Malesky, Edmund J.; Nguyen, T ung-Anh
A icle
His o ical Communis Pa y S eng h and Mode n Pa y
Loyal y. A Replica ion S udy o Ba celó (PNAS, 2021)
Jou nal o Commen s and Replica ions in Economics (JCRE)
Sugges ed Ci a ion: Malesky, Edmund J.; Nguyen, T ung-Anh (2024) : His o ical Communis Pa y
S eng h and Mode n Pa y Loyal y. A Replica ion S udy o Ba celó (PNAS, 2021), Jou nal o
Commen s and Replica ions in Economics (JCRE), ISSN 2749-988X, ZBW - Leibniz In o ma ion Cen e
o Economics, Kiel, Hambu g, Vol. 3, pp. 1-27,
h ps://doi.o g/10.18718/81781.35
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/299994
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/4.0/
Jou nal o Commen s and Replica ions in Economics - JCRE
His o ical Communis Pa y S eng h and Mode n
Pa y Loyal y
A Replica ion S udy o Ba celó (PNAS, 2021)
Edmund J. Malesky*T ung-Anh Nguyen†
Jou nal o Commen s and Replica ions in Economics, Volume 3, 2024-5, DOI: 10.18718/81781.35
JEL: P0, P2, P3, P5, N0
Keywo ds: Con lic , Ci ic engagemen , Vie nam Wa , His o ical legacies
Da a A ailabili y: The R code and da a o ep oduce he esul s o his eplica ion can be downloaded
a JCRE’s da a a chi e (DOI: 10.15456/j1.2023285.1524615125).
Please Ci e As: Malesky, E. J. and Nguyen T. (2024). His o ical Communis Pa y S eng h and
Mode n Pa y Loyal y: A Replica ion S udy o Ba celó (2021). Jou nal o Commen s and Replica ions
in Economics, Vol.3 (2024-5). DOI: 10.18718/81781.35
Abs ac
Does exposu e o iolence c ea e mo e poli ically engaged ci izens? In a PNAS pape , Ba celó (2021)
asks his p o oca i e ques ion and p oposes an in ica e and o iginal socio-psychological heo y
o answe i . Ba celó also employs a esea ch design ha seeks o accoun o e e se causali y o
add ess he ac ha mo e engaged places may ha e been a ge ed. Un o una ely, e en a e a
co ec ion, his s a is ical analysis emains dogged by coding, his o ical, and con ex ual e o s, which
bias esul s and lead o misleading conclusions. When we accoun o hese mis akes wi h simple and
easonable co ec ions, he main esul s no longe hold. Ra he , we show ha he ue unde lying
explana ion is mo e mundane: a eas ha we e sou ces o p e-wa communis insu gency s eng h
we e a ge ed o hei ac i i y du ing he con lic , and hey emain mo e loyal o he egime oday.
*Co esponding au ho : Depa men o Poli ical Science, Duke Uni e si y, NC, USA [email p o ec ed]
†Depa men o Poli ical Science, Duke Uni e si y, NC, USA
Decla a ion: The au ho s decla e no compe ing in e es . Resea ch epo ed in he pape was no he esul s o a o -pay consul ing
ela ionship. Nei he he au ho s no hei employe s ha e a inancial in e es in he opic o he pape .
Recei ed Oc obe 12, 2023; Re ised Ma ch 26, 2024; Accep ed May 09, 2024; Published June 18, 2024.
©Au ho (s) 2024. Licensed unde he C ea i e Common License - A ibu ion 4.0 In e na ional (CC BY 4.0).
1
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
1 In oduc ion
In “The Long-Te m E ec s o Wa Exposu e on Ci ic Engagemen ,” Joan Ba celó a gues ha ci -
izens in Vie nam who expe ienced g ea e bombing du ing he wa wi h he Uni ed S a es (US)
(1966-1974) had highe le els o poli ical pa icipa ion, ins i u ional us , and ci ic engagemen in
2001 han ci izens who we e less impac ed by he con lic (Ba celó, 2021). In his esea ch no e,
we explain why Ba celó’s esul s a e an a i ac o coding, his o ical, and con ex ual mis akes in his
esea ch design. E en a e accoun ing o his ecen co ec ion and a subsequen upda e o he
co ec ion (Ba celó, 2023), he mos likely explana ion o he obse ed co ela ion is ha he US
bombing a ge ed s ongholds o suppo o he communis egime, whe e popula ions emained
loyal o he egime long a e he wa .
In he pape , Ba celó uses a linea es ima ion o demons a e a co ela ion be ween ci ic engage-
men oday and bombing du ing he wa . Howe e , his co ela ion is h ea ened by endogenei y,
he US may ha e s a egically bombed places ha al eady had high le els o p e-wa poli ical
pa icipa ion. To add ess his possibili y, Ba celó exploi s a na u al expe imen in con lic exposu e
caused by dis ance om he 17 h pa allel, sugges ing ha he con lic was he hea ies a ound
he “a bi a y” bo de se a he Gene a Con e ence in 1954.
1
Consequen ly, Ba celó claims i
is plausible ha some ci izens close o he 17 h pa allel we e acciden ally exposed o iolence,
allowing him o ace h ough whe he his con lic exposu e gene a ed he g ea e poli ical pa -
icipa ion o his heo y. Ba celó does no subs an ia e his assump ion, bu elies upon a p e ious
asse ion and measu emen by Miguel and Roland (2011). Based on his logic, Ba celó uses he di e -
ence in la i ude be ween p e-wa p o inces and he 17 h pa allel in a wo-s age leas squa es analysis.
In his no e, we i s demons a e ha Ba celó makes se e al s a is ical e o s ha ende his
esea ch design inapp op ia e and his conclusions p oblema ic. We do so by ep oducing he o iginal
s udy’s esul s and p o iding ou co esponding es ima es a e co ec ing o such e o s. Second, we
show ha e en i hose un o una e e o s we e less consequen ial, he esea ch design is in alida ed
by he ac ha he 17 h pa allel an h ough he s onghold o he Vie namese communis insu gency,
whe e he mos poli ically engaged ci izens we e al eady loca ed. Thi d, we show ha Ba celó’s
su ey-based ou come a iable is no measu ing independen engagemen as he claims, bu is in ac
e lec ing cu en communis pa y s eng h h ough pa icipa ion in en i ely pa y-domina ed mass
o ganiza ions. When we accoun o hese issues in he s a is ical analysis, Ba celó’s p o oca i e
conclusions no longe hold.
