Li, Bingqiang; Nie, Wenjie; Zuo, Xuan; Zuo, Heping
A icle
How does educa ion a ec knowledge and
in e gene a ional social class mobili y in China?
Jou nal o Inno a ion & Knowledge (JIK)
P o ided in Coope a ion wi h:
Else ie
Sugges ed Ci a ion: Li, Bingqiang; Nie, Wenjie; Zuo, Xuan; Zuo, Heping (2025) : How does educa ion
a ec knowledge and in e gene a ional social class mobili y in China?, Jou nal o Inno a ion &
Knowledge (JIK), ISSN 2444-569X, Else ie , Ams e dam, Vol. 10, Iss. 3, pp. 1-13,
h ps://doi.o g/10.1016/j.jik.2025.100678
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/327579
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 ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
How does educa ion a ec knowledge and in e gene a ional social class
mobili y in China?
Bingqiang Li
a
, Wenjie Nie
a
, Xuan Zuo
b
, Heping Zuo
a,*
a
Jiangxi Voca ional Educa ion & Indus ial De elopmen Resea ch Ins i u e, Jiangxi Science & Technology No mal Uni e si y, No. 589, Xue u A enue, Nanchang,
330038, Jiangxi, China
b
School o In e na ional Educa ion, Jiangxi Science & Technology No mal Uni e si y, No. 589, Xue u A enue, Nanchang, 330038, Jiangxi, China
ARTICLE INFO
JEL classi ica ion:
C50
I24
J01
Keywo ds:
Educa ion
Class
In e gene a ional mobili y
In e gene a ional ansmission
ABSTRACT
Educa ion is an impo an ool used o equip knowledge and pu sue sel -imp o emen , which may a ec
in e gene a ional social class mobili y h ough human capi al accumula ion. Using 8014 mixed c oss-sec ion
samples om he 2017, 2018 and 2021 China Gene al Social Su ey (CGSS), an mul i-nominal logis ic eg es-
sion model was adop ed o empi ically analyse he impac o educa ional le els on in e gene a ional social class
mobili y in China. The econome ic esul s indica ed ha (1) highe educa ion conside ably p omo es subjec i e
and objec i e upwa ds in e gene a ional social class mobili ies, wi h he objec i e e ec exceeding he subjec i e
e ec by 3.4 %; (2) he implemen a ion o he compulso y educa ion law and he expansion o highe educa ion
en olmen has inc eased he po en ial o high educa ional le el o d i e upwa ds in e gene a ional social class
mobili y; (3) among he di e en ypes o educa ion, popula ion mig a ion exe s a epelling e ec on objec i e
upwa ds in e gene a ional social class mobili y, whe eas amily backg ound exe s a nega i e e ec on subjec i e
in e gene a ional social class mobili y; and (4) he e ec s o educa ion on in e gene a ional social class mobili y
a e he e ogeneous, wi h women, non-ag icul u al hukou holde s and g oups in de eloped a eas acing g ea e
di icul ies in achie ing social mobili y h ough educa ion. The main con ibu ion o his s udy is i s ini ial
de ec ion o he a ying impac s o educa ional le el on in e gene a ional social class mobili y, as in es iga ed
h ough subjec i e and objec i e pe spec i es.
The p oposed ques ion
The deba e be ween ‘s udying being useless’ and ‘knowledge
changing a e’ (Bel an Tapia & de Miguel Salano a, 2021;
Gai al-Casad´
o e al., 2022; Wu e al., 2023) and be ween he anxie y o
‘success ul people coming ou o poo amily’ and ha o ‘success ul
people no coming om poo amilies’, o a la ge ex en , add esses he
issue o whe he educa ion can b eak down class solidi ica ion h ough
he enhancemen o knowledge. The educa ional scale in China has
apidly de eloped in ecen yea s, e.g. he numbe s o college g adua es
in 2003, 2010, 2020 and 2024 we e 2.13, 6.31, 8.74 and 11.79 million,
espec i ely. Meanwhile, he hinking pa adigm o ‘expec a ions o
hei child en’ po ays Chinese pa en s’ belie ha he social class
mobili y acili a ed by hei o sp ing’s educa ion is an impo an pa h
o b eaking h ough class solidi ica ion and ealising class leap ogging.
I is o en a gued ha educa ion is a ca alys o class mobili y (B een &
Ka lson, 2014; Ca s ensen & Emmenegge , 2023; Gold ho pe, 2014;
Wol , 2020), which is impo an o pe sonal g ow h and p omo ion o
social equi y and jus ice. Fu he mo e, i is no ewo hy ha educa ion
plays a pi o al ole in he acquisi ion and dissemina ion o knowledge.
The educa ional p ocess i sel con ibu es o he de elopmen o c i ical
hinking and p oblem-sol ing skills, which a e impo an componen s
o indi iduals o succeed in a complex social s uc u e. In addi ion,
educa ion can democ a ise access o in o ma ion and knowledge,
po en ially educing in o ma ion asymme y and social inequali ies.
Howe e , he impac o educa ion on social class mobili y is a ec ed by
mul iple ac o s, such as inequi able oppo uni ies owing o esou ce
alloca ion and mobili y he e ogenei y owing o di e ences in he quali y
o educa ion (Mul ey & Li, 2023). The e o e, i is necessa y o in e p e
he impac o educa ion on in e gene a ional social class mobili y in
China. Conside ing he ela ed li e a u e, schola s mainly conduc
esea ch on he di e en aspec s o in e gene a ional mobili y, he ac-
o s a ec ing in e gene a ional mobili y and he in luence o educa ion
on in e gene a ional social class mobili y.
* Co esponding au ho .
E-mail add esses: [email p o ec ed] (B. Li), [email p o ec ed] (W. Nie), [email p o ec ed] (X. Zuo), [email p o ec ed] (H. Zuo).
Con en s lis s a ailable a ScienceDi ec
Jou nal o Inno a ion & Knowledge
jou nal homepage: www.else ie .com/loca e/jik
h ps://doi.o g/10.1016/j.jik.2025.100678
Recei ed 9 No embe 2024; Accep ed 19 Feb ua y 2025
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
A ailable online 8 Ma ch 2025
2444-569X/© 2025 The Au ho s. Published by Else ie España, S.L.U. on behal o Jou nal o Inno a ion & Knowledge. This is an open access a icle unde he CC
BY-NC-ND license (
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ ).
