The Global Impact of Demographic Change

Transcription

1 WP/6/9 The Global Impac of Demographic Change Nicolea Baini, Tim Callen, and Warwick McKibbin

2

3 26 Inernaional Moneary Fund WP/6/9 IMF Working Paper Research Deparmen The Global Impac of Demographic Change Prepared by Nicolea Baini, Tim Callen, and Warwick McKibbin 1 January 26 Absrac This Working Paper should no be repored as represening he views of he IMF. The views expressed in his Working Paper are hose of he auhors and do no necessarily represen hose of he IMF or IMF policy. Working Papers describe research in progress by he auhors and are published o elici commens and o furher debae. The world is in he mids of a major demographic ransiion. This paper examines he implicaions of such ransiion over he nex 8 years for Japan, he Unied Sae oher indusrial counrie and he developing regions of he world using a dynamic ineremporal general equilibrium four-counry model conaining demographics calibraed o he medium varian of he Unied Naions populaion projecions. We find ha populaion aging in indusrial counries will reduce aggregae growh in hese regions over ime, bu should boos growh in developing counries over he nex 2 3 year as he relaive size of heir working-age populaions increases. Demographic change will also affec saving, invesmen, and capial flow implying changes in global rade balances and asse prices. We also explore he sensiiviy of he resuls o assumpions abou fuure produciviy growh and counry exernal risk for he developing counry region. JEL Classificaion Numbers: J11, F21, E21, E22 Keywords: demographic capial flow saving, invesmen Auhors Address: NBaini@imf.org, TCallen@imf.org, and wmckibbin@brookings.edu 1 McKibbin s research in his paper was suppored by he Economic and Social Research Insiue of he Japanese Cabine Office Inernaional Collaboraions Projec hrough he Brookings Insiuion and was prepared as background maerial for he Sepember 24 World Economic Oulook.

4 - 2 - Conens Page I. Inroducion...3 II. Some Background on Global Demographic Change...4 III. Modeling he Economic Impac of Demographic Change...9 IV. How Will Demographic Change Affec he Global Economy?...11 V. Sensiiviy Analysis...18 A. Faser Global Growh hrough More Rapid Technological Convergence...18 B. Improving Capial Marke Access for Developing Counries...2 VI. Conclusions and Policy Implicaions...2 Appendix: The Analyical Approach...23 Figures 1 Global Demographic Transiion, Populaion Srucure, Removing he Demographics a Sylized Represenaion Child Birh and Adul Mauriy Rae Relaive o Seady-Sae Impac of Global Demographic Transiion 4 Region MSG3 Model Implicaions of Faser Produciviy Growh for 5 Years-4 Region MSG3 Model Implicaions of 1 % Poin Decline in Developing Counry Risk 4 Region MSG3 Model...21 References...31

5 - 3 - I. INTRODUCTION The world is in he mids of a major demographic ransiion. No only is populaion growh slowing, bu he age srucure of he populaion is changing, wih he share of he young falling and ha of he elderly rising. Differen counries and region however, are a varying sages of his demographic ransiion. In mos advanced counrie he aging process is already well under way, and a number of developing counries in eas and souheas Asia and cenral and easern Europe will also experience significan aging from abou In oher developing counrie however, he demographic ransiion is less advanced, and working-age populaions will increase in he coming decades. 3 This paper examines he economic implicaions of his demographic ransiion for Japan, he Unied Sae oher indusrial counries largely in Europe, and developing regions of he world using a four-counry version of he MSG3 dynamic ineremporal general equilibrium model or DSGE model from he macroeconomics lieraure exended wih an OLG Blanchard approximaion. This model was developed by McKibbin and Nguyen 24. The demographic ransiion is calibraed o he medium varian of he Unied Naions populaion projecions. The model innovaes upon exising OLG mulicounry models by inroducing a hree-secor producion echnology, rule-of-humb behavior, real and nominal rigidiie and differen ypes of capial ino more basic general equilibrium opimizing OLG seups. In doing his i brings ogeher feaures of real business cycle models wih a fully ariculaed analysis of forward-looking producers and consumers and modern macroeconomeric DSGE models describing he effecs of demand downurns in he face of wage and price sickiness. We find four main resuls: Populaion aging in indusrial counries will reduce growh, beginning in Japan in he nex decade and hen he res of oher indusrial counries by he middle of he cenury. In conra as he relaive size of heir working-age populaions increase developing counries will enjoy a demographic dividend ha should resul in sronger growh over he nex 2 3 year before aging ses in. Demographic change will also affec saving, invesmen, and capial flows. Japan and o a much lesser exen he oher indusrial counries he fases aging counries could see large declines in saving and a deerioraion in heir curren accoun posiions as he elderly run down heir asses in reiremen. 2 The erm developing counries in his paper refers o emerging marke and oher developing counries. 3 An excellen overview of he issue facing developing counries can be found in Birdsall, Kelley, and Sinding 21.