Finally, we demons a e h ough a s a is ical analysis o ou own ha he cen al egions o
Vie nam had highe le els o communis ac i i y be o e he wa and con inued o ha e highe le els
o communis ac i i y a e he wa . The conclusion is clea : Ame ican bombs did no in ensi y ci ic
engagemen ; hey me ely a ge ed p e-exis ing bas ions o communis ac i i y and loyal y.
1
We disag ee wi h e en his ini ial p emise. While i is ue ha he p oposals o dema ca ion a ied om he 13 h
pa allel o he 18 h pa allel a he Gene a Con e ence in 1954, hese p oposals did no e lec nego ia o s h owing da s a a
map. Ra he , he his o ical eco d is clea ha he speci ic sugges ions e lec ed ac o s s aking ou maximalis nego ia ion
posi ions (Asselin, 2011), esponding o changes in mili a y con ol on he g ound (such as he No h’s su p ising ic o y a
Dien Bien Phu), and accommoda ing he in e es s o in e na ional powe s such as he US, USSR, and China (Wai e, 2012).
The selec ion o he 17 h pa allel ensu ed ha bo h mode n comba an s kep hei his o ical powe bases in ac . The F ench
e ained Hue and Da Nang, while he No h Vie namese e ained he impo an Zone IV mili a y a ea (Asselin, 2011, 177),
which we desc ibe in mo e de ail in Sec ion 2.
2
Jou nal o Commen s and Replica ions in Economics - JCRE
2 Replica ion and Co ec ion o E o s
2.1 E o 1: Unclus e ed S anda d E o s and Collinea Co a ia es
In a le e o he PNAS edi o s om Ma ch 2023 ega ding Ba celó’s o iginal a icle (Ba celó, 2021),
we poin ed ou ha Ba celó’s eplica ion iles demons a ed h ee laws in s a is ical analysis among
o he conce ns abou he pape ’s unde s anding o Vie nam’s his o y and geog aphy. These h ee
laws included sys ema ic measu emen e o in his ins umen al a iable (dis ance om he 17 h
pa allel), he ailu e o include clus e ed s anda d e o s as claimed in he ex , and he inclusion o
a con ol a iable (la i ude) ha was a nea ly pe ec ly co ela ed ans o ma ion o his ins umen al
a iable, measu ed as he absolu e alue o he dis ance in deg ees o la i ude om he 17 h pa allel.
We showed ha hese mis akes we e leading o biased esul s and calcula ion o s anda d e o s,
which was d i ing he inaccu a e conclusion ha exposu e o bombing led o g ea e ci ic engage-
men in Vie nam.
Ou esponse was no published, because i was submi ed o PNAS a e he six-mon h deadline
o esponses o published pieces; howe e , Ba celó subsequen ly in oduced a co ec ion (Ba celó,
2023) on July 7 h and an upda e o he co ec ion on Feb ua y 29, 2024
2
ha pu po edly ad-
d essed he conce ns we had aised wi h him and he jou nal’s edi o s. We ha e now e iewed hese
co ec ions along wi h Ba celó’s amended eplica ion iles. While his co ec ion did ix he e o in
he measu emen o dis ance, i , un o una ely, in oduced a new coding e o ha again esul s in
he ailu e o clus e s anda d e o s, does no co ec an inaccu a e coding o whe he a p o ince is
loca ed in Sou h Vie nam, and inco ec ly asse s ha he inclusion o a nea ly pe ec ly collinea
con ol a iable is jus i ied on heo e ical and empi ical g ounds. These undamen al s a is ical
mis akes con inue o lead o inaccu a e esul s.
In his sec ion, we highligh hese s a is ical e o s and hei consequences by ep oducing he
o iginal s udy’s esul s and p o iding ou co esponding es ima es a e co ec ing o hese e o s.
2.1.1 Failu e o include clus e ed s anda d e o s
Equa ion 1 ep oduces he wo-s age ins umen al a iable app oach ha Ba celó uses
3
o es his
heo y ha bombing exposu e leads o g ea e ci ic engagemen oday. In he i s s age, Ba celó
eg esses he ex en o bombing, measu ed as he na u al log o bombs d opped on he p o ince
pe capi a, on dis ance om he 17 h pa allel, measu ed as he absolu e alue o he dis ance om
he la i ude a he cen e o he p o ince o he 17 h pa allel. In he second s age, he eg esses
his measu e o ci ic engagemen on he p edic ed numbe o bombs om he i s s age. Table 1
om Ba celó (2023) p esen s h ee ins umen al a iable models: Model 3 p esen s he bi a ia e
co ela ions, Model 4 adds p o incial-le el con ol o popula ion densi y, and Model 5 in oduces a
ull sui e o indi idual (𝑋𝑖) and p o incial-le el con ols (𝑋𝑝), including la i ude.
2see h ps://www.pnas.o g/doi/10.1073/pnas.2308785120
3
Fo a discussion o he exclusion and s eng h assump ions o he ins umen al a iables app oach and he po en ial
pi alls o causal iden i ica ion in his con ex , see Appendix A.
3
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
No ice ha in he i s s age, bo h bombing exposu e and dis ance om he 17 h pa allel a e
measu ed a he p o incial le el (
𝑝=49
) and do no a y among su eyed ci izens who eside in he
same p o ince. Essen ially, Ba celó is using he same dis ance and bombing measu es epea edly o
di e en ci izens wi hin he same loca ion.
