Pe ez-Sebas ian, Fidel; Ra eh, Ohad; an de Ploeg, F ede ick
A icle — Published Ve sion
Na u al Resou ces, Technology Imp o emen s, and
G ow h
En i onmen al and Resou ce Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Pe ez-Sebas ian, Fidel; Ra eh, Ohad; an de Ploeg, F ede ick (2025) : Na u al
Resou ces, Technology Imp o emen s, and G ow h, En i onmen al and Resou ce Economics, ISSN
1573-1502, Sp inge Ne he lands, Do d ech , Vol. 88, Iss. 8, pp. 2157-2199,
h ps://doi.o g/10.1007/s10640-025-01004-x
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323702
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/
Accep ed: 7 May 2025 / Published online: 24 June 2025
© The Au ho (s) 2025
Fidel Pe ez-Sebas ian, Ohad Ra eh, and F ede ick an de Ploeg con ibu ed equally o his wo k.
Ex ended au ho in o ma ion a ailable on he las page o he a icle
Na u al Resou ces, Technology Imp o emen s, and G ow h
FidelPe ez-Sebas ian1· OhadRa eh2· F ede ick an de Ploeg3,4,5,6
En i onmen al and Resou ce Economics (2025) 88:2157–2199
h ps://doi.o g/10.1007/s10640-025-01004-x
Abs ac
We s udy, analy ically and empi ically, how echnological changes a ec he nexus be-
ween esou ce abundance and economic g ow h. Ou wo-sec o model indica es ha
capi al-augmen ing echnological imp o emen s can be con empo aneously con ac iona y
in esou ce- ich economies, and expansiona y elsewhe e, due o di e ences in he size o
he elas ici y o subs i u ion be ween labo and capi al. In addi ion, such imp o emen s
yield ela i ely s eepe expansiona y pa e ns in esou ce- ich economies in he longe un.
We es hese p edic ions using a panel o U.S. s a es and coun ies. Ou iden i ica ion s a -
egy es s on geog aphically-en enched di e ences in esou ce endowmen s, and he adop-
ion o plausibly exogenous echnology shocks a he na ional le el. Ou co e es ima es
co obo a e ou p edic ions. Fi s , we documen pe sis en di e ences in he elas ici y o
subs i u ion be ween capi al and labo ac oss he na u al- esou ce dimension. Second, we
ind ha an inc ease in TFP is on impac con ac iona y in esou ce- ich s a es, bu is
non-con ac iona y (a wo s ) in esou ce-poo ones. Thi d, we illus a e ha in he longe
e m a posi i e echnology shock expands ou pu and inpu s in esou ce- ich economies
ela i ely mo e s ongly. Ou esul s shed ligh on hi he o o e looked po en ial ad e se
e ec s o na u al esou ce abundance.
Keywo ds Na u al esou ce abundance · Technology shocks · Inpu elas ici ies
JEL classi ica ions Q32 · E32 · O33
1 3
F. Pe ez-Sebas ian e al.
1 In oduc ion
Unde s anding he nexus be ween na u al esou ces and economic g ow h has been o
pe ennial in e es o economis s and policy make s, especially in discussions o he esou ce
cu se. The li e a u e so a has highligh ed a hos o po en ial ansmission channels.1 Li le
a en ion, howe e , has been gi en o he po en ial ole o echnological changes.2 The la e
ep esen s a pi o al phenomenon wi h long-las ing implica ions o he economy. None he-
less, he empi ical li e a u e s ill deba es he size o he con empo aneous impac s.3 This
s udy examines how esou ce abundance in e ac s wi h echnological changes o a ec
g ow h. We hypo hesize, and demons a e using U.S. da a, ha echnology imp o emen s
induce a con empo aneously di e gen ou come on g ow h in ou pu and labo , ac oss he
na u al esou ces dimension. Ou esul s shed ligh on p e iously o e looked po en ial
ad e se e ec s o na u al esou ce abundance, and o e one possible econcilia ion o he
ongoing deba e o e he opposing con empo aneous e ec s o TFP shocks on he economy.
The no ion ha he oil and gas sec o s end o be ela i ely in ensi e in capi al and low-
skilled labo has been documen ed in p e ious s udies (e.g., Michaels e al. 2014). The
well-documen ed capi al-skill complemen a i y hypo hesis (e.g., K usell e al. 2000) hen
sugges s ha he elas ici y o subs i u ion be ween capi al and labo (he ea e , he ESKL)
should be ela i ely highe in esou ce- ich sec o s. This p edic ion has been subs an ia ed
in a numbe o c oss-sec ional s udies. 4 Mo eo e , Ra eh (2020) illus a es ha hese ea-
u es may ansla e o he mac oeconomic le el in economies wi h a dominan oil and gas
sec o , no ing ha esou ce-abundan economies a e consis en ly mo e capi al-in ensi e
o e a pe iod o h ee decades. We hypo hesize ha his may simila ly ex end o he size o
he ESKL, posi ing ha i is pe sis en ly ela i ely highe in ex ac i e indus ies.
Impo an ly, he ela i e size o he ESKL can be cen al o unde s anding he con em-
po aneous impac o echnological changes on economic ac i i y.5 The in ui ion is simple:
a posi i e echnology shock inc eases he p oduc i i y o capi al, bu i also educes he
need o labo h ough ac o subs i u ion. To examine his analy ically, we cons uc a wo-
sec o (ex ac i e and non-ex ac i e) model o economic g ow h wi h capi al adjus men
cos s, and economy-wide labo - and capi al-augmen ing echnological change. The analy i-
cal esul s indica e ha while labo -augmen ing echnological imp o emen s a e simila ly
expansiona y in bo h sec o s, he sho - and long- e m pa e ns may di e o capi al-aug-
men ing imp o emen s. Speci ically, a capi al-augmen ing echnology shock inc eases he
p oduc i i y o capi al while subs i u ing labo , hus gi ing ise o in olun a y unemploy-
men in he sho - un due o capi al adjus men cos s. These cos s esul in a g adual con e -
1 See, e.g., Allco and Kenis on (2018), A ezki e al. (2017), A mand e al. (2020), B ollo e al. (2013), Gyl-
ason e al. (1999), James and Ri e a (2022), To nell and Lane (1999), To ik (2002), and he e e ences
he ein. Van de Ploeg (2011) and Venables (2016) p o ide syn heses o he li e a u e.
2 The li e a u e un eils he po en ial endogenei y o TFP shocks and inno a ion o esou ce abundance (e.g.,
Ku albaye a and S e anski 2013). Howe e , i o e looks he po en ial ole o esou ce abundance in ans-
mi ing he e ec s o echnological changes on he economy.
3 While s anda d ic ionless eal business cycle models p edic sho - e m expansiona y e ec s, o he canon-
ical mac o wo kho se models p edic he opposi e (e.g., Chang and Hong 2006). This inconclusi e e i-
dence, along wi h he o he ela ed li e a u e, is e iewed in mo e de ail in he ollowing sec ion.
4 See, e.g., Caballe o e al. (1995), Young (2013).
5 The no ion ha he ESKL pa ame e is cen al o unde s anding he impac s o echnological changes da es
back o Hicks (1932) and Sa ō (1975). Fo a mo e ecen analysis, see Can o e e al. (2014).
1 3
2158
Na u al Resou ces, Technology Imp o emen s, and G ow h
gence back o ull employmen . Hence, in he sho - e m he impac is con ac iona y i he
ESKL is su icien ly high; ye , in he long- e m he magni ude o he expansion depends on
he ini ial capi al-in ensi y, hus poin ing a he po en ial di e gen ou come ac oss esou ce-
ich and -poo economies.
The model also se es o de i e sec o elas ici ies and capi al- and labo -augmen ing ech-
nology shocks o e ime using U.S. indus y-le el da a. Ou es ima es e eal ha ex ac i e
indus ies indeed exhibi consis en ly highe ESKL han non-ex ac i e sec o s, a e aging
a ound 0.79 compa ed o 0.55 o o he sec o s. The la e unde pins ou subsequen analy-
sis, in which we empi ically examine ou main analy ical p edic ions, using an ex ensi e
panel o U.S. s a es and coun ies, co e ing he pe iod 1963–2015. Employing plausibly
exogenous measu es o na ional echnology shocks— he pu i ied U.S. TFP se ies o Basu e
al. (2006) (hence o h, BFK)—and geologically de e mined c oss-sec ional s a e esou ce
abundance om James (2015), we exploi geog aphic and empo al a ia ion o iden i y he
di e en ial e ec s o echnology imp o emen s ac oss esou ce in ensi ies. Ou empi ical
s a egy hinges on in e ac ing s a e-le el esou ce endowmen s wi h na ional echnology
shocks, acili a ing causal iden i ica ion h ough geog aphic and empo al exogenei y.6
Ou empi ical analysis p o ides obus suppo o ou analy ical p edic ions. We ind
ha posi i e na ional TFP shocks, mos no ably capi al-augmen ing ones, ha e a con em-
po aneously con ac iona y e ec on ou pu and employmen in esou ce- ich s a es, d i en
by labo ma ke impac s, whe eas i is expansiona y o a wo s neu al in esou ce-poo
s a es. Quan i a i ely, ou baseline es ima es sugges ha a s anda d de ia ion inc ease in
TFP dec eases he a e age ou pu o esou ce- ich s a es by app oxima ely 0.1% ela i e o
esou ce-poo s a es in he sho e m. None heless, ou es ima es also show ha wo o i e
yea s ahead, posi i e TFP shocks lead o ela i ely s onge expansions o ou pu and inpu s
in esou ce- ich s a es, aligning wi h ou model’s p edic ions. Finally, we show ha hese
pa e ns a e obus o a ba e y o obus ness es s ha conside di e en ea men s, sample
es ic ions, speci ica ions, and con ols.
Sec ion 2 e iews ela ed li e a u e and places ou con ibu ions wi hin i . Sec ion 3
explains analy ically how esou ce abundance in e ac wi h echnology shocks o a ec
g ow h. Sec ion 4 p esen s he da a, empi ical indings, and obus ness es s. Sec ion 5
concludes.
