Demographic Dynamics and Long‐RunDevelopment ... · debate is concerned with the question of whether the delayed recovery recently experienced reflects a reduction in long‐run

DemographicDynamicsandLong‐RunDevelopment:PerspectivesfortheSecularStagnationDebate*

MatteoCervellatiUniversityofBologna

IZA,Bonn

UweSundeUniversityofMunichCEPR,London

IZA,Bonn

June3,2015

KlausF.ZimmermannIZA,Bonn

UniversityofBonn

CEPR,London

Draft prepared for the IMF Economic Review conference on “SecularStagnation,GrowthandRealInterestRates”attheEUI,18‐19June2015

AbstractThispaper takesaglobalperspectiveon the recentdebateabout secular

stagnation. The analysis is motivated by the observation of the interplaybetween the economic and demographic transition that has occurred in thedevelopedworldoverthepast150years.Totheextentthathighgrowthratesinthepasthavepartlybeentheconsequenceofchangesduringtheeconomicanddemographictransition,growthislikelytobecomemoremoderateasthetransition iscompleted.Atthesametime,asimilartransition ison itsway inmostdeveloping countries,withprofound consequences for theprospects fordevelopment in thesecountries,butalso forglobalcomparativedevelopment.Theevidencepresentedheresuggeststhatthelong‐rundevelopmentdynamicshave potentially important implications for the prospects of human andphysicalcapitalaccumulation,andthequestionofsecularstagnation.JEL‐classification:C54,E10,J11,J13,J18,N30,O10,O40

Keywords: secular stagnation, long‐term development, income growth,demographictransition

*TheauthorsthankLukasRosenbergerforexcellentresearchassistanceandVickiFinnforeditorialcomments.

1

1 Introduction

Theexperienceofthefinancialcrisisof2007anditsfalloutintermsofslowgrowthin

Western countrieshas initiatedanongoingdebateaboutmedium‐runmacroeconomic

outlooks. Inaspeechat the IMF in fall2013,LarrySummersused thephrase “secular

stagnation” when pointing out the surprising resistance of US GDP to return to its

potential, despite substantial financial andmonetary policy interventions. Instead, he

argued, US GDP fell further behind its potential, employment did not increase

substantially, inflation remained low, and capacity utilization did not become tight.

Drawing parallels to Japan’s development since the 1990s, this discussion raised

concernswhether theUS, or indeed the entire developedworld,were up for “secular

stagnation.” Summers’ discussion was mainly concerned about the appropriate

macroeconomic policies. The possibility that the short‐term real interest rate that is

consistent with full employmentmight have fallen to zero or even to negative levels

posesachallengetotraditionalmonetarypolicy.Atthesametime,thismightcallfora

new role of fiscal policy,while the debt crisis put limits on the potential for creating

fiscalstimuli.1

The debate about the “new secular stagnation hypothesis” has gainedmomentum

andmanyfacetshaverecentlybeenadded.2Thecoreofthedebateisaboutwhetherthe

real interest rate that is consistentwith full employment has indeed fallen to zero or

negative levels, focusing on the demand and supply of capital. At a broader level, the

debate is concerned with the question of whether the delayed recovery recently

experiencedreflectsareductioninlong‐rungrowthpotential(aprolongeddropinGDP

below its long‐run potential) , a phenomenon of delayed recovery from cyclical

fluctuations,oraone‐offcrisis‐relateddrop inpotentialoutput(TeulingsandBaldwin,

2014b).MostofthisdebatehasfocusedoncapitalmarketsandinterestratesintheUS

economy,withsomereflectionsonEuropeandJapan.

Thispaperoffersabroaderviewonthesequestionsfromtheperspectiveofunified

growththeory,accordingtowhichthedemographictransitionisthecentralmechanism

1SeeSummers(2014)foradetailedaccountofthisview.2See therecentbookeditedbyTeulingsandBaldwin(2014a), thesessionon“TheEconomicsofSecular Stagnation” at the Allied Social Science Associations Conference 2015, and the OECD(2015).

2

behind the transition from stagnation to growth. Unified growth theories model the

entire process of the endogenous exit from long‐term economic stagnation and the

following transition and convergence to balanced growth. Populations becoming

increasinglymoreeducated,wealthy, andolder characterize this transition.Througha

historical perspective, economic performance has been closely linked to demographic

development. Modern economic growth typically accelerated at the same time as the

demographictransitionoccurred,withitsseculardeclineinmortality,theemergenceof

masseducationandunprecedentedhumancapitalaccumulation.

We adopt theperspective of unified growth theory to obtainnew insights into the

patterns of cross‐country comparative development and to discuss the existence and

determinants of a secular decline in growth. The analysis thereby extends that of

Cervellati and Sunde (2015a),which demonstrated that a simulated prototypeunified

growthmodelnotonlymatchesthenonlinearlong‐rundevelopmentdynamicsofagiven

country but also provides important insights into the patterns of comparative

developmentdifferencesintheworldtoday.

Weputforwardtheargumentthattheexplicitconsiderationofthelong‐termglobal

development patterns may be necessary, or very useful at the least, to get a deeper

understanding of the growth prospects in different countries and regions during the

coming years and decades. The main hypothesis underlying our analysis is that the

nonlinear dynamics of developed countries’ long‐run development also provide an

appropriate qualitative description of less developed countries’ development path,

whose economic anddemographic transition is delayed. As a first step,we revisit the

stylizedfactsonincomegrowthinabroaderperspectivebothintermsoftimehorizon

and units of observation. The cross‐country panel data for the last fifty years suggest

increasingincomesandrelativelystableaveragegrowthratesworldwide.However,the

patterns differ across different world regions. Advanced countries appear to have

experiencedagrowthslowdowninthe lasttwodecadeswhereasdevelopingcountries

inLatinAmericaandAfricahavenot,orhaveseen lessofaslowdown.Thesepatterns

arecoupledwithsimilarandunevendemographicdevelopmentandeducationtrends.

Wethenuse insights fromasimpleprototypeunifiedgrowththeoryandshowthat

thestylizedpatternsarecompatiblewiththeviewthatdifferentcountriesfollowsimilar

3

(nonlinear)economicanddemographicdevelopmentpathsbutdiffersubstantiallyinthe

timing of the takeoff from stagnation. The resulting conceptual framework is used to

derive testable predictions for the evolution ofmortality, human capital accumulation

(and their interactions) and income growth during the different development phases.

Resultsfromcross‐countrypanelregressionssupportthemainpredictions.

Our contribution relates to two papers by Gordon (2012, 2014a) emphasizing the

importanceofaccountingfor“headwinds”inUSgrowthpotentialandgrowthprospects,

meaning unfavorable changes in environmental conditions, particularly demographic

changes, education, and globalization. These headwindsmight unfold negative effects

thatcouldbestrongerthanproductivity improvements.Whilesomehavecriticizedhis

papers for being too pessimistic about the scope for productivity improvements, his

emphasison theotherheadwindshasnot resonated throughout the literature.3 In this

paperwecomplementandextendGordon’sUS‐focusedviewwhiletakingasteptowards

a more global view of comparative development, with a specific focus on the role of

demographicchangeandeducationfromtheunifiedgrowththeoryperspective.

The results document consistent changes in income growth during the process of

long‐termdevelopmentandsuggestthattheobservationofaseculardeclineingrowth

in OECD countries can be related to the global process of long‐term development.

