’WPC6c ;U.J 0ŗx 0D2#v C B… AM¢ %ļU8õU.-Ę[ 0‡! 0JØ 0lņ ^ D+d‚U.ņ?(C@1oŚK!8C!pČC!8I!ČI!pXJ!ČO!pXP!PČU!pV!ˆ[!P\!ph\! Ųa! xb )c ū%8"° : : 0ŠA 0DŲB<Ō6X9`+Courier(q!2–o$¤¤Ż ƒ|«!ŻŃ7€ōōd§7ŃŃ s ŃŻ W ŻÓӀ€€€€ņņŚ  Ś0Ś  Śóó(/Ć$ņņŚ  Ś6Ś  ŚóóHP LaserJet IIIHP3.PRSP F2|,\,\,,š’HĄQȉ- 0 d A<< c’’ō3 ŖHa`$ Line Printer<Ō6X9`+Courier Level 1 Level 2 Level 3 Level 4 Level 5(_2i$££Ż ƒ|«!ŻŃ7€ōōd§7ŃŃ s ŃŻ W Ż("’’$’’ņ ņFigure€Ś  ŚŚ  Śó ó(D|«$””Ń7€XXdČ7ŃŃ s Ń’’HV'ČČČČdxd ’Ngri’ ’’ ’Ngri’ ’ ’ ’N.Ngr’ ’ i’ ’gr’ ’ ’ ’ith’,’heteroskedasticity’,’dowrick’&’autocorrelation’&’T]he’’Std’’Obs’’Curr’ ’sq’ ’w.r.t.’’Intra’’subsample’’natalist’’NAS’’Anarudh’’Agarwal’’Monash’’Cassen’’Barlow’’Blanchet’’LDCs’’DCs’’GDP’’LDC’"’correlational’""’intertemporal’"’Gobin’’Ansley’’Coale’’CBR’’CDR’’Solow’’Koopmans’’Lucus’’Rebelo’’Romer’"’externalities’"’Ngr’’Heston’&’econometrically’&’cols’*’multicollinearity’* ’BR’ ’Wald’’subsamples’’Didier’’Taylor’’Stochastic’’Dominick’’Salvatore’’Menken’’Norton’’Rolf’’Ohlsson’’Springer’’Verlag’’Tjallig’’Pontificia’’Acad’’Sci’’Sergio’’Julian’’Roy’’ESA’’SER.A’’WPDOCS’’AggCorrs’’Lucas’’outliers’’D.E.’’R.B.’’J.A.’’A.J.’’A.C.’’R.M.’’T.C.’’R.D.’’R.E.’’ygr’ ’)’ ’Nt’ ’Xt’ ’t,t’’Zt,t’’Robelo’’Fagerberg’’f[Y’’D.Y’’FEM’’n.s.’’Dns’’DFFITS’’Ngrt,t’’CBRt,t’’CDRt,t’’CDRs’’socio’’R.J.’’Institut’’d'etudes’$’demographiques’$’Aldin’’J.L.’’DaVanzo’’JAI’<Ō6X9`+Courier ’Bob.one’’m.s.’’stochastic’’DFFITS’’socio’’t,t’ ’Nt’ ’Xt’ ’Zt,t’’CDRs’’Clairborne’’Monash’’Cassen’’Dowrick’’Ngr’’LDCs’’DCs’"’correlational’"&’autocorrelation’&"’intertemporal’"’Gobin’’Ansley’’Coale’’T]he’’Blanchet’’CBR’’CDR’’Barlow’ ’)’ ’Solow’’Lucas’’Rebelo’’Romer’"’externalities’"’Fagerberg’’Ngrt,t’’f[Y’’D.Y’’CBRt,t’’CDRt,t’’Boserup’’Barro’’Heston’’Ngrit’ ’i’ ’t’ ’’ ’’ ’Xit’ ’ ’ ’ ’ ’Dit’ ’’ ’N}it’’ it’’FEM’’covariance’’Hausman’’cols’’w.r.t.’’LDC’ ’collinearity’ "’monotonically’" ’quinquennial’ ’OLS’’R.J.’’Jong’’Wha’’NBER’’Institut’’d'etudes’$’demographiques’$’Taylor’’D.E.’’R.B.’’Dominick’’Salvatore’’Aldin’’J.A.’’A.J.’’Menken’’Norton’’A.C.’’Rolf’’Ohlsson’’Springer’’Verlag’’R.M.’’R.D.’’R.E.’’J.L.’’DaVanzo’’JAI’’Dns’’GDP’’Claiborne’’R.K.’’T]hese’ ’Eq’ ’Zt’ ’Dt,t’ ’pathbreaking’ ’Eqs’ ’Ns’ ’Mencken’’Std’’Obser’’vations’FIG1B.WPG’WPC°°0*x 'А3Ų'°°°°,,51’’’3  ,€&ĄĄĄĄ5 3’’’+'’’’ąX4Æ8°$X$€b5b8ą&8 €8b8 ŁTŌļžż’’ ŻŲނԀ’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį-1€ €j ą&j  ŁTjåż’’Y ėĄ Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį0€ €b ŁTjåż’’YAėņŌ€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį1€ €ĻbĻ ŁTjåż’’Ytė%Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį2€ €b ŁTjåż’’Y¦ėWŌ€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį3€ €5b5 ŁTjåż’’YŁėЁԀ’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį4€ €Ÿ8Ÿ’ €8 €—8— € 8  € 8  € 8  €† 8†  € 8  €~ 8~ ’ €ś 8ś  €v 8v  €ņ 8ņ  €n 8n  €é 8é  €e8e €į8į €]8]’ €Ł8Ł €U8U €Ń8Ń €L8L €Č8Č €D8D €Ą8Ą €<8<’ €ø8ø €383 €Æ8Æ €+8+ €§8§ €#8# €Ÿ8Ÿ €8’ €–8– €8 €Ž8Ž € 8  €†8† €8 €~8~ €ł8ł’ €u8u €ń8ń €m8m €é8é €e8e €į8į €\8\ €Ų8Ų’ €T8T €Š8Š €L 8L  €Č 8Č  €C!8C! €æ!8æ! €;"8;" €·"8·"’ €3#83# €Æ#8Æ# €+$8+$ €¦$8¦$ €"%8"% €ž%8ž% €&8& €–&8–&’ ŁTy”Žż’’_‡æ8€Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį0 ŁTr Fü’’> ‡ž 8€Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį2 ŁTk¬żś’’‡}8€Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį4 ŁTd8µł’’ü‡\8€Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį6 ŁT6mų’’ڇ:8€Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį8 ŁTH0:÷’’x‡:8Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį10 ŁTA¼ńõ’’W‡8Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį12 ŁT:H©ō’’6"‡ų"8Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį14 ŁT3Ō`ó’’&‡×&8Ō€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįį16 ŁT• ū’’ķŒØ=ŖŌ€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįįPer€Capita€GDP/N€(U.S.€$1,000,€1985€prices)*Zķ’’’ń’’’’’ŁT¾Dƒ·¹ō Šž1āšŌ€’’ŌŌ-€’’*ÜD4`-Ōį įį’’’įį’’’įįįPredicted€GDP/N€Growth€Rate)%Ŗ€" —˜    Æ P € P † ż  · ~ ~ ś P v . €v . ņ  n é  e į €į ]5 ŁV U~ Ń® Lå €Lå Č" De Ą® <ü øN €øN 3„ Æ +^ §¾ #"€#"Ÿ‡ī–V¾Ž'€Ž' †ų_~Äł'€ł'u‡ńęm@é—eé€eéį6\ŲĀT’Š6€Š6L eČ C!®æ!Å;"Õ€;"Õ·"Ś3#ׯ#Ź+$²¦$€¦$"%`ž%&)%Ŗ€Ÿ—’ m é s€ s†  ® ~ ^ ś  v į €v į ņ ³ n  é w eg į` €į` ]b Łm U Ń˜ L¹ €L¹ Čą D Ą? <v ø³ €ø³ 3ó Æ7+~§É#€#Ÿeµ–YŽ«€Ž« ž†OŸ~īł;€ł;u†ńĶméQe€eįÄ\öŲ#TIŠj€ŠjL ƒČ •C!žæ! ;"™€;"™·"‰3#oÆ#K+$¦$怦$ć"%Ÿž%M%UU’ €Ÿj j  €j —j  €—j  j  € j  j  € j j  € j † j  €† j  j  € j ~ j  €~ j ś j  €ś j v j  €v j ņ j  €ņ j n j  €n j é j  €é j ej  €ej įj  €įj ]j  €]j Łj  €Łj Uj  €Uj Ńj  €Ńj Lj  €Lj Čj  €Čj Dj  €Dj Ąj  €Ąj <j  €<j øj  €øj 3j  €3j Æj  €Æj +j  €+j §j  €§j #j  €#j Ÿj  €Ÿj j  €j –j  €–j j  €j Žj  €Žj j  € j †j  €†j j  €j ~j  €~j łj  €łj uj  €uj ńj  €ńj mj  €mj éj  €éj ej  €ej įj  €įj \j  €\j Ųj  €Ųj Tj  €Tj Šj  €Šj L j  €L j Č j  €Č j C!j  €C!j æ!j  €æ!j ;"j  €;"j ·"j  €·"j 3#j  €3#j Æ#j  €Æ#j +$j  €+$j ¦$j  €¦$j "%j  €"%j ž%j  €ž%j &j  €&j –&j % €‹ 4‹ 8 €Ģ4Ģ$ €°4° €°4°$ ±WLŠēū’’ō · ĀƒŌ€!’’!!ŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįLDCs ±WŹ#ö’’É '‚Ԁ!’’!!ŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįDCs €°$Ń&. ¼Q _ ÷’’«µŠŌ€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįNgr€=€2.00%)  € DhD) €žDH D ’€ ĘųE#  ]V–^gö’’rĶī s ŒŌ€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįDC€Median€Y/N ]VÅÜö’’$ yŹ ƒŌ€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįšš ’€ r ćt  ]V5ö.ū’’Gŗ¦` ‰Ō€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįMedian€Y/N ]Vģ~ēū’’$ ŗo ` ‚Ō€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįLDC ]V†ė^ü’’Ā % Æ ƒŌ€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįšš ¼Q€#‡ü’’ą <µ•Ō€ō’’ōōŌŌ-€’’,ä2¼`-Ōį įį’’’įį’’’įįįMedian€LDC€Ngr€=€2.54%<Ō6X9`+Courier`)Ͼ¦ p`CG Timesa<Ō6X9`+Courier3üŖHa`' Line Printer’<Ō6Xß9`+"CourierItalic<Ō6X9`+ CourierLegal’<Ō6Xß9`+"CourierItalic<Ō6X9`+ CourierLegal’<žž9p`+†2Courier 10RomanSpeedoøT0s4ddd,,ōdģŪž’ž’ž’ģ{{{’’g’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’]Ųž’ž’’’’’TEH)!),,/,)”qž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ŌQ‰?xxx,,`XxģŪž’ž’ž’qqq’’S’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ĪIž’ž’’’’’ŌGk+HHH,,©H1מ’ž’ž’ž’ž’ž’ž’’’’’’’’’’’’’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’ž’’’’’’’T``%7,7,,/7”qž’ž’ž’ž’ž’ž’ā’’ž’’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’TXs4ddd,,ČōdrÜž’ž’ž’ž’ž’ž’ž’’’’’’’’’’’’’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’’’’’’’’’T9!!,,/š!”qž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’”.s4ddd,,ōdģŪž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’ž’ž’’’’’T7x/D8D,,/ōD”qž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ŌS‰?xxx,,<XxrÜž’ž’ž’ž’ž’ž’ž’’’’’’’’’’’’’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’’’’’’’’’A‰?xxx,,wXxģŪž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’ž’ž’’’’’T 8RCR,,/XR”qž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’Ōab,UUU,,°©UIœž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’ž’’’ž’ž’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’ž’ž’ž’ž’ŌV\)PPP,,°PIœž’ž’ž’ž’ž’ž’ž’’’ž’’’’’’’ž’’’’’ž’’’’’ž’’’ž’ž’’’ž’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’’ž’ž’ž’ž’ž’ž’’’’’U•J’’ō,,,’’’’Z’’ššĀ†ų€’d’ׂ’’ō,,,’’’’`’’š±Ā’’’’’dUŠ*’’ō,,,’’’’f’’ššĀš€’d’·’’ō,,,’’’’l’’š±Ā(É2Ē|$„„Ż ƒ|«!ŻŃ7€ōōd§7ŃŃ s ŃŻ W ŻŅ°ŅҰŅŃ(°ŃѰŃŌ€ōģŪōōōģŪŌÓÓŚ  Ś0Ś  Ś.($Ó$‘‘ņņŚ  Ś0Ś  Śóó 0cZ#|H„ūž’"AIP[rƒ’’’’’’’’’’’’’’ﳈE?’’’’’