The e is a la ge quan i y o li e a u e on in e gene a ional mobili y,
which mainly analyses om he pe spec i es o class, income, educa ion
and occupa ion. Some schola s in e p e hese by combining mul iple
dimensions. Rega ding in e gene a ional social class mobili y, He el
and G oh-Sambe g (2019) and DiP e e (2020), Toubol and La sen
(2017) a gued ha in e -class inequali y and social mobili y a e co e-
la ed. Meanwhile, Bukodi and Pasko (2020) ound ha emale class
mobili y is mo e impeded by class ba ie s han male class mobili y,
which a e in luenced by gende di e ences in he le els and pa e ns o
in e gene a ional social class mobili y. Sun e al. (2021) epo ed ha
he e a e di e ences in he le el and end o class and income mobili ies
in China and ha class is gene ally likely o be a be e measu e o
socio-economic s a us han cu en income. Rega ding he in e gene a-
ional mobili y o income, Yaish and K aus (2020) as well as Chow and
Guppy (2021) in es iga ed he e ec o in e gene a ional income
mobili y based on he in e gene a ional income elas ici y index model
p oposed by Solon (1992). Che y e al. (2014) and An e al. (2022)
de eloped a nonlinea model o answe he ques ion o he in insic
co ela ion be ween o sp ing’s income and pa en s’ income. Song e al.
(2022) examined he dynamic in e gene a ional income dispa i ies
h ough li e cycle by c ea ing a comp ehensi e able o in e gene a-
ional income mobili y. Rega ding in e gene a ional educa ional
mobili y, Huq e al. (2021) in es iga ed in e gene a ional educa ional
mobili y in Bangladesh and ound ha indi iduals expe ienced g ea e
mobili y in p ima y and seconda y educa ion han in highe educa ion.
Van de Weide e al. (2024) emphasised ha he uneduca ed pa en s’
o sp ing we e mo e likely o lack he means o ecei e be e educa ion,
whe eas he nex gene a ion o highly educa ed pa en s was mo e
capable o main aining he eli e s a us o he amily (Lin, 2020).
Rega ding in e gene a ional occupa ion mobili y, Lissi sa and
Chachash ili-Bolo in ((2023)) posi ed ha he pa en s’ occupa ional
s a us exe s a mo e de e mina i e e ec on he elucida ion o he
oca ional p e e ences o hei o sp ing han hei educa ional le el.
B une i and Fiaschi (2023) ound ha he a he being in he lowe class
has insu icien incen i es o upwa ds mobili y as pe i s o sp ing. In
addi ion, some schola s compa ed indica o s ela ed o he di e en
ypes o in e gene a ional mobili y and in es iga ed he in e ac ions
be ween hese indica o s. Fo example, Su´
a ez-A besú e al. (2024)
analysed educa ion, occupa ional and income mobili ies using he
ansmission ma ix me hod by ob aining da a om i e coun ies in
A ica. They ound ha educa ion achie ed he highes le el o mobili y
and exhibi ed signi ican di e ences be ween coun ies. Tang (2023)
explo ed he impac o educa ion on in e gene a ional income mobili y
and ound ha upwa ds educa ional mobili y was e ec i e in blocking
he ansmission o in agene a ional income gaps om he pa en o he
indi idual, whe eas downwa ds educa ional mobili y would exace ba e
he in e gene a ional income ansmission.
The e a e mul iple complexi ies in he ac o s a ec ing in e gene -
a ional mobili y, which could be mainly explo ed om he asc ibed and
he acqui ed aspec s. In e ms o indi idual asc ibed ac o s, which
usually include ace, gende , age, pa en ’s educa ional le el and occu-
pa ional s a us (Jha & Wha on, 2023; Lee & Mo ime , 2021; Scandone,
2022; Wi e een & Wes e man, 2023), Lampa d (2007) de eloped a
logis ic eg ession model using he da a om he B i ish Household
Panel Su ey and p oposed ha pa en s’ educa ional le el and occu-
pa ional ca ego y play a i al ole in he o sp ing’s acquisi ion o se ice
posi ion. The indi idual’s acqui ed ac o s mainly include educa ion,
quali ica ion, ma iage, skill and job (Bu a ana & T upp, 2021; Guo,
2022; Kou ellos e al., 2020). Fo example, G onning and K iesi (2022)
poin ed ou ha an indi idual’s gene al skills and knowledge play a
decisi e ole in long- e m upwa ds mobili y. Liu e al. (2023) a gued
ha In e ne skills explici ly a ec class mobili y. In addi ion, he
mac o-en i onmen has an explici impac on in e gene a ional
mobili y, such as he digi al economy, ma ke isa ion p ocess and policy
egime (Xing e al., 2020). Fo example, A un and A un (2023) epo ed
ha digi al jobs a e ypically associa ed wi h highe income le els and
g ea e social mobili y. Ma indale and Lehdon i a (2023) demon-
s a ed ha digi al ans o ma ion o he labou ma ke may ep oduce
di e ences in job oppo uni ies and mi iga e di e ences in employe s’
choice o wo ke s based on class selec ion. Lipp´
enyi and Ge be (2016)
con i med assump ion o a dis inc i e s a e socialis mobili y egime and
a gued ha he s a e ins i u ion would gene a e qui e di e en mobili y
pa e ns. Jackson and E ans (2017) compa ed social mobili y pa e ns
among 13 Eas e n and Cen al Eu opean coun ies in he ea ly 1990s
wi h hose in he la e 2000s, which is he pe iod o he ansmission om
socialism o capi alism. They ound ha he e was a conside able
decline in ela i e social mobili y be ween hese wo pe iods, which is
consis en wi h he indings o Zhou and Xie (2019) in China.
The e a e wo a i udes o educa ion a ec ing in e gene a ional so-
cial class mobili y, namely, unc ional opinion and con lic opinion.
Func ionalism holds ha educa ional le el de e mines occupa ional
heigh and social class and could pe u b social class mobili y by
a ec ing indi idual skills (Kuha e al., 2021). Co esponding esea ch
also showed ha educa ion in luences upwa ds in e gene a ional
mobili y (Dea y e al., 2005; Duan e al., 2022; E. Bukodi & Gold ho pe,
2011; Johnson e al., 2010). Meanwhile, he e ec s o social class so-
lidi ica ion and o sp ing educa ion a e no negligible (Song e al.,
2016). In addi ion, he p obabili y o upwa ds social class mobili y is
highe o he well-educa ed g oup han o he g oups, which was also
demons a ed by B een (2019). Those schola s wi h con lic ing opinions
sugges ed ha educa ion plays an ins umen al ole in ep oducing so-
cial class, i.e. he ‘class solidi ica ion’ e ec (Alb ech & Alb ech , 2010;
Lindley & Machin, 2012). Fo example, Bou dieu (1984) in oduced he
heo y o cul u al capi al ep oduc ion, which s a ed ha he cul u al
mechanism o class ep oduc ion is he hidden in e gene a ional ans-
mission wi hin he amily; his iew shows ha educa ion enables he
in e gene a ional ansmission o class s a us, ein o cing he o iginal
unequal social s uc u e (Wang, 2020). Jo a e al. (2024) con i med
ha he expansion o educa ional oppo uni ies o he equalisa ion o
educa ional access does no subs an ially con ibu e o in e gene a ional
social mobili y in u ban A gen ina. Remes (2022) ound ha educa-
ional le el may con ibu e o he decline in middle class and he
pola isa ion o classes in Finland. Sal anes (2023) s a ed ha educa ion
is no a key ac o in he success o adul s in he labou ma ke and ha
inc easing he educa ional le el o poo child en canno help o imp o e
hei pe o mance in he labou ma ke .