Additionally,thesechangesarenotsolelyabusinesscyclephenomenon,oramatterof

delayedrecoveryfromtherecentcrisis.Thisperspectivecanhelpprovidenewlighton

the question of an existent growth slowdown in middle‐income countries (see, e.g.,

Eichengreen,Park,andShin,2013)by focusingon the interplaybetweentheeconomic

and demographic transition and the emergence of nonlinearities in the development

process.Theanalysisalsosuggeststheneedforfurtherresearchontheglobalstructural

developmentprocess in the economic anddemographicdomains. The implicationsof

thisprocesswill likelygain importanceastheworldcontinuesto integrate intermsof

trade,capitalflows,andhumanmigration.Inshort:integratingtheshort‐term(business

cycle) and the long‐term (unified growth) perspective appears to be a needed, and

potentiallyveryfruitful,directionforacademicresearchandpolicyanalysis.

3See,e.g.,thediscussionsbyGordon(2014b)andMokyr(2014a,2014b).

4

The paper is structured as follows. Section 2 gives an overview of the global

economic and demographic development patterns. Section 3 presents the conceptual

framework and derives empirical implications. Section 4 presents empirical evidence

andSection5concludes.

2 GlobalPatternsofEconomicandDemographic

Development

Asdiscussed in the Introduction, the recentdebateonpossible secular stagnationhas

mainly focused on developed countries, particularly theUS, Europe and Japan. In this

sectionwebeginouranalysisbydocumentingthestylizedpatternsofglobaleconomic

anddemographicdevelopmentinthelasthalfofthecentury.4Figure1plotstheannual

growthratesofGDPpercapitafortheworldovertheperiod1950–2010,togetherwitha

5‐yearmovingaverage.Thefigure’sfirstpanelsuggeststhat,fromaglobalperspective,

the average growth rate has been around 3 percent per year, with sizable variation

duringperiodsof globalboomsand recessions.However, there isno indication foran

obvious downwards trend in growth over the last ten years. Overall average growth

rateshavebeenfairlystableoverthis longperiodandifanything,theyhaveincreased

lately.Also,thevariabilityingrowthrateshasnotsloweddown,infactithasincreased

slightlyduringthelast15years.

The secondpanel replicates the sameexercise, but restricting attention toonly

developed countries (Europe and Western offshoots). In the developed countries,

incomegrowth indeedseemstohaveexperiencedaslowdown. Inaglobalperspective

this slowdown does not appear to be confined to the recent post‐crisis period, even

thoughthecrisishasclearlyacceleratedit.Finally,thefigure’sbottompanelsshowthe

average growth rates for Latin America and Sub‐Saharan Africa where, if anything,

incomegrowthappearstohaveacceleratedinthelastoneortwodecades,withamore

visible trend in Africa. A very similar picture emerges for total factor productivity, as

Figure2documents.

4ThedataaretakenfromthestandardsourceslistedinTable7intheAppendix.

Taki

develop

provide

trajecto

F

ing a broa

pment in s

es a more

oriesoflog

Figure1:G

Figure

der perspe

several diff

e long‐run

GDPperc

lobalGrow

e2:Global

ective of d

ferent econ

perspectiv

apitainlev

5

wthPerform

GrowthPe

evelopmen

nomic and

ve of econ

velsoverth

mance:GDP

erformance

nt, Figure 3

demograp

nomic dev

hesametim

Ppercapita

e:TFP

3 displays

phic dimen

velopment

meperiodf

a

the traject

nsions. Figu

and displ

from1950

tories of

ure 3(a)

lays the

to2010

(using5

havebe

develop

visible

arebyf

themid

Figu

dynami

has bee

1980s.

immigr

America

downw

growth

thebeg

5‐yearaver

eensteadil

ped countr

inLatinAm

farthelow

d‐1990s.

(a)Lo

(c)Life

ure3(b) co

icsinterm

en slowing

Populatio

ation,and

ancountrie

ward trend

ratesarec

ginningfert

rages).Thi

y increasin

ries,which

mericanco

west,eventh

gIncomeper

eExpectancy

Figu

omplement

sofpopula

g down sin

n growth

manyofth

es,populat

began and

comparably

tilitytransit

sfigureillu

ngat least

exhibit th

untries.Th

houghther

rCapita

atBirth

ure3:Globa

ts the econ

ationgrowt

ce the 196

is lowest

hesehavese

tiongrowth

d has cont

yhighand

tion,butal

6

ustratestha

sincearou

ehighest l

hereisless

reappears

alDevelop

nomicdeve

wthrates.W

60s, withm

t in devel

eenfertility

hrateswer

tinued. In

haveonlyr

soduetoe

atonavera

und1960.T

levels of liv

developm

tobeanup

(b)P

(d

mentPatte

elopmentb

Whilepositi

more pron

oped coun

yfallbelow

reathighle

Africa, on

recentlybe

emigration.

ageacrosst

The increa

ving standa

entinAfric

pwardtend

opulationGr

)ShareSkille

erns

y graphing

ve,worldp

ounced red

ntries, des

wreplacem

evelsuntil

the other

eguntodec

Takentog

theglobe,i

se isvisibl

ards, but i

ca,wherei

dencybegin

rowth

ed

g thedemo

population

ductions si

spite consi

mentlevels.

the1960s,

r hand, pop

cline,mostl

gether,thep

incomes

le in the

t is also

incomes

nningin

ographic

ngrowth

ince the

iderable

InLatin

,whena

pulation

ydueto

patterns

7

inFigures3(a)and3(b)appearconsistentwiththeviewthateconomicdevelopmentin

termsof income levelsbeganmuchearlier inWesterncountriesthan inotherpartsof

theworld,withdevelopingregionssuchasLatinAmericaandAfricaexhibitingsimilar

patterns,butwithasubstantialdelay.

Anotherdimensionof long‐rundevelopmentpatternsrefers todemographicrather

thaneconomicdevelopment.Figure3(c)showstheimprovementsin lifeexpectancyat

birthovertheperiod1950–2005.Ataglobalscalemortalityhasfallenacrossallages,for

men andwomen alike, over the past 40 years, with themost pronounced decline for

mortality at young ages. Adult mortality has also fallen, mainly thanks to the

epidemiologicaltransitionleadingtobetterhealthtechnologyandbetteraccesstohealth

care anddiseaseprevention.The figure shows thisbyplotting the standardaggregate

measureoflifeexpectancyatbirth.Whenagainconsideringdifferentpartsoftheworld,

it isclear that the increase in longevity(likethe increase in income)hasbeenaglobal

phenomenon. In OECD countries, life expectancy was already at the highest levels

worldwide in the 1960s. Since then the increase in longevity has continued, but at a

slower pace, in particular due to declines inmortality at older ages; infant and child

mortalityhadalreadybeenapproachingverylowlevelssincetheearly1990s.

LatinAmericadisplaysasimilarpicture,butsomewhatdelayed.Adultaswellaschild

and infantmortalitywereat substantiallyhigher levels than in theOECD in1970,but

since then mortality has fallen substantially. Life expectancy has increased from just

above50inthe1950stoaround70todayandchildandinfantmortalityhaveconverged

toverylowlevels,yetnotquiteaslowasintheOECD.ContrarilyAfricashowsadifferent

picture:adultmortalityratesweretwiceashighandchildmortalityalmostfourtimesas

highasintheOECDin1970.5Lifeexpectancywasaround40yearsinthelate1950s.Also

the dynamics are different. Infant and child mortality have fallen substantially, but

remain at levelsmany timeshigher than those in theOECDor even in LatinAmerica.

Adultmortalityhasnotevenshownamarkeddeclineovertheperiod1970–2010,with

the HIV/AIDS epidemic in the 1990s leading to an increase in adult mortality that is

unparalleledintheworld.Thisepidemicismainlyresponsiblefortheslowdowninthe

increaseinlifeexpectancyinAfricancountries,andevenintheworld,duringthisperiod.

5Foradditionalevidenceongender‐specificmortalityandinfantmortalityrates,seeFigure6.