ž’ĢĢ0³ĖĻ4ĢĻüĄĆü’04/’’’’’’N·’’ą«’ššmpϟ’‰@š2BP€@LL "‚u«Ü,ŁĄ‡a’’’’e’’’’’’^!.22YN!!!2Y!!!!2222222222!!dYd,YH?EJ?;HJ!'F?[JH9HC6?JH^HHA!!!22!,2,2,!222N2222%'22H22,,2,dń222222!222222222H,H,H,H,H,YCE,?,?,?,?,!!!!J2H2H2H2H2J2J2J2J2H2H,J2H2H2H2J292H,H,H,E,E,E,E,J2?,?,?,?,H2H2H2H2H2H2J2J2!!!!I8'F2???>>J2J:J2J2H2H2YHC%C%C%6'6&6'6'???J2J2J2J2J2J2^HH2A,@,A,J2?J2C%6'?H2H2J2H2J2?,!H24!!22!!/ddddddddddddddddddddddddddddddddddddddddddddddddNHHH222!,))22X222YY2#2222Y#!!442Ydd22<CCwh,,,Cw,,,,CCCCCCCCCC,,…w…;w`T[cTO`c,4^Tyc`M`YHTc`~``V,,,CC,;C;C;,CC%%C%hCCCC14%CC`CC;;C;…ńCCCCCC,CCCCCCCCC`;`;`;`;`;wY[;T;T;T;T;,%,%,%,%cC`C`C`C`CcCcCcCcC`C`;cC`C`C`CcCMC`;`;`;[;[;[;[;cCT;T;T;T;`C`C`C`C`C`CcCcC,%,%,%,%aJ4^CT%T%T%S%S%cCcMcCcC`C`Cw`Y1Y1Y1H4H3H4H4T%T%T%cCcCcCcCcCcC~``CV;U;V;cCT%cCY1H4T%`C`CcC`CcCT;,%%`C4,,CC,,/………………………………………………………………………………………………………………………………N```CCC,;66CCuCCCwwC/CCCCw/,,EECw##……CCPPe…C…C`COmmJJw,EC~ /«``………C`……E``ww………C…………åww………wC,,……………`P…w``````````,…C`CJCw…`…YC,C`;CM`hV……``````````……………>…````…………………………………………………``………………mYPh…`………9O````………………Å```C``…``………```;````;````````````````````````````````````````````,```,```,```,`````````````Ų`PTJPJY;T4V;cC`;,%^;YJyJc;YC`;`PMCPCCT;`;YP`JhPPP`PT4cC,%%`;`;;PP;PPPP,,,CCCCCCCPPPPPPPPPPPPPPPPPPPPPP4444444CCCCCCCCCCCCCCCCCCCCCC%%%%%%%%%%%%;;;;;;;;;;;;;;;;;;;PPPPPPPPPPPPPPPPPPPPPPMC;,…ž’"AIP[rƒ’’’’’’’’’’’’’’ﳈE?’’’’’ž’ĢĢ0³ĖĻ4ĢĻüĄĆü’04/’’’’’’N·’’ą«’ššmpϟ’‰@š2BP€@LL "‚u«Ü,ŁĄ‡a’’’’e’’’’’’^%((G>(G((((((((((PGP$G:27;2/:;82I;:.:5+2;:L::4(($($($(((>((((((:(($$($Pń(((((((((((((((:$:$:$:$:$G57$2$2$2$2$;(:(:(:(:(;(;(;(;(:(:$;(:(:(:(;(.(:$:$:$7$7$7$7$;(2$2$2$2$:(:(:(:(:(:(;(;(:,8(22222;(;.;(;(:(:(G:555++++222;(;(;(;(;(;(L::(4$3#4$;(2;(5+2:(:(;(:(;(2$:(4((/PPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPN:::((($!!((F(((GG(((((G))(GPP((00=P(P(:(/BB,,G)(L`«::PPP(:PP*::GGPPP(PPPPåGGPPP G(PPPPP:0PG::::::::::P(:(,(GP:P5((:$(.:>4PP::::::::::PPPPP%P::::PPPPPPPPPPPPPPPPPPP::PPPPPPB50>P:PPP"/::::PPPPPPÅ:::(::P::PPP:::$::::$::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::Ų:02,0,5$24$;(:$8$5,I,;$5(:$:0.(0((2$:$50:,>000:02;(:$:$$00$0000(((((((0000000000000000000000(((((((((((((((((((((($$$$$$$$$$$$$$$$$$$0000000000000000000000.($P"’’/’’6’’=’’’’’’’’’’D’’’’’’zŽ’’õ’’’õ’’ą/)@.€"‚u«Ü,ŁĄ‡a’’’’e’’’’’’^18MSS”‚888S”8888SSSSSSSSSS88§”§J”xir{icx{8Aui—{x`xoYi{xxxl888SS8JSJSJ8SS..S.‚SSSS>A.SSxSSJJSJ§ńSSSSSS8SSSSSSSSSxJxJxJxJxJ”orJiJiJiJiJ8.8.8.8.{SxSxSxSxS{S{S{S{SxSxJ{SxSxSxS{S`SxJxJxJrJrJrJrJ{SiJiJiJiJxSxSxSxSxSxS{S{S8.8.8.8.z]AuSi.i.i.g/g/{S{a{S{SxSxS”xo>o>o>YAZ?YAYAi.i.i.{S{S{S{S{S{SxxSlJkIlJ{Si.{So>YAi.xSxS{SxS{SiJ8..xS488SS88/§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§§NxxxSSS8JDDSS’SSS””S;SSSS”;88VVS”++§§SSdd§S§SxSc‰‰]]”8VSÉ;«xx§§§Sx§§Wxx””§§§S§§§§å””§§§”S88§§§§§xd§”xxxxxxxxxx8§SxS]S”§x§oS8SxJS`x‚l§§xxxxxxxxxx§§§§§M§xxxx§§§§§§§§§§§§§§§§§§§xx§§§§§§‰od‚§x§§§Gcxxxx§§§§§§ÅxxxSxx§xx§§§xxxJxxxxJxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx8xxx8xxx8xxx8xxxxxxxxxxxxxŲxdi]d]oJiAlJ{SxJ8.uJo]—]{JoSxJxd`SdSSiJxJodx]‚dddxdiA{S8..xJxJJddJdddd888SSSSSSSddddddddddddddddddddddAAAAAAASSSSSSSSSSSSSSSSSSSSSS............JJJJJJJJJJJJJJJJJJJdddddddddddddddddddddd`SJ8§ž’"AIP[rƒ’’’’’’’’’’’’’’ﳈE?’’’’’ž’ĢĢ0³ĖĻ4ĢĻüĄĆü’04/’’’’’’N·’’ą«’ššmpϟ’‰@š2BP€@LL "’’/’’6’’=’’’’’’’’’’D’’’’’’zŽ’’õ’’’õ’’ą/)@.€"‚u«Ü,ŁĄ‡a’’’’e’’’’’’^;C]dd²œCCCd²CCCCddddddddddCCȲČY²~‰”~v”CN~µ”s…k~”½‚CCCddCYdYdYCdd88d8œddddJN8ddddYYdYČńddddddCdddddddddYYYYY²…‰Y~Y~Y~Y~YC8C8C8C8”ddddd”d”d”d”ddY”dddd”dsdYYY‰Y‰Y‰Y‰Y”d~Y~Y~Y~Ydddddd”d”dC8C8C8C8’oNd~8~8~8|8|8”d”t”d”ddd²…J…J…JkNlLkNkN~8~8~8”d”d”d”d”d”d½d‚Y€X‚Y”d~8”d…JkN~8dd”dd”d~YC88d4CCddCC/ČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČČNdddCYQQdd°ddd²²dFdddd²FCChhd²44ČČddxx˜ČdČddv¤¤oo²Chd½ńF«ČČČdČČh²²ČČČdČČČČå²²ČČȲdCCČČČČȐxȲCČddod²ČČ…dCdYdsœ‚ČȐČČČČČ]ȐČČČČČČČČČČČČČČČČČČȐČČČČČȤ…xœČČČČUvČČČČČČՐdČČČȐYYCCCCŲx~oxo…Y~N‚Y”dYC8Y…oµo”Y…dYxsdxdd~YY…xoœxxxx~N”dC88YYYxxYxxxxCCCdddddddxxxxxxxxxxxxxxxxxxxxxxNNNNNNNdddddddddddddddddddddd888888888888YYYYYYYYYYYYYYYYYYYxxxxxxxxxxxxxxxxxxxxxxsdYCČž’"BIP[rƒ’’’’’’’’’’’’ń÷Ćó’’’’’’ż’’’’’?’’÷’’’Ļ’’Ļ’ü=Ą//’’’’’N·’’ń’Ź£ü«żššhłŸž &@€ģFRŃĄ@FL .ž’"BIP[rƒ’’’’’’’’’’’’ń÷Ćó’’’’’’ż’’’’’?’’÷’’’Ļ’’Ļ’ü=Ą//’’’’’N·’’ń’Ź£ü«żššhłŸž &@€ģFRŃĄ@FL . !"#$Ż ƒ|«!ŻŃ7€ōōd§7ŃŃ s ŃŻ W ŻŃܰŃѰŃŃ  ŃŌ% € ŌĢҰŅҰŅŌ€XģŪXōōģŪŌĢÓ  ÓAggregate€Population€and€Economic€Growth€Correlations:ĢÓŃÓĢÓ  ÓThe€Role€of€the€Components€of€Demographic€Change*ĢÓ%ÓĢĢĢĢĢÓ  ÓAllen€C.€KelleyĢDepartment€of€Economics€andĢCenter€for€Demographic€StudiesĢDuke€UniversityĢDurham,€NC€€27708ĢĢĢÓxÓÓ  ÓRobert€M.€SchmidtĢE.€Claiborne€Robins€School€of€BusinessĢUniversity€of€RichmondĢRichmond,€VA€€23173ĢÓÓĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢĢ*ņņWe€gratefully€acknowledge€the€financial€support€of€the€WorldŠ ļ&ē) ŠBank€and€the€National€Institutes€for€Child€Health€and€HumanĻDevelopment,€the€research€assistance€of€Joel€R.K.€Moody€andĻMichael€C.€Kelley,€the€editorial€assistance€of€Gail€McKinnis,€theĻfeedback€at€seminars€(Monash€University,€Melbourne;€University€ofĻIndonesia,€Jakarta;€and€East„West€Center,€Honolulu),€and€theĻcomments€by€Robert€H.€Cassen,€Robert€L.€Clark,€Steve€Dowrick,€andĻtwo€referees.ń„ńŠ ”+™$0 ŠCopyright€1995ń„ńóóŠ  H,@%1 ŠŲŲÓÓą@ BģąABSTRACTˆĢThe€results€of€recent€correlations€showing€a€negativeĻimpact€of€population€growth€on€economic€development€inĻcross„country€data€for€the€1980s,€versusĻ"nonsignificant"€correlations€widely€found€for€theĻ1960s€and€1970s,€are€examined€with€contemporaneous€andĻlagged€components€of€demographic€change,€convergence„¼ń„ń¼¼¼¼¼¼¼¼¼¼¼¼¼¼¼ń„ńtype€economic€modeling,€and€several€statisticalĻframeworks.€€The€separate€impacts€of€births€and€deathsĻare€found€to€be€notable€but€offsetting€in€the€earlierĻperiods.€€In€contrast,€the€short„run€costs€(benefits)Ļof€births€(mortality€reduction)€increase€(decrease)Ļsignificantly€in€the€1980s,€and€the€favorable€labor„¼ń„ń¼¼¼¼¼¼¼¼¼¼¼¼¼¼¼ń„ńforce€impacts€of€past€births€are€not€fully€offsetting.ĢŠ  čą Šą ` ąPossibly€the€most€influential€statistical€findingĻthat€has€shaped€the€"population€debates"€in€recentĻdecades€is€the€failure,€in€more€than€a€dozen€studiesĻusing€cross„country€data,€to€unearth€a€statisticallyĻsignificant€association€between€the€growth€rates€ofĻpopulation€and€of€per€capita€output€(Kelley€1988).€ĻRonald€Lee's€(1983)€early€assessment€is€instructive:€Ļ"[T]hese€cross„national€studies€have€not€provided€whatĻwe€might€hope€for:€€a€rough€and€stylized€depiction€ofĻthe€consequences€of€rapid€population€growth,€unless,Ļindeed,€the€absence€of€significant€results€is€itselfĻthe€result"€(1983:54).€€All€observers€agree€that€simpleĻcorrelations€are€difficult€to€interpret;€however,€theĻvirtual€absence€of€a€systematic€relationship€in€theĻface€of€such€strongly€held€beliefs€has€kept€theĻpopulation€debate€alive€and€has€buttressed€theĻ"revisionist"€position€that€emphasizes€a€balanced,€Ļgenerally€"non„alarmist"€assessment€of€the€economicĻimpacts€of€demographic€change€(Kelley€1995).Ģą ` ąPredictably,€even€while€the€(lack€of)€correlationsĻbetween€population€and€economic€development€haveĻconstituted€a€major€force€in€causing€a€reassessment€ofĻthe€impacts€of€population€growth,€the€appearance€ofĻ"new"€correlations€in€several€recent€studies€could€wellŠ ų*š#. Šcause€the€pendulum€to€swing€back€toward€a€more€guarded,€Ļeven€alarmist€interpretation.