2 Rela ed Li e a u e
Ou con ibu ions ela e o h ee s ands o li e a u e. Fi s , na u al esou ce abundance can
be a blessing as well as a cu se.7 A cen al aspec is he po en ial nega i e impac o esou ce
abundance on p oduc i i y and inno a ion. Among he a ious channels p oposed, na u-
al esou ce weal h may dep ess ac o p oduc i i y (e.g., Gyl ason e al. 1999; K ugman
1987; Sachs and Wa ne 2001; To ik 2001; Van Wijnbe gen 1984), lowe human capi al
(e.g., Bha acha yya and Hodle 2010; Gyl ason 2001; S ijns 2006), and induce special-
6 This me hodology is eminiscen o ha adop ed in o he s udies ha ha e also examined he he e ogeneous
local e ec s o agg ega e shocks, by es ing he impac o hei in e ac ion, including Liu and Williams
(2019), Pe ez-Sebas ian e al. (2019), and Ra eh (2020), among o he s.
7 See he su eys in Van de Ploeg (2011) and Venables (2016) o e ec s a he na ional le el, and Van de
Ploeg and Poelhekke (2016) o e ec s a he local le el.
1 3
2159
F. Pe ez-Sebas ian e al.
iza ion ha c owds ou inno a ion and en ep eneu ship (e.g., Ku albaye a and S e anski
2013; Michaels 2011; To ik 2002). In con as , ou analy ical and empi ical se ups conside
(na ional) echnology shocks ha a e exogenous o esou ce abundance a he le el o indi-
idual s a es, and ad ance a no el hypo hesis conce ning he in e ac ion o esou ce abun-
dance and echnological shocks and i s po en ial impac on sho - and long- e m g ow h. We
ind ha imp o emen s in echnology a e con ac iona y on impac p ima ily in esou ce-
ich a eas, and a e mo e expansiona y, ela i e o he emaining a eas, in he longe e m. Ou
esul s shed ligh on hi he o o e looked nega i e impac s o na u al esou ce abundance.
Second, da ing back o he seminal con ibu ions o Kydland and P esco (1982) and
Gali (1999), he ques ion o whe he echnology imp o emen s a e con ac iona y o expan-
siona y has aken a cen al ole in he mac oeconomic li e a u e. While s anda d ic ionless
eal business cycle (RBC) models p edic ha echnology imp o emen s a e expansiona y
in he sho - e m, s icky-p ice models p edic he opposi e.8 The ela ed empi ical li e a-
u e is also inconclusi e. Fo ins ance, in hei seminal wo k BFK show ha echnology
imp o emen s a e con ac iona y on impac due o dec eases in inpu use, mos no ably
labo . Howe e , Ch is iano e al. (2004) demons a e ha hei co ec ion o he BFK ech-
nology measu e yields con empo aneously expansiona y e ec s, ocusing on labo . Mo e
ecen ly, Can o e e al. (2014) highligh ha he ac o -augmen ing na u e o shocks and he
ESKL a e c i ical in de e mining he impac o echnology shocks on he labo inpu in bo h
RBC and s icky-p ice amewo ks.
We con ibu e o his li e a u e by analyzing he ole o na u al esou ces in he ans-
mission o hese shocks. Speci ically, ou analysis o e s one po en ial econcilia ion o
he opposing iews based on he unde lying pe sis en di e ences in he ESKL be ween
ex ac i e and non-ex ac i e indus ies. We obse e ha he sho - e m impac s o ech-
nology imp o emen s a e con ac iona y p ima ily in esou ce- ich a eas, whe e ex ac i e
indus ies hold a signi ican sha e o he economy, bu a e mos ly expansiona y elsewhe e.
We show ha hese pa e ns a ise using BFK’s measu e, ime ame, and me hodology, and
ha hey a e applicable also when implemen ing co ec ions like hose unde aken in Ch is-
iano e al. (2004). Fu he , consis en wi h p e ious s udies, we show ha in he long- e m
posi i e echnology shocks a e expansiona y ac oss all a eas, albei mo e so in esou ce- ich
egions.
Thi d, he e is no sho age o s udies ha p o ide es ima es o he agg ega e ESKL (e.g.,
Do aszelski and Jaumand eu 2013; Klump e al. 2007; Ra al 2019). The e idence summa-
ized in Chi inko (2008) poin a es ima es well below one. Indus y-le el es ima es poin
a simila magni udes, albei wi h some he e ogenei y ac oss sec o s (Balis e i e al. 2003;
Caballe o e al. 1995; Young 2013). No ably, hese s udies es ima e he subs i u ion elas ic-
i y in he ex ac i e indus y o be amongs he highes , and e en he highes unde a ious
speci ica ions, ela i e o he o he indus ies.9
Since es ima es pe ain o speci ic samples, we es ima e he ime se ies o he ESKL in
he oil sec o wi hin ou sample’s ime ame and compa e i o he ime se ies o he a e -
8 Howe e , non-s anda d RBC models can also gene a e a nega i e co ela ion be ween hou s wo ked and
echnology shocks. See, o example, F ancis and Ramey (2005).
9 Fo example, Young (2013) employs he equa ion sys em app oach p oposed by León-Ledesma e al.
(2010) and inds a la ge a e age alue o his elas ici y ac oss he mining and qua ying ac i i ies (0.72)
han ac oss he es o sec o s (0.63). Fu he mo e, he es ima es ha he la ges elas ici y among he o me
ac i i ies is o he oil and gas ex ac ion indus y (0.87).
1 3
2160
Na u al Resou ces, Technology Imp o emen s, and G ow h
age elas ici ies o he es o he economy. We achie e his by employing U.S. indus y da a
o compu e he elas ici y pa ame e di ec ly om ou analy ical amewo k, which inco po-
a es CES p oduc ion unc ions and non-neu al p oduc i i y shocks—bo h essen ial o he
es ima ion o elas ici ies (An as 2004). We ob ain es ima es consis en wi h hose epo ed
in p e ious li e a u e: he es ima es a e below one, and he ESKL in he ex ac i e indus y
is signi ican ly and consis en ly highe han he co esponding elas ici y o he a e age o
he emaining economy.
3 Technology Shocks in a Two-Sec o Economy
He e, we analyze he implica ions o c oss-sec o di e ences in esou ce in ensi ies ( ied o
di e ences in he ESKL) o he sho - and long- e m impac s o p oduc i i y shocks. We
begin by p esen ing he basic elemen s o he model. Nex , he p edic ions a e o ganized in o
p oposi ions, wi h he p oo s elega ed o Appendix A. The sec ion also de i es he es able
p edic ions ha he subsequen empi ical sec ion will ocus on. Finally, we conclude by
pe o ming a s uc u al es ima ion o he sec o al ESKL and he bias o echnology shocks,
and by discussing he implica ions o esou ce- ich and esou ce-poo economies.
3.1 The En i onmen
Conside an economy wi h wo p oduc ion sec o s: ex ac i e (
e
) and non-ex ac i e (
m
).
The non-ex ac i e p oduc is he nume ai e. The economy is inhabi ed by a cons an popu-
la ion o
N
indi iduals ha a e endowed wi h one uni o labo . Each pe iod
, indi iduals
supply hei labo uni inelas ically in exchange o a sala y (
w
). We suppose ha capi al
and ou pu ma ke s a e open o he es o he wo ld. The labo ma ke is closed and labou
is in e na ionally immobile. We ha e a small open economy. The e o e, i
pi
ep esen s he
p ice o ou pu in sec o
i
, and
R
is he g oss e u n o capi al (i.e., he in e es a e plus he
dep ecia ion a e), hen
pm=1
, while he esou ce p ice
pe
and
R
a e cons an and gi en
on wo ld ma ke s.
Ma ke s a e pe ec ly compe i i e and i ms maximize p o i s. In he sho - un, i ms in
sec o
i
, wi h
i∈{e, m}
, employ p oduc i e capi al (
ki
) and labo (
ni
) a ime
acco ding
o he ollowing Leon ie p oduc ion unc ion
yi =Ω
i min {zk ki ,ω
i zn ni },
(1)
whe e
yi
ep esen s ou pu in sec o
i
a pe iod
,
zk
and
zn
p o ide p oduc i i y le els
speci ic o capi al and labo , espec i ely,
Ωi
is a p oduc i i y pa ame e speci ic o sec o
i
,
and
ωi
ep esen s he le el o e ec i e capi al pe uni o e ec i e labo .10
New echnologies b ing labo -augmen ing gains when
zn
ises, and capi al-augmen ing
echnical p og ess when
zk
ises. The new in ages also come wi h pa icula alues o
ωi
and
Ωi
. These alues esul om he long- un, sec o -speci ic ESKL (
εi
) no being equal o
ze o. Speci ically, in he long- un, he p oduc ion unc ion akes he CES o m
10 We could include he s ock o na u al esou ces as an addi ional inpu in he p oduc ion unc ions associa ed
wi h he ex ac i e ac i i y. Ou quali a i e p edic ions would no be a ec ed by his modi ica ion i na u al
esou ces en e he p oduc ion unc ion h ough a Cobb-Douglas o m (see Appendix B).
1 3
2161
F. Pe ez-Sebas ian e al.
y
i =
[
(zk ki )1−1/εi+(zn ni )1−1/εi
]
ε
i
εi−1
.
(2)
We assume ha he alues o
ωi
and
Ωi
esul om imposing he ixed inpu p opo ions
dic a ed by he solu ion o he maximiza ion p oblem aced by i ms unde he long- un
p oduc ion unc ion gi en by Eq. (2).11
Finally, ollowing Caballe o (1994) and BFK, ou amewo k ea u es in es men adjus -
men cos s. In pa icula , he capi al mo ion equa ion is gi en by
ki = (1 −δ)ki −1+xϕ
i ,
(3)
whe e
xi
deno es in es men ,
δ
deno es he dep ecia ion a e,
ki −1
is he p oduc i e capi al
inhe i ed om he p e ious pe iod, and
ϕ∈(0,1)
.12
3.2 P edic ions
Ou i s p oposi ion p esen s he alues o
ωi
and
Ωi
ha d i e sho - un ou pu and shows
how hey depend on capi al-augmen ing p oduc i i y
zk
, bu a e de e mined by he long-
un p oblem.