8

Figures1,3(a),3(b),and3(c)togetherillustratetheexistenceofsimilarpatternsof

economic and demographic development in different regions, withWestern countries

andAfricancountriesattheoppositeextremesofthespectrum.Westerncountrieshave

experiencedaslowdown ineconomicdevelopment,populationdynamics,aswellas in

health improvements;meanwhilethedevelopingcountriesappeartobe justbeginning

theirtransition,showingacceleratingeconomicanddemographicdevelopment.

Historically,Europe’seconomictakeofffromstagnationtogrowthwascloselylinked

not only to increased longevity and a slowdown in population growth, but also to an

accelerationineducationattainment.Theglobalviewreplicatesthispattern,withglobal

economic and demographic development changes going hand in hand with acquiring

education and human capital. Figure 3(d) displays this pattern using the population

shareaged25andolderwithsomeformalschoolingasthebasicmeasureofeducation.

Thefiguredocumentsasubstantialincreaseineducationatthegloballevel.Thisistrue

forthedevelopedcountries,whereinparticular,thesharesofadultswithsecondaryand

tertiary education have been increasing substantially over the past 50 years. Yet the

dynamicsineducationappeartoslowdowntowardstheendoftheobservationwindow,

with almost all individuals above the age of 25 having had some schooling, and the

averageschoolingattainmentis12years.InLatinAmerica,theincreaseineducationis

evenmorepronounced. InAfrica theeducation increaseonlybeganduring the1980s,

andhascontinuedatsubstantialspeed,despitesomeattenuationasoflate.

Very similar patterns emergewhen looking at alternativemeasures of education.6

When considering the developments of education in different parts of the world, the

patterns closely parallel those in mortality and population dynamics. In the OECD

countries,theshareofadultswithoutformalschoolinghadalreadybeenverylowinthe

1950s,buthassincefallentoessentiallyzero.Atthesametime,eventheshareofadults

with only primary education has been falling, while the shares with secondary or

tertiary education have been increasing rapidly,with tertiary education exhibiting an

almostexponentialpattern.EducationlevelsweremuchlowerinLatinAmericain1970.

Alsothere,however,educationhasbeenincreasingsubstantially.InAfrica,theprocess

issubstantiallydelayed.Whereastheshareofadultswithnoformalschoolinghasbeen

decreasingovertime,theshareofadultswithonlyprimaryeducationisstillincreasing.

6SeeFigure7intheAppendix.

Both se

modera

As

longevi

thatthe

the old

parallel

beentr

tothed

the pas

been es

levels.

Adi

consum

elderly

canbes

Purchas

upward

reflects

countri

econdary a

ate,particu

countries

ty,coupled

epopulatio

d‐age depen

lwiththei

endingupw

developedc

st 65 years

stablished,

(a)Shareof

Figure

irectconse

mptionatol

consumpti

seeninFig

sing Powe

dsataglob

s the incre

es,incontr

and tertiar

ularlyforte

undergo

dwiththe

onages.Fig

ndency rat

increasein

wardssince

countriesw

s. In contra

the old‐ag

fthePopulati

4:Old‐age

quenceofp

ldages.Int

ion,reducin

gure4(b),w

er Parity).

ballevel,th

ease in th

rast,exhibi

ry educatio

ertiaryeduc

the demo

reductioni

gure4(a)ill

tio, i.e., the

lifeexpect

e1950.Ifa

wheretheo

ast, inAfric

ge depende

ionAged65+

Dependen

population

tuitively,as

ngthescop

whichdepic

While on

hedevelope

e old‐age

tanincrea

9

on are on

cation.

ographic tr

inpopulati

lustratesth

e share of

tancy,old‐a

anything,th

old‐agedep

ca,where

ency ratio

+

ncyRatiosa

nagingisth

spopulatio

peforcapit

ctstheshar

n average

edcountrie

dependen

asingtrend

the rise, b

ransition,

iongrowth

hisagestru

the popul

agedepend

hereisasl

pendencyr

the demog

has been

(b)Shareof

andGrossC

hegreatern

onsage,lar

talformatio

reofgross

capital fo

esexhibita

cy ratio. L

incapitalf

but the inc

the sustai

hduetolow

uctureimpl

ation aged

dencyratio

ightaccele

atiohasalm

graphic tra

relatively s

f GrossCapita

CapitalForm

needforre

gerfractio

onattheag

capitalfor

rmation h

acleardow

Latin Amer

formation.

crease is st

ined incre

werfertilit

licationby

d 65 and o

osworldwi

eration,ma

mostdoub

ansition ha

stable at v

alFormation

mation

esourcesto

nsofincom

ggregatelev

rmation(at

has been t

wnwardtre

rican and

till very

eases in

ty,imply

plotting

older. In

idehave

inlydue

ledover

s barely

very low

ofinance

megoto

vel.This

tcurrent

trending

endthat

African

10

3 PatternsofLong‐RunDevelopment:UnifiedGrowthand

DelayedTransitions:AConceptualFramework

Howcanwecombinetheseempiricalpatternsinacoherentframeworktoaddressthe

secular stagnation question? This section’s approach is to consider the long‐run

development process at the global level from the perspective of a prototype unified

growth model of the economic and demographic transition. Unified growth models

capture the underlying forces behind different dimensions of economic and

demographic development, typically focusing on one country’s time series evolution.

Whathasbeenlessappreciatedintheliteratureisthat, fromtheperspectiveofsucha

model, comparative development differences across the world can be seen as the

consequenceofdelaysinthetransitionalongotherwiseidenticaldevelopmentpaths.To

illustrate this point we use Cervellati and Sunde’s (2015a) calibrated version of the

prototype unified growth model, which characterizes and simulates the endogenous

evolutionofincome,demographicconditions(particularlymortality)andeducationover

longtimeperiods.

Theanalysisisbasedonasimpledynamicgeneralequilibriumframeworkwithinter‐

temporal spillovers. To fix ideas, we build on Cervellati and Sunde’s (2015a) model,

notingthattheprecisemechanismsbehindtheworkingofthismodelarenotcrucialfor

theargumentandthefollowinganalysis.Thebasisisanoccupationalchoiceframework

in which individuals decide to acquire skilled or unskilled human capital, as well as

decide their fertility.Acquiring skills, aswell asbringingupchildren in termsofbasic

needsandeffortspentoneducatingthechildren,aretime‐intensiveactivities.However,

longevity, the length of adult life, is limited (and individuals are aware of this). Since

becomingskilledinvolvesastrictlylargercostintermsoftimethanbecomingunskilled,

the returns for becoming skilled in terms of income must be sufficiently high for an

individualtobearthisinvestment.Ontheaggregate,skilledandunskilledhumancapital

areusedtoproduceaconsumablecommodity.The technologyused in thisproduction

process is given, and wages for skilled and unskilled human capital are determined

competitively.

11

Thegeneralequilibriumallocationisthengivenbyaggregateconditions,reflectedby

longevity and technology, as well as by individual decisions. The population’s skill

choices and fertility decisions are mutually consistent with quantities and prices, in

terms of aggregate stocks of skilled and unskilled human capital as well as the

correspondingwagesforskilledandunskilledlabor.Duetothedifferentialtimecostfor

education,andbecauseofdecliningmarginalproductivityofeachinput,itturnsoutthat

the equilibrium relationship between the share of skilled individuals and longevity is

nonlinear.Inparticular,forlowlongevity,theshareofskilledissmallasaconsequence

ofthetimecostforbecomingskilled;thisimpliesthatonlyasmallpartofthepopulation

receivessufficientlifetimeearningsfrombecomingskilled,e.g.,becausetheyhavehigher

cognitive skills and thus higher productivity when skilled. Consequentially, large

increasesinlongevityareneededtoinducelargerandlargersharesofthepopulationto

acquire skilled human capital. In this range, the equilibrium relationship between life

expectancy and the share of skilled is convex. If longevity is large, decreasing relative

marginal productivity of skilled human capital compared to unskilled human capital

implies that increasingly larger improvements in longevity are needed to induce even

higherpopulationsharestobecomeskilled.Thisimpliesaconcaverelationshipbetween

longevity and the share skilled, so that overall the equilibrium relationship between

longevityandtheshareskillediss‐shaped.