€€These€studies,€reviewedĻby€Kelley€and€Schmidt€(1994),€seem€to€reveal€a€negativeĻassociation€between€population€and€economic€growthĻbased€on€international€cross„country€data€for€theĻ1980s.€€Even€though€the€authors€of€these€studies€areĻgenerally€cautious€with€respect€to€the€strength€orĻimportance€of€this€finding,€intriguing€questions€arise:€ĻHas€the€impact€of€population€growth€changed?€€If€so,Ļwhy?Ģą ` ąIn€our€recent€study€(Kelley€and€Schmidt€1994),Ļwhich€extends€this€literature,€we€conclude€that€theĻcorrelations€for€the€1980s€indeed€have€changed.€€FigureĻ1€presents€a€representative€regression€model€from€thisĻsurvey,€which€compares€per€capita€output€growth€ratesĻunder€two€alternative€trajectories€for€populationĻgrowth€rate:€€the€median€population€growth€rate€(2.54%)Ļand€a€substantially€lowered€rate€of€2.00%.€€We€findĻthat€the€impact€of€the€rate€of€population€growth€on€perĻcapita€output€growth:Ģ€€€1)€Is€not€statistically€significant€in€the€1960s€orĻą ` ą€1970s,€a€finding€consistent€with€a€wideĢą ` ą€literature;Ģ€€€2)€Is€negative,€statistically€significant,€and€largeŠ ų*š#. Šą ` ą€in€the€1980s;€andĢ€€€3)€Varies€with€the€level€of€economic€development:€Ļą ` ą€it€is€negative€in€the€LDCs€and€sometimes€positiveĢą ` ą€for€the€DCs.ĢModels€exploring€density€and€size,€in€addition€toĻpopulation€growth,€also€tell€us€thatĢ€€€4)€Population€density€is€the€only€demographic€termĢą ` ą€exerting€a€statistically€significant€effectĢą ` ą€(positive)€in€the€1960s€and€1970s;€and€Ģ€€€5)€All€components€of€demographic€change€exert€aĢą ` ą€statistically€significant€effect€in€the€1980s,Ģą ` ą€where€the€effect€of€population€growth€on€economicĢą ` ą€growth€is€negative€and€the€effects€of€bothĢą ` ą€population€size€and€population€density€areĢą ` ą€positive.€€The€"net"€effect€of€all€threeĢą ` ą€components€of€demographic€change€is€negative,Ģą ` ą€on€average,€for€all€countries.Ģą ` ąA€full€explanation€for€the€new€and€statisticallyĻrobust€correlations€is€not€yet€available.€€On€the€oneĻhand,€the€negative€consequences€of€rapid€populationĻgrowth€associated€with€diminishing€returns€to€capitalĻand€the€environment€may€be€emerging€as€more€importantĻforces€than€(say)€the€positive€effects€of€scale,Ļinduced€innovation€or€technical€change,€and/orŠ ų*š#. Šattenuating€feedbacks.€€On€the€other€hand,€the€1980sĻmay€represent€an€anomaly,€plagued€as€they€were€withĻworld€recession,€wars,€droughts,€structuralĻadjustments,€and€declining€and€low€terms€of€trade.€€TheĻadverse€consequences€of€these€events€may€have€beenĻlargest€in€the€countries€with€a€relatively€high€rate€of€Ļpopulation€growth.€€In€the€first€case,€the€negativeĻcorrelations€may€persist;€in€the€second,€they€mayĻrevert€to€the€previous€pattern€of€little€or€noĻassociation.€€Ģą ` ąTo€weigh€these€alternative€hypotheses€requiresĻextensive€and€complex€empirical€modeling,€a€task€beyondĻthe€scope€of€this€paper.€€Instead€we€seek€to€take€aĻfirst€(though€important)€step€by€exposing€the€role€ofĻthe€ņņcomponentsóó€of€demographic€change€in€the€variousĻcorrelational€findings.€€Our€results€at€once€clarifyĻsome€of€the€forces€at€play€and€help€to€set€a€researchĻagenda€for€exploring€the€complex€modeling€required€toĻuntangle€the€underlying€economic„demographicĻrelationships.Ģ€€€€ĢŠ  Ų'Š * ŠÓ  Óą@@č#ģąDEMOGRAPHIC€IMPACTS€IN€THE€SHORT€AND€THE€LONG€RUN:ĻCOMPONENTS€OF€CHANGEĢÓ=Óą ` ąIn€untangling€the€literature€on€the€effects€ofĻpopulation€growth€on€per€capita€output€growth,€a€keyĻprocedure€is€sorting€out€the€short„€and€long„runĻeffects€of€the€components€of€demographic€change.€€Ģą ` ąAn€individual's€economic€impact€varies€over€his€orĻher€lifetime.€€In€the€short€run,€the€effect€of€a€birthĻis€likely€to€be€negative€(i.e.,€children€are€netĻ"resource€users");€yet€at€later€stages€of€the€lifeĻcycle,€the€effect€is€likely€to€be€positive€(workingĻadults€are€net€"resource€creators").€€Countries€withĻrapid€ņņcurrentóó€population€growth€rates€are€likely€to€beĻthose€with€high€ņņpastóó€population€growth€rates€(i.e.,Ļgrowth€rates€show€strong€autocorrelation).€€Therefore,Ļcross„sectional€evidence€using€contemporaneous€dataĻalone€measures,€de€facto,€both€the€negative€impacts€ofĻcurrent€births€and€the€positive€impacts€of€past€births.€ĻAs€a€result,€the€commonly€found€empirical€resultĻshowing€no€measured€impact€of€population€growth€doesĻnot€necessarily€mean€that€demography€is€unimportant:€Ļit€simply€may€imply€that€strong€intertemporalĻā ādemographic€impacts€are€offsetting.Š h)`", Šą ` ąThe€literature€that€has€attempted€to€examine€theseĻā ārelationships€is€both€limited€and€inconclusive.€€SimonĻand€Gobin€(1980)€observed€that€"a€high€populationĻgrowth€rate€this€year€may€be€ņņnegativelyóó€correlated€withĻa€higher€population€growth€rate€in€earlier€decades,€andĻit€is€the€births€in€earlier€decades€that€are€the€causeĻof€the€present€increase€in€economic€growth€due€toĻhigher€population€density"€(1980:190).€€These€authorsĻmake€no€attempt€to€build€lags€into€their€modeling€toĻdirectly€test€the€hypothesis€as€advanced.Ģą ` ąCoale€also€warns€of€interpreting€contemporaneousĻcorrelations€between€population€and€per€capita€economicĻgrowth€because,€as€he€states,€"Populations€that€haveĻexperienced€a€rapid€rate€of€increase€in€per€capitaĻincome€generally€have€lower€mortality€and€lowerĻfertility€than€those€where€per€capita€income€has€grownĻmore€slowly.€€Because€both€fertility€and€mortality€areĻlower€in€such€countries,€the€rate€of€increase€(theĻdifference€between€the€birth€rate€and€death€rate)€canĻbe€either€greater€or€less"€(1986:98).€€Ģą ` ąBloom€and€Freeman,€using€contemporaneous€andĻlagged€data,€are€explicit€in€attempting€to€expose€theĻempirical€relationships.€€They€observe:€"[T]heĻstructure€of€demographic€change€from€1965„1985€has€notŠ ų*š#. Šbeen€uniform€across€country€groups,€an€indication€ofĻthe€potential€importance€of€taking€account€of€theĻcomponents€of€population€growth€in€studying€effects€ofĻpopulation€changes€on€economic€development...[because]Ļcountries€with€similar€population€growth€rates€haveĻdifferent€combinations€of€birth€and€death€rates€andĻthus€may€have€very€different€labor€supply€and€economicĻgrowth€experiences"€(1988:65).€€Ģą ` ąBlanchet€(1988)€also€presents€regression€models€ofĻper€capita€output€growth€that€simultaneouslyĻincorporate€crude€birth€(CBR)€and€death€rates€(CDR).€ĻOn€the€basis€of€cross„country€data,€he€finds€thatĻalthough€the€CBR€has€no€statistically€significantĻeffect,€reductions€in€the€CDR€stimulate€per€capitaĻeconomic€growth.€€Thus€the€form€in€which€populationĻgrowth€occurs€is€important.€€Blanchet's€modeling€doesĻnot€decompose€the€CDR€into€infant€or€child€versus€adultĻdeaths,€where€one€would€expect€different€effects,€norĻdoes€it€expose€any€dynamics€since€his€models€are€basedĻsolely€on€contemporaneous€data.€€The€empirical€resultsĻare€therefore€difficult€to€interpret.Ģą ` ąBrander€and€Dowrick€(1994)€and€Barlow€(1994)Ļrecently€exposed€these€dynamics€more€explicitly.€€BothĻstudies€show€that€past€births€contribute€to€currentŠ ų*š#. Šlabor€force€entry€and€thus€to€economic€growth,€whereasĻcurrent€births€deter€economic€growth€through€adverseĻeffects€on€investment.€€Such€modeling€is€in€the€rightĻdirection;€we€propose€to€extend€this€approach€in€theĻpresent€paper.€ĢĢą@L:ģąTHEORETICAL€MODELING€OF€ˆĢÓ  ÓECONOMIC„DEMOGRAPHIC€RELATIONSHIPSĢÓI*Ӏ€€€€We€begin€by€reviewing€and€briefly€assessing€theĻthree€major€approaches€that€have€dominated€theĻextensive€empirical€literature€on€economic-demographicĻmodeling:€€simple€correlations,€production€function,Ļand€convergence-patterns€studies.