P oposi ion 1 The op imal le el o e ec i e capi al pe uni o e ec i e labo is
ω
i =
[(
zk
R
)
1−εm−1
]ε
i
/(1−ε
m
)
and inc eases in
zk
and dec eases in he use cos o capi al. The in age-speci ic pa am-
e e is
Ω
i =
[
1+ω(1−εi)/εi
i
]ε
i
/(ε
i
−1)
and dec eases in
zk
. Bo h
ωi
and
Ωi
a e independen o
zn
P oo . See Appendix A. □
11 Appendix C analyzes a e sion o he model ha includes skilled labo and accoun s o sec o -speci ic
p oduc i i y shocks. The p oduc ion unc ion displays capi al-, skilled-, and unskilled-biased p oduc i i y
shocks, along wi h capi al-skill complemen a i y wi hin a CES-nes ed-in-CES speci ica ion. The main con-
clusion is ha esul s simila o ou main indings hold. Mo e speci ically, capi al-biased and skilled-biased
p oduc i i y shocks ha e posi i e e ec s on long- un oup u , and hese e ec s can be ampli ied by
εe
. How-
e e , his ampli ica ion is less likely when shocks o igina e om unskilled-biased p oduc i i y. Nega i e
sho - un e ec s on ou pu a e mo e likely o esul om capi al-biased han om skilled-biased p oduc i i y
and, once again, can be ampli ied by
εi
. Impo an ly, unskilled-biased p oduc i i y shocks can ne e cause
sho - un ou pu losses.
12 Fo simplici y, he p esence o in es men adjus men cos s in ou model a e cap u ed h ough he pa am-
e e
ϕ
. The e o e, we ollow a educed- o m app oach. Fo mo e gene al o ms o in oducing in es men
adjus men cos s in mac oeconomic models, see F ancis and Ramey (2005) o example.
1 3
2162
Na u al Resou ces, Technology Imp o emen s, and G ow h
Due o he cons an inpu p opo ions dic a ed by
ωi
, he di e ence be ween he sho and
long un lies in he o al amoun o labo used in p oduc ion. In he sho un, i he capi al-
labo a io inc eases while capi al g ows slowly due o adjus men cos s, he cons an inpu
p opo ions may imply ha o al labo demand alls sho o o al labo supply, hus leading
o in olun a y unemploymen .13 Howe e , in he long un, capi al accumula es un il he
economy eaches ull employmen . The e o e, gi en ha labo mo es owa ds he mos
p oduc i e sec o and labo p oduc i i y inc eases in bo h p oduc ion ac i i ies, echnology
shocks a e always expansiona y in he long un. Fu he mo e, while he impac o labo
augmen ing echnology is independen o he ESKL, capi al-augmen ing echnology shocks
end o gene a e s onge labo -p oduc i i y g ow h as he ESKL inc eases.
The ollowing p oposi ion summa izes he e ec s o echnology shocks in he long un.
P oposi ion 2 An inc ease in capi al-augmen ing p oduc i i y
zk
o in labo -augmen ing
p oduc i i y
zn
leads o an inc ease in labo p oduc i i y in bo h sec o s. Fu he mo e, in
he long un, which is cha ac e ized by ull employmen , an inc ease in
zk
o
zn
yields
la ge le els o ou pu . Addi ionally, he change p oduced by labo -augmen ing p oduc i i y
zn
is independen o he ESKL. I capi al-augmen ing p oduc i i y
zk
ises, he induced
g ow h o labo p oduc i i y in he ex ac i e ac i i y inc eases wi h
εe
i
(
z
k
/R
)1−ε
m>
2
and
εm<1
, o i
zk /R
is su icien ly close o
1
and ei he
εm>1
o bo h
εe
and
εm
a e
smalle han 1.
P oo See Appendix A. □
A co olla y o P oposi ion 2 is ha , unde he s a ed pa ame e es ic ions, i
εe>ε
m
, he
g ow h o labo p oduc i i y caused by an inc ease in
zk
will be la ge in he ex ac i e
sec o han in he non-ex ac i e sec o . This esul sugges s ha economies in which he
ex ac i e ac i i y con ibu es su icien ly mo e o o al ou pu will end o g ow mo e in he
long un in esponse o an inc ease in
zk
i
εe>ε
m
. As men ioned p e iously, in he sho
un, since capi al and labo en e he p oduc ion unc ion in ixed p opo ions and capi al
does no mo e o he new long- un equilib ium immedia ely, employmen can be below
N
. Impo an ly, i he economy does no ha e ull employmen , he impac o a capi al-
augmen ing echnology shock can be con empo aneously con ac iona y.
The ollowing p oposi ion summa izes hese sho - un e ec s.
P oposi ion 3 A posi i e labo -augmen ing echnology shock
zn
can ne e cause a all in
ou pu . Howe e , because op imal in es men in capi al dec eases wi h
ϕ
, i
ϕ
is su icien ly
low, a posi i e capi al-augmen ing echnology shock
zk
can con empo aneously educe
ou pu . In he non-ex ac i e ac i i y, his occu s i
εm>1
, bu i
εm<1
,
ym
always
inc eases wi h
zk
. In he ex ac i e sec o , su icien condi ions o
ye
o all wi h
zk
a e
εe=εm>1
, o
εm<ε
e<1
and
zk /R
is su icien ly close o 1.
13 In p inciple, indi iduals a e willing o wo k a any wage because he labo supply is inelas ic. Howe e ,
ω
i
≡z
k
k
i
zn ni
is ixed a i s long- un alue, as es ablished by P oposi ion 1, which implies ha labo demand
equals
n
i
=z
k
k
i
zn ωi
. The e o e, i
ωe
o
ωm
ha e inc eased su icien ly, whe eas
ke
o
km
g ow ela-
i ely slowly, he sum
ne +nm
can be lowe han
N
in he sho un, whe e
N
deno es he ull-employ-
men le el.
1 3
2163
F. Pe ez-Sebas ian e al.
P oo . See Appendix A. □
To be e unde s and P oposi ion 3, no e ha adjus men cos s in oduce diminishing e u ns
o in es men . Fu he mo e, o e y low alues o
ϕ
, hese diminishing e u ns can be so
s ong ha capi al may expe ience only negligible inc eases. Thus, in he sho un, i he
inc ease in
zk
causes a su icien ly la ge educ ion o labo pe uni o capi al, and
ϕ
is su -
icien ly low, he all in he labo demand due o he la ge
ωi
can domina e he posi i e
e ec o a highe
zk
, esul ing in a dec ease in ou pu .14 Rega ding he sho - un e ec o a
change in
zn
, ecall ha i nei he a ec s
ωi
no
Ωi
. So, a highe
zn
only se es o inc ease
he op imal s ock o capi al, and as a esul , ou pu canno all in he sho un.
3.3 Tes able Hypo heses
The analysis indica es ha a capi al-augmen ing inc ease in p oduc i i y boos s ou pu in
he long- un bu may educe ou pu in he sho - un i he ESKL is la ge in he ex ac i e
sec o han in he non-ex ac i e sec o . This inding leads o wo es able hypo heses. The
i s one is ha a echnology imp o emen in capi al in a sec o whe e i is easie o subs i-
u e labo o capi al will lead o (1) highe ou pu g ow h in he long- un, and (2) a smalle
inc ease o e en educ ion o ou pu in he sho - un. The second es able p edic ion is ha
(3) a labo -augmen ing imp o emen in p oduc i i y boos s ou pu bo h in he sho and in
he long un.
Ou s a egy o es hese hypo heses consis s o wo s eps. We i s es ima e he ESKL
and ind ha he ESKL is highe o he ex ac i e han o he non-ex ac i e sec o s. A med
wi h his insigh , we hen p oceed o es hese hypo heses employing a sample o economies
ha di e in hei deg ee o na u al esou ce abundance.
3.4 Sec o al Elas ici ies o Subs i u ion Be ween Labo and Capi al
Ou analysis hus poin s a a p ima y igge ing p imi i e o he sign o echnology-shock
e ec s on he economy, namely c oss-sec ional di e ences in he ESKL. As no ed ea lie ,
he e is al eady some c oss-sec ional e idence ha suppo s his hypo hesis (e.g., Young
2013).15 We now explo e he hypo hesis ha he ESKL is pe sis en ly highe in ex ac i e
indus ies.
Fo his, simila o Caselli and Coleman (Caselli e al. 2006), we i s de i e exp essions
om he model ha allow eco e ing he p oduc i i y and elas ici y pa ame e s
zk
,
zn
,
εe
, and
εm
, and hen es ima e hem using c oss-indus y U.S. da a. We s a om equa ion
(2) and, o he es ima ion, allow
R
,
w
and
εi
o a y ac oss ime and sec o s, while
zk
14 Fo example, he condi ions s a ed in P oposi ion 3 o an inc ease in
ym
and a dec ease in
ye
as a con-
sequence o an inc ease in
zk
a e ul illed o
εm=0.6
,
εe=0.8
,
zk =0.5
, and
R=0.2
. The alues
o
εm
and
εe
a e he a e age es ima es ob ained in Sec . 3.4. The alue o
zk
is he maximum es ima e
ob ained in Sec . 3.5. Finally,
R=0.2
co esponds o he his o ical S&P 500 a e age e u n o
0.10
( see,
e.g., h p s : / / w w w . s e n . n y u . e d u / a d a m o d a / p c / d a a s e s / h i s e S P . x l s .) minus a 0.02 a e age annual in la ion
a e calcula ed using da a om Bu eau o Labo S a is ics o he U.S. consume p ice index om 1998 o
2015, and an a e age dep ecia ion a e o 0.12 calcula ed o he same pe iod om Table 1 in Esc ibá-Pé ez
e al. (2023).
15 Al hough we do no conside c oss-sec o al di e ences in adjus men cos s, empi ical e idence indica es
ha adjus men cos s in ex ac i e indus ies a e signi ican ly highe han in o he indus ies (e.g., G o h and
Khan 2010), hus s eng hening he sugges ed mechanism.
1 3
2164
Na u al Resou ces, Technology Imp o emen s, and G ow h
ploymen a e and hus dec eases he change in labo , ye in he pe iods he ea e hey
expand i . Columns (2) and (3) epo esul s when he sample is spli in o esou ce- ich
and -poo s a es, espec i ely. The ou come o he con empo aneous impac shows ha o
he esou ce- ich sub-sample, echnological imp o emen s a e con ac iona y, bu o he
esou ce-poo sample such imp o emen s a e expansiona y. Hence, he na u al esou ces
dimension is po en ially an impo an aspec in he in e p e a ion o he key esul s o BFK.33
In addi ion, he esul s in columns (2) and (3) u he indica e ha in he longe e m, labo
expands mo e s ongly in he esou ce- ich sample, despi e he ini ial d op.
