wo king
alone. We also do no obse e signi ican di e ences in he amoun o ime spen
wi h o he s ou side o wo k be ween pa ne ed employed and unemployed pe sons.
Hence, di e ences in con ac a ailabili y do no subs an ially con ibu e o he di -
e ence in expe ienced well-being be ween employed and unemployed indi iduals
wi h pa ne s. Singles, in con as , enjoy spending hei wo k ime wi h o he s sig-
ni ican ly mo e han wo king alone. This bene i is los when becoming unemployed.
We also obse e ha single unemployed pe sons spend mo e ime wi h o he s
ou side wo k han employed singles, which pa ially o se s he educ ion in well-
being om losing wo k con ac s. Hence, we do no ind a signi ican o e all con i-
bu ion o changes in social con ac s o he well-being di e ences be ween employed
and unemployed singles, ei he . These indings imply a mo e di e en ia ed iew o
Jahoda’s (1981) hypo hesis ha one o he la en bene i s o wo k is he es ablish-
men o social con ac s ou side he amily. While we ind e idence ha employed
pe sons ha e mo e social con ac s ou side hei amily, he obse a ion ha his does
no seem o con ibu e o highe emo ional well-being aises doub s abou whe he
his should be conside ed a la en bene i o wo k.
The pape is s uc u ed as ollows. In he nex sec ion, we b ie ly e iew he ela ed
li e a u e. In Sec ion3, we desc ibe he da a. Sec ion4 examines di e ences in ime
use, social con ac s, and well-being be ween he employed and he unemployed. In
Sec ion5, we decompose hese di e ences in a ious channels. Sec ion6 concludes.
2 Rela ed li e a u e
P io esea ch has ex ensi ely demons a ed he ad e se impac o unemploymen on
cogni i e well-being, as measu ed by b oad men al heal h indica o s o li e sa is ac ion
(Cla k and Oswald 1994; Winkelmann and Winkelmann 1998; Cla k e al. 2001, 2010;
Knabe and Rä zel 2010; Shields e al. 2009). Some s udies ha e also examined how
an indi idual’s employmen s a us a ec s hei e e yday emo ional expe iences. These
in es iga ions sugges ha unemploymen in luences expe ienced well-being h ough
wo channels. On he one hand, he unemployed expe ience lowe le els o well-being
han he employed when engaged in he same kind o ac i i ies (Knabe e al. 2010;
K uege and Muelle 2012). On he o he hand, employed pe sons a e wo king among
he leas pleasan ac i i ies du ing hei days (Kahneman e al. 2004; Whi e and Dolan
2009; B yson and MacKe on 2017; Hoang and Knabe 2021a; Wol e al. 2022). Since,
by de ini ion, he employed ha e o spend mo e ime wo king han he unemployed, his
wo sens hei a e age emo ional expe ience o e he day. The i s channel is e e ed
o as he saddening e ec o unemploymen , and he second as he ime-composi ion
e ec (Knabe e al. 2010). The wo e ec s wo k in opposi e di ec ions, such ha i is
unclea , a p io i, which o he wo g oups eels be e o e he en i e day.
Va ious s udies ind ha he saddening and ime-composi ion e ec s o se each
o he , such ha unemploymen is no nega i ely ela ed o day-a e age expe ienced

Social con ac s, unemploymen , andexpe ienced well‑being:… Page 5 o 28 26
well-being. This is shown by Knabe e  al. (2010), Lawes e  al. (2023), and Wol
e  al. (2022) o Ge many, Flèche and Smi h (2017) o F ance, and Hoang and
Knabe (2021a) o he UK. Fo he Uni ed S a es, K uege and Muelle (2012) ind
ha he saddening e ec domina es when looking a he speci ic eelings o sad-
ness and pain, whe eas Dolan e al. (2017) do no ind di e ences be ween agg ega e
emo ional well-being measu es o he employed and he unemployed.2
The e a e also some s udies looking a how people spend hei ime and how hey
expe ience i depending on he p esence o o he people. These s udies gene ally
suppo he iew ha social con ac s posi i ely a ec emo ional well-being. Kahne-
man e al. (2004) epo ha socializing and in ima e ela ions ank among he bes
expe iences o he day based on hei a ec a ings. B yson and MacKe on (2017),
using sel -collec ed da a om a sma phone app, also epo ha people a e happi-
es when hey spend ime wi h o he s, pa icula ly wi h iends o hei spouse/pa -
ne . This posi i e e ec is, howe e , diminished when people a e wo king. In hei
analysis o Ame ican ime-use da a, Flood and Genadek (2016) ind ha , con olling
o hou s o wo k, dual-ea ne couples spend simila amoun s o ime oge he as
couples in which one pa ne s ays a home. They also ind ha ime spen oge he
is associa ed wi h mo e happiness and less s ess. Musick e al. (2016) and Flood
e al. (2020) show ha pa en s enjoy hei leisu e ime much mo e when hey a e
wi h hei child en o o he adul s han alone. A alay (2024) inds ha e alua i e
and expe ienced well-being is lowe when people a e alone du ing hei non-wo k
ime and ha he sha e o ime spen alone has inc eased in he US in he las wo
decades. Han and Kaise (2024) show ha bo h men and women ha e lowe expe-
ienced well-being on days when hey spend mo e o hei ime alone, bu ha he
nega i e impac is s onge o women. Hence, e en hough hey ind ha he Co id
pandemic a ec ed he sha e o ime spen alone simila ly o bo h gende s, women
su e ed a la ge educ ion in expe ienced well-being om i . Using Ame ican and
B i ish ime-use da a, Gimenez-Nadal e al. (2023) examine how wo ke s pe cei e
ime alone and ime spen wi h amily membe s. Thei indings show ha espond-
en s p e e oge he ness o e solo leisu e ac i i ies.
No all social con ac s a e equally enjoyable. Kahneman e al. (2004) ind ha
indi iduals’ emo ional expe iences di e depending on who hey spend ime wi h.
While people a e gene ally e y happy when being wi h hei iends, hey epo
s ong nega i e emo ions while being wi h hei boss. Uziel and Schmid -Ba ad
(2022) p esen e idence om an expe imen al su ey and an expe ience-sampling
s udy illus a ing he signi icance o being able o choose one’s con ac s. Being wi h
o he s by choice is gene ally much mo e enjoyable han being wi h o he s due o
ex e nal ci cums ances.