€€Our€approach€followsĻthe€third€of€these€traditions,€but€draws€on€the€othersĻas€well.€Ģ€€€€€€Simple€correlations€studies€hypothesize€that€perĻcapita€output€growth€(ņņY/Nņņgróóóó)€is€influenced€by€variousŠ    Šdimensions€of€demography,€Ģ€€€€€€€€€€€€€€€€€€€€€ņņY/Nņņgróóóó€=€ņņf(D)óó,€€€€€€€€€€€€€€€€€€(1)Š (# $ Šwhere€ņņDóó€is€usually€contemporaneous€population€growthŠ ø$°& Š(ņņNņņgróóóó),€sometimes€density,€size,€and/or€age€structure,Š H&@( Šand€occasionally€births€and/or€deaths.€€TheseĻcorrelations€provide€at€most€a€first-pass€assessment€ofĻthe€effects€of€demography;€they€are€difficult€toŠ ų*š#. Šinterpret€because€they€suppress€the€specific€channelsĻthrough€which€population€affects€the€economy.€€€Ģ€€€€€Production-function€studies€are€based€onĻestimating€variants€of€a€model,€€€€€€€€€€€€€€€€€€€€€€€€€Ļ€€€€€€€€€€€€€€€€€ņņYóó€=€ņņgóó(ņņK,€L,€H,€Róó,€š)š),€€€€€€€€€€€€€€(2)Š H @ Šwhere€output€is€produced€by€the€ņņstocksóó€of€variousĻfactor€inputs:€physical€capital€(ņņKóó),€labor€force€(ņņLóó),Š h`  Šhuman€capital€(ņņHóó;€education€and€health),€resources€(ņņRóó;Š ųš  Šland,€minerals,€and€environment),€and€technology€(š)š).€ĻBecause€data€on€these€stocks€are€difficult€to€compile,ĻEq.€(2)€is€usually€transformed€into€growth„rate€terms,Ļin€which€attention€is€focused€on€more€easily€observableĻfactor€ņņflowsóó€such€as€the€growth€of€physical€capitalĻ(net€investment€=€gross€investment€less€depreciation).€ĻDemography€is€then€linked€to€the€growth€of€factorĻinputs.€€Ģą ` ąThese€models,€however,€face€formidable€challengesĻto€empirical€implementation:€€estimates€of€depreciationĻand€depletion€are€€difficult€to€compile,€and€technologyĻand€scale,€considered€central€to€economic€growth,€haveĻproved€exceptionally€elusive€to€assess.€€Therefore,€itĻis€necessary€to€impose€constraining€assumptions€thatĻresult€in€simpler€renderings€than€desired.€€PotentiallyĻimportant€demographic€linkages€through€scale,Š ų*š#. Šdiminishing€returns,€and€technical€change€are€sometimesĻcombined€into€a€"residual."€€(See€Brander€and€DowrickĻ1994,€and€others€they€cite€in€this€tradition.)€€Thus,Ļalthough€aggregate€production€functions€representĻpromising€analytical€frameworks,€the€empiricalĻrenderings€are€quite€limited€in€scope.Ģą ` ąConvergence-patterns€or€"technology„gap"€modelsĻbuild€on€the€production„function€framework€and€exploreĻthe€relationships€between€economic€growth€and€theĻņņlevelsóó€of€economic€development.€€They€take€the€formĢ€€€€€€€€€€€ņņY/Nņņgróóóóņņ(ņņt,t+nóó)óó€=€ņņg(Y/Nņņtóóóó,€ņņXņņtóóóó;€ņņZņņ(t,t+n)óóóó),€€€€€€€€(3)Š Ø  Šwhere€the€ņņY/Nņņgróóóó€over€the€interval€(ņņt,t+nóó)€isŠ 80 Šhypothesized€to€vary€with€the€ņņinitialóó€level€of€perĻcapita€income€(ņņY/Nņņtóóóó)€and€with€ņņXņņtóóóó,€as€well€as€with€ņņZņņt,t+nóóóóŠ XP Šover€the€interval€(ņņt,t+nóó).€€The€initial€state€ofŠ čą Šeconomic€development,€Y/Nņņtóó,€encompasses€many€of€thoseŠ xp Šinfluences€which€are€difficult€to€measure€for€theĻproduction„function€rendering,€such€as€capital„to„laborĻratios,€technology,€and€human€capital.€€ņņXóó€variablesŠ (# $ Šprovide€additional€detail,€when€available,€on€theĻinitial€state€(e.g.,€educational€attainment€andĻpopulation€density).€€ņņZóó€variables€represent€factorsŠ Ų'Š * Šinfluencing€the€economic€environment€as€well€as€changesĻin€the€stocks:€€saving,€political€stability,€investmentŠ ų*š#. Šreturns,€and€the€like.€€Demographic€change€in€turn€canĻinfluence€or€serve€as€a€proxy€for€those€flow€variables.€ĻCertain€aspects€of€population€growth€directly€affectĻthe€size€of€the€labor€force,€while€the€associatedĻdependency€of€youths€and€the€aged€may€deter€saving€andĻinvestment.Ģą ` ąThe€relationship€between€ņņY/Nņņgróóóó€and€ņņY/Nņņtóóóó€is€variedŠ h`  Šand€complex.€€ņņY/Nņņtóóóó€can€be€interpreted€as€a€proxy€forŠ ųš  Škey€omitted€or€difficult„to„measure€factors€ofĻproduction.€€Insofar€as€there€are€diminishing€returnsĻto€these€factors,€the€sign€on€ņņY/Nņņtóóóó€should€be€negative.€Š Ø  ŠIndeed,€neoclassical€models€beginning€with€Solow€(1956)Ļimply€that€growth€rates€will€converge€across€countries.€Ļą ` ąThis€prediction€has€not€gone€unchallenged.€€LucasĻ(1988)€and€Rebelo€(1991)€posit€constant€returns€toĻreproducible€capital,€which€is€defined€broadly€toĻinclude€human€capital.€€As€a€result,€there€may€be€noĻassociation€between€ņņY/Nņņgróóóó€and€ņņY/Nņņtóóóó.€€Romer€(1986)Š ˜!" Šsuggests€that€per€capita€output€growth€even€mayĻincrease€with€the€level€of€per€capita€income€because€ofĻexternalities€related€to€technical€change.€€In€short,Ļthere€are€numerous€empirical€possibilities,€and€dataĻare€required€to€determine€the€importance€of€the€variousĻcompeting€forces.€€Flexible€functional€forms€betweenŠ ų*š#. ŠņņY/Nņņgróóóó€and€ņņY/Nóó€(as€illustrated€in€Figure€1)€areŠ  Štherefore€justified.Ģą ` ąIn€this€popular€convergence€literature,€the€impactĻof€demographic€factors€on€ņņY/Nņņgróóóó€is€spotty€and€mixed.€Š ø ° ŠIn€a€review€of€more€than€two€dozen€studies,€FagerbergĻ(1994)€finds€that€population€growth€is€omittedĻone-third€of€the€time.€€Where€ņņNņņgróóóó€is€included,€itsŠ h`  Šestimated€effect€is€split€equally€between€significantlyĻnegative€and€statistically€insignificant.€€No€studyĻexamines€explicitly€and€in€appropriate€detail€theĻcombined€and€separate€impacts€of€the€components€of€ņņNņņgróóóó;Š Ø  Šwe€attempt€to€remedy€that€deficiency€here.Ģ€€€€€Our€analysis€is€in€the€tradition€of€theĻconvergence-patterns€literature.€€In€contrast€to€mostĻprevious€renderings,€we€allow€1)€the€effects€ofĻdemography€to€vary€by€stage€of€economic€development;Ļ2)the€effects€of€contemporaneous€and€past€componentsĻof€births€and€deaths€to€be€revealed;€and€3)€the€fullĻimpact€of€the€1980s€to€be€calculated€by€extending€theĻstudies€from€1985€to€1990.€€We€conduct€our€empiricalĻmodeling€in€two€steps:€€1)€we€begin€with€a€demographicĻrendering€of€the€convergence„patterns€model,€and€2)€toĻā āthis€formulation€we€add€several€growth„determiningŠ h)`", Švariables€found€in€the€literature.€€Both€models€takeĻthe€form:Ģą ` ąņņY/Nņņgróóņņt,t+nóóóó€=€ņņfóó[ņņY/Nņņtóóóó,€ņņXņņtóóóó,€ņņZņņtóóóó,€{ņņDņņt,t+nóóóó,€ņņDņņt,t+nóóóó€x€ņņY/Nņņtóóóó}],€ą€hqŠ ą(4)ˆŠ (  Šwhere€ņņDóó€is€separately€the€contemporaneous€CBRņņņņt,t+nóóóó€andŠ ø ° ŠCDRņņņņt,t+nóóóó,€as€well€as€CBR€lagged€15€years.€€CBR€and€CDRŠ H @ Šare€netted€of€infant€deaths,€which€are€unlikely€toĻaffect€ņņY/Nņņgróóóó€but€which€vary€systematically€with€ņņY/Nóó.€Š h`  ŠThis€modification€facilitates€a€comparison€of€ourĻresults€with€those€in€recent€studies€(Barlow€1994;ĻBrander€and€Dowrick€1994).Ģą ` ąThe€contemporaneous€CBR€captures€mainly€theĻnegative€resource-using€effects€of€births€(especiallyĻon€saving),€although€these€effects€may€be€confused€inĻestimation€with€the€positive€(labor€force)€impacts€ofĻpast€births€because€birth€rates€are€correlated€overĻtime.€€Therefore€we€append€the€model€with€CBRņņ„15óó,Š xp Šbirths€lagged€15€years.€€Estimates€of€the€negativeĻresource-using€effects€of€contemporaneous€births€thusĻshould€increase;€and€the€estimated€effect€of€laggedĻbirths€should€be€less€negative€than€that€ofĻcontemporaneous€births,€and€it€could€even€be€positive.Ģą ` ąIn€contrast,€the€contemporaneous€CDR€captures€bothĻpositive€and€negative€effects€on€economic€growth:€ĻņņY/Nņņgróóóó€is€slowed€by€the€deaths€of€workers,€who€also€haveŠ ų*š#. Ša€relatively€high€propensity€to€save,€and€is€enhancedĻby€the€deaths€of€dependents.