Howe e , Ch is iano e al. (2004) add essed conce ns ela ed o po en ial endogenei y
o he BFK measu e. In addi ion, hey conside ed he le el o ( a he han changes in) he
labo inpu measu e, and ound a con empo aneous posi i e impac o TFP imp o emen s
on labo inpu . Ch is iano e al. (2004) hus ound ha TFP shocks a e con empo aneously
expansiona y.
In ou s a e-le el se ing conce ns ela ed o he endogenei y o TFP shocks a e mi iga ed
gi en hei na ional pe spec i e. Hence, o illus a e ha co ec ing o he e ec o he na u-
al esou ces dimension may ep esen a econcilia ion be ween he indings o BFK and
Ch is iano e al. (2004), we e-es ima e ou esul s o columns (2) and (3) when he depen-
den a iable (i.e., he unemploymen a e) is in le els a he han in changes. The esul s
appea in columns (4)-(5). They a e simila o hose epo ed in columns (2)-(3): echnology
imp o emen s a e con ac iona y in esou ce- ich and expansiona y in esou ce-poo s a es.
33 The po en ial ele ance o he na u al esou ces dimension o he in e p e a ion o BFK’s indings has been
implied by Bils (1998), who poin ed a he po en ial o e -es ima ing e ec o he oil p ice ins umen used in
BFK’s analysis. None heless, as will be no ed in ou main analysis, we illus a e ha he obse ed pa e ns
ex end o a ious measu es, and a e no speci ic o hose used in BFK.
Table 1 TFP shocks and he unemploymen a e o esou ce- ich and esou ce-poo s a es, 1980–1996
( e isi ing BFK)
1 3
2171
F. Pe ez-Sebas ian e al.
4.4 Co e Resul s on Impac o TFP Shocks
We now u n o ou co e esul s on he he e ogeneous impac s o TFP shocks and how hese
depend on esou ce weal h in a mo e comple e and igo ous se ing, and wi h an expanded
sample. We es ima e a ious e sions o Eq. (4), o each o he h ee
ou come
a iables.
These co e esul s a e p esen ed in Table 2.
4.4.1 E ec s on Ou pu
S a ing wi h ou pu , measu ed by he G oss S a e P oduc (GSP), Column (1) ep esen s
ou co e speci ica ion and p o ides suppo o ou main hypo heses. The es ima ed alue
o
θ
is nega i e and s a is ically signi ican , which indica es ha con empo aneous echnol-
ogy imp o emen s indeed induce a s onge nega i e impac in esou ce- ich s a es han in
esou ce-poo s a es, subs an ia ing he main analy ical p edic ion. In e ms o magni ude,
unde he mean endowmen o na u al esou ces, a one s anda d de ia ion inc ease in TFP
con ac s a e age ou pu o esou ce- ich s a es by
0.1%
ela i e o ou pu o esou ce-poo
s a es.34 Unde an al e na i e in e p e a ion, gi en an a e age esou ces le el, a echno-
logical imp o emen ha amoun s o he annual mean le el (i.e., he a e age imp o emen
unde aken each yea o e ime) con ac s he ou pu le el o esou ce ich s a es ela i e
o hei esou ce poo coun e pa s by 0.007%, being he equi alen o abou $800 million
di e ence be ween he wo s a e g oups.35
No ably, he magni ude o his es ima e o
θ
sugges s ha he ou come is no only in ela-
i e e ms (no ing ha some s a es ha e no esou ce endowmen s). Fu he mo e, i poin s a
a di e gen ou come (c . Table 1). This is also illus a ed by columns (2) and (3) o Table 2.
In he la e , we es ima e a e sion o Eq. (4) which excludes
ν
and
θ
and examines he di ec
impac o TFP shocks ia
δ
unde he wo sepa a e sub-samples. This a emp s o examine
whe he he main ou come is he esul o a ela i e e ec ( esou ce- ich ela i e o esou ce-
poo s a es), o a di ec one d i en by esou ce in ensi y. We ocus on examining he sign,
in e p e ing he magni ude wi h cau ion due o he exclusion o he ime ixed e ec s. Fol-
34 This is compu ed by mul iplying he es ima ed alue o
θ
by he mean esou ce endowmen and he s an-
da d de ia ion o TFP, and examining he change ha his induces in he mean ou pu measu e.
35 Simila o he p e ious case, his is compu ed by mul iplying he es ima ed alue o
θ
by he mean esou ce
endowmen and he annual mean TFP le el, and di iding by he mean ou pu le el. The mone a y alue is
compu ed by conside ing he 0.007% amoun o he a e age annual s a e ou pu le el.
Table 2 Resou ce endowmen s and e ec o echnology imp o emen s, 1963–2015
1 3
2172
Na u al Resou ces, Technology Imp o emen s, and G ow h
lowing he p e iously ou lined di ision, he sub-sample in column (2) includes s a es wi h
a pe -capi a esou ce endowmen abo e he 25 h pe cen ile while he sub-sample in column
(3) includes he emaining s a es wi h li le o no esou ce endowmen s.
We obse e ha con empo aneous TFP shocks ha e a nega i e and s a is ically signi ican
impac on ou pu i he e a e some na u al esou ce endowmen s. Con e sely, i esou ce
endowmen s a e sca ce, he impac becomes s a is ically imp ecise wi h a magni ude close
o ze o. These ou comes cla i y he sou ce o he obse ed ela i e di e ence. They poin a
a dis inc di e ging ou come, simila o he pa e ns no ed p e iously in ou BFK exe cise
epo ed in Table 1, and consis en wi h ou analy ical p edic ions.
4.4.2 E ec s on Unemploymen and Capi al
Columns (4)-(6) and (7)-(9), p esen an analysis simila o ha p esen ed in columns (1)-(3)
ye wi h he labo o capi al inpu p oxies (namely, he unemploymen a e o in es men ,
espec i ely) as
ou come
. Columns (4) and (7) examine he comple e sample whe eas col-
umns (5)-(6) and (8)-(9) conside he spli samples based on he same di ision used be o e.
Fo he case o labo , he es ima ed alue o
θ
in column (4) poin s a a simila ou come
as obse ed unde ou pu . Speci ically, echnology imp o emen s con ac he labo ma ke
mo e s ongly i na u al esou ce endowmen s a e high. Simila pa e ns a e also obse ed in
columns (5)-(6), since he es ima ed alues o
δ
indica e ha he con ac iona y e ec occu s
only in he g oup o s a es ha a e endowed wi h signi ican na u al esou ces.
The ou comes in columns (8)-(9) show ha in es men con ac s simila ly in bo h ypes
o en i onmen s ollowing a posi i e TFP shock. This is u he con i med by he ou come in
column (7), which poin s a no s a is ically dis inguishable impac o TFP shocks on in es -
men ac oss esou ce in ensi y le els. These pa e ns, in conjunc ion wi h hose obse ed
o he e ec s on labo , a e consis en wi h ou analy ical p edic ions gi en he p e iously
es ablished sys ema ic di e ences in elas ici ies o subs i u ion be ween labo and capi al
o esou ce- ich and -poo s a es, o he ex en ha he TFP shocks a e capi al-augmen ing.
Nex , we examine his analy ical p edic ion.
4.5 E ec s o Capi al- and Labo -Augmen ing TFP Shocks
Ou analy ical p edic ions indica e ha capi al-augmen ing shocks igge con empo ane-
ous subs i u ion be ween capi al and labo wi h a magni ude ha depends on he size o
he elas ici ies o subs i u ion be ween labo and capi al, hence con ac ing he labo inpu
in esou ce- ich s a es. He e we examine he di e en ial impac o capi al- and labo -aug-
men ing TFP shocks. We do so by employing he
zn
and
zk
pa ame e s compu ed and
ou lined p e iously in Sec . 3.5, co esponding o labo - and capi al-augmen ing shocks,
espec i ely. Gi en he scope o he unde lying U.S. indus y da a, he compu ed pa ame e s
a e a ailable annually o he pe iod 1998–2015.
We es ima e ou baseline speci ica ion, as pe column (1) o Table 2, whe e now he
zn
and
zk
measu es en e in lieu o
p
, sepa a ely. The esul s a e p esen ed in in Table 3.
Columns (1)-(3) and (4)-(6) examine he case o
zk
and
zn
, espec i ely. In each case, he
i s , second, and hi d column examine he
ou come
ela ed o he ou pu , labo , and capi al
measu e, espec i ely.
1 3
2173
F. Pe ez-Sebas ian e al.
These esul s a e consis en wi h ou analy ical p edic ions. The es ima ed alues o
θ
indica e ha he di e en ial impac ac oss esou ce in ensi y le els is obse ed only unde
capi al-augmen ing shocks, and mos no ably wi h espec o ou pu and labo inpu . This is
consis en wi h he iew ha capi al-augmen ing shocks induce subs i u ion be ween capi al
and labo mo e s ongly in s a es whe e his subs i u ion is s onge , i.e., esou ce- ich s a es,
as sugges ed by ou model.
4.6 Robus ness Tes s
We now conduc a ious obus ness es s o see whe he ou co e indings su i e i we
allow o o he sec o s han na u al esou ces, di e en agg ega e shocks, o di e en TFP
measu es, and when he esou ce measu e is in e ac ed wi h he wo ld oil p ice. We also
examine he le el o U.S. coun ies, and es o obus ness using di e en sample es ic-
ions, con ols, and speci ica ions.
Tables 4 and 5 p esen he obus ness esul s when we allow o o he sec o s han na u al
esou ces, o di e en agg ega e shocks, espec i ely. The o he obus ness esul s a e p e-
sen ed in Table 6. All speci ica ions ollow he co e speci ica ion, unless o he wise speci ied,
and hey co e di e en ime pe iods (depending on da a a ailabili y), as s a ed in he able.
4.6.1 Resul s wi h O he Sec o s han Na u al Resou ces
Ou co e analysis has ocused on one dimension o he indus ial composi ion o s a es, i.e.,
esou ce abundance. To u he mo i a e his ocus, we also examine he ole o o he majo
sec o s. We hus conside he GSP sha e o ou majo agg ega e sec o s: manu ac u ing,
se ices, ag icul u e, and wholesale ade.36 To examine how hey migh a ec he impac o
36 Each, as es ima ed by Young (2013), wi h a signi ican ly lowe ESKL, compa ed o ha es ima ed o he
oil and gas sec o ( aking manu ac u ing as an a e age o i s sub-sec o s).