Now we consider this model in an overlapping generations setting. Through

intergenerationalspillovers,theparentgeneration’seducationcompositionaffectstheir

respectivechildren’s longevityandtechnology, intermsoftherelative importanceand

productivity of skilled and unskilled human capital. In such a setting, endogenous

improvementsinlongevityandskill‐biasedtechnologicalchangearelinkedtotheshare

ofskilled inthepreviousgeneration.Asaconsequenceoftheshapeoftheequilibrium

relationshipbetween theshareskilledand longevity, it spansadynamical system that

exhibitsnonlinearities.

Inasituationwheretechnologyisnotsufficientlydeveloped,skilledhumancapitalis

notveryproductiverelativetounskilledhumancapital,implyingrelativelylowreturns

for becoming skilled. At the same time longevity is low, imposing a large cost for

becomingskilled.Asaconsequence, theequilibriumshareofskilledindividuals is low,

12

and thedynamic spillovers arenot very strong. If over time technology improves and

thusincreasesthereturnstoskills,theeducationcompositionshiftsveryslowlytowards

a higher share of skilled across generations. Ultimately, however, at some point the

subsequentandmutuallyreinforcing improvements inskillcomposition, longevityand

technology trigger a transition in the education composition. Large shares of the

population start becoming skilled, with corresponding bidirectional feedbacks with

longevityandtechnology.Thistransitionaffectsalldimensionsofindividualdecisions,in

termsof skill acquisitionand fertility choice, andwithin fewgenerations theeconomy

converges toabalancedgrowthpath,which resembles thatof a standardendogenous

growthmodel.

The model exhibits no corner solutions and the dynamic equilibrium path’s

simulationisrelativelystraightforward.Thesimulatedmodelgeneratesdatathatallow

for a systematic quantitative investigation of long‐run development in all central

dimensions,includingeducation,fertility,longevityandincomepercapita.Inparticular,

asCervellati andSunde (2015a) show, thisdevelopmentpath’spatternscloselymatch

thehistorical time series for countries such as SwedenorEngland. From our paper’s

perspective, however, the more important fact is that differences in country‐specific

features can lead to a transition delay, without affecting the transition dynamics

qualitatively. Cervellati and Sunde (2015a) give the example of differences in geo‐

climatologicalconditionsthatimplydifferencesinthediseaseenvironmentthatgoverns

longevity.Everythingelseequal,alowerbaselinelongevity(withanyhealthtechnology

andotherfactorsrelatedtodevelopmentabsent) leadstoadelayinanotherwisevery

similartransitioncontext.Figure5illustratesthisbycontrastingtwomodeleconomies,

abaselineeconomy(correspondingtoSweden)andasyntheticeconomythatisidentical

to thebaselineeconomyexcept that itexhibitsahigherextrinsicmortality,and thusa

latertransition.7Thedelayinthetransitionissubstantial.

7Theextrinsic longevity, theaverage longevityof anadultwhohas survived toage5without

access to any health care or other factors—thus resembling the situation in an entirelyundeveloped state of the world—is assumed to be 45 years in the baseline model. The highmortalityscenariocorrespondstothesameenvironment,butwithanextrinsiclongevityofonly40years.SeeCervellatiandSunde(2015a)forthecalibrationandsimulationdetails.

Figure

unified

develop

Firs

can be

dimens

relative

slowdo

country

(interm

the tim

resultin

differen

develop

earlier

5 provides

growththe

pmentpatt

Figure

t,thedyna

interprete

sionsthate

elyrapidtr

wn in the

ygroupsor

msof incom

ming and fo

ng from th

ntworldre

pment.Fou

will also e

s four mai

eoryandfr

terns.

5:TheRole

micevolut

ed as resu

eventually,

ransitionto

processof

rglobalreg

me,mortal

or some reg

he observe

egionswere

urth,andcr

experience

in insights

romapplyin

eofTransit

ionofincom

ulting from

afteralong

osustained

f convergin

gionsappea

ityandedu

gions of th

ed delays,

eindiffere

rucially,the

e an earlie

13

that one

ngthemto

tionDelays

me,popula

m dynamic

gperiodof

dgrowth.U

ng to theb

artofollow

ucation),a

heworld, s

it appears

entphases

esedifferen

r slowdow

can extrac

theinvest

sforCompa

ationdynam

interactio

f slow(quas

Ultimately,

alancedgr

wverysimi

althoughw

substantial

s plausible

ofapotent

ntphasesim

wn in trans

ct from the

igationofg

arativeDev

mics,morta

ons betwee

si‐stagnant

thetransit

rowthpath

ilardevelop

ithsubstan

delays in

e that alrea

tiallyquite

mplythatc

sition dyna

e recent re

globalcomp

velopment

alityanded

en these d

t)growth,l

tionexper

h.Second,d

pmenttraj

ntialdiffere

the takeof

ady in 19

esimilarpr

countriesta

amics, part

esults in

parative

ducation

different

leadtoa

iencesa

different

ectories

ences in

ff. Third,

60 the

rocessof

akingoff

ticularly

14

regarding income growth rates. The slowdown occurswhen the region is completing

demographic and education transitions and approaching a balanced growth path

characterizedbyuniversaleducation,lowmortality,andagingpopulationswithgreater

longevity.

Qualitatively, the simulateddatapatterns closely resemble those in the actualdata

shown in the previous section. Of course, there are discrepancies in the quantitative

match and the simulateddata do not account for the existence of shocks and random

events like crises. Nonetheless this should not distract attention from the more

important point that follows from the qualitative similarity of the long‐term

developmentpatterns.Inthisregard,Figure5putstheglobalgrowthanddevelopment

patterns from Figure 3 into perspective. In particular, it allows us to interpret the

observed patterns of theOECD countries’ growth slowdown, aswell as LatinAmerica

andAfrica’sacceleration,ascompatiblewithaglobalprocessoflong‐termdevelopment.

4 EmpiricalImplications

Irrespective of the reason why it occurred, Figure 5 shows that the delay in the

transition to sustained growth is an important element for understanding global

developmentpatternsingeneral,andparticularlytheprospectswithinthecontextofthe

secular stagnation question. The insights from the conceptual framework, and the

simulation,presentedintheprevioussectionhaverelevantimplicationsforthesecular

stagnationdebateinWesterncountriesandsuggestsometestablepredictions.

Theconceptualframework’sfirstempiricalimplicationisthateducationandhuman

capital, which are main growth engines in the long‐term growth literature, exhibit

nonlinear dynamics during the different phases of economic and demographic

development.As illustratedinFigure5(d),andconsistentwiththedata inFigure3(d),

thechangeinhumancapitalislargeintheearlydevelopmentphases(aftertheexitfrom

Malthusianstagnation)andslowsdownastheeconomydevelops.Averysimilarremark

appliestotheimprovementsinlifeexpectancy,whichisthecentralstatevariablebehind

the demographic and economic transition. The recent literature in economics and

demographyhasmadethepointthatlookingattheleveloflifeexpectancyattainedata

15

certain point in time is the bestway to identify different economic and demographic

developmentphases.8

The simplest way of testing the prediction that human capital increases at a

decreasing rate (that is in a concave fashion) during the development process is to

combine thedynamicsof life expectancyandeducationas suggestedby the simulated

dataillustrationreportedinFigure5.Thehypothesis is thatthehigherthe levelof life

expectancy,thesmallerthesubsequentchangeinacountry’shumancapital.Thisissince

better educated populations have less scope for further improvements in education

attainment.Wetestthishypothesisbyregressingdifferentmeasuresofeducationonlife

expectancy while allowing for a nonlinear (quadratic) relationship. For education

measures,weusetheaverageyearsofschooling,ahumancapitalindexandtheshareof

individualswithschoolingasalternativevariablesofinterest.9PanelAinTable1reports

theresultsforPooledOLSspecificationsfor131countriesovertheperiod1950–2010in

five‐yearfrequencies.Higherlifeexpectancyisassociatedwithsubsequentincreasesin

education,butatadecreasingrate.Theresultsholdalsoaccountingforthepastlevelof

GDPpercapita.