€€The€net€impact€of€the€CDRĻthus€varies€by€the€age€composition€of€deaths;€this€inĻturn€varies€systematically€and€in€complex€ways€withĻņņY/Nóó,€which€is€entered€directly€in€the€empirical€model.€Š H @ Š(Ideally,€age-specific€mortality€effects€should€beĻanalyzed€explicitly,€but€this€possibility€isĻconstrained€by€the€availability€of€acceptable€data€overĻtime€for€a€sufficient€number€of€LDCs.)Ģą ` ąIn€the€demographic€model,€density€is€included€toĻrepresent€arable€land€stock€per€capita€(an€ņņXóó„typeŠ Ø  Švariable)€and,€on€the€basis€of€the€rich€literature€inĻthe€Boserup€(1965)€tradition,€to€signify€positiveĻinducements€to€technical€change€(a€ņņZóó„type€variable).€Š XP ŠThe€resulting€sign€is€the€combined€influence€ofĻdiminishing€returns€to€land€(negative)€and€positiveĻeffects€on€technical€change.Ģą ` ąThe€interaction€of€ņņDóó€with€ņņY/Nóó€allows€the€impact€ofŠ ˜!" Špopulation€growth€(and€its€components)€to€vary€by€levelĻof€economic€development.€€Most€of€the€competingĻhypotheses€are€consistent€with€a€declining€(increasing)Ļnegative€(positive)€effect€of€population€growth€asĻcountries€develop.€€Such€a€trend€derives€from€anĻassociation€at€early€development€with€1)highŠ ų*š#. Šdependency€rates,2)€poorly€developed€market€andĻpolitical€systems,€and€3)the€relative€unimportance€ofĻaggregate€demand.€€€This€trend€can€be€countered€by€theĻeffects€of€technology€as€embodied€in€new€investments,Ļespecially€in€the€low„income,€low„capital€stockĻsetting.€€Ģą ` ąTo€augment€the€demographic€model,€the€literatureĻprovides€a€wide€selection€of€variables.€€Our€somewhatĻlimited€goal€is€to€ascertain€whether€our€demographicĻresults€are€modified€by€adding€a€defendable€set€of€suchĻvariables.€€Accordingly€we€append€a€set€of€variablesĻfrom€the€widely€discussed€Barro€(1991)€model.€€TheseĻallow€for€the€effects€of€primary€and€secondaryĻenrollment€rates€at€time€ņņtóó€(taken€as€rough€proxies€forŠ XP Šincreases€in€the€stock€of€human€capital),governmentĻexpenditure€shares€over€the€period€(taken€asĻpotentially€affecting€productivity€and€private€saving),Ļand€distortions€in€the€price€of€capital€(absoluteĻdeviations€from€the€sample€mean€purchasing€power€parityĻindex€of€investment).€€€Ģą ` ąFurther€clarification€of€the€above€models€wouldĻrequire€dynamic€modeling;€this€rapidly€can€becomeĻcomplex,€even€when€supplied€with€constrainingĻassumptions€to€enable€mathematical€tractability.€€AŠ ų*š#. Špathbreaking€model€is€provided€by€Lee€(1974),€who€showsĻthat€in€a€framework€incorporating€demographic€andĻeconomic€lags,€economies€and€populations€may€followĻnumerous€uneven€trajectories.€€Lee's€results€show€(anĻimportant€point€for€us)€that€a€cross-sectionalĻempirical€examination€of€such€relationships€must€beĻinterpreted€carefully€because€it€is€difficult€toĻdistinguish€between€long-run€equilibria€and€adjustmentsĻto€these€equilibria.€ĢĢą@Ä;ģąEMPIRICAL€MODELING€OF€ˆĢÓ  ÓECONOMIC„DEMOGRAPHIC€RELATIONSHIPSĢÓqVÓą ` ąOur€data€set€consists€of€a€panel€of€three€growthĻperiods€(1960„1970,€1970„1980,€1980„1990)€and€89Ļcountries€with€populations€exceeding€1€million€in€1960,Ļwhich€Summers€and€Heston€(1994)€classify€as€market„¼ńƒń¼ńƒńoriented€and€for€which€they€provide€data€on€grossĻdomestic€product€in€constant€purchasing€power€parityĻdollars.€€(We€exclude€countries€with€missing€data,Ļextensive€dependency€on€resources,€and€problems€withĻdata€definitions.€€For€details,€see€Kelley€and€SchmidtĻ1994.)€€Ģą ` ąWe€pooled€the€three€periods€in€most€of€ourĻestimation€to€take€advantage€of€the€time„series€natureŠ ų*š#. Šof€the€data.€€Alternative€estimation€€approaches€can€beĻillustrated€by€the€following€rendering€of€Eq.€(4):ĢÓ  ÓņņY/Nņņgróóņņitóóóó€=€ššņņņņióóóó€+€ššņņņņtóóóó€+€šš€ņņlnóó(ņņY/Nóó)ņņņņitóóóó€+€šš€ņņXņņitóóóó€+€š š€ņņZņņitóóóó€+Š (  ŠÓ‘YӀ€€€€€€€€€€€€€š š€ņņDņņitóóóó€+€šš€{ņņDóó€x€ņņY/Nóó}ņņņņitóóóó€+€š šņņņņitóóóó;€€€€€€€€ą€hqŠ ą(5)ˆŠ Ę ¾ Šwhere€the€variables€are€described€above€in€conjunctionĻwith€Eqs.€(3)€and€(4).€€As€specified,€Eq.€(5)€providesĻfor€individual€terms€for€each€country€(ššņņņņióóóó)€as€well€asŠ „|  Šeach€time€period€(ššņņņņtóóóó).€€Assumptions€made€about€theseŠ   Šterms€dictate€the€method€of€estimation.€€If€one€assumedĻthese€terms€to€be€constant€across€all€countries€andĻperiods,€then€a€single€intercept€would€suffice€andĻordinary€least€squares€regression€would€be€appropriate.€Ļ(Below€we€refer€to€this€approach€as€"pooled"Ļestimation).Ģą ` ąIn€contrast,€fixed„effects€modeling€(FEM€orĻcovariance€modeling)€provides,€in€our€case,€for€89Ļseparate„country€intercepts€and€three€individual„periodĻintercepts.€€The€country€binaries€control€for€ņņY/NņņgróóóóŠ Ā!ŗ" Šinfluences€that€vary€across€countries€but€areĻreasonably€constant€over€time€within€countries.€€SuchĻinfluences€might€include€natural€resources€or€culturalĻattitudes€toward€work€and€saving.€€In€the€convergenceĻliterature,€these€influences€typically€are€includedĻamong€the€"conditioning"€variables.€€The€time€binariesŠ "+$. Šcontrol€for€global,€period„specific€influences€such€asĻthe€oil€shocks€of€the€1970s€and€the€worldwide€recessionĻof€the€1980s.Ģą ` ąFinally,€random„effects€modeling€(REM€or€errorĻcomponents€modeling)€treats€ššņņņņióóóó€and€ššņņņņtóóóó€as€separableŠ H @ Šcomponents€of€the€error€term,€which€adhere€to€theĻclassical€assumptions.€€REM€is€more€efficient€than€FEMĻbecause€it€models€the€many€intercepts€as€two€stochasticĻterms.€€REM,€however,€can€introduce€bias€andĻinconsistency€into€the€estimates€if€theĻcharacteristic(s)€modeled€by€ššņņņņióóóó€and€ššņņņņtóóóó€are€correlatedŠ ¶® Šwith€other€independent€variable(s);€this€is€aĻpossibility€in€our€case.€€For€example,€if€naturalĻresources€are€part€of€ššņņņņióóóó€and€are€correlated€withŠ tl Šņņln(Y/N)óó,€then€the€REM€coefficient€estimate€will€beŠ ü Šbiased.Ģą ` ąWhich€modeling€approach€is€preferable?€€A€priori,ĻFEM€edges€out€REM€in€situations€in€which€the€sampleĻrepresents€a€sizable€proportion€of€the€population,€aĻfeature€of€our€specific€case.€€In€any€event,Ļstatistical€tests€are€available€for€comparing€models.ĻFor€our€models€and€data,€FEM€consistently€dominatesĻboth€pooled€(as€per€chi„square€and€F„tests)€and€REM€(asĻper€the€Hausman€test)€estimation.€€Š + $. Šą ` ąThree€additional€modeling€issues€merit€attention.€ĻThe€first€two,€simultaneity€and€persistence,€areĻrelated.€€With€respect€to€possible€simultaneity€betweenĻleft„€and€right„hand„side€variables,€most€authors€wouldĻagree€that€births€and€deaths€are€influenced€by€theĻņņlevelóó€rather€than€by€short„term€ņņchangesóó€in€ņņY/N;óó€thus,Š ŲŠ  Šexplicit€modeling€to€account€for€simultaneity€may€notĻbe€necessary.€€On€the€other€hand,€when€data€areĻaggregated€over€long€growth€periods,€differential€ņņY/NņņgróóóóŠ ˆ€  Šgrowth€rates€can€alter€ņņY/Nóó€enough€to€substantivelyŠ  Šinfluence€the€pace€of€demographic€change.€€As€a€roughĻassessment€of€this€possibility,€we€compared€one„,€five„¼ńƒń¼ńƒń,€and€10„year€periods€and€found€the€resulting€estimatesĻto€be€relatively€insensitive,€at€least€up€to€our€10„¼ńƒń¼ńƒńyear€aggregation€(Kelley€and€Schmidt€1994,€29„33).€ĻAnalytically€one€could€reduce€problems€withĻsimultaneity€by€reducing€the€period€to€(say)€fiveĻyears,€but€at€the€potential€cost€of€encounteringĻproblems€with€"persistence"„„matching€smoothly€changingĻ(long„term)€demographic€trends€with€relatively€erraticĻ(short„term)€variations€in€ņņY/Nņņgróóóó.