Table 3 E ec s o capi al- and labo -augmen ing TFP shocks, 1998–2015
1 3
2174
Na u al Resou ces, Technology Imp o emen s, and G ow h
TFP shocks, we in e ac hem each wi h
p
and add hem sepa a ely, and hen concu en ly,
o he co e speci ica ion. We will ocus on ou pu .
The esul s a e p esen ed in Table 4. In columns (1)-(4) we add each o he addi ional in e -
ac ion e ms sepa a ely, in conjunc ion wi h ou in e ac ion e m o in e es ,
esou ce ∗ p
.
The ou come in each case indica es ha ou co e esul s a e obus o hese inclusions, i.e.,
he es ima ed alue o
θ
main ains i s sign and p ecision. The obus ness o
θ
is u he
obse ed in he demanding speci ica ion unde aken in column (5), in which all in e ac ion
e ms a e added concu en ly. In e es ingly, while na u al esou ce in ensi y e ains i s ole
in he e ec s o echnology shocks, none o he o he majo sec o s exhibi simila cha -
ac e is ics. In all cases he es ima ed coe icien s on he addi ional in e ac ion e ms ha e
close o ze o magni udes and no s a is ical signi icance,37 hus ea i ming he ole o na u al
esou ces in unde s anding he e ec s o TFP shocks on s a e-le el ou comes.
37 Se ices and ag icul u e yield ma ginally p ecise pa e ns, bu appea in only one o he speci ica ions.
Hence, hese e ec s a e no obus .
Table 4 Resul s wi h o he sec o s han na u al esou ces, 1963–2015
Table 5 Resul s wi h di e en agg ega e shocks
1 3
2175
F. Pe ez-Sebas ian e al.
4.6.2 Resul s wi h Di e en Agg ega e Shocks
Ou main hypo hesis pe ains o he impac o TFP shocks. Fo iden i ica ion pu poses, we
conside agg ega e, na ional TFP shocks, as hey a e exogenous o each indi idual s a e.
None heless, hei empo al a ia ion may co ela e wi h o he concu en na ional shocks
wi h he e ogeneous impac s ac oss he na u al esou ce dimension, especially as TFP shocks
cap u e noise in agg ega e p oduc ion by cons uc ion. To add ess his, we conside addi-
ional na ional shocks, ocusing on he impac o hei in e ac ion wi h he esou ce mea-
su e, in conjunc ion wi h he e ec o ou main in e ac ion e m o in e es .
We conside ou addi ional shocks a he agg ega e, na ional U.S. le el: mone a y policy
shocks, ede al ax shocks, news shocks, and business cycles. Fo mone a y shocks, we
conside he da a se ies cons uc ed by Ten ey o and Thwai es (2016); conside ing ede al
ax shocks, we employ he na a i e-based ede al ax changes om Rome and Rome
(2010); news shocks a e ep esen ed by he In es men Speci ic Technology (IST) news
shocks o Ben-Zee (2018); las , business cycles a e examined ia an indica o ha cap u es
whe he he U.S. is in a ecession, as de ined by he U.S. Fede al Rese e.38 We in e ac each
measu e wi h esou ce, and add each sepa a ely o he baseline speci ica ion, as pe as pe
Column (1) o Table 2.39
The esul s a e p esen ed in Table 5. Columns 1–4 epo he ou comes unde each agg e-
ga e shock, in he o de ha appea s abo e, espec i ely. The ea e , column 5 p esen s
he esul s o a speci ica ion in which all shocks a e concu en ly included. The es ima ed
coe icien s indica e ha , consis en wi h Pe ez-Sebas ian e al. (2019) and Pe ez-Sebas-
ian and Ra eh (2019), esou ce- ich s a es ha e be e abso p ion o con ac iona y ed-
e al ax changes, and in e es ingly also o na ional ecessions. In addi ion, consis en wi h
Ra eh (2020), esou ce- ich s a s a e impac ed mo e s ongly by con ac iona y mone a y
38 Each is a ailable o a di e en sample pe iod, as ou lined in he Da a Appendix; hence, sample size di e s
ac oss cases.
39 Simila o
TFP
, each (non-in e ac ed) agg ega e shock is cap u ed by he yea ixed e ec s, and hence
excluded.
Table 6 Fu he obus ness es s
1 3
2176
Na u al Resou ces, Technology Imp o emen s, and G ow h
shocks.40 Impo an ly, howe e ,
θ
main ains i s sign, signi icance, and magni ude, in all
cases, including in he one in which all in e ac ion e ms a e conside ed concu en ly, hus
u he mo i a ing he ocus on na ional echnology changes in he analysis.
4.6.3 Resul s a he U.S. Coun y Le el
While he a ailabili y o some o ou da a is limi ed a he mo e g anula coun y le el,
examining ou hypo heses unde he measu es ha a e a ailable a he coun y le el enables
us o exploi a signi ican ly la ge sample o mo e han 3,000 coun ies. To measu e c oss-
sec ional esou ce endowmen s a he coun y le el, we employ he plausibly exogenous
esou ce measu e cons uc ed by James and Smi h (2017). This p o ides a geologically-
based indica o o coun ies wi h ese es o shale gas. We examine he impac o
p
,
zk
,
and
zn
on coun y pe -capi a ou pu , by in e ac ing each o hese wi h he coun y esou ce
measu e. Resul s a e p esen ed in columns (1)-(3) o Table 6, espec i ely. They indica e
ha ou co e esul is obus a he coun y le el, as we obse e di e en ial e ec s on ou pu
in he case o
p
and
zk
, bu none o
zn
.
4.6.4 Resul s o Th ee Di e en Al e na i e TFP Measu es
Columns (4)-(6) o Table 6 p esen es ima ion esul s o h ee di e en ypes o TFP mea-
su es. Ou co e es ima es we e done wi h he BFK measu e, p ima ily in an a emp o
c ea e a mo e di ec compa ison o he BFK esul s. The li e a u e, howe e , o e s a ious
measu es o echnology shocks, each wi h hei own me i s and limi a ions. To examine
he alidi y o ou esul s, we conside h ee addi ional da a sou ces o TFP shocks: he
FORD se ies (F ancis e al. 2014), he BS se ies (Ba sky and Sims 2011), and he JPT se ies
(Jus iniano e al. 2011). Each o he TFP ypes is in e ac ed wi h ou
esou ce
measu e and
we use hese ins ead o ou baseline measu e
p
. Fo each o hese h ee al e na i e TFP
measu es, he es ima ed alue o
θ
main ains i s sign and p ecision. Ou co e esul s a e hus
obus o using hese di e en ypes o TFP shocks.
4.6.5 Resul s wi h Di e en Sample Res ic ions, Con ols, and Speci ica ions
Columns (7)-(14) o Table 6 p esen he esul s o some addi ional obus ness es s ha
include di e en sample es ic ions, con ols, and speci ica ions. Fi s , mo i a ed by he
BFK exe cise, we e-es ima e ou co e speci ica ion p io o 1997 and a e 1996, sepa-
a ely. This se es o es he applicabili y o he BFK case unde he comple e speci ica ion
(no di ec ly examined in he p e ious ela ed sub-sec ion), and examine whe he ou co e
esul s depend on ha pe iod. The es ima ed alues o
θ
in columns (7)-(8) indica e ha ou
co e esul s a e appa en in bo h pe iods, and ha i in ensi ies in magni ude in he pos -
1996 pe iod. This is consis en wi h he no ion ha echnology imp o emen s become mo e
capi al-o ien ed o e ime.
Nex , we add a ious addi ional basic con ols ha may a ec he impac o echnol-
ogy shocks indi ec ly, conside a speci ica ion wi h s a e-by-yea ixed e ec s, and es he
baseline speci ica ion unde a di e en clus e ing me hod. In column (9), we include as con-
ols go e nmen ax e enues pe capi a, go e nmen expendi u e pe capi a, go e nmen
40 The impac o IST news shocks is no obus , as he e ec is uns able ac oss speci ica ions.
1 3
2177
F. Pe ez-Sebas ian e al.
deb pe capi a, s a e unemploymen a e, pa y a ilia ion o s a e go e no , s a e inequali y,
and s a e popula ion size. These con ols accoun o a ious po en ial con ounding ac-
o s a he s a e le el, including he size o s a es (popula ion), e iciency and size o s a es’
public sec o ( axes, expendi u es, deb ), s a es’ business cycles (unemploymen ), poli ical
incen i es o s a es’ go e no s (pa y a ilia ion), and s a es’ income pola iza ion (inequal-
i y). In column (10) we conside a qua e ly-based sample, in which bo h
ou come
and
esou ce ∗ p
a e qua e ly-based (i.e., in his speci ica ion
ep esen s a qua e -by-yea
cell). Such a speci ica ion enables adding s a e-by-yea ixed e ec s, which con ol o all
s a e-by-yea changes, including he s a e indica o s employed in column (9), as well as
addi ional unobse ed ones. The sample pe iod in his case is 2005–2015,
υ
ep esen s
qua e ixed e ec s, and s a e-by-yea ixed e ec s a e included in lieu o
ηi
.41 In Column
(11) we hen e-es ima e ou co e speci ica ion wi h a wo-way clus e ing me hod, whe e
s anda d e o s a e clus e ed by s a e and yea . The ou comes in all cases indica e ha ou
main esul is obus o hese sensi i i y examina ions.
In columns (12)-(14) we e-es ima e ou model wi h h ee es ic ed samples. In column
(12) we exclude Mon ana, No h Dako a, and Wyoming. Figu e 4 indica es ha hese s a es
a e ou lie s in e ms o hei esou ce ichness, hence his exclusion enables us o exam-
ine he ex en o which ou co e esul s a e a ec ed by hem. In column (13) we exclude
s a es wi h ze o esou ce endowmen s (e.g., Delawa e, Maine, New Hampshi e, and Rhode
Island). This add esses he po en ial conce n ha ou co e esul s a e d i en by s a es wi h no
esou ces. In Column (14) we exclude Cali o nia, New Yo k, and Texas om he sample o
es he obus ness o ou esul s when he h ee la ges s a es a e excluded. This es ic ion
add esses he conce n ha ou esul s may be d i en by he dominan s a es. The es ima ed
alues o
θ
in all hese cases indica e ha ou co e esul s a e obus o hese es ic ions on
he sample.