Being socially in eg a ed plays an essen ial pa in p edic ing cogni i e and a ec-
i e well-being. P io li e a u e has shown, howe e , ha he e is a nega i e associa-
ion be ween joblessness and social in eg a ion. The unemployed ha e been ound
o eel lonely (Mo ish and Medina-La a 2021) and socially excluded (Pohlan 2019)
2 See Hoang and Knabe (2021b) o an empi ical in es iga ion o he di e ences be ween hese wo
s udies.
T.T.A.Hoang, A.Knabe
26 Page 6 o 28
and o pa icipa e less in social ac i i ies (B and and Bu ga d 2008; Paugam and
Russell 2000; Dieckho and Gash 2015; Gallie e al. 1994). They belong o dis-
ad an aged socioeconomic g oups who a e a isk o social isola ion and ha e less
social con ac (Simone e al. 2022; Bu cha d e al. 1999; Röh e al. 2022). They
su e om he nega i e implica ions and he cumula ed disad an ages o being
unemployed (Gangl 2006), he eeling o shame and s igma iza ion (McFadyen
1995; Eales 1989; Paugam and Russell 2000; Pe e ie e al. 2019b), and he pe cep-
ion o unemploymen as hei aul (Lis e e al. 1996; Pe e ie e al. 2019a). Fu -
he mo e, he li e a u e sugges s ha he unemployed a e caugh in a icious ci cle:
hei ulne abili y o social isola ion ein o ces hei economic disad an ages (Gallie
1999; Mo ish and Medina-La a 2021), which in u n inc eases hei isk o social
isola ion u he (Eckha d 2018). Some mo e ecen s udies ha use panel da a and
can obse e people be o e and a e becoming unemployed p o ide mo e di e en i-
a ed insigh s. Using Ge man panel da a, K ug and P echsl (2022) obse e ha he
s uc u e o he social ne wo k does no change a e becoming unemployed. Simi-
la ly, Röze e al. (2020) ind ha , excep o people o e 50, unemploymen gene -
ally has no ad e se e ec s on he s eng h o he social ne wo k. In hei Swiss panel
da a, hey ind some indica ion o shi s in he ne wo k s uc u e, whe e con ac s
wi h colleagues o dis an acquain ances a e eplaced by iends and amily.
In mos s udies, he in ensi y o social pa icipa ion has been measu ed by b oad
social indica o s, such as how equen ly one a ends cul u al e en s, how o en one
isi s iends and ela i es, o he numbe o iends and social con ac s one has. Such
measu es and indica o s can cap u e he equency o magni ude o di e en ypes o
social pa icipa ion o e mo e ex ended pe iods. Howe e , hey do no e lec he ime
in ensi y (leng h) and he quali y (pleasan ness) o he social con ac s. Fu he mo e,
while wo k may p o ide mo e oppo uni ies o es ablish social con ac s, employed
pe sons a e also mo e ime-cons ained, since hey canno spend much ime wi h o h-
e s ou side he wo kplace, as well as mo e choice-cons ained, since hey ypically do
no choose hei con ac s a wo k (Uziel and Schmid -Ba ad 2022). The unemployed,
in con as , migh ha e a smalle numbe o con ac s ou side hei household, i.e., a
lowe con ac equency, bu mo e ime o alloca e o hei exis ing con ac s, i.e., a
highe con ac in ensi y, and in p inciple mo e eedom o shi hei a ailable ime
owa ds people whose company hey enjoy. S udies by Paugam and Russell (2000)
and Kunze and Suppa (2017) e eal ha he unemployed pa icipa e less in o mal
(clubs, o ganiza ions, ins i u ions) o public social ac i i ies (a ending cul u al e en s
o going o he mo ies), bu ha e highe le els o in o mal o p i a e social pa icipa-
ion (mee ing iends and amily, helping ou neighbo s, e c.).
3 Da a
We analyze da a om he 2014/2015 Uni ed Kingdom Time-Use Su ey (UKTUS).
The UKTUS is a na ionally ep esen a i e su ey ha collec s in o ma ion on how
people use hei ime (Ge shuny and Sulli an 2017). Responden s comple e abula
dia ies o up o 2days, du ing which hey p o ide de ailed in o ma ion abou wha
ac i i ies hey engaged in and whe e hey sojou ned a each poin in ime du ing he
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 7 o 28 26
day. Two addi ional pieces o in o ma ion collec ed in he su ey a e pa icula ly
ele an o his s udy. Fi s , esponden s e eal whom hey we e wi h du ing each
epo ed episode (spouse/pa ne , pa en s, child en, o he household membe s, o he
known pe sons, o being alone), so we ha e de ailed in o ma ion on he ypes, e-
quencies, and iming o hei social con ac s. Second, a subsample o hese espond-
en s is asked o a e how much hey enjoyed each epo ed ac i i y. To be p ecise,
esponden s we e asked abou each episode: “How much did you enjoy his ime?”
and could answe on a scale om 1 (“no a all”) o 7 (“ e y much”). We in e p e
his esponse as a measu e o a pe son’s a ec i e/expe ienced well-being.3 Respond-
en s in his su ey p o ide dia ies o 2 sepa a e days. The exac days esponden s
should ill ou he dia ies a e andomly de e mined. Gene ally, one is a weekday,
and he o he is a Sa u day o a Sunday. The da ase con ains 16,550 dia ies col-
lec ed om 9388 indi iduals aged eigh o olde li ing in 4239 households. We use
he dia ies om all household membe s and om weekdays and weekends. We use
he UKTUS dia y weigh s ha accoun o he unequal numbe o weekdays and
weekend days. We also calcula e clus e ed s anda d e o s o accoun o he non-
independence o dia y-le el obse a ions om pe sons om he same household.
Ou analysis ocuses on how di e ences in he equency and ype o social con-
ac s con ibu e o he well-being gap be ween he employed and he unemployed.
How many oppo uni ies a pe son has o be in con ac wi h o he s migh no only
depend on hei employmen s a us bu is also hea ily in luenced by hei pa ne -
ship s a us (Röh e al. 2022). Ma ied o cohabi ing pe sons ha e much mo e social
con ac “by de aul ” (wi h hei pa ne ), whe eas singles ha e o ac i ely seek con-
ac i hey do no wan o be alone in hei ee ime. We conduc ou analyses sep-
a a ely o ma ied/cohabi ing pe sons and singles o accoun o his c i ical di -
e ence. We hus es ic he sample o he subse s o esponden s who a e ei he
ma ied/cohabi ing o single (ne e ma ied/di o ced/widowed). Fu he mo e, we
use a ailable da a abou how di e en household membe s a e ela ed o alida e
he consis ency o he ac ual li ing a angemen and he epo ed ma i al s a us. I a
con lic is de ec ed (e.g., when a esponden answe s o be single when asked abou
he ela ionship s a us bu epo s o be he pa ne o ano he household membe in
ano he ques ion), we emo e he obse a ion om he sample.