€€Ten„yearŠ H&@( Šaggregations€attenuate€(though€they€do€not€eliminate)Ļthis€difficulty.€€Decennial€periods€do,€however,€embodyĻmore€"real"€demographic€information€because€five„yearŠ ų*š#. Šaggregations€rely€on€the€assumptions€inherent€inĻextrapolations€between€decennial€censuses.Ģą ` ąThe€third€modeling€issue€concerns€the€selection€ofĻcontrol€variables€and€their€functional€form.€€OurĻapproach€was€to€explore€two€aspects€of€a€"convergence"Ļmodel„„the€functional€form€of€ņņY/Nóó,€and€the€set€ofŠ ŲŠ  Šconditioning€(ņņXóó€and€ņņZóó)€variables.€€We€examined€threeŠ h`  Šfunctional€forms€for€ņņY/Nóó:€€linear€(as€in€Barro€1991),Š ųš  Šlog€(as€in€Barro€and€Lee€1993),€and€cubic€(as€in€KelleyĻand€Schmidt€1994).€€Adjusted€Rņņ2óó€was€consistentlyŠ  Šhighest€with€the€log€form,€with€no€notable€differencesĻin€demographic€effects.€€Consequently€we€employ€ņņln(Y/N)óóŠ 80 Šin€our€estimation.€€Among€his€control€variables,€BarroĻ(1991)€included€1)primary€and€secondary€enrollmentĻrates,€2)government€expenditure€shares,€and€3)a€proxyĻfor€distortions€in€investment€prices.€€Initially€weĻincluded€these€in€our€modeling€as€well,€but€theyĻcomplicated€the€model€and€their€inclusion€barelyĻchanged€the€demographic€coefficients€or€ņņtóó„values.€Š (# $ ŠConsequently,€we€present€largely€demographic€models.€Ļ(Results€from€these€initial€runs€and€any€other€resultsĻcited€here€are€available€from€the€authors.)ĢŠ  h)`", ŠÓ  ÓRESULTSĢÓpoÓą ` ąTable€1€summarizes€our€findings€for€threeĻdecennial€periods.€€Columns€1„3€present€three€fixed„¼ńƒń¼ńƒńeffects€models€(FEM)€of€increasing€demographic€detail;ĻColumns€4„7€replicate€the€model€in€Column€3,€first€withĻpooled€cross„sectional€data€(Col.€4)€and€then€withĻcross„sections€by€decade€(Cols.€5„7).Ģą ` ąOverall,€1)the€preferred€FEM€regressions€areĻconsistent€with€the€neoclassical€prediction€of€economicĻgrowth„rate€convergence€(the€sign€on€ņņln(Y/N)óó€isŠ  Šnegative);€2)density€appears€to€exert€a€positiveĻeffect€on€economic€growth€(Boserup„type€influences€mayĻdominate€Malthusian€diminishing€returns);€andĻ3)demography€matters€considerably,€but€in€complexĻways.Ģą ` ąAt€the€simplest€level,€for€the€30„year€panel,Ļpopulation€growth€exerts€a€large€negative€effect€onĻeconomic€growth€(Col.€1,€joint€tests,€FEM).€€TheĻpartial€derivative€of€ņņY/Nņņgróóóó€with€respect€to€ņņNņņgróóóó,Š (# $ Ševaluated€at€the€LDC€median€ņņY/Nóó,€is€„.64.€€This€impliesŠ ø$°& Šthat€a€large€(unit)€reduction€of€ņņNņņgróóóó€from€its€median€ofŠ H&@( Š2.54€to€1.54€would€increase€ņņY/Nņņgróóóó€from€(for€example)Š Ų'Š * Šits€median€of€1.36€to€2.00.€€This€result€contrasts€withĻthe€nonsignificant€findings€in€the€literature€for€theŠ ų*š#. Š1960s€and€1970s,€and€conforms€to€the€new€findings€forĻthe€1980s.Ģą ` ąTo€more€fully€understand€these€results,€it€isĻuseful€to€examine€the€separate€effects€of€theĻcomponents€of€population€growth„„specifically,Ļincreases€(decreases)€in€the€CBR€and€CDR€(Col.€2).€€OnĻthe€one€hand,€an€increase€in€the€CBR€strongly€reducesĻeconomic€growth,€possibly€through€a€negativeĻdependency„rate€impact€on€saving.€€This€effect€isĻinvariant€to€the€level€of€income.€€On€the€other€hand,€aĻdecrease€in€the€CDR€increases€economic€growth,Ļespecially€in€the€LDCs,€where€mortality€reduction€isĻconcentrated€in€the€younger€and/or€working€ages.€€InĻcontrast,€in€the€DCs€such€gains€have€occurred€in€theĻretired€cohort,€whose€members€not€only€are€unproductiveĻin€generating€income,€but€also€incur€relatively€highĻhealth€expenditures.€Ģą ` ąModel€3€further€exposes€the€effects€of€demographyĻby€attempting€to€account€for€the€impacts€of€past€birthsĻ(lagged€15€years).€€The€results€conform€with€our€aĻpriori€expectations.€€First,€the€negative€effect€ofĻcurrent€births€increases€when€purged€of€some€of€theĻcollinearity€with€past€births.€€Although€theĻinteraction€term€is€insignificant,€this€negative€effectŠ ų*š#. Šappears€to€be€relatively€stronger€in€the€LDCs,€possiblyĻbecause€of€weaker€capital€markets€and€thus€of€strongerĻdependency€effects.€€Second,€the€positive€effect€ofĻdeath„rate€reductions€increases€as€well,€especially€inĻthe€LDCs.€€Finally,€past€births€exert€a€somewhatĻsmaller€negative€effect€than€contemporaneous€birthsĻ(especially€in€the€LDCs),€plausibly€because€theĻpositive€effects€of€labor€force€entry€offset€theĻnegative€resource„using€effects€of€youth€dependency.€ĢņņChanges€over€Timeóó€€Ģą ` ąThe€last€four€columns€of€Table€1€present€cross„¼ńƒń¼ńƒńsectional€results,€first€pooled€and€then€by€decade.€ĻPooled€estimation€is€fundamentally€different€from€FEM,Ļwhich€controls€for€unspecified€period„€and€country„¼ńƒń¼ńƒńspecific€effects€on€ņņY/Nņņgróóóó.€€Nevertheless,€our€majorŠ čą Šconclusions€with€respect€to€the€impacts€of€demographyĻare€invariant€to€estimation€procedure€of€Columns€3€vs.Ļ4.€€This€finding€partially€justifies€our€examining€theĻdecennial€cross„section€patterns€(Cols.€5„7)€in€anĻattempt€to€ferret€out€temporal€trends€of€interest.Ģą ` ąThe€results€are€intriguing,€and€they€stronglyĻreinforce€our€strategy€of€demographic€decompositionĻwith€some€attention€to€variations€by€level€of€income.€Ļ(The€full€set€of€demographic€terms€are€jointlyŠ ų*š#. Šsignificant€at€the€1%€level€in€each€period,€while€theĻinteractions€with€Y/N€are€jointly€significant€at€the€5%Ļlevel€in€the€1970s€and€1980s.)€€In€contrast€to€theĻfindings€based€on€FEM€(Col.€1),€which€show€an€overallĻnegative€effect€of€ņņNņņgróóóó€over€the€30„year€period,€theŠ H @ Šeffect€of€ņņNņņgróóóó€in€comparable€cross„sections€(not€shown)Š ŲŠ  Šis€statistically€insignificant€in€the€1960s€and€1970s,Ļand€strongly€negative€in€the€1980s.€Ģą ` ąCross„sectional€Columns€5„7€parallel€FEM€Column€4Ļin€extending€the€simplistic€ņņNņņgróóóó€rendering.€€To€assessŠ  Šthe€sources€and€magnitudes€of€these€demographicĻeffects,€Table€2€presents€the€partial€derivatives€ofĻņņY/Nņņgróóóó€with€respect€to€the€three€demographic€componentsŠ ČĄ Šof€evaluated€at€LDC€and€DC€median€ņņY/Nóós€for€each€period.€Š XP ŠThese€statistics€are€comparable€both€over€time€andĻacross€demographic€components€because€the€metrics€areĻthe€same€(births€and€deaths€per€1,000),€and€eachĻcontrols€for€interactions€of€demography€with€ņņY/Nóó.€Š ˜!" ŠAlthough€some€of€the€effects€in€Table€1€are€notĻstatistically€significant€for€a€given€decade,Ļadditional€testing€(not€shown€here€because€of€spaceĻlimits)€reveals€statistically€significant€coefficientĻchanges€over€time€for€each€of€the€combined€pairs€ofĻCBR,€CDR,€and€CBRņņ„15óó.Š ų*š#. Šą ` ąAmong€the€LDCs,€three€trends€are€apparent.€€First,Ļthe€effect€of€the€CBR€becomes€increasingly€negativeĻover€time,€with€a€sizable€jump€in€the€1980s.€€We€canĻspeculate€that€saving/dependency€effects€may€haveĻbecome€increasingly€powerful€over€time,€and€especiallyĻin€the€1980s,€a€period€of€capital€shortages,Ļlanguishing€aggregate€saving,€high€real€interest€rates,Ļand€exceptionally€rapid€debt€accumulation.€€Ģą ` ąSecond,€although€death„rate€reductions€contributedĻpositively€in€each€decade,€this€contribution€declinedĻmonotonically€over€time.€€This€apparently€bafflingĻresult€is€clarified€somewhat€by€a€comparison€of€medianĻCDRs€across€time€and€level€of€development.€€By€theĻ1970s€the€CDR€in€the€median€LDC€had€fallen€to€the€levelĻof€the€median€DC;€by€the€1980s€the€median€LDC€was€wellĻbelow€that€level.