Panel B in Table 1 replicates the same exercise but exploiting within country

variation over time in specifications with country and period fixed effects. Besides

identification issues, thisspecificationhas theadvantage that it isclosely linkedto the

conceptual ideaof identifyingnonlinearitiesalongagivencountry’sdevelopmentpath

that lies behind the simulation in Figure5.The results confirm themain findings and

also suggest that income levels do not significantly affect future changes in education

once one accounts for the nonlinear relation behind life expectancy and education

during different development phases. As economies get to the mature stages of the

demographic transition, the structural slowdown in the process of human capital

formationisbyitselfaninterestingobservationregardingthequestionontheexistence

8Theevolutionof lifeexpectancyatbirthovertimehasbeenshowntomostprecisely identify

the timing of the demographic transition onset, with demographers suggesting a threshold ofapproximately50years(see,e.g.,Chesnais,1992,p.19,andthediscussioninCervellatiandSunde,2011b).

9Wetakeyearsofschoolingandtheshareofindividualsaged25+withsomeformalschoolingfromBarroandLee (2010),whereas thehumancapital index is fromthePennWorldTable (seeFeenstra,Inklaar,andTimmer,2015).

16

of a secular slowdown in growth.The results even indicate anon‐monotonicity in the

effect,atleastwhenconsideringthehumancapitalindexortheshareofthepopulation

withsomeschooling,withamaximumeffectoccurringbetween60and75yearsoflife

expectancy.

Table1:RelationbetweenLifeExpectancyandChangesinEducationDependentvariable:Changein: YearsSchooling HCIndex SharewithSchooling

(1) (2) (3) (4) (5) (6)

LifeExpectancyatBirth

PanelA

0.056*** 0.060*** 0.015*** 0.016*** 0.80*** 0.84***

(0.01) (0.01) (0.00) (0.00) (0.08) (0.10)

LifeExpectancyatBirth(sq.) ‐0.00040*** ‐0.00044*** ‐0.00012*** ‐0.00013*** ‐0.0074*** ‐0.0077***

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

logGDPp.c. 0.012 0.0057* 0.017

(0.02) (0.00) (0.17)

CountryFE No No No No No No

YearFE No No No No No No

Observations 1441 1253 1261 1261 1441 1253

R2 0.118 0.095 0.073 0.076 0.110 0.112

AdjustedR2 0.117 0.093 0.071 0.073 0.109 0.109

LifeExpectancyatBirth

PanelB

0.034*** 0.031** 0.014*** 0.011*** 0.59*** 0.47***

(0.01) (0.01) (0.00) (0.00) (0.10) (0.13)

LifeExpectancyatBirth(sq.) ‐0.00014 ‐0.00017 ‐0.00010*** ‐0.000079*** ‐0.0048*** ‐0.0034***

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

logGDPp.c. 0.053 0.0029 ‐0.32

(0.03) (0.01) (0.33)

CountryFE Yes Yes Yes Yes Yes Yes

YearFE Yes Yes Yes Yes Yes Yes

Observations 1441 1253 1261 1261 1441 1253

R2 0.093 0.111 0.048 0.086 0.031 0.038

AdjustedR2 0.092 ‐0.003 ‐0.065 ‐0.033 ‐0.067 ‐0.086

OLSestimates (PanelA), fixedeffectsestimates (PanelB), standarderrors shown inparentheses. *, **, and*** indicate statisticalsignificanceatthe10%,5%,and1%level,respectively.Observationsare5‐yearcountryobservationsfortheperiod1950–2010.Alltime‐varyingexplanatoryvariablesarelaggedby5years.DataSources:SeeTable7.

Another direct implication of the conceptual unified growth framework, illustrated

byFigure5’ssimulateddata,isthattheeffectofeducationongrowthisalsononlinear.

Most importantly, the effect of education changes during the different development

17

phases. This can be illustrated by jointly considering Panels (a) and (b) of Figure 5.

Having in mind a canonical representation of a standard human capital augmented

neoclassical production framework, one can decompose growth performance by

applyingagrowthaccountingexercisealong the linesofBenhabibandSpiegel (1994).

Then,incomepercapitacanbewrittenas

lnyit=α+βg(Hit)+γlnkit+ΓX’it+εit(1)

where yi,t is real per capita GDP, y, in country i in period t,.g(hi,t) is a function of the

average stock of human capital per capita (using a measure h), ln ki,t is the stock of

physical capital per capita, and the vector X includes other controls that have been

consideredintheempiricalgrowthliterature.Thecorrespondingcoefficientvectorsare

αand include theproductivity component,βandγ forhumanandphysical capital, as

wellasΓforthevectorofcontrols.10

The standard decomposition in (1) clarifies that, everything else equal, income

growthdependsonthedynamicsofhumancapitalaccumulation.Thespecification(1)is

typically derived from the balanced growth path of (non‐unified) endogenous growth

models, however. The simulation in Figure 5, that plots the predicted long‐term

evolutionofaprototypeeconomyundergoingtheeconomicanddemographictransition,

suggests theexistenceofpossiblenonlinearities in the return tohumancapitalduring

thedifferentdevelopmentphases.Astheeconomymatures,themarginalcontributionof

further years of schooling (or education) is expected to be reduced. Specifically, as

impliedbythesimulateddatashowninFigure5,therelationshipbetweeneducationand

future income ispositive and convex in theearlydevelopmentphases (duringgrowth

takeoff)butconcaveinthelaterphases(duringtheconvergencetothebalancedgrowth

path).11

10 To derive this empirical model, suppose that aggregate income, Y is given by a neoclassicalproductionfunctionYit=Ktγ(AtHδt)thatusesphysicalcapitalK,humancapitalHandproductivityA,wherehumancapitalisgivenbyHit=eg(hit)LitwithhasaverageyearsofschoolingandwhereListhepopulation. Dividing by population and taking log differences, one can derive an estimationequationasin(1);foradifferencedversionseealsoBenhabibandSpiegel(1994).11Afterthedemographictransitiononset,greaterlongevityacceleratestheexpansionofeducation,thereby reducing fertility andpopulation growth,with positive effects on incomeper capita. SeeCervellatiandSunde(2011a,2011b,2015b)foramorestructuredderivationofthesepredictions.

18

Accounting for these implications and testing these predictions is not

straightforward.A first, purely statistical attempt,may involveextending the standard

specification (1) to the inclusion of nonlinear (e.g. quadratic) effects of education on

incomegrowth.Columns (1) inTable2 report the resultsof thenonlinear (quadratic)

versionofthestandardlinearspecification(1)foryearsofschooling,thehumancapital

index and the share of population with some formal education in Panels A, B and C,

respectively. While the results suggest the existence of some nonlinearities, it is not

obvioustointerpretthepatterns,whichappeartochangeslightlydependingondifferent

humancapitalmeasures.