The UKTUS con ains in o ma ion abou a pe son’s sel -decla ed wo k s a us,
labo ma ke his o y, and u u e wo k in en ions. We apply he ILO de ini ion o
iden i y employed and unemployed pe sons.4 We es ic ou a en ion o su ey pa -
icipan s aged 16yea s and abo e.5 To be conside ed employed, a pe son has o ha e
3 Unlike some o he ime-use su ey, such as he Ame ican Time-Use Su ey (ATUS), he UKTUS does
no collec in o ma ion on a a ie y o posi i e and nega i e a ec s. Ins ead, esponden s a e only asked
abou hei “enjoymen ” o each episode. This p ecludes disagg ega ed analyses o posi i e and nega i e
a ec s. Howe e , Fleche and Smi h (2017) ha e shown ha a summa y measu e o episode enjoymen is
highly co ela ed wi h common composi es o sepa a e a ec s (e.g., he ne a ec o he U-index).
4 Ou main esul s do no change when, ins ead o he ILO de ini ion o (un)employmen , we use
esponden s’ sel - epo ed employmen s a us.
5 We do no apply an uppe age limi in he main analysis. The e a e in o al 111 employed pe sons and
one unemployed pe son olde han 65yea s old who a e included in ou sample. Ou indings emain
unchanged also when conside ing only esponden s who a e be ween 16 and 65yea s old.
T.T.A.Hoang, A.Knabe
26 Page 8 o 28
been in paid wo k o 7days, ending he Sunday p eceding he in e iew, o be hold-
ing a job om which hey we e aking ime o a he ime o he in e iew. Unem-
ployed pe sons a e hose who ha e no been wo king o pay on any o he 7days
ending he Sunday p eceding he in e iew, a e no jus empo a ily away om a job
o business, ha e been sea ching o a job in he ou weeks p eceding he in e iew,
and would be able o s a wo king immedia ely when gi en a job o e . We include
ull- ime and pa - ime employees and sel -employed indi iduals bu exclude ull-
ime s uden s. We exclude all non-employed pe sons who do no sa is y he ILO c i-
e ia o unemploymen . To a oid an undue in luence o ou lie s, indi iduals whose
household income (adjus ed by he modi ied OECD equi alence scale) belongs o
he op and bo om 1% o he dis ibu ion a e emo ed om he analyses. Finally,
since we analyze he ole o social con ac s o indi idual well-being, we exclude
dia ies con aining enjoymen sco es o ewe han h ee waking episodes. Al hough
each esponden is supposed o comple e wo dia ies, he e a e cases whe e only one
o he wo dia ies con ains enjoymen a ings o a leas h ee waking episodes. We
keep his dia y bu emo e he o he dia y (wi h ewe han h ee enjoymen - a ed
episodes) om he sample. All he dia ies con ain comple e ime-use in o ma ion.
Wi h hese adjus men s, we ob ain a sample o 3492 indi iduals who epo 6923
dia ies wi h su icien enjoymen in o ma ion. The sample includes 2,555 ma ied/
cohabi ing pe sons (2499 employed and 56 unemployed wi h 4960 and 111 dia ies,
espec i ely) and 937 single pe sons (869 employed and 68 unemployed wi h 1720
and 132 dia ies, espec i ely). A ca ea o ou analysis is ha he numbe o unem-
ployed esponden s is ela i ely small; he e o e, s anda d e o s a e likely la ge.6
As men ioned ea lie , some dia ies migh be incomple e in he sense ha hey
con ain seamless sequences o ac i i ies, bu enjoymen sco es o some o hese
ac i i ies a e missing. This conce ns oughly en pe cen o obse a ions (22,370 ou
o 224,490 waking episodes). These episodes can be used o analyze ime use bu
no well-being.
One impo an aspec o UKTUS 14/15 is ha all membe s o he same house-
hold epo on he same dia y days. This allows o impu ing some missing obse -
a ions o co-p esen ( amily) con ac s. Fo example, i a husband’s dia y has
missing in o ma ion on co-p esen pe sons du ing some ime in e al on a dia y
day (so he nei he epo s any con ac s no being alone), bu he dia y o his wi e
epo s a ha exac ime o he day ha she was wi h he husband, he missing
con ac in o ma ion o ha episode in he husband’s dia y would be impu ed as
“wi h he spouse.” Ou sample has 24,698 waking episodes wi h missing co-p e-
sen con ac in o ma ion. We can impu e con ac in o ma ion o 5594 episodes
wi h his me hod. I no u he de ails abou episodes wi h missing con ac in o -
ma ion can be acqui ed, hey a e classi ied as spen alone.7 This is a conse a-
i e assump ion. I hese episodes we e ac ually spen wi h someone else, and
6 The small numbe o unemployed esponden s pa ly e lec s he ela i ely low unemploymen a e in
he UK in 2016 (4.9 pe cen , ONS 2023).
7 This assump ion is no c i ical o ou indings. They emain unchanged also i we igno e all episodes
whose missing social-con ac in o ma ion canno be impu ed using o he household membe s’ dia ies.