4.6.6 Resul s When Resou ce Measu e is In e ac ed wi h he Oil P ice
Nex , we in e ac he c oss-sec ional measu e o esou ce endowmen (used in ou baseline)
wi h he oil p ice which is plausibly exogenous (James 2015). This measu e hen en e s he
es ima ed equa ion ins ead o ou co e esou ce measu e. We examine he e ec s o TFP
shocks ac oss s a es, bu also wi hin hem ac oss ime. Column (15) p esen s he esul s.
The es ima ed alue o
θ
main ains i s sign and signi icance unde his in e ac ed measu e.
This indica es ha he impac o echnology shocks does no only depend on he exis ence
o esou ce endowmen s, bu also on hei alue.
4.6.7 Resul s Unde S a e-Le el TFP Shocks
As a inal obus ness es , we examine he assump ion abou he homogenei y o he echno-
logical shocks ac oss U.S. s a es. The baseline empi ical analysis adop s his assump ion in
an a emp o add ess po en ial endogenei y conce ns, no ing ha agg ega e echnological
changes a e plausibly exogenous o in a-s a e indica o s. Ou analy ical amewo k akes
a simila pe spec i e unde he assump ion ha TFP shocks a e co ela ed ac oss s a es
due o he in a- ede al se ing. To examine he obus ness o he esul s o his homogene-
41 The sample pe iod in his case is limi ed by he a ailabili y o da a on G oss S a e P oduc a he qua e -
by-yea le el.
1 3
2178
Na u al Resou ces, Technology Imp o emen s, and G ow h
i y assump ion, we employ da a om Caliendo e al. (2018). The la e es ima ed changes
in measu ed TFP ac oss U.S. s a es du ing wo ime ames, 2002–2007, and 2007–2012.
Fi s , we no e ha he co ela ion o es ima es ac oss hese wo pe iods is highly posi i e,
mo i a ing he assump ion abou c oss-s a e simila i ies in TFP pa e ns. Second, we adop
hei es ima es o he i s pe iod, as i is close o he mid-poin o ou sample pe iod, and
conside hem as c oss-sec ional di e ences in he abso p ion a e o na ional TFP changes,
hence p o iding a c oss-sec ional dimension o he baseline na ional TFP measu e.42 In e -
ac ing his c oss-sec ional TFP measu e wi h he baseline na ional TFP changes yields a
s a e-le el measu e, wi h a ia ions ac oss bo h dimensions, namely s a es and ime. We
hen employ his measu e in lieu o
p
in he baseline speci ica ion. The esul s in column
(16), unde his modi ied speci ica ion, indica e ha he main esul s hold and inc ease in
an o de o magni ude, hus poin ing a he obus ness o he analysis o he homogenei y
pe spec i e.
4.7 Longe -Te m E ec s o TFP Shocks
Ou ocus has been on he con empo aneous e ec s o TFP shocks. Howe e , ou analysis
in Sec ion 3 also gi es insigh s conce ning he dynamic pa e ns o e ime. Speci ically, we
ind ha esou ce- ich s a es should expand mo e s ongly beyond he con empo aneous
e ec . Hence, we es ima e and p esen he dynamic he e ogeneous e ec s o TFP shocks
ac oss s a es wi h di e en le els o na u al esou ces o e he cou se o i e yea s.43 We
employ he me hod o local-p ojec ions o Jo da (2005).
The me hod o local p ojec ions gi es us es ima es o impulse esponse unc ions sepa-
a e eg essions o each lead o e he o ecas ho izon. The e ec o TFP shocks a
+h
wi h
h=0,1,...,4
is es ima ed by eg essing dependen a iables a
+h
on shocks and
co a ia es a ime
. Responses hus do no ely on nonlinea ans o ma ions o educed-
o m pa ame e s as in VARs.44 We de ine
∆ −1xi, +h≡xi, +h−xi, −1
and es ima e he
sequen ial equa ions
∆
−1
(
ou come
)
i, +h
=
α
h+
β
h(
ou come
)
i, −1
+
γ
h(
esou ce
)
i
+
δh
(
p
) +
θh
(
esou ce
∗
p
)i, +
ηh
i+
νh
+
ϵ
i, +h
. (6)
The dependen a iable is cumula i e g ow h o he
ou come
a iable,
∆ −1(ou come)i, +h
,
o di e en alues o
h
. Ou main coe icien s o in e es a e he ones on he
esou ce ∗ p
in e ac ion a iable, i.e.,
θh
o he con empo aneous e ec
h=0
and he di e en leads
h=1,...,4
. These 5 pa ame e s shape he impulse esponse unc ion, and hence enable us
o ace he ime p o ile o he e ec o TFP shocks.
Figu e 5 plo s he impulse esponse unc ions o each o he ou come a iables, oge he
wi h
95%
con idence in e als. Fo ou pu , he g adual inc ease in he es ima ed alue o
42 Fo ins ance, du ing he 2002–2007 pe iod he change in measu ed in TFP in O egon was wice he size o
ha in Oklahoma, and h ee imes ha in New Hampshi e.
43 The leng h o he examined ho izon is based on a 1-yea ex ension o he BFK amewo k in which he
e ec s o TFP shocks a e obse ed, and measu able o e he medium- e m ho izon o app oxima ely ou
yea s.
44 In Appendix B we p esen VAR es ima es, and illus a e ha he obse ed pa e ns a e obus o he es ima-
ion me hod.
1 3
2179
F. Pe ez-Sebas ian e al.
θh
, as he lead
h
inc eases, indica es ha a e he con empo aneous nega i e e ec o TFP
shocks, echnology imp o emen s become mo e expansiona y in esou ce- ich s a es, mos
no ably s a ing in he second yea , ela i e o hose in esou ce-poo s a es. The impulse
esponse unc ions o labo and capi al inpu s pain a simila pic u e. This is e iden om
he g adually dec easing (inc easing) pa e ns in he unemploymen a e (in es men ), indi-
ca ing again ha posi i e TFP shocks a e mo e expansiona y in esou ce- ich en i onmen s,
s a ing in he second yea .
These pa e ns lend suppo o ou analy ical p edic ions, and impo an ly, hey a e also
consis en wi h he ou comes no ed in he ini ial BFK exe cise in which he obse ed pos -
con empo aneous expansiona y impac s (no ed as well, unde he gene al sample) we e
s onge o esou ce- ich s a es han o esou ce-poo s a es.
5 Conclusion
We ha e examined, bo h analy ically and empi ically, how echnological shocks in e ac
wi h na u al esou ce abundance o a ec g ow h in ou pu and inpu s. We o e ed a wo-
sec o g ow h model wi h non-neu al echnical p og ess and adjus men cos s o show ha
c oss-sec o di e ences in he deg ee o subs i u ion be ween capi al and labo can induce
co esponding di e ences in he con empo aneous and long- un eac ions o echnology
imp o emen s, mos no ably capi al-augmen ing ones. Using ou model and U.S. indus y
da a, we ha e compu ed elas ici y pa ame e s o di e en sec o s, e ealing ha he ESKL
is pe sis en ly highe in ex ac i e indus ies. We ha e also compu ed ime se ies o he
capi al- and labo -augmen ing echnology pa ame e s and employed hese in ou empi ical
analysis.
Fig. 5 Impulse esponse unc ions o he e ec o TFP shocks. The igu e p esen s he impac o echnol-
ogy shocks in e ac ed wi h esou ce endowmen s on he na u al loga i hms o eal pe capi a GSP, he
unemploymen a e, and eal pe -capi a in es men o e a 5-yea ho izon, wi h
95%
con idence in e als,
ollowing he me hod o local p ojec ions o Jo da (2005). The sample includes he 48 con inen al U.S.
s a es and co e s he pe iod 1963–2015
1 3
2180
Na u al Resou ces, Technology Imp o emen s, and G ow h
p
iy1/εi
i
[
(zki ki )1−1/µi+(zsi si )1−1/µi
]
µ
i
µi−1
(1
−
1
/εi
)
−
1
(zki ki )−1/µizki =
R,
(26)
p
iy1/εi
i
[
(zki ki )1−1/µi+(zsi si )1−1/µi
]
µi
µi−1
(1
−
1
/εi
)
−
1
(zsi si )−1/µizsi =ws
,
(27)
piy1
/εi
i (zni ni )
−
1
/µi
zni =wn ,
(28)
whe e
yi
is gi en by Eq. (24).
Combining he op imali y condi ions (26) and (27) gi es he op imal a io o physical
capi al o skilled labo , so ha
θ
i ≡zki k
∗
i
z
si
s
∗
i
=
(
zki
z
si
ws
R
)µ
i
(29)
Equa ion (29) implies ha
θi
inc eases in
zki
, dec eases in
zsi
, and is independen o
zni
.
In he same way, he op imali y condi ions (26) and (28) deli e he op imal a io o
physical capi al o unskilled labo , which gi es
ω
i ≡zki k∗
i
zni n∗
i
=
(
zki
zni
wn
R
)
εi
[
1+
(
zki
zsi
ws
R
)
1−µi
]
ε
i
−µ
i
µi−1
.
(30)
I is easy o deduce om equa ion (30) ha
ωi
inc eases in
zsi
and dec eases in
zni
. In
con as , since
zki
appea s wice in Eq. (30), we ob ain he i s de i a i e o know how
ωi
is a ec ed by changes in
zki
, namely
∂ωi
∂zki
=
(
zki
zni
wn
R
)
εi
zki
[
1+
(
zki
zsi
ws
R
)
1−µi
]
ε
i
−µ
i
µi−1−
1[
εi+µi
(
zki
zsi
ws
R
)
1−µi
].
(31)
This las exp ession is always posi i e.
To summa ize, gi en ha
εi>µ
i
, bo h equa ions (29) and (30) indica e ha he op i-
mal amoun o an inpu ela i e o ano he inpu , bo h measu ed in e iciency uni s, a ies
di ec ly wi h ela i e inpu p oduc i i y le els and in e sely wi h ela i e inpu p ices.
The p ices o physical capi al and skilled labo a e gi en exogenously. Howe e , he
wage a e o skilled labo is endogenous. F om equa ions (28) o (30), applied o he non-
ex ac i e sec o (
m
), we can de i e an exp ession o he wage o unskilled labo exp essed
as a unc ion o exogenous a iables and pa ame e s,
1 3
2187
F. Pe ez-Sebas ian e al.
w
n =znm
1−
(zkm
R)εm
[
1+(zkm
zsm
ws
R)1−µm
]
εm−µm
µm−1
1−1/µ
m
+
(zsm
ws )εm[(zsm
zkm
R
ws )1−µm
+1
]εm−µm
µm−1
1−1/µm
µm
µm−1
εm−1
εm
1
1−εm
.