€€Part€of€this€disparity€is€due€toĻaging€in€the€DCs,€but€another€portion€is€the€result€ofĻreductions€in€infant€and€youth€mortality€in€the€LDCsĻover€the€1960s€and€1970s.€€Many€of€the€more€easilyĻachieved€growth„enhancing,€mortality„reducing€advancesĻmay€have€been€largely€exhausted€in€most€of€the€LDCs€byĻthe€1980s.Ģą ` ąFinally,€the€impact€of€lagged€births,€initiallyĻnegative,€becomes€positive€and€is€large€in€the€1980s.€Š ų*š#. ŠIndeed,€in€the€1980s,€the€positive€(presumably€labor„¼ńƒń¼ńƒńforce)€effects€of€past€births€reduce€the€negativeĻeffect€of€contemporaneous€births€(possibly€on€saving)Ļby€half€(„.34€versus€.17).€€This€result€should€put€toĻrest€the€notion€that€births€have€uniformly€negativeĻeffects€on€growth.€€The€assessment€clearly€depends€onĻwhen€the€impacts€are€evaluated€or,€stated€differently,Ļwhether€one€takes€a€short„term€or€a€longer„term€view.€ĻLonger„term€assessments€are€necessary€for€a€reasonablyĻcomplete€and€balanced€perspective.€€It€is€unclear€whyĻthe€positive€effects€of€past€births€increase€over€time,Ļalthough€the€previously€noted€health€gains€among€youthsĻand€the€gains€in€education€plausibly€account€for€aĻportion€of€this€trend.€€After€all,€labor€force€entrantsĻin€the€1980s€were€healthier€and€better€educated€thanĻthose€in€the€1960s,€with€a€resulting€positiveĻcontribution€to€economic€growth.Ģą ` ąA€somewhat€different€picture€emerges€from€the€DCĻresults.€€Negative€effects€of€birth€rate€are€also€foundĻhere;€they€are€large€in€the€1960s€and€1970s,€but€fairlyĻsmall€in€the€1980s.€€The€positive€effects€of€CDRĻreduction€diminish€over€time,€and€in€fact€becomeĻnegative€in€the€1980s.€€Finally,€the€effects€of€pastĻbirths€are€small.€€These€results€support€the€strategyŠ ų*š#. Šof€assessing€demographic€impacts€by€stage€ofĻdevelopment,€most€clearly€seen€in€the€results€for€theĻCDR.€€Reductions€in€the€CDR€may€well€be€concentrated,Ļrelatively,€in€the€older€cohorts€in€DCs,€resulting€in€aĻnegative€impact€on€economic€growth€due€to€highĻexpenditures€(especially€for€health€care)€and€lowĻearnings€among€the€aged.ĢņņRobustness€of€Resultsóó€€€Ģą ` ąHow€robust€are€our€various€findings?€€To€assessĻthis€question,€we€reran€the€results€in€Table€1,Ļ1)excluding€countries€identified€by€the€difference€inĻfits€(DFFITS)€statistic€as€exerting€unusual€influenceĻon€the€results,€2)using€data€aggregated€byĻquinquennial€(rather€than€decennial)€periods,€andĻ3)employing€OLS€(rather€than€White„corrected)€ņņtóó„Š čą ŠńƒńŠ čą Šńƒństatistics€for€the€pooled€and€cross„sectionalĻregressions.€€Generally€our€findings€are€highly€robustĻwith€respect€to€these€modeling€variations:€€90%€of€theĻparameters€found€to€be€statistically€significant€inĻTable€1€remain€so€in€the€modeling€variants.€€In€most€ofĻthe€cases€where€changes€are€observed,€the€statisticalĻsignificance€increases€and€this€strengthens€our€majorĻfindings.€ĢŠ  ų*š#. Šą@ˆ;ģąSUMMARY€AND€CONCLUSIONSˆĢą ` ąThe€catalyst€for€this€paper€was€an€empiricalĻfinding,€based€on€cross„country€data€and€now€wellĻestablished€in€the€literature,€that€the€rate€ofĻpopulation€growth€does€not€appear€to€have€a€notableĻimpact€on€per€capita€output€growth„„at€least€in€theĻ1960s€and€1970s.€€Recent€studies,€however,€show€that€aĻstrong€and€robust€negative€impact€emerged€in€the€1980s.€ĻTwo€questions€immediately€arise:€€1)Are€theseĻsignificant€negative€correlations€a€short„termĻaberration,€specific€to€the€conditions€of€the€1980s?€Ļ2)Is€the€change€to€a€negative€correlation€the€resultĻof€changes€in€the€population„growth€coefficients€and/orĻchanges€in€the€components€of€population€growth?€€TheĻfirst€question€is€beyond€the€scope€of€this€paper,€butĻwe€addressed€the€second€by€decomposing€net€populationĻgrowth€into€its€birth,€death,€and€labor€force€growthĻcomponents.€€We€found€that€indeed,€the€effects€of€theĻcomponents€changed€in€the€1980s.Ģą ` ąThe€results€from€Tables€1€and€2,€taken€as€a€whole,Ļare€consistent€with€the€following€interpretation€of€theĻeffects€of€demographic€change€on€per€capita€outputĻgrowth.€€In€the€1960s€and€1970s,€the€ņņnetóó€impact€ofĻpopulation€growth€appeared€to€be€ņņnilóó,€but€the€ņņseparateóóŠ ų*š#. Šeffects€of€the€components€of€this€change€were€ņņsizableóó.€ĻIndeed,€the€negative€costs€of€high€birth€rates€wereĻlargely€offset€by€the€positive€contributions€ofĻmortality€reduction.€€In€the€1980s,€however,€the€short„¼ńƒń¼ńƒńterm€costs€of€high€birth€rates€increased,€especially€inĻthe€LDCs.€€Also,€even€though€the€favorable€effects€ofĻpast€births€on€current€labor€force€growth€wereĻpositive,€they€were€not€sufficient€to€offset€theĻcurrent€forgone€growth„rate€costs€of€births.€€TheĻresulting€net€impact€on€population€growth€becameĻnegative.€€This€change€was€reinforced€by€a€reduction€inĻthe€impact€of€mortality€declines;€the€declines€wereĻconcentrated€in€the€early€ages,€where€the€effect€onĻcurrent€labor€force€entry€was€nil.Ģą ` ąThese€results€must€be€viewed€in€perspective.€€OnĻthe€one€hand,€aggregate€correlations€are€difficult€toĻinterpret€and€must€be€considered€as€only€one€piece€ofĻevidence€on€the€impacts€of€demographic€change.€€On€theĻother€hand,€these€correlations€have€profoundly€affectedĻthe€"population€debates."€€There€is€no€reason€toĻbelieve€that€their€influence€will€wane,€especially€inĻthe€face€of€the€increasing€challenge€of€generatingĻsupport€for€population€programs€during€a€period€ofĻdeclining€birth€rates.€€As€a€result,€the€"new€negativeŠ ų*š#. Šcorrelations"€predictably€will€attract€excessiveĻattention,€just€as€the€"old€no€correlations"€did€inĻpast€decades.€€Therefore,€it€is€important,€to€gainĻinsight€into€the€possible€meaning€of€the€new€results.€ĻAssessing€the€underlying€components€of€change,€and€howĻtheir€effects€have€changed€over€time,€is€a€first€stepĻin€this€direction.€€Ģą ` ąThis€analysis€also€provides€a€more€balancedĻperspective€on€the€consequences€of€demographic€changeĻbecause€it€rests€on€the€proposition€that€populationĻgrowth€has€both€negative€and€positive€effects,€and€thatĻthese€effects€vary€over€the€short€and€the€long€term.€ĻEmbracing€such€a€perspective,€more€than€any€otherĻfactor,€will€place€the€population€debates€on€a€firmerĻfooting,€more€conducive€to€advancing€knowledge€in€whatĻhas€become€an€inordinately€contentious€area€ofĻscientific€dialogue.€€As€this€paper€demonstrates,Ļpopulation€growth€is€not€all€good€or€all€bad€forĻeconomic€growth:€€it€contains€both€elements,€which€canĻand€(as€in€the€results€presented€above)€do€change€overĻtime.€€An€examination€of€the€underlying€nature€of€theĻnew€correlations€provides€a€useful€case€study€of€thisĻā āproposition,€which,€if€accepted,€can€go€a€long€wayŠ h)`", Štoward€transforming€the€heat€of€the€population€debatesĻinto€some€useful€light.Š  ˜ Šą@”AģąREFERENCESˆĢBarlow,€R.€1994.€"Population€Growth€and€EconomicĢ€ą ` ąGrowth:€€Some€More€Correlations."€€ņņPopulation€andŠ (  Šą ` ąDevelopment€Reviewóó€20:153„65.ĢBarro,€R.J.€€1991.€€"Economic€Growth€in€a€Cross€SectionĢ€ą0 ` ąof€Countries."€€ņņQuarterly€Journal€of€EconomicsóóŠ ŲŠ  Š106:407„44.Š ` Š ` Š  ŠBarro,€R.J.€and€J.€Lee.€€1993.€€"Losers€and€Winners€inĢą0 ` ąEconomic€Growth."€€Working€Paper€4341.€€Cambridge,ĻMA:€€National€Bureau€of€Economic€Research.Š ` Š ` Š  ŠBlanchet,€D.€1988.€"A€Stochastic€Version€of€theĢą0 ` ąMalthusian€Trap€Model:€€Consequences€for€theĻEmpirical€Relationship€between€Economic€Growth€andĻPopulation€Growth€in€LDC's."€€ņņMathematicalĻPopulation€Studiesóó€1(1):79„99.