Fi s s age: Bombs𝑝=𝛼0+𝛼1Dis 17𝑝+𝑋𝑖+𝑋𝑝+𝑒𝑖 𝑝,
Second s age: Ci ic𝑖 𝑝 =𝛽0+𝛽1ˆ
Bombs𝑝+𝑋𝑖+𝑋𝑝+𝑢𝑖 𝑝,(1)
This exagge a es he s a is ical signi icance o he unde lying co ela ion by a i icially educing he
size o he s anda d e o s (Came on, Gelbach, and Mille , 2008). To see why his happens, le ’s
s ep away om he speci ic analysis and ocus on a simple hough expe imen . Imagine we wan ed
o calcula e he ela ionship be ween he median income o US s a es and he pe cen age o each
s a e ha u ned ou o o e in he 2020 P esiden ial Elec ion using eg ession analysis. Ob iously,
he sample size (
𝑛
) o his ela ionship, measu ed by
ˆ
𝛽1
, is 50: we ha e one obse a ion o median
income and one obse a ion o agg ega e u nou o each s a e. In u n, he s anda d e o (
𝜎
) o
he co ela ion should be calcula ed using 𝑛=50:
ˆ
𝜎=√︃ˆ
𝛽1/50 −2
The au ho ’s mis ake is akin o eg essing agg ega e s a e u nou o e e y o ing-eligible ci izen in
e e y s a e on he median income o ha s a e. This is equi alen o calcula ing he s anda d e o
o he co ela ion using a sample size equal o he o al o ing-eligible popula ion in he Uni ed
S a es (app oxima ely 255 million). Wi h he new la ge numbe , he denomina o o he s anda d
e o is now much la ge . This will a i icially educe he s anda d e o by o de s o magni ude:
ˆ
𝜎=√︃ˆ
𝛽1/255,457,000 −2
By no clus e ing in his PNAS analysis, Ba celó has essen ially aken 49 obse a ions o he ou come
and independen a iables bu applied hem o 875 indi idual-le el su ey obse a ions. Econome-
icians ha e long unde s ood ha such clus e ed da a is common in s a is ical analysis (i.e. child en
nes ed inside ea ed schools) and ha e wo ked o co ec he dange s o his alse p ecision and
de la ed s anda d e o s. The solu ion is clus e ing he s anda d e o s o add ess he co ela ion
(in Ba celó’s case, pe ec co ela ion) o obse a ions wi hin he highe -le el uni (Abadie e al.,
2023). Clus e ing co ec s o he alse p ecision by calcula ing s anda d e o s wi hin each g ouping,
inc easing he size o he s anda d e o s and leading o mo e accu a e es s o s a is ical signi i-
cance. To be su e, he e a e se ious s a is ical deba es abou clus e ing and ques ions abou he
accu acy o he app oach when he numbe o highe -le el uni s o he numbe o obse a ions
wi hin highe -le el uni s is small (Came on, Gelbach, and Mille , 2008). Howe e , none o hese
apply in Ba celó’s si ua ion, which is he canonical case o when clus e ing is necessa y, because he e
is ze o wi hin-p o ince a ia ion in bo h he dependen and independen a iables o his i s -s age
eg ession model.
Ba celó demons a es awa eness o he issue. In bo h his o iginal a icle and July 2023 co ec ion,
he w i es ha he ables epo , “P o ince-le el he e oskedas ic clus e ed-SEs in pa en hesis.” In he
o iginal pape , howe e , he mis akenly epo ed only obus s anda d e o s by ailing o de ine he
clus e ing a iable in his R-sc ip . He co ec ed he o iginal mis ake bu in oduced a new one, which
4
Jou nal o Commen s and Replica ions in Economics - JCRE
un o una ely leads o he exac same ou come o a i icially small s anda d e o s.
4
His calcula ion
is shown in he sc eensho below.
Figu e 1: Sc eensho om Ba celó’s R-sc ip Showing S anda d E o s We e No
Clus e ed (line 192-7 in p o ided eplica ion ile, las upda ed May 13, 2023)
When Ba celó applied co ec ions o he s anda d e o s, he used he command co HC, which
is designed o calcula e obus s anda d e o s and does no allow o a clus e a gumen . As a esul ,
he clus e a gumen shown in he sc ip (“clus e =
∼
p o ince_wa ime”) is igno ed and e e y
model in his analysis de aul s o obus s anda d e o s. To clus e , he needed o use he command
co CL, which akes a clus e a gumen . The e is no doub ha Ba celó’s code is no co ec ing o
clus e ing, because he s anda d e o s he p oduces wi h his calcula ion in e e y model a e smalle
han no clus e ing a all.
To illus a e his poin in Table 1 below, we ep oduce Ba celó’s es ima es om he i s able in
his co ec ions wi h esul s using ou new co ec ed code ha accu a ely calcula es he s anda d
e o s by en e ing co CL in o Ba celó’s o iginal command. We use M o deno e he models used
in Ba celo’s 2023 co ec ion able (i.e. M3=Model 3), while e e ing o ou own es ima es using
column numbe s a he op o he able. Beginning wi h he i s s age es ima es in column 1 a
he bo om o he able, Ba celó’s M1 displays a coe icien o -.04 (a one-uni change in dis ance
is associa ed wi h a 4% educ ion in bombing) and a s anda d e o o .0232 which he epo s is
signi ican a he 0.1 le el. The F- es o his i s s age is 3.66, which is smalle han he cu en
s anda d o 10 and ecommended sco e o 104.7 o ep esen ins umen s eng h (Lee e al., 2022).
In he second s age, his leads o a coe icien o bombing on he ci ic engagemen o 1.19 wi h a
s anda d e o o .65, which Ba celó also epo s as signi ican a he .1 le el.
4
In he Feb ua y 2024 upda e o he July 2023 co ec ion, he able no es on s anda d e o s we e edi ed o:
"He e oskedas ici y-consis en s anda d e o s in pa en hesis. Main models a e obus o using p o ince-le el he e oskedas ic
clus e ed s anda d e o s." This indica es ha Ba celó is awa e ha he applied co ec ions o he s anda d e o s only
accoun ed o obus ness o he e oskedas ici y and no clus e ing a he p o ince le el. Addi ionally, we show in ou Table 1
ha wo ou o he h ee main models we e, in ac , no obus o using p o ince-le el he e oskedas ic clus e ed s anda d
e o s as he able no es claim.