As epo ed by exis ing li e a u e, mos schola s ocus on a single
dimension o in e gene a ional educa ional mobili y o social class
mobili y. Howe e , his s udy ook educa ion as he main ac o
a ec ing class mobili y and used he China Gene al Social Su ey
(CGSS) da abase o de elop he heo e ical model. I explo ed he di ec
causali y be ween educa ion and class mobili y om he pe spec i e o
absolu e in e gene a ional mobili y. Second, al hough some schola s
ha e in es iga ed he long- e m impac o educa ion on social class
mobili y by acking da a, hey igno ed he ole o ‘educa ion in la ion’
in di e en his o ical pe iods. The p esen s udy di ided he samples
in o h ee di e en g oups acco ding o he landma k e en s so ha he
esul s would be mo e possible. Thi d, schola s examined he objec i e
class mobili y and he subjec i e class iden i y sepa a ely, whe eas he
p esen s udy combined bo h o make a co esponding compa ison and
simul aneously examined he media ing and mode a ing e ec s o
popula ion mig a ion and amily backg ound. The main con ibu ion o
his s udy is i s empi ical analysis o he he e ogenei y o he impac o
educa ion on in e gene a ional social class mobili y in he objec i e and
subjec i e dimensions.
Model es ablishmen and esea ch hypo hesis
Model es ablishmen
In e gene a ional mobili y usually has h ee ypes, namely, upwa ds
mobili y, in e gene a ional ansmission and downwa ds mobili y,
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
2
which we e used as explana o y a iables in his s udy. Taking he
alues 0 and 1 and could no sa is y he de ini ion o bina y ca ego ical
a iables. Hence, we used he mul i-nominal logis ic eg ession model
(MLRM) (Yi & Yao, 2023), which is a model ha uses uno de ed
mul i-ca ego ical a iables as explana o y a iables. I s ini ial model is
he ollowing:
Logi (Y /Yi) = β0+β1⋅X1+β2⋅X2+.... +βn⋅Xn(1)
He e, deno es h ee di e en ypes o in e gene a ional mobili y
pa e ns, and hus, Y
is se o own h ee alues, −1, 0 and 1, which
ep esen downwa ds mobili y, in e gene a ional ansmission and up-
wa ds mobili y, espec i ely. As he MLRM needs o de ine a ce ain
alue as he benchma k in he eg ession analysis, his s udy used
in e gene a ional ansmission (i.e. Y
=0) as he benchma k o
compa a i e analysis. Conside ing ha his s udy explo es he impac o
educa ion on in e gene a ional social class mobili y, i is necessa y o
add he a iable o educa ional le el and o in oduce explana o y and
con ol a iables. Hence, Eq. (2) is exp essed as ollows:
Mobili yi=Ln[P(Yi= )/P(Yi=0)]=β0+β1⋅Edui+∑βi⋅Xi+
ε
i(2)
Eq. (2) is he benchma k model o his s udy. He e, Mobili yi deno es
he in e gene a ional mobili y mode; Edui, he o sp ing’s educa ional
le el, which is ca ego ised in o p ima y, seconda y and highe educa-
ion. We ake seconda y educa ion as he benchma k and epo he
co ela ion coe icien s o p ima y and high educa ion; Xi deno es he
con ol a iable and
ε
i ep esen s he pe u ba ion e ms.
The modes o in e gene a ional social class mobili y include he
objec i e and subjec i e dimensions. As pe he objec i e dimension,
in e gene a ional social class mobili y could be cha ac e ised in e ms o
he di e ence in socio-economic s a us be ween he o sp ing and he
pa en s. I is mainly quan i ied by he socio-economic s a us index o an
indi idual’s posi ion in he socio-economic s uc u e, which could be
measu ed by he a e age educa ional le el and income (Iam akul e al.,
2022; Tan e al., 2020). As pe he subjec i e dimension, he ac o s
a ec ing indi idual pe cep ions o class include educa ional le el, in-
come s a us, occupa ional s a us, amily backg ound, social ne wo k and
wel a e. The e o e, in e gene a ional social class mobili y could be
mainly ob ained by compa ing he class o he o sp ing wi h ha o he
a he , so ha di e ences in class iden i y and psychological ac o s lead
o he he e ogenei y o subjec i e class pe cep ions (Gal an e al., 2023;
Meli a e al., 2021; Moss e al., 2023). Acco dingly, subjec i e and
objec i e in e gene a ional social class mobili ies exhibi signi ican
di e ences o easons o in o ma ion asymme y, social compa ison
and cul u al ac o s. Thus, i is necessa y o explo e he impac o
educa ional le el on subjec i e and objec i e in e gene a ional social
class mobili ies, so Eq. (2) can be expanded o Eqs. (3) and (4):
Omi=Ln[P(Yi= )/P(Yi=0)]=β0+β1⋅Edui+∑βi⋅Xi+
ε
i(3)
Smi=Ln[P(Yi= )/P(Yi=0)]=β0+β1⋅Edui+∑βi⋅Xi+
ε
i(4)
He e, Omi and Smi deno e he objec i e and subjec i e in e gene a-
ional social class mobili y, espec i ely; hese wo models a e used o
analyse he subsequen media ing and mode a ing e ec s.
Resea ch hypo hesis
Occupa ional s a us a ainmen is one o he mos impo an in-
dica o s o measu ing in e gene a ional mobili y. Since Blau and
Duncan (1967) p oposed he s a us a ainmen model, educa ional le el
has been he main ocus o esea ch on s a us a ainmen (Bukodi &
Gold ho pe, 2011; Hou , 1988). Educa ion can imp o e indi idual
knowledge o cogni i e abili y and help es ablish a wide ange o social
ne wo ks, achie ing s onge compe i i eness by accumula ing human
capi al le el (Schul z, 1961). The e o e, he indi idual would own mo e
choices in he labou ma ke , which is he key o ob aining high-income
occupa ions and imp o ing social s a us. In iew o he ac ual si ua ion,
he accele a ion o indus ial ans o ma ion and a i icial in elligence
ad ancemen led o highe equi emen s o he comp ehensi e quali y
o wo ke s, making i mo e di icul o people wi h low educa ional
le el o ealise he class leap. Highe educa ion is an impo an ini ia i e
o ealise he c oss-domain o social classes. In ac , acco ding o he
signal sc eening heo y (Spence, 1973) and he eli e heo y (Bühlmann
e al., 2022), unde he condi ion o in o ma ion asymme y, employe s
usually sc een job applican s based on hei educa ion, which would
help indi iduals wi h highe educa ion o ob ain highe posi ions and
be e ea men . Fu he mo e, a high educa ional le el would be mo e
con enien o ob ain p o essional s a us, s eng hen he possibili y o
upwa ds class mig a ion and p omo e an upwa ds end o in e gene -
a ional social class mobili y. The e o e, hypo hesis (H) 1 is p oposed:
Educa ion enhances he p obabili y o upwa ds in e gene a ional social
class mobili y, concu en ly mi iga ing he p opensi y o s a i ica ion
o descending mobili y.