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 15 o 28 26
The well-being li e a u e has ound e idence ha unemploymen p oduces s igma
and shame. The esul s in his analysis can be in e p e ed as p elimina y e idence
ha hese ad e se psychological e ec s o unemploymen a e isible in he cogni-
i e dimension o well-being (li e sa is ac ion). A he same ime, hey do no seem o
a ec unemployed people’s abili y o enjoy hei ime and con ac wi h o he people
in e e yday li e e y much. We do no see ha unemployed people would wi hd aw
om hei social con ac s o ha hey could no enjoy mee ing wi h o he s. Qui e o
he con a y, he unemployed enjoy spending ime wi h o he s e y much. The ime
spen wi h o he acquain ances e en belongs o he mos enjoyable momen s o hei
day. Hence, we do no ind e idence ha he unemployed would be less able han he
employed o enjoy he ime hey spend wi h o he s. Ou ime-use in es iga ion sug-
ges s, howe e , ha hey canno eplace mos o he social con ac s hey lose a wo k
wi h addi ional con ac s o amily membe s o acquain ances in hei non-wo k ime.13
5 Decomposing hewell‑being di e ences be weenemployed
andunemployed pe sons
In he p e ious sec ion, we ha e illus a ed ha he main eason why he unem-
ployed spend, on a e age, mo e ime alone han he employed is because hey lose
con ac wi h o he people a wo k, which hey canno adequa ely eplace by o he
con ac s du ing non-wo k ac i i ies. We also obse e ha bo h he unemployed and
he employed gene ally enjoy any ac i i y mo e i i is no done alone bu wi h o h-
e s. This obse a ion sugges s ha pa o he saddening e ec o unemploymen ,
i.e., he obse a ion ha he unemployed o en ha e lowe expe ienced well-being
han he employed e en when bo h engage in he same kind o ac i i ies, migh be
explained by he ac ha he unemployed a e mo e o en alone in hese ac i i ies. In
his sec ion, we explo e his hypo hesis in mo e de ail and examine o which ex en
he loss o social con ac s a wo k and i s pa ial eplacemen by o he con ac s con-
ibu e o he expe ienced well-being o he employed and he unemployed.
As a s a ing poin , we decompose he well-being di e ence be ween he
employed and he unemployed in o a ime-composi ion e ec and a saddening
e ec . This wo-s ep decomposi ion has al eady been applied in his con ex by
Knabe e al. (2010). In his s udy, we ex end his decomposi ion by u he b eak-
ing down hese wo e ec s, aking di e ences in he p e alence o social con-
ac s in o accoun . This leads o a i e-s ep decomposi ion o he di e ence in
expe ienced well-being be ween he employed and he unemployed, cap u ing
di e ences in he composi ion and du a ion o ac i i ies, he composi ion and
equency o social con ac s, and he a ings o expe ienced well-being. In he
13 We also conduc eg ession analyses o li e sa is ac ion and day-a e age enjoymen on employmen
s a us, he ex en o social con ac s, and o he co a ia es. The esul s a e p esen ed in he Supplemen a y
Ma e ial (TablesA5 and A6). These eg essions p o ide u he suppo o ou inding ha he ex en o
social con ac s is gene ally posi i ely associa ed wi h li e sa is ac ion and wi h expe ienced enjoymen .
We also ind ha con ac o pe sons om ou side one’s household a e less bene icial o well-being i
hey ake place a wo k.

T.T.A.Hoang, A.Knabe
26 Page 16 o 28
ollowing subsec ions, we i s ep oduce he wo-s ep decomposi ion and hen
desc ibe and conduc he i e-s ep decomposi ion. In he las subsec ion, we es
he obus ness o ou indings when we eweigh he da a using en opy balanc-
ing, which makes he subsamples o employed and unemployed pe sons mo e
simila ega ding ele an co a ia es.
5.1 Two‑s ep decomposi ion
Table3 con ains he esul s o he wo- and i e-s ep decomposi ions, sepa a ed by
ela ionship s a us (ma ied/cohabi ing on he le , single on he igh ). The i s
line in he able shows he a e age enjoymen o pa ne ed and single employed
pe sons (5.271 and 5.338, espec i ely). To de e mine he ime-composi ion
e ec , we hypo he ically ealloca e he ime use o an a e age employed pe son o
di e en ac i i ies o ma ch ha o he a e age unemployed. The enjoymen le els
Table 3 Two-s ep and i e-s ep decomposi ions wi h social con ac s
Ma ied/cohabi a ing Single
5.271 A e age employed pe son 5.338
(0.023) (0.036)
-0.024**Con ac loss a wo k -0.072 ***
(0.012)
5.24
75
.266
(0.026) (0.045)
0.262 *** 0.187 *** Ac i i y-composi ion e ec 0.329 *** 0.326 ***
(0.027)(0.032) (0.043)(0.033)
5.43
45
.596
(0.034) (0.042)
0.099 *** 0.068 ***
(0.007) (0.008)
(0.032) (0.041)
-0.080 *** -0.033 *
(0.021) (0.018)
-0.224 5.45
45
.631 -0.060
(0.146)(0.043) (0.041)(0.123)
-0.14
4-
0.027
(0.147) (0.123)
5.309 A e age unemployed pe son 5.604
(0.146) (0.116)
"Saddening e ec "
Con ac shi when unemployed
"Saddening e ec "
Adjus ed saddening e ec
" ce enoi isopmoc-emiT"
"Time-composi ion e ec "
Con ac gain in non-wo k
ac i i ies
5.53
45
.664
S anda ds e o s om (clus e ed) boo s apping wi h 5000 eplica ions in pa en hesis. Signi icance le el:
* p < 0.10; ** p < 0.05; *** p < 0.01
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 17 o 28 26
in each ype o ac i i y emain unchanged. This shi would esul in a la ge and
s a is ically signi ican inc ease in expe ienced well-being (+ 0.262 o he pa -
ne ed, + 0.326 o he singles). This illus a es ha he oppo uni y o use one’s
ime eely ins ead o ha ing o wo k could, as such, be bene icial o well-being.
Howe e , he unemployed gene ally do no enjoy hei ee ime as much as
he employed. The saddening e ec cap u es how he day-a e age enjoymen o
employed and unemployed pe sons di e s, assuming hey could spend hei ime
simila ly. Ou esul s o ma ied o cohabi ing couples show his would lead o a
educ ion in enjoymen (− 0.224), which is, howe e , no s a is ically signi ican .
The ime-composi ion and saddening e ec s o se each o he , such ha we only ind
a small and s a is ically insigni ican di e ence in day-a e age enjoymen be ween
employed and unemployed pe sons wi h pa ne s. Fo he singles, he saddening
e ec is much smalle and no s a is ically signi ican . A e age enjoymen is signi i-
can ly la ge o unemployed han o employed singles.
5.2 Fi e‑s ep decomposi ion
Using he in o ma ion wi h whom people spend each episode, we u he decom-
pose he obse ed well-being di e ences o illumina e he ole o social con ac s in
explaining he well-being gap be ween employed and unemployed people.