According to the theoretical predictions, the effect should not just be nonlinear but

should also depend on the different stages of the demographic. To explicitly test this

specificprediction,theremainingcolumnsofTable2moveonestepfurthertoallowfor

thepossibilitythatthehumancapitaleffectisnonlinearandchangesdependingonthe

development stages. This is done in two ways. First, we investigate the (nonlinear)

education effect on income growth by splitting the sample depending on whether

countries are forerunners (early developers) or latecomers in terms of their

demographic development as of 1960. The results in Columns (2) show that for the

countriesthathadalreadystarteddevelopingin1960(labeledas“Early”inTable2),the

education effect on income growth in the next half a century is positive but concave.

Paralleltothis,Columns(3)showthattheeffectispositiveandconvexforthecountries

that were still underdeveloped as of 1960 (labeled “Late”). Alternatively, one can

evaluate a country’s position during its development process period by period by

verifyingwhetheracountryispre‐orpost‐transitionalaccordingtothecriteriaadopted

bydemographerstoidentifythedemographictransitiononset.12Theresultsreportedin

Columns (4) and (5) indeed suggest that the relationship betweenhuman capital and

growthispositivebutconcaveincountriesthatdevelopedearlierorthatwerealready

post‐transitional, while the relationship is convex for late developers and pre‐

transitional countries. Once one explicitly accounts for the different demographic

development stages, the results become consistent for all the different human capital

measures,asseenfromColumns(2)to(5)inPanelsA,BandCofTable2.

12Wefollowtheusualcriteriausedintheliterature(see,e.g.,Reher,2004).

19

Table2:TheDemographicTransition:HumanCapitalonIncomeGrowth

Dependentvariable:Sample

logREALGDPp.c.(2005USD)

Full Early Late Post‐Trans. Pre‐Trans.

(1) (2) (3) (4) (5)

PanelA:EffectofYearsofEducation

Av.YearsofSchooling 0.030* 0.044** ‐0.020 0.024 ‐0.055*

(0.02) (0.02) (0.03) (0.02) (0.03)

Av.YearsofSchooling(sq.) 0.0016 ‐0.0050*** 0.0082*** ‐0.0016 0.0097*** (0.00) (0.00) (0.00) (0.00) (0.00)

logCapitalp.c. 0.58*** 0.50*** 0.58*** 0.52*** 0.66*** (0.02) (0.03) (0.03) (0.03) (0.04)

CountryFE Yes Yes Yes Yes YesYearFE Yes Yes Yes Yes YesObservations 1316 592 724 735 581R2 0.736 0.856 0.652 0.820 0.686AdjustedR2 0.703 0.838 0.602 0.789 0.610

PanelB:EffectofHumanCapita(Index)

HumanCapitalIndex(SH) 0.098 0.71*** ‐0.63*** 0.36** ‐0.87***

(0.14) (0.16) (0.22) (0.16) (0.25)

HumanCapitalIndex(SH)(sq.) 0.056* ‐0.15*** 0.26*** ‐0.043 0.28*** (0.03) (0.03) (0.05) (0.03) (0.05)

logCapitalp.c. 0.57*** 0.50*** 0.56*** 0.52*** 0.63*** (0.02) (0.03) (0.03) (0.03) (0.05)


PanelC:ShareofPopulationwithFormalSchooling

%withformalschooling ‐0.0060*** 0.020*** ‐0.016*** 0.0053* ‐0.024***

(0.00) (0.00) (0.00) (0.00) (0.00)

%withformalschooling(sq.) 0.000075*** ‐0.00011*** 0.00015*** ‐0.000023 0.00021***

(0.00) (0.00) (0.00) (0.00) (0.00)

logCapitalp.c. 0.59*** 0.50*** 0.60*** 0.52*** 0.63*** (0.02) (0.03) (0.03) (0.03) (0.04)


Fixedeffectsestimates,standarderrorsshowninparentheses.*,**,and***indicatestatisticalsignificanceatthe10%,5%,and1%level,respectively.Observationsare5‐yearcountryobservationsfortheperiod1950–2010.Alltime‐varyingexplanatoryvariablesarelaggedby5years.DataSources:SeeTable7.

20

The aforementioned theoretical investigations on the role of demographic

developmenthasanotherimplicationrelatingtotheroleofchanginglifeexpectancyfor

dependency ratios and, accordingly, for savings and physical capital investments. To

explore these issues, we proceed in two steps. First, we investigate how population

aging,measuredintermsoflifeexpectancyatbirth,affectsold‐agedependencyratios.13

Table3displaystheresults.

Table3:LifeExpectancyandOld‐AgeDependencyRatio

Dependentvariable:Sample

Old‐AgeDependencyRatio(SharePop.65+)


(1) (2) (3) (4) (5) (6)

LifeExpectancyatBirth ‐0.68*** ‐0.53*** ‐1.07*** ‐0.33*** ‐1.03*** ‐0.31***

(0.04) (0.05) (0.14) (0.05) (0.15) (0.06)

LifeExpectancyatBirth(sq.) 0.0052*** 0.0043*** 0.0077*** 0.0032*** 0.0088*** 0.0027***

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

logCapitalp.c. 1.29***(0.17)

logGDPp.c. 0.16 1.68*** 0.40*** 1.13*** 0.67***

(0.16) (0.27) (0.12) (0.22) (0.16)

CountryFE Yes Yes Yes Yes Yes YesYearFE Yes Yes Yes Yes Yes YesObservations 1771 1365 544 821 673 692R2 0.375 0.481 0.649 0.226 0.688 0.325AdjustedR2 0.307 0.406 0.597 0.100 0.619 0.159

Fixedeffectsestimates,standarderrorsshowninparentheses.*,**,and***indicatestatisticalsignificanceatthe10%,5%,and

1% level, respectively. Observations are 5‐year country observations for the period 1950–2010. All time‐varying explanatory

variablesarelaggedby5years.DataSources:SeeTable7.

Inlinewiththepredictionsfromunifiedgrowth, lifeexpectancyhasaconvexeffect

onpopulation aging in termsof theold‐agedependency ratio. For low life expectancy

levels (anddemographicdevelopment), an increase in lifeexpectancy initially leads to

longerlifeduringworkingages(andincreasedfertilityandpopulationgrowth),thereby

reducing the old‐age dependency ratio. At later demographic development stages, life

expectancymostly increases in old (non‐working) ages; the effect of further longevity

increasesdoesnotnecessarilyfuellargerworkingagepopulationsbutratherintohigher

13Datafortheold‐agedependencyratioarefromtheUnitedNations(2015).

21

sharesofretiredindividualsinthepopulation.14Asaresult,theeffectofincreasesinlife

expectancyeventuallyturnspositiveasagingleadstoamorethanproportionalincrease

in the share of elderly.15 On average the empirical reversal of the effect is around life

expectancyatbirthof60–64years.TheresultsinTable3showthatanonlineareffectis

presentevenforcountriesthatwerestillunderdevelopedasof1960,althoughtheeffect

tendstobestrongerforthosethatexperiencedthedemographictransitionearlyon(and

thatarenowlargelypost‐transitional).

Thesecondsteplinksthepopulation’sagestructuretosavingsandcapitalformation.

Table4 investigates thepossibility thatahigherold‐agedependencyratiomayreduce

gross capital formationandshows that this indeedappears tobe the case.Asone can

expecttheeffectisstronger,andmorestatisticallysignificant,intheagingcountriesthat

underwent their economic and demographic transition early on and are largely post‐

transitional.

Table4:PopulationAgingandCapitalAccumulationDependentvariable:Sample

Sharegrosscapitalformation(currentPPP)


(1) (2) (3) (4) (5) (6)

Old‐AgeDependencyRatio ‐0.0040*** ‐0.0041*** ‐0.0056*** ‐0.0036 ‐0.0056*** ‐0.0032

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

logCapitalp.c. 0.0023(0.01)

logGDPp.c.