The impac o educa ion on in e gene a ional social class mobili y
signi ican ly a ies ac oss economic and social con ex s (Mullins e al.,
2021). Fo example, business cycle luc ua ions in he economy exe a
push-and-pull e ec on he social class o wo ke s (Becke e al., 2024).
Social change and ma ke luc ua ions a ec he enac men and imple-
men a ion o educa ion policies, which in u n lead o di e en occu-
pa ional equi emen s o g oups wi h he same educa ional le el o e
ime, implying ha he ma ke aises he co esponding educa ion,
c ea ing skilled jobs and enhances he occupa ional dis ibu ion h ough
skill-biased echnological change (Jephco e e al., 2021; Rausche ,
2015). Howe e , some schola s a gued ha he e exis s a
non-syne gis ic in e ac ion be ween social mobili y and sho - e m
luc ua ions in he economic cycle o labou ma ke (T inh & Bukodi,
2022). In China, wo policies, he enac men o he compulso y educa-
ion law in 1986 and he expansion o highe educa ion en olmen in
1999, ha e gained conside able a en ion, and Guo e al. (2019) ound
ha hese policies enhanced in e gene a ional educa ional mobili y in
u ban a eas bu wo sened ha in u al a eas. In addi ion, he dep eci-
a ion o diplomas induced by educa ion in la ion leads o conside able
he e ogenei y in he impac o educa ion on social class mobili y
(G uij e s, 2022). The e o e, H2 is p oposed: In he dynamic con ex o
economic and socie al e olu ion, he impac o educa ion on in e gen-
e a ional social class mobili y is cha ac e ised by a nonlinea pa e n.
F om he pe spec i e o human capi al, indi iduals wi h high skills
and educa ional le els ha e g ea e ini ia i e in selec ing jobs, and he e
a e s ong co ela ions be ween hem (A aki, 2023; A e e al., 2021;
T inidad e al., 2023). In ac , hose who a ain highe educa ion may
exe a s onge push-and-pull e ec on popula ion mig a ion
(Ha haway, 1960), such as ob aining be e jobs, be e heal h ca e and
be e educa ion o hei child en, o achie e he end o ou wa d
mig a ion o shi ing o he ci y. F om he pe spec i e o economic and
social de elopmen , ci ies end o ha e high-quali y esou ces, good
secu i y, bene i policies, e c., which os e alen con e gence wi hin
he u ban a ea. Based on he needs o hei de elopmen , g oups wi h
middle and high educa ional le els would choose occupa ions wi h high
social s a us and high oppo uni y po en ial; hus, a la ge numbe o
high-quali y people would mo e o ci ies o de eloped egions, o ming
a mo e ob ious siphon e ec (Luo e al., 2024). Based on he Toda o
model, i is o en assumed ha an icipa ed emune a ion in u ban sec-
o s o locales su passes non- u al depa men o egions, and he
inducing a che e ec o psychological implica ion is inclined o allu e
indi iduals wi h ad anced educa ional quali ica ion o ie o posi ions
wi h g ea e po en ial and angible bene i s. When an indi idual un-
de goes ou wa d mig a ion, hei employmen oppo uni ies, sou ces o
income and social secu i y migh be supe io o he o iginal le el, which
migh p omp a highe p obabili y o in e gene a ional social class
mobili y. The e o e, H3 is p oposed: Du ing he u banisa ion p ocess,
popula ion mig a ion se es as a pi o al in e media e ac o acili a ing
he impac o educa ion on in e gene a ional social class mobili y.
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
3
The educa ional le els a ained by he o sp ing as well as he
educa ional esou ces a hei disposal is in ica ely associa ed wi h hei
amilial backg ound (Cos a Filho & Rocha, 2022; E ikson, 2016). The
well-es ablished ac ha high educa ion and eli e deg ees a e inc eas-
ingly dependen on amily backg ound, pa icula ly on pa en al inan-
cial esou ces, sugges s ha amily backg ound can ha e an explici ly
mode a ing e ec on he educa ional ou comes o child en (Se ediak &
Helland, 2023; Yan & Gao, 2024). In u n, i would lead o a decline in
in e gene a ional educa ional mobili y, e en gene a ing educa ion
inequali y (Da is-Kean e al., 2021). When he expansion o highe ed-
uca ion inc eases, he eposi o y o he social eli e conside ably expands,
he signalling unc ion o educa ion declines, and he amily backg ound
o o igin would play a la ge ole, leading o ‘nepo ism’, so ha social
s a us ep oduc ion equen ly occu s. Pa en s wi h s ong backg ounds
and high s a us ha e access o a wide ange o social esou ces, and hese
could link hei o sp ing o be e employmen oppo uni ies (To &
F iedman, 2021) and mo e s able ca ee s (To , 2019). A ela i ely
s able amily backg ound se es as a sub le mechanism o he o -
sp ing’s pe cep ion o in e gene a ional social class mobili y. In addi-
ion, a he s wi h high educa ional le el a e also likely o ha e high
social class, esul ing in hei o sp ing’s p esen a ion o be e cogni i e
abili y and ewe pe sonal beha iou al p oblems (Cano, 2022; Ming
e al., 2021; Zhang e al., 2020), which would also impac he pe cep ion
o in e gene a ional social class mobili y. The e o e, H4 is p oposed: The
endu ing amilial con ex ope a es as a clandes ine de e minan shaping
he p ogeny’s consciousness ega ding he dynamics o in e gene a-
ional mo emen wi hin he social hie a chy.
Da a esou ce and index implemen a ion
Da a sou ce and desc ip ion
This s udy used he CGSS da abase. The sample consis ed o China’s
31 p o inces, au onomous egions, municipali ies di ec ly unde he
Cen al Go e nmen and special adminis a i e egions. Since he
occupa ion code used be o e 2016 in he CGSS is ISCO-88, bu no he
la es occupa ion code ISCO-08, he e o e, he blended c oss-sec ional
da a consis ing o he yea s 2017, 2018 and 2021 we e selec ed as he
da a sou ce. Conside ing he ep esen a ion o he a he in he amily
ole, he in o ma ion o he a he s was used o cha ac e ise he basic
si ua ion o he pa e nal gene a ion (Donnelly e al., 2016; Wang e al.,
2019). A e excluding ou lie s and missing alues, omi ing hose who
ha e ne e been educa ed and who we e unemployed and e aining he
sample o 16–65-yea -olds, he inal sample sizes ob ained in he yea s
2017, 2018 and 2021 we e 3510, 3414 and 1946, espec i ely, which
o als 8870. A e con e ing he occupa ion codes o he a he and he
o sp ing o In e na ional Socio-Economic Index (ISEI) alues, he o al
sample size ob ained a e adjus men was 8014 owing o he p esence o
unma ched samples.