5
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
Table 1: O iginal Model Insigni ican Rega dless o App oach o Clus e ing
Va iables Ci ic Engagemen (log) in 2001
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Wi hou geog aphic con ols
O iginal/Co ec ed
Model in Ba celó (2023) Table 1
O iginal
M3
O iginal
M4
O iginal
M5
Co ec ed
M3
Co ec ed
M4
Co ec ed
M5
O iginal
M5
Co ec ed
M5
Co ec ed
M5
Bombs, pe km2 (log) 1.19* 0.44** 0.23** 1.19 0.44 0.23** -0.72** -0.72 -0.58
( esidence p e-1975) (0.65) (0.12) (0.03) (3.65) (0.46) (0.06) (0.31) (1.33) (0.97)
Female=1 -0.07** -0.07** -0.02 -0.02 -0.02
(0.04) (0.03) (0.07) (0.10) (0.09)
Age -0.002 -0.002 0.0003 0.0003 0.001
(0.001) (0.002) (0.002) (0.003) (0.003)
Educa ion 0.02* 0.02* 0.07** 0.07 0.06
(0.01) (0.01) (0.03) (0.10) (0.07)
Popula ion densi y -0.22** -0.11** -0.22 -0.10* 0.44** 0.44 0.36
(1960 o 61) (’000) (0.11) (0.03) (0.23) (0.06) (0.20) (0.86) (0.63)
A e age p ecipi a ion -0.22** -0.14 1.54** 1.54 1.30
(’00) (2.10) (0.21) (0.65) (2.79) (2.10)
La i ude (’0) -0.02 -0.17
(0.17) (0.30)
Sou h -0.29* -0.45*
(0.16) (0.26)
Cons an -3.23* -1.15** -0.15 -3.23 -1.15 0.04 -1.29** -1.29 -2.65
(1.66) (0.28) (0.31) (9.45) (1.05) (0.48) (0.50) (2.04) (3.64)
Clus e ed SEs (“HC0”) N N N Y Y Y N Y Y
E hnici y FEs N N N N N N N N Y
Fi s S age
Dis ance o 17 h Pa allel -0.04* -0.11** -0.57** -0.04 -0.11 -0.59** 0.07** 0.07 0.08
(0.02) (0.023) (0.027) (0.12) (0.12) (0.1) (0.02) (0.10) (0.10)
F-S a is ic 2.84 24.6 683.5 0.1 0.84 38.9 9.37 0.43 0.59
Obse a ions 875 875 862 875 875 862 862 862 856
Clus e s 49 49 48 49 49 48 48 48 48
No es: S anda d e o s in pa en heses (** p<0.05, * p <0.1). Using he code p o ided in he eplica ion package, we
we e unable o eplica e Ba celó (2023, 2)’s F-s a is ics in Columns 1-3 o his h ee IV models, which we e epo ed o
be 3.66, 23.2, and 464.4, espec i ely. M3, M4, and M5 deno e he model numbe s in Table 1 o he 2023 co ec ion.
Column numbe s ep esen he o de o ou es ima ions.
6
Jou nal o Commen s and Replica ions in Economics - JCRE
The second s age esul implies a 10% inc ease in bombing is associa ed wi h a 12% inc ease
in ci ic engagemen . Howe e , cau ion should be aken in in e p e ing economic e ec om hese
coe icien s as he es ima es a e widely a ying. Reading ac oss his o iginal i e es ima es in Table 1
o he 2023 co ec ion Ba celó (2023), we ind ha a 10% inc ease in bombing had an 0.8% (M1),
1.2% (M2), 12% (M3), 4.4% (M4), o 2.3% (M5) impac on ci ic engagemen . When we emo e
he con ols o Sou h and La i ude, he e ec is a 7.2% educ ion in ci ic engagemen .
Column 4 ep oduces he column 1 (M3) es ima es wi h he co ec s anda d e o s. We eco e
he exac same coe icien size, howe e , he s anda d e o s a e e y di e en . The i s -s age
and second-s age s anda d e o s a e .123 and 3.65 espec i ely, o e i e imes he size o he
o iginal Ba celó es ima es. The new es ima es imply i s -s age and second-s age p- alues o 0.75
and 0.74, which a e ex emely a om s a is ical signi icance. The same pa e n is ue o he
popula ion-densi y adjus ed es ima es ha Ba celó p esen s in column 2 (M4) and we e-analyze in
column 5.
Co espondingly, Ba celó’s e o impac s his de e mina ion o ins umen s eng h, whe e an
F- es o a leas 10 is needed o demons a e an ins umen is a s ong enough p edic o o he
ea men a iable ha i will no gene a e bias in he second s age. New wo k, howe e , sugges s a
F- es h eshold o 104.7. (Lee e al., 2022). When app op ia e clus e ing is applied, Ba celó’s F- es
in he i s s age d ops om 2.84 o 0.1 in his M3, om 24.6 o 0.84 in M4, and om 683.5 o 38.9
in M5. The bo om line is ha when co ec p ocedu es a e applied, nei he he i s no second
s ages o Ba celó’s es ima es a e s a is ically signi ican . The e is no obse able co ela ion be ween
dis ance om he 17 h pa allel and bombs d opped, and no s a is ical ela ionship be ween exposu e
o bombs and ci ic engagemen , in alida ing his i s s age and he e o e his esea ch design.
Analys s disag ee on he ma hema ical nuances o co ec ing o clus e ing. To a oid conce ns
ha ou conclusions may be a ec ed by eliance on a pa icula ype o clus e ing es ima ion, in
Table 2 we e-es ima e Ba celó’s s anda d e o s using ou di e en es ima ion ypes (HC0, HC1,
HC2, and HC3) o clus e ing included in R’s sandwich package (Zeileis, Köll, and G aham, 2020).
All lead o he unmis akable conclusion ha he es ima es in M3 and M4 om he co ec ion’s Table
1 a e no longe s a is ically signi ican . In o he wo ds, he e is no ela ionship be ween dis ance and
bombing o be ween bombing and ci ic engagemen in hese es ima ions.