Le
sk
i
,
j
be he a e age sha e o waking ime ha indi iduals in g oup
k∈{E,U}
(whe e E and U deno e employed and unemployed pe sons, espec i ely) spend in
ac i i y
i∈{1, ..., I}
wi h con ac s
j∈{a, ,o}
(deno ing alone, amily, and o he s,
espec i ely). The o al sha e o ime ha an a e age indi idual in g oup
k
spends in
ac i i y
i
is hen gi en by
sk
i
=
∑j∈{a
,
,
o}
s
k
i,j
, wi h
∑I
i=1
sk
i
=
1
o each
k∈{E,U}
. Le
ac i i y
i=1
be con ac ed ime (“wo k”) and ac i i ies
i≥2
be all non-wo k ac i i-
ies. The a e age enjoymen expe ienced by indi iduals o g oup
k
in ac i i y
i
wi h
con ac
j
is deno ed by
𝜀k
i,j
.
As in he wo-s ep composi ion, we s a wi h a ep esen a i e employed pe son
whose enjoymen a ings and sha e o ime-use by ac i i ies and social con ac s
e lec he a e age o all employed pe sons in he sample ( i s line in Table 3),
which is gi en by
In he i s s ep, we a e in e es ed in he impo ance o mee ing o he people
a wo k o indi idual well-being. We calcula e how he a e age enjoymen o an
employed pe son would change i she supposedly had o spend he en i e wo king
ime alone (con ac loss a wo k). We cons uc a hypo he ical pe son whose epi-
sodes wi h con ac s a wo k a e ea ed as wo k episodes alone. The expe ienced
well-being would hen be
(1)
∑I
i=
1
∑j∈{a
,
,
o}
sE
i,j𝜀E
i,
j
(2)
s
E
1𝜀E
1,a+
∑I
i=2∑j∈{a, ,o}
sE
i,j𝜀E
i,
j
T.T.A.Hoang, A.Knabe
26 Page 18 o 28
As seen in Table3, he educed ime spen wi h con ac s a wo k con ibu es neg-
a i ely o he day-a e age enjoymen o ma ied/cohabi ing people (− 0.024). How-
e e , he magni ude o he es ima ed con ibu ion is ela i ely small. Fo he singles,
we ind a subs an ially la ge nega i e e ec (− 0.072). This con as sugges s ha
wo k con ac s migh be mo e impo an o singles, pe haps because hey need o
ac i ely socialize in non-wo k ime i hey wan o spend ime wi h someone else,
bu less so o he ma ied/cohabi ing who, by de aul , ha e immedia e access o
p i a e con ac s in daily li e.
The second s ep o he decomposi ion iden i ies an ac i i y-composi ion e ec .
While he employed ha e o wo k, he unemployed can choose mo e eely wha
ac i i ies hey engage in o e he day. We examine wha would happen o he a e -
age expe ienced well-being i he pe son cons uc ed in he p e ious s ep, who
spen all wo king hou s alone, could shi his ime o non-wo k ac i i ies bu would
s ill ha e o engage in hem alone. Fo his decomposi ion, we assume ha , a e
his shi , he o al ime sha e in each ac i i y co esponds o he a e age obse ed
among he unemployed. We main ain he ime sha es spen wi h amily and o he s a
he a e age le el o he employed and ea he esiduals as ime spen alone.14 The
expe ienced well-being would hen be
This change in he ac i i y composi ion esul s in a s a is ically signi ican
inc ease in enjoymen , which is also he single-la ges channel we iden i y in ou
i e-s ep decomposi ion (+ 0.187 o pa ne ed people, + 0.329 o singles). Ou
esul sugges s ha i is he ype o ac i i ies ha people engage in ha ma e s mos
o hei expe ienced well-being. One can also see ha he ac i i y-composi ion
e ec is much s onge o singles han o ma ied/cohabi ing pe sons. As seen in
he p e ious sec ion, single unemployed pe sons engage mo e in ee- ime ac i i-
ies, while pa ne ed ones de o e mo e o commi ed ime. Since leisu e is gene ally
mo e enjoyable han o he kinds o ac i i y o indi iduals o all conside ed sub-
g oups, his migh explain why he ac i i y-composi ion e ec is much la ge o
singles han pa ne ed indi iduals.
A his poin in ou decomposi ion, he hypo he ical employed pe son has shi ed
all wo king hou s o non-wo k ac i i ies and assumedly spends he addi ional non-
wo k ime alone. Howe e , his employed pe son migh wan o use some o he
addi ional ime o mee o he people. The hi d s ep es o es he ela i e con ac dis-
ibu ion o he employed wi hin each ac i i y bu applies i o he o al sha e o ime
spen by he unemployed on his ac i i y. The expe ienced well-being hen becomes
(3)
∑I
i=2[(
sU
i−
∑j∈{ ,o}
sE
i,j
)
𝜀E
i,a+
∑j∈{ ,o}
sE
i,j𝜀E
i,j
]
(4)
∑
I
i=2∑j∈{a, ,o}
s
E
i,j
sE
i
sU
i𝜀E
i,
j
14 In ou da a, he esidual is posi i e o all non-wo k ac i i ies, i.e.,
sU
i
−
∑
j∈{ ,o}s
E
i
,
j
>0∀i≥
2
.
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 19 o 28 26
The well-being e ec o his (po en ial) con ac gain in non-wo k ac i i ies is pos-
i i e and s a is ically signi ican o bo h g oups. Since ime wi h o he s is gene ally
pe cei ed as mo e enjoyable han ime alone, inc easing he amoun o ime spen
wi h o he s leads o mo e enjoymen . We ind a la ge e ec o he pa ne ed g oup
(+ 0.099) han he singles (+ 0.068), pe haps because li ing wi h a pa ne gene ally
enhances he a ailabili y o social con ac s, pa icula ly om one’s household and a
spouse’s ci cle o iends.
The i s h ee s eps o he i e-s ep decomposi ion join ly cons i u e he ime-
composi ion e ec o he wo-s ep decomposi ion. The ime-composi ion e ec cap-
u es he hypo he ical well-being change ha would a ise i employed people could
shi hei wo king ime o non-wo k ac i i ies bu main ain hei a e age enjoymen
le els in each kind o ac i i y. Fo mally, his co esponds o he expe ienced well-
being o
∑I
i=2
sU
i
𝜀
E
i
, which is he same as he one in Eq. (4) since
𝜀
E
i=
∑
j∈{a, ,o}
�
sE
i,j∕sE
i
�
𝜀E
i,
j
. In he i e-s ep decomposi ion, we decompose his
change in o a pa a ibu able o he pu e change in ypes o ac i i ies and ano he
pa ha can be a ibu ed o he di e ences in he p e alence and ypes o social
con ac s people ha e in hese ac i i ies. Ou esul s sugges ha pa ne ed and single
people bene i om less ime wo king. The con ac e ec s, howe e , di e be ween
he wo g oups. Fo hose wi h pa ne s, losing con ac s a wo k does no ma e
much o well-being, bu he po en ial abili y o spend mo e ime wi h iends and
amily in he gained non-wo k ime could ha e a posi i e and s a is ically signi ican
impac on well-being. Combining bo h e ec s shows ha he la e signi ican ly
domina es he o me e ec . Singles lose enjoymen when hey canno mee o he
people a wo k anymo e. They also bene i om he po en ial enhancemen o p i-
a e con ac s. Bo h e ec s a e o simila magni ude bu ope a e in di e en di ec-
ions, so he combined e ec is no s a is ically di e en om ze o.