Index implemen a ion
Table 1 p esen s i e classi ica ions ega ding he selec ion, de ini-
ion and assignmen o a iables, including dependen , independen ,
media ing, mode a ing and con ol a iables.
Dependen a iable
(1) Objec i e in e gene a ional social class mobili y. This s udy used
he ISEI o measu e he social class o he a he and he o sp ing.
This index could measu e an indi idual’s objec i e socio-
economic s a us owing o he combina ion o mul iple ac o s
o anking and assignmen (Ganzeboom & T eiman, 1996;
Ganzeboom e al., 1992). The ISCO08 codes o he o sp ing’s
cu en occupa ion and he a he ’s occupa ion a he age o 14
yea s we e con e ed in o ISEI alues (Newlands & Lu z, 2024),
and hen he social classes we e combined in o h ee ca ego ies, i.
e. he uppe social class (ISEI g ea e han 61) was assigned he
alue o 2; middle social class (ISEI be ween 31 and 60), 1; and
lowe social class (ISEI less han 30), 0. The social class o he
o sp ing was sub ac ed om ha o he pa en . A di e ence o
o e 0 be ween he wo indica es an upwa ds in e gene a ional
mobili y and is assigned he alue o 1. A di e ence o 0 indica es
an in e gene a ional ansmission and is assigned he alue o 0.
A di e ence o less han 0 indica es a downwa ds in e gene a-
ional mobili y and is assigned he alue o −1.
(2) Subjec i e in e gene a ional social class mobili y. D awing on he
ea men made by Zhang e al. (2016), he ques ion ‘In his
cu en socie y, wha le el o socie y a e you pe sonally in?’ was
used as a subjec i e class a ing o he o sp ing, whe eas he
ques ion ‘On which le el do you hink you amily was a when
you we e 14 yea s old?’ was used as a subjec i e class a ing o
he a he . The o me sco es we e sub ac ed om he la e
sco es o p oduce a con inuous a iable, wi h alues anging om
−9 o 9. He e, samples g ea e han 0 we e ca ego ised as up-
wa ds in e gene a ional mobili y and assigned he alue o 1,
samples equal o 0 we e ca ego ised as in e gene a ional ans-
mission and assigned he alue o 0 and samples less han 0 we e
ca ego ised as downwa ds in e gene a ional mobili y and
assigned he alue o −1.
Independen a iable
Educa ional le el. Following he ea men made by Zhu and Zhang
(2024), he esponden s’ highes educa ion in he o iginal ques ionnai e
could be di ided in o 11 le els; he samples ha did no ecei e edu-
ca ion we e al eady dele ed. ‘Elemen a y school’ was de ined as p ima y
educa ion and assigned he alue o 1, ‘junio high school, senio high
school, oca ional seconda y school, echnical school and oca ional
Table 1
Desc ip ion o a iables.
Va iable De ini ion Measu emen me hod
Independen
a iable
Edu Educa ional le el o
o sp ing
P ima y educa ion =1
Seconda y educa ion =2
Highe educa ion =3
Dependen
a iable
Sm Subjec i e
class mobili y
In e gene a ional upwa ds
mobili y =1
In e gene a ional
ansmission =0
In e gene a ional
downwa ds mobili y = − 1
Om Objec i e
class mobili y
In e gene a ional upwa ds
mobili y =1
In e gene a ional
ansmission =0
In e gene a ional
downwa ds mobili y = − 1
Con ol
a iable
Age A pe son’s age Yea o su ey minus yea
o bi h
Gende Dis inguishing
be ween he sexes
Male =1 Female =0
Pa y Poli ical p o ile CPC membe =1
Non-CPC membe =0
Fedu Le el o pa e nal
educa ion
P ima y educa ion =1
Seconda y educa ion =2
Highe educa ion =3
Fpa y Pa e nal poli ical
p o ile
CPC membe =1
Non-CPC membe =0
Media ing
a iable
Mig Popula ion mig a ion Mig a ed =1 No mig a ed
=0
Mode a ing
a iable
Bac Le el o household
economic capi al
Well below a e age =1
Below a e age =2
A e age =3
Abo e a e age =4
Well abo e a e age =5
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
4
senio high school’ we e de ined as seconda y educa ion and assigned
he alue o 2 and ‘highe oca ional college, bachelo ’s deg ee, mas e ’s
deg ee and abo e’ we e de ined as high educa ional le el and assigned
he alue o 3.
Media ing a iable
Popula ion mig a ion. The ques ions ‘Wha was you hukou egis-
a ion s a us a bi h?’ and ‘Wha is you cu en hukou egis a ion
s a us?’ we e used. The di e ence be ween hese wo is cha ac e ised by
assigning he alue o 1 i he indi idual’s hukou mig a es and
0 o he wise.
Mode a ing a iable
Le el o household economic capi al. The ques ion ‘Whe e does you
amily’s economic s a us all in he locali y?’ was used as he cha ac-
e isa ion a iable o household economic capi al, which was di ided
in o i e g ades, namely, ‘ a below a e age’, ‘below a e age’, ‘a e age’,
‘abo e a e age’ and ‘well abo e a e age’ and assigned he alues o 1 o
5, espec i ely.
Con ol a iable
The age, gende and poli ical p o ile o he o sp ing as well as he
educa ional le el and poli ical p o ile o he a he we e included in he
model as con ol a iables. Among hem, a he o sp ing le el, he age
a iable was he yea o su ey minus he yea o bi h. The gende
a iable was assigned he alue o 1 o he male and 0 o he emale.
The poli ical p o ile a iable was assigned he alue o 1 o CPC
membe s and 0 o non-CPC membe s. As o he aspec o he a he , he
assignmen was done in he same manne as ha o he o sp ing.
Desc ip i e analysis
The sample used in his s udy consis ed o 8014 indi iduals, o whom
44 % we e men and 56 % we e women. The p opo ions o young (aged
16–34 yea s), middle-aged (aged 35–59 yea s) and elde ly (aged 60
yea s and abo e) indi iduals we e 40 %, 58 % and 2 %, espec i ely. In
addi ion, he a io o u ban o u al pa icipan s was 37:63. The
desc ip i e esul s o he main a iables a e p esen ed in Table 2. The
di e ence be ween he mean alues o subjec i e and objec i e in e -
gene a ional social class mobili ies was ela i ely small, app oxima ely
40 %, he o al in e gene a ional social class mobili y a e was 64.97 %
and he in e gene a ional ansmission a e was 35.03 %, indica ing ha
he cu en mode o in e gene a ional social class mobili y in China
mainly consis s o upwa ds mobili y and he o e all deg ee o openness
is ela i ely high. The mean alue o educa ional le el is 2.353,
demons a ing ha mo e han 90 % o he o al samples ha e seconda y
and high educa ion, indica ing ha China’s cu en policy o 9-yea
compulso y educa ion and he expansion o highe educa ion ha e
yielded good esul s, which implies ha he educa ional le el o he
whole popula ion has been e ec i ely imp o ed. In e ms o popula ion
mig a ion, app oxima ely 25 % o he esponden s ha e mig a ed hei
hukou, whe eas he es ha e u al hukou and ha e no mig a ed,
sugges ing ha he cu en a e o popula ion mobili y and mig a ion in
China s ill needs o be imp o ed. In e ms o he le el o household
economic capi al, mo e han 80 % a e middle- o lowe -income house-
holds, indica ing ha Chinese esiden s’ pe cep ion o household in-
come le el has no imp o ed wi h he inc ease in pe capi a disposable
income o he na owing o he u ban– u al income gap. In e ms o he
in e ac ion e m be ween he educa ional le el and he le el o house-
hold economic capi al, he decen alised ea men is aken, wi h a
minimum alue o −1.702 and a maximum alue o 2.357.