7
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
Table 2: Replica ion o Ba celó’s Table 1 wi h New Models Using Clus e ed S anda d E o s
Va iables
Ci ic Engagemen (log) in 2001
(1) (2) (3) (4) (5) (6)
O iginal
Model 4
Non- obus
SEs
Clus e ed
SEs
(HC0)
Clus e ed
SEs
(HC1/STATA)
Clus e ed
SEs
(HC2)
Clus e ed
SEs
(HC3)
Bombs, pe km2 (log) 0.44** 0.44** 0.44 0.44 0.44 0.44
( esidence p e-1975) (0.12) (0.11) (0.46) (0.47) (0.50) (0.54)
Popula ion densi y -0.22** -0.22** -0.22 -0.22 -0.22 -0.22
(1960 o 61) (’000) (0.06) (0.06) (0.23) (0.23) (0.26) (0.33)
Cons an -1.15** -1.15** -1.15 -1.15 -1.15 -1.15
(0.28) (0.26) (1.05) (1.05) (1.13) (1.23)
Fi s S age
Dis ance o 17 h Pa allel -0.11** -0.11** -0.11 -0.11 -0.11 -0.11
(0.02) (0.02) (0.12) (0.12) (0.13) (0.15)
F-s a is ic 24.6 70.4 0.84 0.84 0.73 0.6
Obse a ions 875 875 875 875 875 875
No es: S anda d e o s in pa en heses (** p <0.05, * p <0.1). Using he code p o ided in he eplica ion package, we
we e unable o eplica e Ba celó (2023, 2)’s F-s a is ics in Column 1 o his IV Model 4, which was epo ed o be 23.2.
2.1.2 Reg ession wi h highly collinea co a ia es
I is wo h no ing ha in ou e-es ima ions in Table 1, he coe icien s in he i s and second s ages
om column 3 (M5) emain signi ican in column 6. The key di e ence is ha in column 3, Ba celó
includes a ull sui e o indi idual and p o incial con ol a iables. Why do hese new con ols change
he esul s o he es ima ion?
Mos immedia ely, Ba celo’s sou h con ol is miscoded o he wo sou he n p o inces closes o
he 17 h pa allel. As Figu e 2 shows, bo h Thua Thien-Hue and Quang T i a e coded as No h, when
hey a e below he 17 h pa allel and we e pa o Sou h Vie nam. In ou analysis abo e and in he
es o he pape , we p ope ly code hese p o inces.
8
Jou nal o Commen s and Replica ions in Economics - JCRE
0.45
1.13
2.35
2.62
2.17
4
3.67
6
0
2
4
6
P o inces Ou side Zone IV 6 P o inces Inside Zone IV
A e age Numbe o Cen al Commi ee Membe s
pe P o ince
Second Cen al Commi ee (1951)
Thi d Cen al Commi ee (1960)
Nin h Cen al Commi ee (2001)
Twel h Cen al Commi ee (2016)
Figu e 4: Pa y S eng h was Highe in Zone IV Be o e Vie nam Wa and Remains
Highe Today
C i ical o ou al e na i e heo y is ha Zone IV was a ge ed by US mili a y planne s, because
his unde lying pa y s eng h made i a massi e sou ce o manpowe and esou ces. To demons a e
his poin , Table 4 documen s sigh s om Zone IV lis ed in he ecen ly declassi ied publica ion o
he op 94 mili a y a ge s selec ed by he US Join Chie s o S a o bombing aids in 1964, known
as he JCS 94-Ta ge Lis (Kamps, 2001). Jus unde a qua e (23 a ge s) we e di ec ly in Zone IV,
including he mili a y ba acks Chanh Hoa, Dong Hoi, Vi Thu Lu, Vinh, and Chap Le whe e ec ui s,
soldie s, and leade ship o he No he n A my and Vie Cong insu gency ained and li ed (Kamps,
2001). This is a s ong indica ion ha he bombing did no c ea e pa y s eng h and loyal y, bu
ha Zone IV was delibe a ely a ge ed o i s p e-exis ing egime suppo .
15
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
Table 4: 23 Ta ge s om he Join Chie o S a 94-Ta ge Lis Tha A e in Zone IV
Ta ge numbe Ta ge desc ip ion
4 {R} Dong Hoi ai ield [limi ed je -capable]
5 {R} Vinh ai ield [limi ed je -capable]
14 Thanh Hoa ail oad/highway b idge
24 {R} Chanh Hoa ba acks SE and di ision headqua e s
33 {R} Dong Hoi ba acks WNW (p obable di ision headqua e s)
36 {B}{R} Vi Thu Lu ba acks/s o age a ea (gue illa s aging a ea)
38 Vinh ba acks and headqua e s mili a y egion IV
39 {R} Chap Le ba acks NW
41 {R} Phu Van ammuni ion depo E (majo depo )
42 {R} Phu Van POL s o age and ammuni ion depo NE
50 Vinh POL s o age
52 {R} Vinh supply depo E
53 {R} Phu Van supply depo SE
54 Thien Linh Dong supply depo S
55 {R} Vinh Son supply depo SW/SE
64 {R} Xom Bang ammuni ion depo
71 {R} Ben Thuy po acili ies/ ansshipmen cen e
74 Quang Khe Po app oaches
85/86 Road/Rail Rou e 1 (Ham ong o Hanoi)
89 Rou e 7 (Laos/No h Vie nam bo de )
90 Rou e 8 ( icini y Nape, Laos o Roa Qua)
91 Rou e 12 (Laos/No h Vie nam bo de o Xom Ma Na)
To illus a e ou empi ical conce ns, Figu e 5 looks a di e ences be ween he six Zone IV
p o inces and all o he s on Ba celó’s main ou come a iable (ci ic engagemen ), he con ounde
abo e ha we showed is biasing esul s (pa y membe ship), he sugges ed ea men a iable
(logged bombs pe capi a), and he ins umen al a iable (dis ance om he 17 h pa allel). His o ical
pa y membe ship is s ongly co ela ed wi h all ou a iables. Zone IV p o inces ha e highe
le els o ci ic engagemen and d ama ically mo e pa y membe s in he Wo ld Values Su ey sample,
expe ienced g ea e bombing, and a e close o he 17 h pa allel.