We now u n o s eps 4 and 5, which u he decompose he saddening e ec o
he wo-s ep decomposi ion. In s ep 3, we assumed ha he con ac composi ion in
he addi ional non-wo k ime could be he same as ha in he obse ed non-wo k
ime (i.e., o each ac i i y
i
we imposed he sha es
sE
i,j
∕s
E
i
on he en i e ime sha e
sU
i
). Fo a ious easons, his migh no be he case. Po en ial con ac s, i.e., amily
and iends, migh no be a ailable du ing he day ime o he unemployed migh no
wan o mee o he people as much. In ei he case, he con ac composi ion o an
ac ual unemployed pe son migh no be he same as ha o he hypo he ical
employed pe son wi h a ully lexible schedule pos ula ed in s ep 3. Hence, he
ou h s ep applies he ac ual con ac dis ibu ion o he unemployed, such ha expe-
ienced well-being is
This e lec s he enjoymen change ha would ake place i he p e alence o
social con ac s co esponded o i s ac ual alues among he unemployed. This ou h
s ep (con ac composi ion when unemployed) shows ha bo h g oups would expe i-
ence signi ican ly lowe a e age enjoymen han hypo hesized in s ep 3. This sug-
ges s ha he unemployed su e om a con ac loss in he sense ha hey canno
(5)
∑I
i=
2
∑j∈{a
,
,
o}
sU
i,j𝜀E
i,
j
T.T.A.Hoang, A.Knabe
26 Page 20 o 28
enjoy hei day as much as hey would i hey could combine he mo e ad an a-
geous con ac s uc u e o an employed pe son wi h he ime a ailabili y o he
unemployed.
In he i h and las s ep, we now accoun o he di e ences in he enjoymen ha
employed and unemployed people pe cei e e en i bo h pe o m he same ac i i ies
and ha e he same social con ac s. When assigning he ac ually obse ed enjoymen
le el o unemployed people o he hypo he ical pe son cons uc ed in s ep 4, we
a i e a he ac ual a e age o unemployed pe sons in he sample whose day-a e age
well-being is gi en by
The change in a e age enjoymen ob ained in his las s ep can be e e ed o as
an adjus ed saddening e ec . I is he o iginal saddening e ec om he wo-s ep
decomposi ion bu cleansed o he impac o changes in he composi ion o social
con ac s. E en hough he poin es ima es o his e ec a e sizable (especially o he
ma ied/cohabi ing g oup), nei he o he wo es ima es is s a is ically signi ican .15
Summing up, he ex ended decomposi ion shows ha he mos impo an d i e
o he gap in expe ienced well-being be ween he employed and he unemployed
is he di e ence in how hey alloca e hei ime o di e en ac i i ies. The mo e
ime indi iduals spend on leisu e, he mo e hey enjoy hei days on a e age. This
also explains why he unemployed wi hou pa ne s, who alloca e he la ges sha e
o ime o leisu e among he ou g oups, ha e he highes day-a e age enjoymen
sco e, e en hough hey epo he lowes li e sa is ac ion. In e es ingly, while he e
a e subs an ial di e ences be ween he employed and he unemployed ega ding
who hey mee du ing he day and o how long, hese di e ences in social con ac s
con ibu e only li le o he o al well-being gap.
When looking a single and pa ne ed pe sons sepa a ely, we ind ha ma ied/
cohabi ing people enjoy con ac a wo k only a li le, so hey do no seem o su e
much when losing hem due o unemploymen . E en hough hey would bene i i
hey used he addi ional ime o mee o he people, we do no see an inc ease in he
amoun o non-wo k ime hey spend wi h o he s. Hence, he o al e ec o combin-
ing all h ee social-con ac channels is small and no s a is ically di e en om ze o.
Fo singles, we obse e ha employed people would lose mo e expe ienced well-
being i hey had o gi e up spending ime wi h o he s a wo k. They could ully
(6)
∑I
i=2∑j∈{a, ,o}
sU
i,j𝜀U
i,
j
15 One could also imagine di e en o de ings o he speci ic decomposi ion s eps. Howe e , hese seem
in ui i ely less appealing o would p oduce inconsis encies han he chosen o de . Fo example, one
could swi ch he o de o he ime-composi ion and he saddening e ec . Howe e , since no enjoymen
in o ma ion is a ailable o wo k ac i i ies o he unemployed, one could no assign he enjoymen le els
o unemployed pe sons o all conside ed ac i i ies. The “con ac loss a wo k”-e ec has o be placed
be o e he ac i i y-composi ion e ec . O he wise, he e would no be any wo k episodes le o which
he enjoymen con ibu ion o ha ing con ac s a wo k could be de e mined. Placing he “con ac gain
in non-wo k ac i i ies”-e ec be o e he “ac i i y composi ion”-e ec is p oblema ic because one would
ha e o assign con ac s wi h iends and amily ha ac ually ake place du ing nonwo k ac i i ies o imes
when esponden s a e s ill obse ed a wo k. Fo hese easons, we belie e ha ou p oposed o de o
decomposi ion is bes sui ed o e eal he sepa a e con ibu ions o con ac s and ac i i ies.

Social con ac s, unemploymen , andexpe ienced well‑being:… Page 21 o 28 26
o se his loss i hey used hei expanded ee ime o mee o he s. Howe e , we
only see a sligh inc ease in he amoun o non-wo k ime du ing which singles mee
o he people. This inc ease su ices o p oduce a s a is ically signi ican inc ease in
he enjoymen de i ed om ha ing mo e non-wo k con ac s.