Now we u n o analyse he subjec i e and objec i e in e gene a-
ional social class mobili y ke nel densi y, wi h he de ailed esul p e-
sen ed in Fig. 1. F om he objec i e pe cep ion, he ke nel densi y cu e
o he a he ’s social class is mo e dis ibu ed in he low le el and de-
c eases in a s ep-by-s ep manne , whe eas he o sp ing’s social class is
concen a ed in he middle le el. F om he subjec i e pe cep ion, he
ke nel densi y cu e o he o sp ing’s social class also ends o be
concen a ed in he middle le el, which indica es ha compa ed wi h
he a he ’s social class, he o sp ing’s social class ends o be mo e
in e media ed. This demons a es ha he p opo ion o he lowe -
middle class emains, whe eas ha o he middle class inc eases,
which is also in line wi h he cu en social class s uc u e o China’s ‘Tu’
shape. Compa ing he subjec i e and objec i e in e gene a ional social
class mobili ies in Fig. 2, i can be in e ed ha , on he one hand, he
end o upwa ds in e gene a ional social class mobili y is mo e ob ious
han ha o in e gene a ional ansmission o downwa ds mobili y, and
he subjec i e pe cep ion is la ge , which migh be ela ed o he ac
ha indi iduals belie e hey could b eak h ough he class ba ie s o
make he leap based on co esponding ac o s, such as he cu en social
ai ness and cul u al sys em. On he o he hand, he objec i e ac o
downwa ds in e gene a ional mobili y is lowe han he subjec i e
pe cep ion, indica ing ha al hough some esponden s ag ee wi h he
phenomenon o downwa ds social class mobili y compa ed wi h hei
a he , he e has no been a subs an ial decline in objec i e socio-
economic s a us, while opposi e conclusion is obse ed in he sample
o in e gene a ional ansmission. Fo example, 30 % o he esponden s
pe cei ed ha he e a e basically no gap in class be ween he wo
gene a ions, bu class solidi ica ion migh be ac ually qui e se ious.
O e all, he e is a signi ican de ia ion ega ding subjec i e and objec-
i e in e gene a ional social class mobili ies.
Empi ical analysis
To sys ema ically analyse he e ec o educa ional le el on in e -
gene a ional social class mobili y, i is p oposed o analyse he esul s o
he main e ec o educa ional le el on in e gene a ional social class
mobili y om he aspec s o baseline eg ession, he esul s o he
media ing e ec on objec i e dimension om he ansmission mecha-
nism o popula ion mig a ion, he esul s o he in e ac ion in subjec i e
dimension om he mode a ing e ec o he amily backg ound and he
esul s o he e ogenei y om he h ee dimensions o gende , hukou and
egion and o epo he esul s o he obus ness es .
Baseline eg ession analysis
The MLRM could be used o p edic he p obabili y ha an obse -
a ion belongs o each ca ego y gi en he alue o he independen
a iable. Mos ela ed s udies usually used he yea o educa ion o
educa ion pe cen ile o ank as a cha ac e ising a iable o educa ional
le el (Azomahou & Yi ba ek, 2021; Wu & Ma ois, 2024), while he
p esen s udy adop ed a di e en hie a chical ea men o yea o ed-
uca ion. Educa ion is ca ego ised in o p ima y, seconda y and highe
educa ion, wi h seconda y educa ion se as he benchma k o compa e
he impac o educa ional le el di e ences on in e gene a ional social
class mobili y.
We used he MLRM o examine he e ec o educa ional le el on
subjec i e and objec i e in e gene a ional social class mobili ies, wi h
Table 2
Desc ip i e esul s.
Numbe o
obse a ions
A e age
alue
S anda d
e o
Min Max
Om 8014 0.401 0.660 −1 1
Sm 8014 0.442 0.713 −1 1
Edu 8014 2.353 0.589 1 3
Mig 8014 0.242 0.429 0 1
Bac 8014 2.742 0.685 1 5
Edu*Bac 8014 0.076 0.408 −1.702 2.357
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
5
he econome ic esul s p esen ed in Table 3. As discussed in he p e-
ious sec ion, he es ima ion esul s o Eq. (2) we e based on he
benchma k o ‘in e gene a ional ansmission’ o highligh he di e -
ence be ween upwa ds and downwa ds mobili ies, and all o hese add
i e con ol a iables, including he o sp ing and a he dimensions. To
be consis en wi h he esul s o he obus ness es , econome ic esul s
o he con ol a iables a e epo ed bu no analysed o s a is ical
signi icance. In con as , in all o he aspec s o he econome ic esul s,
only whe he con ol a iables a e aken o no is conside ed, bu he
econome ic esul s a e no epo ed. In pa icula , columns (1) and (2)
Fig. 1. Subjec i e and objec i e in e gene a ional class mobili y ke nel densi y map.
Fig. 2. His og am o subjec i e and objec i e in e gene a ional class mobili ies.
Table 3
Resul s o educa ional le el on subjec i e and objec i e mobili ies.
(1) Objec i e downwa ds
in e gene a ional mobili y
(2) Objec i e upwa ds
in e gene a ional mobili y
(3) Subjec i e downwa ds
in e gene a ional mobili y
(4) Subjec i e upwa ds
in e gene a ional mobili y
Edu P −0.225 −0.272** −0.891*** −0.005
(0.259) (0.009) (0.000) (0.960)
H−0.249** 0.277*** −0.175** 0.109*
(0.011) (0.000) (0.046) (0.072)
Age 0.011** −0.010*** 0.003 0.008***
(0.009) (0.000) (0.336) (0.001)
Gende 0.019 0.154** −0.071 0.029
(0.819) (0.002) (0.345) (0.573)
Pa y −0.087 0.172** −0.038 0.368***
(0.461) (0.019) (0.743) (0.000)
Fedu 0.684*** −0.701*** 0.134** −0.283***
(0.000) (0.000) (0.042) (0.000)
Fpa y 0.510*** −0.676*** 0.400*** −0.222**
(0.000) (0.000) (0.000) (0.002)
_cons −3.108 1.590 −1.118 0.711
N 8014 8014
No e: (1) P- alues in pa en heses; (2) *, ** and *** indica e passing 10 %, 5 % and 1 % signi icance es s, espec i ely; (3) he ollowing akes he same ea men .