(32)
The e o e, he unskilled-labo wage a e depends only on a iables and pa ame e s
ela ed o he non-ex ac ed sec o . I can be easily shown ha
wn
inc eases in
znm
,
zkm
,
and
zsm
. Recall ha , due o capi al-skill complemen a i y, he op imal e iciency uni s o
physical capi al and skilled labo ela i e o hose o unskilled labo inc ease in
zkm
and
zsm
. This, along wi h he ac ha he e is always some deg ee o complemen a i y be ween
he h ee inpu s, implies ha an inc ease in he p oduc i i y o any inpu aises he wage a e.
Subs i u ing he ixed inpu p opo ions implied by Eqs. (29) and (30) in o he long- un
p oduc ion unc ion, gi en by equa ion (24), we ob ain
y
i =
[ω
µi−1
µi
i +(ωi
θi )
µi−1
µi]
µi
µi−1
(
εi−1
εi
)
+1
εi
εi−1
zni n∗
i
,
(33)
o , al e na i ely,
y
i =
(
1+θ
1−µi
µi
i
)
µi
µi−1
(
1−1
εi
)
+ω
1−εi
εi
i
εi
εi−1
zki k∗
i
.
(34)
Mo eo e , compa ing his las exp ession o he sho - un p oduc ion unc ion, gi en
now by Eq. (23), we deduce ha
Ω
i =
(
1+θ
1−µi
µi
i
)
µi
µi−1
(
1−1
εi
)
+ω
1−εi
εi
i
εi
εi−1
.
(35)
Equa ion (35) indica es ha
Ωi
dec eases in bo h
θi
and
ωi
. The e o e, Eqs. (29) and
(30) imply ha
Ωi
dec eases in
zki
, as in he benchma k case. Howe e , in he new e sion
o he model, bo h
zni
and
zsi
also a ec i — he o me wi h a posi i e impac , while he
la e has an ambiguous impac .
In wha ollows, we assume ha
z
ki
zni
w
n
R>1
and
z
ki
zsi
w
s
R>1
, since
R
is cons an , bu
wages end o inc ease wi h he economy. We i s analyze he e ec s o p oduc i i y shocks
on long- un ou pu . In he long un, all he economy’s a ailable labo is used in p oduc ion.
The e o e, simila o wha happens in he benchma k model, because labo is di ec ed o
he mos p oduc i e ac i i y, gains in any inpu p oduc i i y mus esul in gains in ou pu .
Howe e , unlike in he benchma k model, since p oduc i i y shocks a e now sec o -speci ic,
1 3
2188
Na u al Resou ces, Technology Imp o emen s, and G ow h
long- un ou pu will end o inc ease mo e signi ican ly in he sec o whe e p oduc i i y
pa ame e s expe ience a la ge ise.
Ou nex ask is o de e mine how hese long- un ou pu gains induced by p oduc i i y
shocks a y wi h he elas ici ies o subs i u ion be ween inpu s (
εi
and
µi
). The e ec s on
long- un ou pu can be in e ed om Eq. (33), which is he ele an one because labo goes
back o ull employmen in he long un, ha is,
n∗
e +n∗
m =N
. The e o e, we can assume
ha
n∗
i
does no a y signi ican ly. We ocus on he elas ici ies o he ex ac i e ac i i y,
as we a e in e es ed in knowing, o example, how he di e en e ec s change when
εe
inc eases away om
εm
. F om Eqs. (29) and (30), we see ha
εe
does no a ec
θe
bu
ampli ies he posi i e e ec o
zke
on
ωe
. While his sugges s a s onge posi i e e ec , he
p esence o
εi
wi hin he CES s uc u e o he p oduc ion unc ion in oduces an addi ional
channel—he ea e e e ed o as he CES channel— h ough which changes in he elas ici y
o inpu subs i u ion in luence he o e all ou come, leading o ambigui y.
This ambigui y implies ha , as obse ed in he benchma k model, he e ec ope a ing
h ough inpu p opo ion esponses may domina e o some subse s o he inpu elas ici y
pa ame e s. Mo eo e , in mos cases, when analyzing he impac o
εe
and
µe
, he CES
channel will deli e he same ou come in ou analysis. The e o e, ou discussion will hence-
o h ocus p ima ily on he impac on inpu p opo ions, while keeping in mind ha he
esul s hold o speci ic alues o he elas ici y pa ame e s.
Hence, he esul s so a a e simila o hose o he benchma k model. An inc ease in he
elas ici y o subs i u ion be ween he capi al-skill bundle and unskilled labo in he ex ac-
i e indus y has a clea e ec on he inpu p opo ion esponses, con ibu ing o a s onge
posi i e impac o long- un ou pu o changes in
zke
. Howe e , his esul may hold only
o ce ain alues o
εe
.
In con as , an inc ease in
µe
makes
∂θe/∂zke
mo e posi i e and
∂ωe /∂zke
less posi-
i e. Equa ion (33) hen sugges s a weake posi i e e ec o capi al-augmen ing p oduc i -
i y on long- un ou pu when he elas ici y o subs i u ion be ween capi al and skilled labo
inc eases in he ex ac i e sec o .
I is also easy o show ha changes in he inpu elas ici y pa ame e s a ec he long- un
ou pu esponse o skill-biased p oduc i i y shocks (
zse
) in quali a i ely he same way as
hey do o an inc ease in
zke
.,47 Rega ding he esponse o
zne
, as
εe
ises,
ωe
alls mo e,
while
θe
does no a y. As a esul , he impac o
zne
weakens as
εe
inc eases, since capi al
ises less due o i s g ea e subs i u abili y wi h unskilled labo . Con e sely, he impac o
zne
on long- un ou pu channeled h ough inpu p opo ions s eng hens as
µe
ises.
We can hus d aw he ollowing conclusions. Fi s , i he ex ac i e indus y exhibi s a
highe elas ici y o subs i u ion be ween capi al and unskilled labo (i.e.,
εe>ε
m
), he e
may exis alues o
εe
and
εm
o which long- un ou pu in he ex ac i e sec o ises mo e
han in he non-ex ac i e indus y as a esponse o he same capi al- o skill-biased p oduc-
i i y shock. Second, i he ex ac i e indus y has a highe elas ici y o subs i u ion be ween
capi al and skilled labo (i.g., i
µe>µ
m
), he opposi e e ec may occu —namely, long-
un ou pu may inc ease less in he ex ac i e sec o han in he non-ex ac i e indus y o
47 This can be easily deduce by compa ing Eq. (30), which gi es
ωi
θi
=
(
zsi
zni
wn
ws
)
εi
[
1+
(
zsi
zki
R
ws
)
1−µi
]
ε
i
−µ
i
µi−1
.
(36)
1 3
2189
F. Pe ez-Sebas ian e al.
ce ain alues o
µe
and
µm
as a esponse o an inc ease in
zke
o
zse
. Thi d, highe alues
o
εe
a e less likely o ampli y he long- un ou pu impac when p oduc i i y shocks esul
om unskilled-biased echnical change. Finally, since all hese p oduc i i y shocks a e sec-
o speci ic, he impac will be s onge in he sec o expe iencing he g ea es p oduc i i y
inc ease.
Sho -Run E ec s
Le us now analyze he sho - un e ec s o which Eq. (34) is now ele an . The maximiza-
ion p oblem ha de e mines he op imal in es men in physical capi al is
max
xi
{
piΩi zki ki −R
∑
j=1
(1 −δ) −jxij −wn ni −ws si
}
(37)
subjec o he mo ion o capi al, which is de e mined by Eq. (3), and he cons an inpu
p opo ion condi ions gi en by Eqs. (29) and (30). Sol ing his p oblems deli e s he same
solu ion as he benchma k model, gi en by Eq. (20), bu now
Ψ
i
=p
i
Ω
i
−w
n
ωi zni −w
s
θi zsi
.
The e o e, ollowing he same logic as in he p oo o P oposi ion 3 in Appendix A, we can
ocus on he pa icula case whe e capi al adjus men cos s a e so s ong (i.e.,
ϕ
su icien ly
small) ha he change in he capi al inpu is negligible. In his scena io,
Ωi zki
gi es he
esponse o sho - un ou pu o p oduc i i y shocks.
F om Eqs. (29), (30), and (35), when
zki
inc eases, bo h
θi
and
ωi
inc ease, pushing
Ωi
down. This implies ha bo h skilled and unskilled labo dec ease pe uni o capi al.
The e o e, as in he benchma k model, capi al-augmen ing p oduc i i y shocks can educe
ou pu in he sho un i he dec ease in
Ωi
in absolu e alue is la ge han he inc ease in
zki
.
Skill-biased echnical change, on he o he hand, is less likely— hough no impossible—
o cause a dec ease in sho - un ou pu . As
zsi
ises,
ωi
inc eases, bu
θi
alls, making
he o e all impac on
Ωi
unce ain. This occu s because, while he amoun o unskilled
labo pe uni o capi al alls, he amoun o skilled labo inc eases. Finally, in he sho
un, he impac o
zni
on sec o
i
’s ou pu can only be posi i e—no e ha as
zni
ises,
ωi
dec eases,
θi
does no a y, and hen
Ωi
inc eases.
We also aim o examine how inpu elas ici ies a ec he s eng h o a po en ial nega i e
sho - un e ec on ou pu . The e o e, we ocus on how he impac o
zke
and
zse
a ies
wi h
εe
and
µe
. As
εe
inc eases,
∂θ
e
∂zke
emains unchanged, bu
∂ω
e
∂zke
inc eases, po en ially
s eng hening he nega i e e ec on sho - un ou pu . The in luence o
εe
on he sho - un
impac o
zse
is quali a i ely iden ical o ha o
zke
.
Con e sely, an inc ease in
µe
, aises
∂θ
e
∂zke
bu dec eases
∂ω
e
∂zke
, hus implying ha he
in luence o
µe
on he sho - un esponse o a
zki
shock is ambiguous. In con as ,
µe
has
a well-de ined impac on inpu p opo ion esponses o changes in
zsi
. When
µe
inc eases,
∂θ
e
∂zse
becomes mo e nega i e and
∂ω
e
∂zse
less posi i e, making a posi i e e ec o
zsi
on
sho - un ou pu mo e likely.