10
This p e ious his o y o his o ical
communis ac i i y, a ge ing by he US, and highe le els o pa y- ela ed ac i i y oday appea s o
be a clea exclusion iola ion and is he e o e likely o bias esul s.
10See also Appendices E, G1, and G2.
16
Jou nal o Commen s and Replica ions in Economics - JCRE
−0.22
0.15
−0.3
−0.2
−0.1
0.0
0.1
0.2
Ou side Zone IV Inside Zone IV
Ci ic Engagemen (log) o P o inces
Ci ic Engagemen
23.53
60.45
0
20
40
60
Ou side Zone IV Inside Zone IV
Sha e o Pa y Membe s (%)
in P o inces
Pa y Membe s in 2001
2.48
3.09
0
1
2
3
Ou side Zone IV Inside Zone IV
Bombs pe squa e KM (log)
Bombs D opped
4.92
2.23
0
2
4
Ou side Zone IV Inside Zone IV
La i ude Di e ence
om 17 h Pa allel
Dis ance om Bo de
Range ba s ep esen 95% CI wi h SEs clus e ed a wa ime p o ince le el
Figu e 5: His o ical Pa y S eng h is S ongly Associa ed wi h All Measu es
2.3
E o 3: Cons uc ing an Index o Cu en Pa y S eng h and No Ci ic Engage-
men
Ba celó’s nex con ex ual e o occu s du ing he assembly o his ou come a iable, ci ic engagemen ,
whe e he cons uc s an index using esponden s’ sel - epo ed pa icipa ion ( olun ee wo k and
membe ship) in 14 ypes o social g oups and olun ee associa ions om he 2001 Wo ld Values
Su ey (WVS). The choice o he 2001 WVS is odd; he P o incial Adminis a i e Pe o mance Index
(PAPI) su ey is a a la ge , mo e na ionally and p o incially ep esen a i e, and be e - ansla ed
su ey o public opinion and go e nance ha has been p oduced since 2009 by he Uni ed Na ions
De elopmen P og am and he Cen e o Communi y Suppo and De elopmen S udies. Ba celó
claims o use he WVS because i allows him o ack esponden s who mig a ed a e he bombing.
Howe e , his g oup cons i u es only 12.6% o he 1000 su ey esponden s, as mos esponden s
emained in he p o ince whe e he su ey was conduc ed.
Beyond ansla ion e o s ha ha e been aised be o e (Ku zman, 2014), he e a e wo addi ional
p oblems wi h using his measu e om WVS. Fi s , i does no p o ide a na ionally ep esen a i e
sample, as he da ase se e ely o e samples membe s o he VCP. Th oughou he coun y in 2001,
he e we e abou 2.5 million egis e ed pa y membe s (Phong ien ong hop, n.d.), comp ising 3%
o he popula ion. Howe e , in he 2001 WVS, he a e age pa y membe ship is 23.8%. Limi ing he
sha e o he 878 esponden s who answe ed he p e-1975 loca ion ques ion as Ba celó does, hose
17
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
who say hey a e membe s o a poli ical pa y is 29.2% (V43, WVS), and hose who say hey do
olun a y wo k o he pa y is 24.3% (V58, WVS). As such, hese su ey esponden s’ answe s abou
ci ic engagemen , pa icipa ion, and con idence canno be conside ed ep esen a i e o he 97% o
non-pa y membe s who we e a ec ed by he wa . These esponden s, as a whole, a e unusually
poli ical engaged.
Secondly, he au ho ’s measu e o ci ic engagemen , which was cons uc ed om WVS, is no
independen o poli ical engagemen (o pa y membe ship), which we ha e shown abo e was
high in Zone IV bo h be o e and a e he wa , because 12 ou o hese 14 o ganiza ions a e pa y
mass o ganiza ions ha all unde he Vie namese Fa he land F on (VFF) and con olled by he
pa y-s a e (See Table 5). Ba celó claims ha local poli ical o ganiza ions a e decen alized and
independen o na ional o ganiza ions (Ba celó, 2021, 5). Howe e , he p o ides no e idence o his
asse ion, which con adic s exis ing schola ship (Bui, 2016). In Vie nam’s Leninis single-pa y s a e
s uc u e, VFF is an umb ella g oup o mass o ganiza ions including he Communis Pa y i sel and
some o icially sanc ioned eligious g oups. Many o he go e nmen ’s social p og ams a e conduc ed
h ough VFF and i is in ended o supe ise go e nmen al o ganiza ions and ac i i ies. In eali y,
VFF’s a ious o ganiza ions a e made up o mos ly pa y membe s and he F on i sel is chai ed by a
high- anking VCP o icial, usually a Cen al Commi ee membe . Membe ship and pa icipa ion in
you h, emale, labo , and communi y o ganiza ions we e no olun a y in Vie nam in 2001 a he
ime o he WVS su ey. Mos membe ship a his ime would ha e occu ed h ough Vie nam’s pa y
mass o ganiza ions. I was no un il 2003 ha he Vie namese go e nmen p omulga ed Dec ee 88
on Associa ions, which p o ides a guiding egula o y documen o he o ma ion and ope a ions
o associa ions and ela ed g oups (Sidel, 2010). Essen ially, Ba celó’s nai e models a e simply
eg essing a measu e o his o ical pa y s eng h (bombing) on cu en pa y s eng h.
18
Table 5: 14 Types o Social G oups and Volun ee O ganiza ions Used in he WVS. Fo he Vie namese names o he VFF
o ganiza ions, see Appendix D
Wo ding in English O iginal Vie namese wo ding Mos likely VFF o ganiza ions
% o WVS
esponden s
as membe s
1 Social wel a e se ices
Các dịch ụ phúc lợi xã hội
cho người già àn ậ
à người úng hiếu
Red C oss Vie nam,
Vie nam Blind Associa ion,
Vie nam Assoc. o Vic ims o Agen O ange,
Assoc. in Suppo o Vie nam Handicapped and O phaned,
Vie nam Relie Assoc. o Handicapped Child en
26.5%
2 Religious o ganiza ion Các ổ chức ôn giáo
Vie nam Buddhis Sangha,
Commi ee o Solida i y Ca holics o Vie namese,
Vie namese P o es an Assoc.