0.0046 0.019* 0.00074 ‐0.00064 0.0049 (0.01) (0.01) (0.01) (0.01) (0.01)

CountryFE Yes Yes Yes Yes Yes YesYearFE Yes Yes Yes Yes Yes YesObservations 1705 1542 602 940 783 759R2 0.031 0.039 0.066 0.041 0.073 0.040AdjustedR2 ‐0.080 ‐0.085 ‐0.057 ‐0.095 ‐0.099 ‐0.173


14Thepreciserelationshipbetweenincreases in lifeexpectancyandretirementdependsonthe

age‐specificmortalitychanges(seeD’Albis,Lau,andSanchez‐Romero,2012).15Theeffectisreinforcedbythefactthatforhighlongevitythefertilitytransitionalsoreducethe

inflowofnewbornandoftheworkingagepopulation.

22

Inthecontextofthesecularstagnation,thesefindingsoverallimplythat,astheless

developedcountriesenterandproceedwiththeireconomicanddemographictransition,

theywill experience a fast growth rate phase similar towhat the advanced countries

experiencedinthepast.Intermsofinterestratesandcapitalflows,thismeansthatthe

supplyofcapital thatemerges fromthedevelopingworld’s increasing incomes,paired

with populations exhibiting higher aggregate savings,might well lead to low interest

rates in themedium run. The ratesmay remain low until population aging induces a

slowdownincapitalformationandcapitalsupplyproducedindevelopingcountries.

Another implication following from the preceding discussion is that human and

physicalcapitalmighthaveheterogeneouseffectsonpost‐transitionalcountries’growth

rates,aswellasoncountriesstillamidstthetransition.Theroleofphysicalcapitalmight

change as the economydevelops fromamore industry‐focused to amore innovation‐

drivenregime.16

Theothercentralfactorbehindgrowthinthesimpleempiricalframework(1)refers

tochangesintechnologyandproductivity.Theresultsfromthetotalfactorproductivity

onlifeexpectancyregressionsshowninTable5revealthatincreasesinlifeexpectancy

significantly affect productivity dynamics, particularly during the early demographic

transitionphases.

Table5:LifeExpectancyandProductivityDependentvariable:Sample

TotalFactorProductivity


(1) (2) (3) (4) (5) (6)

LifeExpectancyatBirth ‐0.048*** ‐0.048*** 0.0053 ‐0.061*** ‐0.017 ‐0.062***

(0.01) (0.01) (0.01) (0.01) (0.02) (0.02)

LifeExpectancyatBirth(sq.) 0.00040*** 0.00040*** ‐0.000066 0.00051*** 0.00013 0.00063*** (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

CountryFE Yes Yes Yes Yes Yes Yes

YearFE Yes Yes Yes Yes Yes YesObservations 959 959 499 460 572 387

R2 0.184 0.184 0.327 0.142 0.278 0.271AdjustedR2 0.066 0.066 0.233 ‐0.015 0.144 0.059


16SeeCervellatiandSunde(2015c)foradetailedtheoreticalandempiricalanalysisofthispoint.

23

This implies that health improvements have a profound effect on output through

productivityincreasesincountriesthatareundergoingthedemographictransition.The

effectevenappearstobeconvex,sothat,consistentwithaMalthusianperspective, life

expectancy must increase to sufficiently high levels before it starts to materialize

productivity gains.17 However, as seen in Columns (3) and (5), this effect loses

momentum (and significance) in countries that have completed the demographic

transition,aswellasinthosewherepopulationaginghasbegun.

Asa final step,Table6presentsresults for the implicationsofpopulationaging, in

termsoftheold‐agedependencyratio,forproductivity.Theresultssuggestthatthereis

noreasontoexpectproductivitytosufferfromaging.Infact,thefindingspointtowards

a positive effect that shows no clear differences between countries at various

demographicdevelopmentstages.Thisimpliesthateventhoughtheincreaseinhuman

capital might slow down as the demographic transition reaches its final stages, the

human capital embodied in the population, even an aging one, does not necessarily

becomeunproductive.

Table6:PopulationAgingandProductivity

Dependentvariable:

Sample TotalFactorProductivity


(1) (2) (3) (4) (5) (6)

OldAgeDependencyRatio 0.027*** 0.027*** 0.019*** 0.039*** 0.019*** 0.017**

(0.00) (0.00) (0.00) (0.01) (0.00) (0.01)

CountryFE Yes Yes Yes Yes Yes YesYearFE Yes Yes Yes Yes Yes YesObservations 1090 1090 562 528 658 432R2 0.270 0.270 0.452 0.174 0.384 0.291AdjustedR2 0.178 0.178 0.385 0.043 0.288 0.109


17TheresultsinColumns(4)and(6)implythequadraticrelationship’sminimumis60yearsoflifeexpectancy.

24

5Outlook:Will“SecularStagnation”Occur?

Eversince theendof the19thcentury, theworldhaswitnessedfasteconomicgrowth

and economic development has made its ways from Europe across the globe. More

recently,growthrateshavestabilizedandevenfallenintheWesternworld.Inanalogy

to the famous Club of Rome study in 1972, economists have recently resumed the

discussionabouttheprospectsforeconomicgrowth.Thispapertakesaunifiedgrowth

theory perspective, according to which the demographic transition is the central

mechanismbehindthetransitionfromstagnationtogrowth.Ithasarguedthatpasthigh

growth rates have to some extent been the consequence of singular, non‐recurrent

mortalityandfertilitychanges,coupledwithdramaticdemographicchangeintermsof

populationdynamics,educationattainment,andlaborforceparticipation.Theseone‐off

developments took several generations to take effect, but as they abate, growth

eventuallyslowsdownduringthelatertransitionalstagestothebalancedgrowthpath.

An important aspect for the context of secular stagnation is the observation of

capital flowsandcapitalaccount imbalancesdue todifferential timing indevelopment

across countries,with the consequence of asynchronous savings patterns. Forerunner

countries that are now more developed might enjoy increased capital inflows from

countries where increased longevity has just set in, resulting in increased savings

rates.18 Indeed, international capital flows across countries appear to be linked to

demographics,particularlyagingpatterns(see,e.g.DomeijandFloden,2006).19Thereis,

however, also an effect on the differential timing of economic and demographic

development that is more directly linked to the population aging process. Aging

populations in regions with an increased dependency ratio should expect to deplete

theirsavingsforold‐ageconsumption,whichdirectlyreducestheinternalavailabilityof

resourcesforphysicalcapitalinvestments.

In terms of demographic and economic transition, the forerunners have

incidentallybeenabletobalancetheslowdownandcontinuetogrowatrelativelyhigh

18 See Börsch‐Supan, Ludwig, andWinter (2006) for a simulation exercise that illustrates this

mechanism.19Thisdepresses thereturnsoncapitaland thus real interest rates, inparticular if savingsare

massiveduetocountrieswithlargepopulationsthatfacerapidlyagingpopulations,suchasChina.By themselves, capital flows might affect the development dynamics by interfering with theconvergencetothebalancedgrowthpath(see,e.g.,Birchenall,2007).