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
6
epo he e ec o educa ional le el on objec i e in e gene a ional so-
cial class mobili y, whe eas columns (3) and (4) epo he e ec on
subjec i e in e gene a ional social class mobili y.
As can be seen om Table 3, indi idual educa ional le el is an
impo an a iable ha a ec s in e gene a ional social class mobili y.
Speci ically, columns (1) and (2) show ha p ima y educa ion dec eases
he p obabili y o a ec ing objec i e upwa ds in e gene a ional social
class mobili y by 1.31 (he e, i is e
0.272
) and highe educa ion inc eases
by 1.32 (he e, i is e
0.277
) compa ed wi h in e gene a ional ansmission.
This indica es ha indi iduals wi h highe educa ion ha e a highe
p obabili y o upwa ds class mobili y and a lowe p obabili y o
downwa ds class mobili y o class en enchmen . Combined wi h he
ma ginal e ec shown in Table 4, i o he a iables a e cons an ,
compa ed o he benchma k o seconda y educa ion, p ima y educa ion
leads o an inc ease in he p obabili y o objec i e in e gene a ional
ansmission by 6.3 % and a dec ease in he p obabili y o upwa ds
mobili y by 5.3 %, bu highe educa ion leads o a dec ease in he
p obabili y o objec i e downwa ds in e gene a ional mobili y by 3.1 %,
a dec ease in he p obabili y o in e gene a ional ansmission by 4.2 %
and an inc ease in he p obabili y o upwa ds in e gene a ional mobili y
by 7.3 %, which implies highe educa ion is associa ed wi h highe
mobili y (Ojaleh o e al., 2023).
Fu he mo e, compa ing columns (3) and (4) wi h columns (1) and
(2), i can be obse ed ha he es ima ed coe icien s a e in he same
di ec ion be ween he subjec i e and he objec i e pe cep ions. The
abo e esul s no only imply ha he e ec o educa ional le el on
subjec i e social class mobili y is u he co obo a ed by he objec i e
pe cep ion bu also ha indi iduals wi h high educa ional le el a e mo e
subjec i ely inclined o expe ience upwa ds in e gene a ional social
class mobili y compa ed wi h in e gene a ional ansmission. Fo
example, in Table 4, high educa ional le el leads o a 2.8 % dec ease in
he p obabili y o subjec i e downwa ds in e gene a ional mobili y, a
1.1 % dec ease in he p obabili y o in e gene a ional ansmission and a
3.9 % inc ease in he p obabili y o upwa ds in e gene a ional mobili y.
The compa ison shows ha educa ional le el has a g ea e impac on he
objec i e han on he subjec i e dimension, which migh be ela ed o
he ac ha he ac o s a ec ing he pe cep ion o in e gene a ional
social class mobili y a e mo e complex and ha i is di icul o p ecisely
highligh he ole o educa ion (F iedman e al., 2021). The combined
econome ic esul s indica e ha high educa ional le el has he ad an-
age o enhancing upwa ds in e gene a ional social class mobili y,
which indica es ha indi iduals a e empowe ed by access o quali y
educa ion esou ces, he eby inc easing hei chances o being in he
highe class. Hence, H1 is pa ially e i ied.
To explo e he di e en ial e ec s o educa ional le el on upwa ds
in e gene a ional social class mobili y in di e en his o ical pe iods in
China, based on he s udy by Guo e al. (2019), he enac men o he
compulso y educa ion law in 1986 and he expansion o highe educa-
ion en olmen in 1999 we e aken as he landma k e en s o ouching
an impo an e ec on educa ional le el. A he same ime, ‘Wha yea
did you ob ain you highes deg ee?’ is di ided in o h ee s ages: be o e
1986 o S age I, 1986–1999 o S age II and a e 1999 o S age III, wi h
assigned alues o 1, 2 and 3, espec i ely. Fu he mo e, he educa ional
le el o he same coho is con olled o enhance compa abili y. The
econome ic esul s indica e ha downwa ds in e gene a ional mobili y
is no signi ican enough o be s a is ically analysed. Thus, he esul s o
he subjec i e and objec i e upwa ds in e gene a ional social class
mobili ies a e e ained, wi h he de ailed esul p esen ed in Table 5.
A e emo ing he in la iona y e ec o educa ion, implemen a ions
o bo h policies signi ican ly in luence he e ec o high educa ional
le el on upwa ds class mobili y a he objec i e le el, and he es ima ed
coe icien s a e conside ably la ge . In S age I, which is be o e he
enac men and implemen a ion o he compulso y educa ion law,
educa ional le el does no subs an ially a ec he subjec i e and
objec i e upwa ds in e gene a ional social class mobili ies, p obably
because ecei ing educa ion equi es a la ge amoun o inancial e-
sou ces, which hinde s o e en dep i es he o sp ing’s igh o ecei e
an equal educa ion (Liu & Lei, 2023) o a ce ain ex en . In S age II,
which is he pe iod om he enac men o he compulso y educa ion law
o ha o he expansion o highe educa ion en olmen , p ima y edu-
ca ion does no ha e a conside able e ec . On he one hand, i migh be
because he policy o comple ing he 9-yea compulso y educa ion o all
school-age child en has no been ully implemen ed in China, pa icu-
la ly in he u al a eas. On he o he hand, he o sp ing who ecei e
highe educa ion a e mo e likely o mo e o highe s a a o he hie -
a chy and subjec i ely ag ee wi h he exis ence o a b eak h ough be-
ween gene a ions; hus, highe educa ion is mo e e ec i e in p omo ing
upwa ds social class mobili y in he subjec i e and objec i e dimensions
han seconda y educa ion (Li e al., 2023). In S age III, which is a e he
expansion o highe educa ion en olmen , he ad an ageous impac o
highe educa ion on upwa ds class mobili y exhibi ed a downwa ds
end, sugges ing ha he expansion o highe educa ion en olmen
enhances he p obabili y o upwa ds in e gene a ional social class
mobili y by inc easing he oppo uni ies o college en olmen . Howe e ,
i leads o he p oblem o misma ch o supply and demand in he labou
ma ke , which esul s in lowe high educa ion e u ns, makes i mo e
di icul o ob ain sui able occupa ions and ewa ds han hose wi h he
same deg ee be o e he expansion and induces i s unc ion o ela i ely
decline due o he high le el o educa ion (Ou & Hou, 2019); hus, he
es ima ed coe icien o upwa ds in e gene a ional mobili y in S age III
is lowe han ha in S age II. In e ms o subjec i e pe cep ion, he
expansion o highe educa ion en olmen enabled he o sp ing o ob ain
mo e employmen oppo uni ies by imp o ing educa ional le el, which
indica es he inc easing di icul y o he disad an aged class o b eak
h ough he social class ep oduc ion by upg ading hei academic
quali ica ions. Mo eo e , he obs acles o hei o sp ing in ealising
class-c ossing a e also inc easing, he eby p esen ing a sha p decline o
e en becoming nonsigni ican e ec s. The e o e, H2 is e i ied.