10.4%
3Educa ion, a s, music o
cul u al ac i i ies
Các hoạ động giáo dục,
nghệ huậ , âm nhạc hoặc ăn hoá
Vie nam Union o Li e a u e and A s Assoc.,
Vie nam Assoc. o Lea ning P omo ion 17.3%
4 Labo unions Các liên đoàn lao động Vie nam Gene al Con ede a ion o Labo 11.3%
5 Poli ical g oups o o ganiza ions Các ổ chức, đoàn hể chính ị Vie nam Communis Pa y 28.5%
6Local communi y ac ions
on social issues
Hoạ động ại cộng đồng địa phương
ề các ấn đề như nghèo khổ,
iệc làm, nhà cửa, bình đẳng chủng ộc
Vie nam Women’s Union 26.2%
7Thi d wo ld de elopmen
o human igh s Các ổ chức phá iển hế giới hứ 3 1.5%
8Conse a ion, en i onmen ,
animal igh s g oups
Các nhóm bảo ồn, môi ường,
quyền động ậ Vie nam Assoc. o Sa e wa e and En i onmen 7.6%
19
9 P o essional associa ions Các hiệp hội nghề nghiệp
Vie nam Fa me ’s Union,
Vie nam Union o Science and Technology Assoc.,
Vie nam Coope a i e Alliance,
Vie nam Lawye s Assoc.,
Vie nam Jou nalis Assoc.,
Vie nam Ga dening Assoc.,
O namen al Oganism Science Assoc.,
Vie nam Medical Assoc.,
Vie nam His o y Science Assoc.,
Vie nam Gems ones, Jewel y, and A c a Assoc.,
Ve e an Teache Assoc. o Vie nam,
Vie nam Publishe s Assoc.,
Vie nam Fishe ies Socie y,
Vie nam Public Heal h Assoc.,
Ve e an Volun ee ing You h Assoc.,
Vie nam Assoc. o In alids and Disabili ies En e p ise,
Vie nam Assoc. o Small and Medium En e p ises,
Vie nam O ien ally T adi ional Medicine Assoc.,
Young Business People Assoc.,
Vie nam Ba Fede a ion
13.3%
10 You h wo k (scou s, guides,
you h clubs, e c.)
Công ác hanh niên
( ức là các câu lạc bộ hanh niên,
hướng dẫn iên, hướng đạo sinh )
Ho Chi Minh Communis You h Union,
Vie nam You h Fede a ion 15.4%
11 Spo s o ec ea ion Thể hao hoặc giải í 19.2%
12 Women’s g oups Các nhóm phụ nữ Vie nam Women’s Union 28.4%
13 Peace mo emen s Phong ào ì hoà bình Vie nam Union o F iendship O ganisa ions 9.2%
14 Volun a y o ganiza ions
conce ned wi h heal h
Các ổ chức ình nguyện
liên quan đến sức khoẻ Family Planning Assoc. 14.8%
20
Jou nal o Commen s and Replica ions in Economics - JCRE
−0.34
0.25
−0.4
−0.2
0.0
0.2
Non−Pa y Membe s Communis Pa y Membe s
Ci ic Engagemen (log)
Ci ic Engagemen
0.28
0.33
0.0
0.1
0.2
0.3
Non−Pa y Membe s Communis Pa y Membe s
Pa icipa o y alues (Index)
Pa icipa ion
0.34
0.4
0.0
0.1
0.2
0.3
0.4
Non−Pa y Membe s Communis Pa y Membe s
Pa icipa ion in Communi y
Mo e Voice in Communi y
0.19
0.25
0.0
0.1
0.2
0.3
Non−Pa y Membe s Communis Pa y Membe s
Pa icipa ion in Go e nmen
Mo e Voice in Go e nmen
Range ba s ep esen 95% CI wi h SEs clus e ed a wa ime p o ince le el
Figu e 6: Pa y Membe s a e Signi ican ly Mo e Engaged and In e es ed in
Poli ical Pa icipa ion han Non-Membe s
In sho , pa y membe ship is highly co ela ed wi h he o ganiza ions used o cons uc he
ci ic engagemen index. To demons a e his poin , Figu e 6 s udies di e ences be ween pa y and
non-pa y membe s on Ba celó’s main ou come a iable o ci ic engagemen , a logged index o
pa icipa ion in he g oups lis ed in Table 5, and his h ee main mechanism measu es (pa icipa o y
alues, p e e ence o oice in communi y, and p e e ence o oice in go e nmen ). I is immedia ely
ob ious ha esponden s, who a e membe s o he communis pa y, sco e signi ican ly highe on all
o Ba celó’s ou come measu es. Pa y membe s epo pa icipa ion in 3.5 non-pa y o ganiza ions
om Ba celó’s lis compa ed o 1.4 o ganiza ions o non-pa y membe s (p<0.0001). Simila ly,
pa y membe s demons a e signi ican ly highe le els o pa icipa o y alues and willingness o
engage in he communi y and go e nmen . Figu e 7d s udies he ela ionship om he p o incial
le el— he le el o he ea men a iable, demons a ing he s ong bi a ia e co ela ion be ween
a e age pa y membe ship and pa icipa ion in he Women’s Union and e en Spo s associa ions
using bo h WVS 2001 and PAPI 2020. Combined, he wo igu es clea ly demons a e ha Ba celó’s
measu e o “ci ic engagemen ” is eally jus a p oxy o unde lying pa y s eng h in 2001. P o inces
wi h la ge sha es o pa y membe s also had la ge popula ion sha es in o he seemingly un ela ed
mass o ganiza ions.