We can summa ize he sho - un esul s when he e a e s ong adjus men cos s o he
capi al inpu as ollows. Fi s , capi al-biased p oduc i i y shocks can nega i ely a ec sec-
1 3
2190
Na u al Resou ces, Technology Imp o emen s, and G ow h
o al ou pu in he sho un. Fu he mo e, as in he benchma k model, o ce ain alues o
εe
and
εm
, an inc ease in
εe
can ampli y his nega i e sho - un impac . The in luence o
µe
is less clea , as i a ec s he capi al-unskilled-labo and he capi al-skilled-labo a ios in
opposi e di ec ions. Second, skilled-biased echnical change is less likely o gene a e sho -
un ou pu losses due o he inc ease in he amoun o skilled labo . Thi d, unskilled-biased
echnical change canno educe ou pu in he sho un. Finally, since p oduc i i y shocks
a e sec o -speci ic, all hese e ec s will be mo e p onounce i p oduc i i y gains impac he
ex ac i e indus y mo e signi ican ly.
D. Es ima ing ESKL and he P oduc i i y Pa ame e s
We in oduce ime and sec o subindices when necessa y o allow
R
,
w
and
εi
o a y
ac oss ime and sec o s. Hence, Eqs. (2) and (10) become
y
e =
[
(zk ke )1−1/εe +(zn ne )1−1/εe
]
εe
εe −1
,
(38)
ke
n
e
=
(
zk
z
n )ε
e
−1(
we
R
e )ε
e
,
(39)
y
m =
[
(zk km )1−1/εm +(zn nm )1−1/εm
]
εm
εm −1
,
(40)
km
n
m
=
(
zk
z
n )ε
m
−1(
wm
R
m )ε
m
.
(41)
We can use hese o ob ain
z
n =yi
n
i (
wi ni
p
i
y
i )
εi
εi −1
, o i=
e, m,
(42)
and
z
k =yi
k
i (
Ri ki
p
i
y
i )
εi
εi −1
, o i=
e, m.
(43)
Finally, equalizing Eqs. (42) and (43) ac oss he wo sec o s deli e s
εm
εm −1=
ln
(
ym /km
ye /ke
)
−
ln(R
e
k
e
pe ye
)
ln(we ne
pe ye )ln
(
ym /nm
ye /ne
)
ln(Re ke
pe ye )
ln
(
we ne
pe ye )
ln
(
wm nm
ym
)−
ln
(
Rm km
pm ym
)
(44)
and
1 3
2191
F. Pe ez-Sebas ian e al.
εe
εe −1=
ln
(
ym /km
ye /ke
)
+εm
εm −1ln
(
Rm km
pm ym
)
ln
(
Re ke
pe ye
).
(45)
Using da a o
ye
,
ym
,
ke
,
km
,
ne
,
nm
,
we
,
wm
,
Re
and
Rm
, we can hus ob ain
an es ima e o
εm
om Eq. (44). Then, aking
εm
in o Eq. (45) gi es
εe
. Finally, subs i u -
ing
εm
and
εe
in o (42) and (43) gi es
zn
and
zk
. We can hus ob ain he ESKL o each
o he wo sec o s and he common labo -augmen ing and capi al-augmen ing p oduc i i-
ies, a any gi en poin in ime.
Mo e speci ically, he EUKLEMS p o ides da a o he a iables equi ed o pe o m he
es ima ion. Ou pu a iables
ye
and
ym
, and he s ocks
ke
,
km
,
ne
and
nm
a e di ec ly
a ailable in he da ase . We compu e he sala ies
we
and
wm
by di iding o al labo com-
pensa ions by o al hou s wo ked (
ni
). We compu e
Re
and
Rm
by di iding o al capi al
compensa ions by he o al capi al s ock (
ki
). This is consis en , o example, wi h he com-
pu a ions unde aken by Caselli and Coleman (Caselli e al. 2006).
E. Da a
We use an annual s a e-le el panel ha , unless o he wise speci ied, co e s he 48 con inen al
U.S. s a es o he pe iod 1963–2015. Real a iables a e exp essed in 2009 p ices. Desc ip-
i e s a is ics o all a iables a e p esen ed in Table A1.
Table A1 Desc ip i e s a is ics
1 3
2192
Na u al Resou ces, Technology Imp o emen s, and G ow h
Va iable de ini ions
Resou ce endowmen : Reco e able s a e s ocks o oil and na u al gas (c oss-sec ional),
no malized by a e age s a e income (a e aged o e 1958–2008). Alaska and Hawaii a e
excluded. Sou ce: James (2015)
Real pe -capi a G oss S a e P oduc (GSP): Real G oss S a e P oduc di ided by s a e popu-
la ion. Sou ce: U.S. Census Bu eau.
Popula ion: S a e popula ion. Sou ce: U.S. Census Bu eau.
Real pe -capi a ax a es: S a e ax e enues di ided by s a e popula ion. Sou ce: U.S. Cen-
sus Bu eau.
Real pe -capi a go e nmen expendi u es: To al expendi u es o s a e go e nmen di ided
by s a e popula ion. Sou ce: U.S. Census Bu eau.
1 3
2193
F. Pe ez-Sebas ian e al.
Real pe -capi a s a e ou s anding deb : To al ou s anding deb o s a e go e nmen di ided
by s a e popula ion. Sou ce: U.S. Census Bu eau.
Pa y a ilia ion o go e no : An indica o o he pa y o he go e no ; 0 = Republican,
1 = Democ a , 0.5 = non-majo pa y go e no . Sou ce: Ma y and G ossman (2016).
Inequali y: Theil Index measu e o income inequali y. Sou ce: F ank (2009).
Unemploymen a e: S a e unemploymen a e. Sou ce: U.S. Census Bu eau.
Real pe -capi a capi al s ock: S a e capi al s ock di ided by s a e popula ion. Sou ce:
Ga o alo and Yama ik (2002), including an ex ension o i a ailable a he second au ho ’s
homepage.
Fe nald TFP shocks: Agg ega e, na ional TFP shocks, agg ega ed o an annual le el.
Sou ce: Fe nald (2014).
Fede al ax shocks: Na a i e-based ede al ax shocks, agg ega ed o an annual le el, and
no malized by U.S. GDP; a ailable up o 2007. Sou ce: Rome and Rome (2010).
Business cycles: An indica o o whe he he U.S. economy is in a ecession. Sou ce: U.S.
Fede al Rese e.
News shocks: In es men Speci ic Technology (IST) news shocks, agg ega ed o an annual
le el; a ailable up o 2011. Sou ce: Ben-Zee (2018).
Mone a y shocks: Mone a y policy shocks a-la Rome and Rome (2004); a ailable o
1969–2007. Sou ce: Ten ey o and Thwai es (2016).
JPT TFP shocks: Se ies o TFP shocks de i ed om Jus iniano e al. (2011).
BS TFP shocks: Se ies o TFP news shocks de i ed om Ba sky and Sims (2011).
FORD TFP shocks: Se ies o TFP shocks de i ed om F ancis e al. (2014).
Zk: Capi al-augmen ing echnology shocks. Compu ed om he model, as desc ibed in he
ex , is-à- is da a om he EUKLEMS da ase . Sou ce: O’Mahony and Timme (2009).
Zn: Labo -augmen ed echnology shocks. Compu ed om he model, as desc ibed in he
ex , is-à- is da a om he EUKLEMS da ase . Sou ce: O’Mahony and Timme (2009).
GSP sha e o manu ac u ing: Sha e o s a e manu ac u ing sec o in G oss S a e P oduc .
Sou ce: U.S. Census Bu eau.
1 3
2194
Na u al Resou ces, Technology Imp o emen s, and G ow h
GSP sha e o se ices: Sha e o s a e se ices sec o in G oss S a e P oduc . Sou ce: U.S.
Census Bu eau.
GSP sha e o ag icul u e: Sha e o s a e ag icul u e sec o in G oss S a e P oduc . Sou ce:
U.S. Census Bu eau.
GSP sha e o wholesale: Sha e o s a e wholesale ade sec o in G oss S a e P oduc .
Sou ce: U.S. Census Bu eau.
F. VAR analysis
In Sec . 4.7 o he pape , we ha e examined he dynamic pa e ns ollowing he me hod o
local p ojec ions (Jo da 2005). To examine he obus ness o he obse ed pa e ns o he
ype o es ima ion me hod, we unde ake an equi alen es ima ion unde a VAR amewo k.
Speci ically, we es ima e
∆(ou come)
i,∆( −1, )
=α+β(ou come)
i, −1
+γ( esou ce)
i
+δ( p) +
j=4
∑
j=0
θj( esou ce ∗ p)i, −j+ηi+ν +ϵi, . (46)
He e
ou come
again deno es each o he h ee ou come a iables. The esul s a e p e-
sen ed in Table A2. Columns (1)-(3) examine he cases o GSP, unemploymen , and capi al,
espec i ely. The obse ed pa e ns a e quali a i ely simila o hose unde he Jo da (2005)
me hod ou lined in Sec . 4.7. We no e ha upon impac , TFP shocks con ac ou pu and
labo mo e s ongly in esou ce- ich han esou ce-poo s a es, bu he e a e no such di e -
en ial impac s on capi al. Howe e , speci ically om abou he second o hi d yea s, ou pu ,
labo , and capi al expand mo e s ongly in hose same, ini ially con ac ed, esou ce- ich
s a es. These esul s indica e ha he main obse ed pa e ns a e obus o his al e na i e
es ima ion me hod.
Table A2 Resou ce endowmen s and echnology imp o emen s (VAR analysis)
1 3
2195
F. Pe ez-Sebas ian e al.
Funding Open access unding p o ided by Heb ew Uni e si y o Je usalem. Open access unding p o ided
by Heb ew Uni e si y o Je usalem. Fidel Pe ez Sebas ian acknowledges unding om he Spanish Minis y
o Science and Inno a ion, P ojec s PID 2019-111208 GB-I00 and PID2023-153032NB-I00.
Da a a ailabili y The da a used in his s udy is a ailable om he au ho s upon eques .
Decla a ions
Con lic o In e es The au ho s ha e no con lic s o in e es o disclose.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License,
which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as
you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons
licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e
included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial.
I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed
by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he
copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
Re e ences
Acemoglu D (2002) Technical change, inequali y and he labo ma ke . J Econ Li 40:7–72
1 3
2196