Analysis o he media ing e ec
The media ion e ec ollows he KHB media ion e ec es p oposed
by Kohle e al. (2011). The KHB es has he ad an age o sa is ying he
‘con inuous omission assump ion’ and has been widely employed o
assess he media ing a iables in he MLRM. The econome ic esul
indica ed ha popula ion mig a ion did no pass he signi icance es as
pe he e ec o educa ional le el on objec i e downwa ds in e gene -
a ional social class mobili y; hus, he esul was no epo ed. The
econome ic esul s a e p esen ed in Table 6.
A e decomposi ion using he KHB analysis me hod, he esul
indica ed ha educa ion is an impo an ac o in luencing popula ion
mig a ion, which is consis en wi h he indings o Zou and Deng (2022).
Table 4
Econome ic esul s o he ma ginal e ec .
Objec i e Subjec i e
Downwa ds ansmission Upwa ds Downwa ds ansmission Upwa ds
P−0.010 0.063** −0.053** −0.080*** 0.031 0.050**
(0.549) (0.010) (0.018) (0.000) (0.190) (0.039)
H−0.031*** −0.042*** 0.073*** −0.028*** −0.011** 0.039**
(0.000) (0.001) (0.000) (0.002) (0.360) (0.003)
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
7
Speci ically, he e ec s o high educa ion on objec i e downwa ds
in e gene a ional mobili y a e all posi i e, whe eas he e ec s o p i-
ma y educa ion a e all nega i e. I can be conside ed ha popula ion
mig a ion is an impo an ansmission mechanism in he e ec o
educa ional le el on objec i e in e gene a ional social class mobili y,
and high educa ion can e ec i ely p omo e popula ion mig a ion, hence
ealise upwa ds in e gene a ional social class mobili y. On he one hand,
ecei ing highe educa ion can imp o e skills, knowledge and income,
he eby p omo ing indi iduals o mig a e o economically de eloped
a eas wi h mo e employmen oppo uni ies, whe eas ecei ing p ima y
educa ion makes i di icul o mee he di e si ied demand o high-skill
jobs; hus, i exe s a signi ican nega i e e ec on popula ion mig a ion.
On he o he hand, high educa ional le el can e ec i ely link he social
ne wo k o help indi iduals ob ain la ge in o ma ion esou ces and
mo e ex ensi e job oppo uni ies in he p ocess o choosing o mig a e,
he eby p omo ing upwa ds mobili y o in e gene a ional s a a.
Meanwhile, o g oups wi h a p ima y educa ion le el, he mig a ion
may des oy hei o iginal social links and suppo sys ems and e en
solidi y he exis ing en i onmen ; hence, p ima y educa ional le el does
no exe a posi i e e ec on upwa ds in e gene a ional mobili y.
The e o e, H3 is e i ied.
Analysis o he mode a ing e ec
The econome ic esul o he mode a ing e ec s is p esen ed in
Table 7, whe e columns (1) and (2) a e he o iginal o al e ec s wi hou
in e ac ion e m and columns (3) and (4) a e he mode a ing e ec s wi h
in e ac ion e m. A e adding he in e ac ion e m, in subjec i e up-
wa ds in e gene a ional social class mobili y, he eg ession coe icien
o he in e ac ion e m is −0.167, and he eg ession coe icien o
highe educa ion dec eases o 0.051, which shows ha amily back-
g ound nega i ely alle ia es he e ec o educa ional le el on subjec i e
upwa ds in e gene a ional social class mobili y. Tha is, he s onge he
amily backg ound, he weake he e ec o educa ional le el on up-
wa ds in e gene a ional social class mobili y, and his is pa icula ly
signi ican in high educa ional le el. This could be a ibu ed o he
complexi y and a iabili y o subjec i e iden i y. Excessi e amily
in ol emen in he educa ion o o sp ing is also de imen al o he
ad an ageous ole o educa ion in p omo ing upwa ds class mobili y.
F om ano he pe spec i e, s ong economic suppo in he amily could
educe he p essu e on educa ion, so hey a e no wo ied much abou
he a e o e u n o educa ion, he eby educing he u gency o
educa ional le el in imp o ing social s a us. The e o e, H4 is e i ied.
Table 5
Econome ic esul s o he elimina ion o educa ion in la ion.
S age I S age II S age III
(1) Objec i e upwa ds
mobili y
(2) Subjec i e upwa ds
mobili y
(3) Objec i e upwa ds
mobili y
(4) Subjec i e upwa ds
mobili y
(5) Objec i e upwa ds
mobili y
(6) Subjec i e upwa ds
mobili y
P ima y −0.261* −0.040 −0.197 0.095 −0.800** −0.337
(0.095) (0.804) (0.274) (0.630) (0.014) (0.311)
High 0.480 0.437 0.434** 0.335** 0.328*** 0.130
(0.115) (0.163) (0.002) (0.002) (0.000) (0.123)
Age Yes Yes Yes Yes Yes Yes
Gende Yes Yes Yes Yes Yes Yes
Pa y Yes Yes Yes Yes Yes Yes
Fedu Yes Yes Yes Yes Yes Yes
Fpa y Yes Yes Yes Yes Yes Yes
_cons 1.272 0.866 1.030 0.471 0.623 −0.118
N 1444 2312 4258
Table 6
Resul s o KHB decomposi ion.
In e media y a iable Implici a iable Agg ega e e ec Di ec e ec Indi ec e ec P opo ion
P ima y Mig Objec i e upwa ds in e gene a ional mobili y −0.272** (0.009) −0.242** (0.019) −0.030** (0.011) 10.96 %
High 0.277*** (0.000) 0.247*** (0.000) 0.030** (0.010) 10.89 %
Table 7
Econome ic esul s o he mode a ing e ec .
(1) Subjec i e downwa ds
in e gene a ional mobili y
(2) Subjec i e upwa ds
in e gene a ional mobili y
(3) Subjec i e downwa ds
in e gene a ional mobili y
(4) Subjec i e upwa ds
in e gene a ional mobili y
Edu P ima y −0.891*** −0.005 −0.938*** −0.131
(0.000) (0.960) (0.000) (0.253)
Highe −0.175** 0.109* −0.148* 0.051
(0.046) (0.072) (0.097) (0.405)
Bac −0.140** 0.401***
(0.011) (0.000)
Edu*Bac 0.019 −0.167**
(0.845) (0.010)
Age Yes Yes Yes Yes
Gende Yes Yes Yes Yes
Pa y Yes Yes Yes Yes
Fedu Yes Yes Yes Yes
Fpa y Yes Yes Yes Yes
_con −3.108 1.590 −0.772 −0.316
N 8014 8014
B. Li e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100678
8