Taking all h ee e ec s oge he , we ind ha he con ac loss a wo k is a leas
pa ially o se by mo e non-wo k con ac s, such ha he o al e ec is no s a is i-
cally di e en om ze o. Hence, nei he o pa ne ed no o single pe sons do we
ind e idence ha he di e ences in he con ac composi ion be ween he employed
and he unemployed con ibu e o di e ences in hei expe ienced well-being.16
I is wo h no ing he simila i y be ween ou decomposi ion app oach and a
s anda d Ki agawa–Blinde –Oaxaca decomposi ion (KBO). In ac , when we un a
eg ession o episode enjoymen on a ully in e ac ed se o ac i i y- ype and con-
ac - ype dummies, weigh ing he obse a ions in he eg ession using episode du a-
ions and c oss-sec ional dia y weigh s and hen conduc a KBO o mean enjoymen
di e ences based on ha eg ession, we ob ain esul s ha a e e y simila o he
ones p esen ed in Table3. One di e ence be ween ou app oach and a KBO a ises
om he ea men o episodes wi h missing enjoymen in o ma ion. In ou decom-
posi ion, we i s de e mine a e age enjoymen le els in he di e en ac i i y-con ac
combina ions and hen assign hese o hypo he ical dai y ypes de e mined om all
obse ed episodes wi h ime-use in o ma ion on ac i i ies and con ac s. Con a y
o a s anda d KBO, ou app oach hus also akes in o accoun he ime-use in o -
ma ion con ained in obse a ions wi hou epo ed enjoymen . Ano he di e ence is
ha ou decomposi ion dis inguishes be ween he (po en ial) “con ac gain in non-
wo k ac i i ies” and he “con ac shi when unemployed.” Sepa a ely accoun ing o
bo h e ec s allows linking ou decomposi ion o he wo-s ep app oach, whe eas he
s anda d KBO would only yield an es ima e o he sum o he wo e ec s.
5.3 Ma ching employed andunemployed pe sons using en opy balancing
To his poin , we ha e analyzed he di e ences be ween employed and unemployed
pe sons. Clea ly, people a e no andomly assigned o he wo g oups. As shown in
Table1, he employed and he unemployed di e in hei demog aphic and socioeco-
nomic cha ac e is ics. Fo example, he unemployed a e gene ally younge and ha e
less o en ob ained uni e si y deg ees han he employed. Hence, we canno necessa ily
in e p e he obse ed well-being di e ences be ween he wo g oups as causal e ec s
o unemploymen in he sense ha he obse ed di e ences a e he bes p edic o o
wha would happen o a ep esen a i e employed pe son when becoming unemployed.
16 The bene i s o wo k migh no only be los when becoming unemployed, bu also when lea ing he
labo ma ke o o he easons. In he supplemen a y ma e ial o his a icle, we p esen he esul s om
decomposing he di e ences be ween he employed and all non-employed indi iduals (Table A2). To
summa ize he indings, he non-employed enjoy hei days, on a e age, mo e han he employed. The
decomposi ion sugges s ha he e is no nega i e saddening e ec o he non-employed. Mo eo e , ma -
ied o cohabi ing non-employed pe sons ha e mo e con ac s o people ou side he wo kplace han he
employed, which u he con ibu es o hei enjoymen . Non-employed singles, on he o he hand, do no
seem o ha e mo e non-wo k con ac s han employed singles.
T.T.A.Hoang, A.Knabe
26 Page 22 o 28
In his subsec ion, we will ex end ou analysis by ma ching he unemployed and
he employed o elimina e he impac o di e ences in obse able cha ac e is ics.
The ma ching is conduc ed using en opy balancing (Hainmuelle 2012), which
eweigh s all obse a ions in he con ol g oup ( he employed) in a way ha hei
obse able cha ac e is ics co espond o hose o he ea men g oup ( he unem-
ployed). We choose age, household size, he numbe o child en li ing in he house-
hold, gende , and he highes a ained le el o quali ica ion as he a iables on which
he ma ching is pe o med.17 En opy balancing calcula es weigh s such ha he dis-
ibu ion o hese a iables (in e ms o i s and second momen s) is he same in he
ea men and he con ol g oup. A e his eweigh ing, he employed and he unem-
ployed in he sample a e simila in e ms o age, gende s uc u e, educa ion, and
household size while di e ing om each o he in hei employmen s a us.18
Table4 shows he decomposi ion analysis using he en opy-balanced samples. The
esul s a e e y simila o hose ob ained wi h he unbalanced samples. In pa icula , o
ma ied/cohabi ing people, we ind only a small (and he e e en s a is ically insigni i-
can ) enjoymen e ec o he loss o con ac s a wo k. A la ge and s a is ically signi i-
can inc ease in well-being is associa ed wi h he adjus ed ac i i y composi ion (mo e
ime spen (alone) on non-wo k ac i i ies). Conce ning he con ac composi ion du -
ing non-wo k ime, people in his g oup could bene i subs an ially om spending ime
wi h o he s i hey spen he addi ional non-wo k ime wi h o he s o he same ex en as
hei ac ually obse ed non-wo k ime. We ind a s a is ically signi ican enjoymen gain
associa ed wi h he inc eased non-wo k con ac s when unemployed. On he o he hand,
jus as in he unbalanced analysis, we see ha he unemployed spend mos o hei addi-
ional non-wo k ime alone, so he e is no indica ion o well-being di e ences be ween
he pa ne ed employed and pa ne ed unemployed when conside ing he o al change
in con ac composi ion. The adjus ed saddening e ec , which emains a e accoun ing
o changes in ime use and social con ac s, is no s a is ically signi ican .
Fo singles, he esul s a e also e y simila o hose o he unbalanced analysis.
The e is a small bu s a is ically signi ican educ ion in well-being om he loss o
con ac s a wo k. Di e ences in he composi ion o ac i i ies con ibu e posi i ely o
expe ienced enjoymen . The combined hi d and ou h s eps (so he impac o he
ac ual change in non-wo k con ac s) a e posi i e and s a is ically signi ican a he
10% le el. This posi i e e ec pa ly o se s he nega i e impac o losing con ac s
a wo k. The combina ion o all con ac e ec s ( he i s , hi d, and ou h) is nega-
i e bu small and s a is ically insigni ican . Hence, also wi h en opy balancing, we
do no ind e idence ha he expe ienced well-being o singles would be a ec ed by
changes in hei con ac composi ion when becoming unemployed.