Š ` Š ` Š  ŠBloom,€D.E.€and€R.B.€Freeman.€1988.€"EconomicĢą0 ` ąDevelopment€and€the€Timing€and€Components€of€ą @x ą¼ńƒń¼ńƒńPopulation€Growth."€€ņņJournal€of€Policy€ModelingóóĻ10(1):57„81.Š ` Š ` Š  ŠBoserup,€E.€€1965.€€ņņThe€Conditions€of€AgriculturalĢą0 ` ąGrowth:€€The€Economics€of€Agrarian€Change€underĻPopulation€Pressureóó.€€Chicago:€Aldin€E.Š ` Š ` Š  Šā āBrander,€J.A.€and€S.€Dowrick.€1994.€"The€Role€ofŠ h)`", Šą0 ` ąFertility€and€Population€in€Economic€Growth:€Ļā āEmpirical€Results€from€Aggregate€Cross„NationalĻData."€€ņņJournal€of€Population€Economicsóó€7:1„25.Š ` Š ` Š  ŠCoale,€A.J.€1986.€"Population€Trends€and€EconomicĢą0 ` ąDevelopment."€€Pp.€96„104€in€ņņWorld€Population€andĻU.S.€Policy:€€The€Choices€Aheadóó,€edited€by€J.€ą @š ą¼ńƒń¼ńƒńMencken.€€New€York:€€Norton.Š ` Š ` Š  ŠFagerberg,€J.€€1994.€€"Technology€and€InternationalĢą0 ` ąDifferences€in€Growth€Rates."€€ņņJournal€of€EconomicĻLiteratureóó€32:1147„75.Š ` Š ` Š  ŠKelley,€A.C.€1988.€"Economic€Consequences€of€PopulationĢą0 ` ąChange€in€the€Third€World."€€ņņJournal€of€EconomicĻLiteratureóó€27:1685„1728.Š ` Š ` Š  Š___________.€1995.€"Revisionism€Revisited:€€An€Essay€onĢą0 ` ąthe€Population€Debate€in€Historical€Perspective."€ĻForthcoming€in€ņņPopulation,€Development€andĻWelfare:€€The€Nobel€Jubilee€Symposium€inĻEconomicsóó,€edited€by€R.€Ohlsson.€€Berlin:€ĻSpringer„Verlag.Š ` Š ` Š  ŠKelley,€A.C.€and€R.M.€Schmidt.€1994.€"Population€andĢą0 ` ąIncome€Change:€€Recent€Evidence."€€World€BankĻDiscussion€Paper€249.€€Washington,€DC:€€WorldĻBank.Š ` Š ` Š  ŠLee,€R.D.€1974.€€"The€Formal€Dynamics€of€ControlledŠ ų*š#. Šą0 ` ąPopulations€and€the€Echo,€the€Boom,€and€the€Bust."€ĻņņDemographyóó€11:563„85.Š ` Š ` Š  Š________.€1983.€"Economic€Consequences€of€PopulationĢą0 ` ąSize,€Structure€and€Growth."€€ņņInternational€UnionĻfor€the€Scientific€Study€of€Population€Newsletteróó€Ļ17:43„59.Š ` Š ` Š  ŠLucas,€R.E.€1988.€"On€the€Mechanics€of€EconomicĢą0 ` ąDevelopment."€€ņņJournal€of€Monetary€Economicsóó€22:3„¼ńƒń¼ńƒń42.Š ` Š ` Š  ŠRebelo,€S.€1991.€"Long„Run€Policy€Analysis€and€Long„RunĢą0 ` ąGrowth."€€ņņJournal€of€Political€Economyóó€99:1001„37.Š ` Š ` Š  ŠRomer,€P.M.€1986.€"Increasing€Returns€and€Long„RunĢą0 ` ąGrowth."€€ņņJournal€of€Political€Economyóó€94:1002„¼ńƒń¼ńƒń1037.Š ` Š ` Š  ŠSimon,€J.€and€R.€Gobin.€1980.€"The€Relationship€betweenĢą0 ` ąPopulation€and€Economic€Growth€in€LDCs."€€Pp.€215„¼ńƒń¼ńƒń34€in€ņņResearch€in€Population€Economicsóó,€Vol.€2,Ļedited€by€J.L.€Simon€and€J.€DaVanzo.€€Greenwich,ĻCT:€€JAI.Š ` Š ` Š  ŠSolow,€R.M.€1956.€"A€Contribution€to€the€Theory€ofĢą0 ` ąEconomic€Growth."€ņņQuarterly€Journal€of€EconomicsóóĻ70(1):65„94.Š ` Š ` Š  Šā āSummers,€R.€and€A.€Heston.€1994.€"Data€Update€5.5,"Š h)`", Šą0 ` ącomputer€diskette€based€on€ņņThe€Penn€World€TableĻā ā(Mark€5):€An€Expanded€Set€of€InternationalĻComparisons,€1950„1988óó.€€Cambridge,€MA:€€NationalĻBureau€of€Economic€Research.€Š ` Š ` Š  ŠŠ  H @ ŠÓČÓŅ„ŅŅXŅŌ€ōģŪōöXXģŪŌŃ7€ōōd§ōōd§7ŃTable€1.€€Kelley/Schmidt€Basic€Model€with€Components€of€Demographic€Change:€€DependentŠ  Š€€€€€€€€€€Variable,€Y/NņņgróóŠ ˜ ŠŌ  ŌŌ" ’Č Ą( Č Ų'`Ą`"Ōņņą€€%…€%ąóóˆĢŌ" ’ų šX Pų Ų'`š`"Ōņņ€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€óóĢÓ€ÓĢÓ€ÓÓ-„4€ Ģd°ü!°›X-Óą 4 ąņņ€€€€€€€€Panel€Estimation€€€€€€€óóą  ąņņ€€€€€€€€Cross„Sectional€Estimation€€€€€€€€óóŠ “ ¬  Šą 4 ąņņ€€Fixed„Effects€Modeling€(FEM)€óóą  ąņņ€€Pooled€óóą d ąņņ€1960„70€óóą ° ąņņ€1970„80€óóą ü! ąņņ€1980„90€óóŠ |t  Šą 4 ą€€€(1)ą € ą€€€(2)ą Ģ ą€€€(3)ą  ą€€€(4)ą d ą€€€(5)ą ° ą€€€(6)ą ü! ą€€€(7)Š D<  Šņņą€€%…€%ąóóˆĢĢln(Y/N)ą 4 ą€-4.03*ą € ą€-6.32*ą Ģ ą€-5.72*ą  ą€-0.42ą d ą€€0.14ą ° ą€€0.29ą ü! ą€-2.01*Š œ”  ŠĢNņņgróóą 4 ąš1š-0.54€š7šŠ ,$  ŠNņņgróó€x€Y/Ną 4 ąš%š-0.09€š+šŠ ōģ  ŠĢCBRą 4 ąą € ąš1š-2.29*š7šą Ģ ąš1š-2.98*š7šą  ąš1š-2.07*š7šą d ąš1š€1.75€š7šą ° ą€-1.67ą ü! ąš1š-3.91*š7šŠ „| ŠCBR€x€Y/Ną 4 ąą € ąš%š€0.00€š+šą Ģ ąš%š€0.17€š+šą  ąš%š€0.25€š+šą d ąš%š-0.62*š+šą ° ą€-0.02ą ü! ąš%š€0.32€š+šŠ LD ŠĢCBRņņ„15óóą 4 ąą € ąą Ģ ąš1š-0.01€š7šą  ąš1š€0.88€š7šą d ą€-2.54*ą ° ą€€1.29ą ü! ą€€2.10Š ÜŌ ŠCBRņņ„15óó€x€Y/Ną 4 ąą € ąą Ģ ąš%š-0.34*š+šą  ąš%š-0.30*š+šą d ą€€0.49*ą ° ą€-0.17ą ü! ą€-0.24Š ¤œ ŠĢCDRą 4 ąą € ąš1š-1.72€š7šą Ģ ąš1š-2.18€š7šą  ą€-0.88ą d ąš1š-2.96*š7šą ° ą€-2.48ą ü! ą€-0.55Š 4, ŠCDR€x€Y/Ną 4 ąą € ąš%š€0.43*š+šą Ģ ąš%š€0.64*š+šą  ą€€0.03ą d ąš%š-0.07€š+šą ° ą€€0.10ą ü! ą€€0.21Š üō ŠĢDensity€ą 4 ą€€2.97*ą € ą€€2.64*ą Ģ ą€€2.58*ą  ą€€0.92*ą d ą€€1.12*ą ° ą€€0.72*ą ü! ą€€0.92*Š Œ„ ŠĢConstantą 4 ą€€5.68*ą € ą€13.31*ą Ģ ą€16.49*ą  ą€€6.87*ą d ą€€9.30*ą ° ą€€6.28*ą ü! ą€€6.13*Š  ŠĢRņņ2óó€ą 4 ą€€0.73ą € ą€€0.75ą Ģ ą€0.77ą  ą€€0.23ą d ą€€0.37ą ° ą€€0.39ą ü! ą€€0.52Š ¬¤! ŠAdj€Rņņ2óóą 4 ą€€0.58ą € ą€€0.61ą Ģ ą€0.63ą  ą€€0.20ą d ą€€0.31ą ° ą€€0.33ą ü! ą€€0.47Š t l" ŠĢStd€Errorą 4 ą€€1.56ą € ą€€1.49ą Ģ ą€1.47ą  ą€€2.16ą d ą€€1.72ą ° ą€€1.84ą ü! ą€€1.65Š "ü$ ŠĢ#€of€Obser„ą 4 ą€€€267ą € ą€€€267ą Ģ ą€€267ą  ą€€€267ą d ą€€€€89ą ° ą€€€€89ą ü! ą€€€€89Š ”#Œ& ŠvationsĢņņą€€%…€%ąóóˆĢŌ€©Iœ ©ōōģŪŌŌ€ ©1שĩ©IœŌÓ–ČÓNotes:€€Bracketed€items€are€jointly€significant€at€the€.05€level.€€Standard€errors€in€the€cross„sectional€estimates€areŠ ģ%ä) ŠWhite„corrected€for€possible€heteroscedasticity.€€In€all€cases,€FEM€dominates€pooled€(as€per€chi„square€and€F„tests)€andĻREM€(as€per€the€Hausman€test)€at€ņņpó󀚚.01.€€Per€capita€GDP€(Y/N)€is€in€1985€US$1,000€and€was€obtained€from€Summers€andĻHeston€(1994).€€Crude€birth€and€death€rates€(CBR,€CDR)€are€measured€per€100€population€in€this€table€for€scalingĻconvenience,€and€represent€1992€United€Nations€estimates.€€Population€and€density€(1000€population€per€square€kilometer)Ļvalues€are€from€the€World€Bank€research€database.€€Growth€rates€are€continuous€rates€in€percentage€terms.€€BecauseĻdemographic€data€are€unavailable€before€1950,€lagged€values€for€the€1960s€apply€to€1950„1955;€for€the€1970s,€to€1955„¼ńƒń¼ńƒń1965;€and€for€the€1980s,€to€1965„1975.€€Additional€detail€is€provided€in€Kelley€and€Schmidt€(1994).ĢĢ*€ņņpó󀚚.05€(two„tailed€test)ĢĢĢŠ  ō,ģ%5 ŠŅ„ŅŅXŅĢŌ€ ō1שq© ©1×ŌŌ€ ōģŪōq© ō1×ŌTable€2.€€Partial€Derivatives€of€Y/Nņņgróó€with€respect€to€DemographyŠ ž– ŠŌ" ’ŌĢ4,ŌŲ'`Ģ`"Ōņņ€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€óóĢŌ" ’ jb Ų'``"Ōņņ€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€ĻóóĢ€€€€€€ņņPartial€Derivativesó󀀀ņņ€€€€€€€€€€€Variable€Medians€€€€€€€€€€€óóĢĢ€€€€€€€CBR€€€CDR€€CBRņņ„15ó󀀀€Y/Nņņgróó€€Y/N€€€Dns€€€Nņņgró󀀀CBR€€€CDR€€CBRņņ„15óóŠ b Z Šņņ€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€ĻóóYearsĢÓÓą@<:ģąLess€Developed€CountriesˆĢ1960s€€€.11€€„.30€€„.20€€€€2.44€€1.02€€.02€€2.58€€38.3€€11.2€€38.5Ģ1970s€€„.17€€„.24€€€.11€€€€1.93€€1.24€€.03€€2.53€€37.3€€€9.3€€38.7Ģ1980s€€„.34€€„.02€€€.17€€€€„.67€€1.59€€.04€€2.54€€35.1€€€7.8€€37.7ĢĢą@6=ģąDeveloped€CountriesˆĢ1960s€€„.18€€„.34€€€.03€€€€3.88€€5.79€€.08€€€.79€€17.7€€€9.0€€18.8Ģ1970s€€„.19€€„.17€€„.01€€€€2.35€€8.20€€.09€€€.82€€14.9€€€9.4€€17.7Ģ1980s€€„.04€€€.17€€„.05€€€€1.95€10.79€€.09€€€.44€€12.7€€€9.1€€16.6ĢҰŅņņ€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€óóĢŌ€ Iœ ō ōģŪŌNote:€These€results€are€based€on€regression€coefficients€in€Columns€5„7€of€Table€1.Š ÜŌ ŠŠ    ŠŌ€ XģŪX? IœŌŌ€XģŪXöX XģŪŌFigure€1.€€GDP/N,€Growth€vs.€Level,€1980„1990,Š  ŠKelley/Schmidt€Basic€ModelĢŠ  ` X ŠFigure€1ĢĢĢß}€I95x (  `Š0*x€@@A£š#Œ ’’}ßā   (#(#āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā     āŠ  Šā (#(#  āĢ