25

rates.Theyhavedonesobyexpanding theirmarkets throughglobalization,effectively

gaining capital and effectively reaching demographic development potential from the

less developed countries. Ultimately, however, as globalization reaches even themost

remote parts of the world and as latecomer countries undergo the demographic

transition,thetransitionalgrowthsourceslosemomentumatthegloballevel.However,

thisdoesnotmeanthattheworldeconomywillnecessarilystagnate.Whentheinfluence

of transitionaldynamicsoneconomicdynamicsbecomeless important,growthwillbe

basedmoreexclusivelyonproductivityimprovementsandtechnologicalchangethanin

the past. The scope for future productivity changes is more difficult to quantify and

predict.Opinionsdisagreeastothescopeforproductivityimprovements,andwhether

innovationswillstagnate,accelerateordecelerate(see,e.g.,Gordon,2014b,andMokyr,

2014).However,ifthehumancapitalstockisimportantforinnovation,asinmodelsof

humancapitaldrivenendogenousgrowth,theinnovationpotentialmightremainhighor

increase even post‐transition.20 Similarly, the expected human capital reallocation

around the world (through migration), mobility’s contributions to resource

optimization, and the potential economic benefits from ethnic diversitymay all affect

andstimulategrowthinwayswehavenotdiscussedhere.21

Thispaper’saimwas two‐fold: togiveaqualitativeassessmentofdifferent factors’

contribution to growth, including demographic factors, market size effects, and

innovation,aswellastopointoutimplicationsforfuturedevelopment.Thefindingsare

broadlyinlinewiththepredictionsfromaprototypeunifiedgrowthmodelthatexhibits

nonlinear development patterns like the demographic transition. To the extent that

many forecasts disregard these inherent development dynamics nonlinearities, the

findingsshedsomenewlightontherecentsecularstagnationdebate.

20Forevidenceon thedistinct roleofhumancapital changesandstocks in termsofgrowth, see,e.g.,SundeandVischer(2015).21ZaicevaandZimmermann(2014)provideareviewoftheissuesrelatedtoagingandmigration.

26

References

Barro,R.,andJ.‐W.Lee,2010,“ANewDataSetofEducationalAttainmentintheWorld,

1950‐2010,”JournalofDevelopmentEconomics,Vol.104,pp.184–98.

Benhabib,J.,andM.Spiegel,1994,“Theroleofhumancapitalineconomicdevelopment:

Evidencefromaggregatecross‐countrydata,”JournalofMonetaryEconomics,Vol.

34,No.1,pp.143–73.

Birchenall, J.A., 2007, “Equilibrium, convergence, and capital mobility in neoclassical

modelsofgrowth,”EconomicsLetters,Vol.99,pp.10–13.

Börsch‐Supan,A.,A.Ludwig,andJ.Winter,2006,“Ageing,PensionReformandCapital

Flows:AMulti‐CountrySimulationModel,”Economica,Vol.73,No.292,pp.625–

58.

Cervellati, M., and U. Sunde, 2011a, “Disease and Development: The Role of Life

ExpectancyReconsidered,”EconomicsLetters,Vol.113,pp.269–72.

, 2011b, “LifeExpectancyandEconomicGrowth:TheRoleof theDemographic

Transition,”JournalofEconomicGrowth,Vol.16,pp.99–133.

,2015a,“TheEconomicandDemographicTransition,Mortality,andComparative

Development,”AmericanEconomicJournal:Macroeconomics,Vol.7,No.3,pp.1–39.

,2015b,“TheEffectofLifeExpectancyonEducationandPopulationDynamics,”

EmpiricalEconomics,Vol.48,No.4,pp.1445–78.

, 2015c, “The Timing of the Demographic Transition and the Determinants of

Growth,”mimeo.

Chesnais, J.‐C., 1992, The Demographic Transition: Stages, Patterns and Economic

Implications. A Longitudinal Study of Sixty‐Seven Countries Covering the Period

1720–1984.(Oxford,ClarendonPress).

D’Albis,H.,S.‐H.P.Lau,andM.Sanchez‐Romero,2012,“MortalityTransitionand

DifferentialIncentivesforEarlyRetirement,”JournalofEconomicTheory,Vol.147,

No.1,pp.261–83.

Domeij,D.,andM.Floden,2006,“PopulationAgingandInternationalCapitalFlows,”

InternationalEconomicReview,Vol.47,No.3,pp.1013–32.

27

Eichengreen,B.,D.Park,andK.Shin,2013,“GrowthSlowdownsRedux:NewEvidenceon

the Middle‐Income Trap,” NBER Working Paper No. 18673 (Cambridge,

Massachusetts,NationalBureauofEconomicResearch).

Feenstra,R.C.,R.Inklaar,andM.P.Timmer,2015,“TheNextGenerationofthe

PennWorldTable,”AmericanEconomicReview,forthcoming.

Gordon,R. J.,2012,“IsU.S.EconomicGrowthOver?FalteringInnovationsConfront the

Six Headwinds,” NBER Working Paper No. 18315 (Cambridge, Massachusetts,

NationalBureauofEconomicResearch).

, 2014a, “The Demise of U.S. Economic Growth: Restatement, Rebuttal, and

Reflections”,NBERWorkingPaperNo.18315(Cambridge,Massachusetts,National

BureauofEconomicResearch).

, 2014b, “The turtles progress: Secular stagnation meets the headwinds,” in

Secular Stagnation: Facts, Causes, and Cures, ed. by Coen Teulings and Richard

Baldwin(London,CEPRPress).

Lucas,R.E.,1988,“OntheMechanicsofEconomicDevelopment,”JournalofMonetary

Economics,Vol.22,No.1,pp.3–42.

Mokyr, J., 2014a, “Secular Stagnation? Not in your life,” in Secular Stagnation: Facts,

Causes,andCures,ed.byCoenTeulingsandRichardBaldwin(London,CEPRPress).

, 2014b, “Secular Stagnation: US hypochondria, European disease?,” in Secular

Stagnation: Facts, Causes, and Cures, ed. by Coen Teulings and Richard Baldwin

(London,CEPRPress).

OECD,2015,EscapingtheStagnationTrap:PolicyOptionsfortheEuroAreaandJapan.

OrganizationforEconomicCooperationandDevelopment.

Reher,D.,2004,“TheDemographicTransitionRevisitedasaGlobalProcess,”Population,

SpaceandPlace,Vol.10,No.1,pp.19–41.

Summers,L.,2014,“U.S.EconomicProspects:SecularStagnation,Hysteresis,andthe

ZeroLowerBound,”BusinessEconomics,Vol.49,No.2,pp.65–73.

Sunde,U.,andT.Vischer,2015,“HumanCapitalandGrowth:SpecificationMatters,”

Economica,forthcoming.

28

Teulings,C., andR.Baldwin,2014a, “Introduction,” inSecularStagnation:Facts,Causes,

and Cures, ed. by Coen Teulings and Richard Baldwin (London, CEPR Press).

,2014b,SecularStagnation:Facts,Causes,andCures.(London,CEPRPress).

UnitedNations,2015,“WorldPopulationProspects:The2012Revision,”UnitedNations,

DepartmentofEconomicandSocialAffairs,PopulationDivision.

Zaiceva,A.,andK.F.Zimmermann,2014,“MigrationandtheDemographicShift,”IZA

DiscussionPaperNo.8743(Bonn,Germany,InstitutefortheStudyofLabor).

A Additi

Fig

ionalFi

Figure6:I

gure7:Glo

iguresa

mproveme

balEducat

29

andTab

entsinHeal

tionAttainm

bles

lth:Mortal

ment:Diffe

ityPattern

rentMeasu

s

ures

30

B DataSources

Table7:DataandSourcesVariable DescriptionandSource

GDP,Capital,Population,HumanCapitalindex

PennWorldTable(PWT)Version8.1,Source:http://www.ggdc.net/pwt

Fertility(TFR,ageatfirstbirth) UNWorldFertilitydata,Source:http://www.un.org/esa/population/publications/WFD2012/MainFrame.html

Mortality(mortalityrates,lifeexpectancyatdifferentages) UNMortalityData,Source:

http://unstats.un.org/unsd/demographic/sconcerns/mortality/mort2.htm

Education(populationshareswithdifferenteducationlevels,yearsofschooling)

BarroandLee(2010), Source:http://www.barrolee.com

OldAgeDependencyRatio UNWorldPopulationProspects:(2012Revision),Source:https://data.un.org/Data.aspx

Documents

Demographic Dynamics and Long‐RunDevelopment ... · debate is concerned with the question of whether the delayed recovery recently experienced reflects a reduction in long‐run