O e all, mos esul s emain unchanged when we conduc ou analyses on he
ma ched sample. This sugges s ha he obse ed di e ences a e no d i en by he
17 We choose hese balancing a iables because he subsamples o employed and he unemployed pe -
sons ha e di e en dis ibu ions o hese a iables (see Table 1), bu he a iables a e a guably exog-
enous, i.e. no a ec ed by unemploymen i sel . Fo ha eason, we do no use income o li e sa is ac ion
as balancing a iables, as bo h migh be endogenous. The e is su icien o e lap in he co a ia e dis ibu-
ions. The Bha acha yya coe icien is la ge han .95 o all co a ia es.
18 Summa y s a is ics on he i s and second momen s o he balancing a iables be o e and a e
en opy balancing a e epo ed in he Supplemen a y Ma e ial (TableA1).
Social con ac s, unemploymen , andexpe ienced well‑being:… Page 23 o 28 26
di e en composi ions o he obse ed cha ac e is ics o he wo g oups. This does
no p eclude he possibili y ha he e emain di e ences in unobse ed cha ac e is-
ics be ween he employed and he unemployed, such ha we do no wan o sugges
ha hei obse ed well-being di e ences can necessa ily be equa ed wi h he causal
impac o unemploymen .
6 Conclusion
In his s udy, we in es iga e di e en channels ha con ibu e o he subjec i e well-
being o employed and unemployed pe sons as hey expe ience i daily, using da a
ha combine ime-use wi h ac i i y- and con ac -speci ic well-being in o ma ion.
While o he s udies ha e used such da a o examine he expe ienced well-being
o pa en s o couples depending on hei wo k a angemen s (Flood and Genadek
2016; Musick e al. 2016), his is— o he bes o ou knowledge— he i s s udy ha
Table 4 Two-s ep and i e-s ep decomposi ions (wi h en opy balancing)
Ma ied/cohabi a ing Single
5.285 A e age employed pe son 5.382
(0.026) (0.048)
-0.014 Con ac loss a wo k -0.053 *
(0.015)
5.27
15
.329
(0.029) (0.057)
0.273 *** 0.183 *** Ac i i y-composi ion e ec 0.303 *** 0.314 ***
(0.029)(0.034) (0.049)(0.039)
5.45
45
.632
(0.036) (0.051)
0.104 *** 0.064 ***
(0.009) (0.012)
(0.033) (0.050)
-0.090 *** -0.030 *
(0.022) (0.017)
-0.249* 5.46
85
.666 -0.092
(0.147)(0.045) (0.050)(0.127)
-0.15
9-
0.062
(0.147) (0.126)
5.309 A e age unemployed pe son 5.604
(0.146) (0.116)
"Saddening e ec "
Con ac shi when unemployed
"Saddening e ec "
Adjus ed saddening e ec
" ce enoi isopmoc-emiT"
"Time-composi ion e ec "
Con ac gain in non-wo k
ac i i ies
5.55
85
.696
S anda ds e o s om (clus e ed) boo s apping wi h 5000 eplica ions in pa en hesis. Signi icance le el:
* p < 0.10; ** p < 0.05; *** p < 0.01
T.T.A.Hoang, A.Knabe
26 Page 24 o 28
speci ically ocuses on he ole o social con ac s o he well-being o unemployed
pe sons. Using ich da a on ime use and enjoymen o e he en i e day, we examine
he hypo heses ha he unemployed ha e less con ac wi h o he people han he
employed, ha hey a e less able o enjoy he emaining ime hey spend wi h o he
people, e.g., due o shame and pe cei ed s igma iza ion, and ha his aken oge he
is e lec ed in lowe expe ienced well-being o he unemployed. We de elop an
ex ended decomposi ion app oach o iden i y he deg ee o which he abili y o allo-
ca e ime ac oss ac i i ies eely, he di e ences in enjoymen a ings, and he access
o social con ac s can explain he di e ences in du a ion-weigh ed enjoymen sco es
o he employed and he unemployed.
The s a ing poin o ou analysis is he claim ha one o he la en bene i s o
employmen is he es ablishmen o social con ac s ou side he amily (Jahoda
1981). Ou esul s pa ially suppo his claim. Fi s , people gene ally enjoy being
wi h o he s. Fo bo h wo k and non-wo k ac i i ies, we ind ha expe ienced well-
being is ypically highe when people engage in hese ac i i ies wi h o he people
han i hey a e alone, and being wi h people om ou side he household is o en
a ed e en mo e enjoyable han being wi h o he household membe s. Second, he
unemployed spend mo e ime alone han he employed. As expec ed, he employed
gain a la ge sha e o hei social con ac s a wo k, o which he unemployed do no
ha e access. We also ind ha , al hough he unemployed ha e mo e a ailable ee
ime and a e enjoying he ime hey spend wi h o he s, in pa icula wi h non-house-
hold acquain ances, he unemployed a e no able o ully compensa e o he ime
he employed spend mee ing o he s a wo k by inc easing he amoun o ime hey
spend wi h o he con ac s. One eason o his could be he ime cons ain s o o he
people. E en hough he unemployed ha e a ela i ely high deg ee o ime lexibil-
i y, we obse e ha hey mee o he people mo e o en du ing ypical a e -wo k
hou s o on he weekend.
Ou decomposi ion shows, howe e , ha hese subs an ial di e ences in he
ime spen wi h o he s do no ma e much o he day-a e age expe ienced
well-being o employed and unemployed pe sons. The employed ha e con ac s
a wo k, while he unemployed do no , bu hese con ac s seem o be o li le
alue o he expe ienced well-being o he employed. Among pa ne ed indi-
iduals, he e is only a small and no always s a is ically signi ican posi i e con-
ibu ion o he a e age enjoymen o wo king wi h o he s compa ed o wo king
alone. Unemployed pe sons wi h pa ne s also spend abou he same amoun o
ime wi h non-wo k con ac s as he employed. Since ha ing wo k con ac s is no
associa ed wi h highe expe ienced well-being among he employed, while he
unemployed do no spend less ime wi h non-wo k con ac s han he employed,
we do no ind ha di e ences in social con ac s a e ela ed o well-being di e -
ences be ween employed and unemployed pe sons wi h pa ne s. Among singles,
in con as , we see a sligh ly la ge and s a is ically signi ican di e ence in he
expe ienced well-being when wo king wi h o he s compa ed o wo king alone,
bu he magni ude is s ill ela i ely small. A he same ime, we obse e ha
unemployed singles can spend subs an ially mo e ime (+ 1 h) wi h non-wo k
con ac s han employed singles, which bene i s hei expe ienced well-being