21
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
Figu e 7: Pa y Membe ship Is S ongly Co ela ed wi h Membe ship in Women’s
and Spo s O ganiza ions
(a) Spo s clubs and Communis Pa y membe ship using Wo ld Values 2001 Su ey da a
(b) Spo s clubs and Communis Pa y membe ship using PAPI 2020 da a
(c) Women’s union and Communis Pa y membe ship using Wo ld Values 2001 Su ey da a
(d) Women’s union and Communis Pa y membe ship using PAPI 2020 da a
A e accoun ing o bo h E o 2 and E o 3 in Appendix B by co ec ly clus e ing s anda d
e o s by wa ime p o inces, d opping he s uc u ally collinea geog aphical con ols o sou h and
la i ude, as well as adding in addi ional ea men a iables o Zone IV and numbe o Second
Cen al Commi ee membe s (1950), we show ha he main OLS esul s a e also s a is ically
indis inguishable om ze o - he e is no e idence ha bombing caused ci ic engagemen . We u he
discuss hese co ec ions and es ou heo y in he nex sec ion.
3 Ou Modi ica ion
Thus a , we ha e demons a ed ha he au ho employs a esea ch design ha con uses his o ical
pa y s eng h wi h exposu e o bombing, es s his heo y on a lawed sample o he Vie namese
popula ion, and con la es pa y s eng h wi h ci ic engagemen . Toge he , hese e o s lead o he
e oneous conclusion ha bombing caused ci ic engagemen , when i only shows ha communis
22
Jou nal o Commen s and Replica ions in Economics - JCRE
bas ions we e a ge ed o bombing.
We es he heo e ical ex en o ou iews abou he exclusion es ic ion iola ion in Table 6. We
ha e a gued ha dis ance is eally cap u ed by his o ical pa y s eng h, because he 17 h pa allel an
di ec ly h ough Zone IV, he epicen e o communis ac i i y. To es his di ec ly, Table 6 includes a
di ec measu e o whe he his o ically communis communi ies con inue o be mo e hea ily in ol ed
in pa y ac i i ies oday. To do his, we un a linea model ha eplaces he bombing measu e wi h a
dummy a iable o whe he a p o ince is loca ed in Zone IV. In line wi h he adjus men s, we show
he o iginal con olled OLS speci ica ion in Model 1 and Model 2 (wi hou he geog aphic con ols
sou h and la i ude) while d opping pa y membe s in Models 3-4.
Table 6 inds a obus ela ionship be ween his o ical pa y membe ship and ci ic engagemen
oday. In he ully-speci ied o iginal Model 1, loca ion in mili a y Zone IV is associa ed wi h 34%
highe ci ic engagemen , while i is 37% in Model 2 wi h he immed con ols. D opping pa y
membe s in Models 3 and 4 educes he size o he coe icien s, which is in line wi h ou heo y o
he exclusion es ic ion iola ion ha his o ical pa y s eng h is co ela ed wi h pa y domina ion
oday. Remo ing pa y membe s educes he pa h dependency o pa y loyal y and con ol. In sho ,
we obse e he legacy o his o ical pa y s eng h: places ha we e mo e ac i e and in luen ial in he
p e-wa communis mo emen emain mo e so oday.
23
E. J. Malesky & T. Nguyen – His o ical Communis Pa y S eng h (Replica ion). JCRE (2024-5)
Table 6: His o ical Pa y S eng h is Associa ed wi h Poli ical Ac i i y Today
Va iables
Ci ic Engagemen (log) in 2001
Using Full Da ase D opping Pa y Membe s
(1) (2) (3) (4)
OLS OLS OLS OLS
Zone IV=1 0.34** 0.37** 0.27** 0.27**
(0.07) (0.06) (0.06) (0.04)
Female=1 -0.05* -0.05* -0.05 -0.05
(0.03) (0.03) (0.03) (0.03)
Age -0.001 -0.001 -0.002 -0.002
(0.001) (0.001) (0.001) (0.001)
Educa ion 0.03** 0.03** 0.005 0.01
(0.01) (0.01) (0.01) (0.01)
Popula ion densi y 0.02 0.02 0.04* 0.05**
(1960 o 61) (’000) (0.03) (0.03) (0.02) (0.02)
A e age p ecipi a ion 0.14 0.07 0.06 0.05
(’00) (0.13) (0.08) (0.11) (0.06)
La i ude (’0) -0.29 -0.09
(0.23) (0.20)
Sou h -0.22 -0.05
(0.23) (0.23)
Cons an 0.08 -0.39* -0.21 -0.36**
(0.46) (0.22) (0.36) (0.18)
Obse a ions 862 862 608 608
R-squa ed 0.09 0.08 0.06 0.05
RMSE 0.51 0.52 0.39 0.39
No es: S anda d e o s, clus e ed a wa ime p o ince le el, in pa en heses (** p <0.05, * p <0.1). Model 1 eplica es
he ully-speci ied OLS model 2 in Table 1 o he o iginal a icle while eplacing bombing wi h a dummy a iable o
whe he a p o ince is in Zone IV. Model 2 d ops he geog aphical con ols Sou h and La i ude. Models 3-4 a e simila
o Models 1-2 bu uses da a ha excludes membe s o he communis pa y.
4 Discussion and Conclusion
Does exposu e o bombing c ea e mo e poli ically engaged ci izens? The e idence does no suppo
an a i ma i e answe . None o he indings p esen ed in Ba celó (2021, 2023) hold in he ace o
igo ous sc u iny, and much o i is he a i ac o elemen a y s a is ical e o s. Impo an conclusions
should no be based on such lawed ounda ions. Ra he , we show ha he ue unde lying explana-
ion is mo e mundane: a eas ha we e sou ces o communis insu gency s eng h om be o e he
wa we e a ge ed o hei ac i i y du ing he con lic , and hey emain mo e loyal o he egime
oday.
Beyond he immedia e indings in he pape , ou econs uc ion should lead schola s o exe cise
cau ion when employing sophis ica ed causal in e ence ools using his o ical da a om coun ies in
which hey a e no expe s. Such analyses ha e become inc easingly popula , as schola s sca enge
he wo ld looking o his o ical na u al expe imen s o es social science heo ies. This is impo an
wo k, bu unde s anding hese expe imen s equi es g ea con ex ual nuance, his o ical esea ch,
and language abili y. Wi hou hese, e idence can be easily misin e p e ed. The damage is mo e
han academic: s a is ical manipula ions can obscu e undamen al mis akes, leading o e oneous
conclusions and misguided policy ecommenda ions.
24