6 - 4 - Resuls are sensiive o assumpions made abou produciviy growh and exernal risk premia. The paper is organized as follows. Secion II summarizes key aspecs of he curren and expeced global demographic ransiion. Secion III reviews he recen lieraure ha has explored he macroeconomic implicaions of demographic change and describes he model ha we employ. Secion IV presens our baseline resuls esimaing he impac of demographic change from 25 unil 28, and Secion V examines how sensiive hese are o he assumpions made abou growh and risk assessmen in developing counries. Conclusions and policy implicaions follow Secion VI. Deails of he model and he calibraion/esimaion are conained in he appendix. II. SOME BACKGROUND ON GLOBAL DEMOGRAPHIC CHANGE The world is in he mids of a hisorically unprecedened demographic ransiion ha is having and will coninue o have profound effecs on he size and age srucure of is populaion Figure 1. Before 19, world populaion growh was slow, he age srucure of he populaion was broadly consan, and relaively few people lived beyond age 65. This began o change during he firs half of he wenieh cenury as rising life expecancy boosed populaion growh, alhough iniially here was lile change in he age srucure of he populaion. 4 The second half of he wenieh cenury saw he sar of anoher phase in his ransiion. Feriliy raes declined dramaically by almos one-half causing populaion growh o slow, he share of he young in he populaion o decline, and he share of he elderly o increase. The share of he working-age populaion, however, changed lile. These global developmens mask considerable variaion beween counries and regions ha are he resul of very differen feriliy, moraliy, and migraion rends. For example, alhough feriliy raes have fallen almos universally in recen decade hey remain much higher in developing han in advanced counrie where hey are generally below he replacemen rae. 5 Even among developing counrie considerable differences exis feriliy raes are high in Africa and he Middle Ea bu are below replacemen raes in eas Asia and cenral and easern Europe. Likewise, while life expecancy has risen across he globe over he pas 5 years and he larges gains have generally been made in developing counries life expecancy sill remains much higher in advanced counries. Excepions o he generalized increase in life expecancy are Africa where as a resul of he HIV/AIDS 4 In a number of European counrie he demographic ransiion began much earlier. Lee 23 daes he beginning of he decline in moraliy in norhwes Europe o abou Replacemen level feriliy is esimaed o be 2.1 birhs per woman in indusrial counries and 2.4 birhs per woman in he developing counries. The level exceeds 2 in par because more boys are born han girls and in par because some children die before hey reach reproducive age.

7 - 5 - Figure 1. Global Demographic Transiion, Life Expecancy years a birh Toal Feriliy Rae birhs per woman Toal Populaion Growh percen Populaion Aged percen of oal populaion Populaion Under 15 percen of oal populaion Populaion over 65 percen of oal populaion Source: Lee 23.

8 - 6 - pandemic, life expecancy has declined by more han 25 percen in some counries and he Commonwealh of Independen Saes CIS counries. Lasly, ne immigraion has made an imporan conribuion o populaion growh in recen years in Norh America, bu much less so in Europe and Japan. As a consequence of all of hese rend populaion growh is much higher in developing counries paricularly in Africa and he Middle Eas han in advanced counries. Indeed, in Japan and Europe, populaion growh is close o zero. The share of he young in he oal populaion is also higher in developing counrie while he elderly accoun for a larger share of he populaion in advanced naions. Looking ahead, he Unied Naions curren populaion projecions which exend o 25 envisage ha feriliy raes in low-feriliy counries will recover modesly, ha feriliy in oher counries will coninue o decline, ha furher gains in life expecancy will be made in boh advanced and developing counry region and ha migraion will make an increasingly imporan conribuion o oal populaion growh in advanced counrie bu will only modesly reduce populaion growh in developing counries. 6 This has he following consequences Figure 2. Global populaion growh will coninue o slow. By 25, global populaion growh is projeced o be only ¼ percen a year, compared wih 1¼ percen a presen. The populaion in a number of counries is acually expeced o decline over he nex 5 year including by over 3 percen in some cenral and easern European counrie by 22 percen in Ialy, and by 14 percen in Japan. In developing counries paricularly in Africa and he Middle Ea bu also pars of Asia populaion growh, alhough slowing, will remain robu reflecing heir higher feriliy raes. The world s populaion will coninue o age. The elderly will accoun for an increasing share of he populaion alhough he pace and iming of aging varies widely beween counries and regions and he median age of he world is expeced o increase by over 1 years during 2 5 o 37 years. The elderly dependency raio which shows he populaion aged 65 and older as a share of he working-age aged populaion is projeced o rise dramaically in Japan and Europe, wih lesser increases anicipaed in he Unied Saes. 7 Among he developing counry 6 The projecions discussed in his secion refer o he medium varian of Unied Naions These elderly dependency raios are only an approximaion o he suppor needs of an elderly populaion. Some people coninue o work afer hey have reached age 65, while no everyone in he age group is in employmen hey may be sill in school, coninued

9 - 7 - Figure 2. Populaion Srucure, Developing Economies Developing average Cenral and easern Europe Lain America Middle Eas and cenral Asia Africa Asia Advanced Economies Advanced average Europe Japan Unied Saes Ohers Populaion Growh percen per year Populaion Growh percen per year Elderly Dependency percen of working-age populaion Elderly Dependency percen of working-age populaion Working-Age Populaion percen of oal populaion Working-Age Populaion percen of oal populaion Working-Age Populaion 75 percen of oal populaion India 7 Russia China Nigeria Brazil Working-Age Populaion percen of oal populaion Germany Spain France 6 Canada 55 Ialy Source: Unied Naion World Populaion Prospecs: The 22 Revision 23. unemployed, or ouside he labor force. Furher, in some counrie children younger han 15 are in full-ime employmen. I is possible o develop a measure of economic dependency ha adjuss for hese facor bu his alernaive measure is difficul o calculae, paricularly for developing counries.

10 - 8 - region aging is already under way in cenral and easern Europe, a process ha is expeced o accelerae from abou 215. Aging will also begin o accelerae in Asia and Lain America around his ime wih China experiencing paricularly rapid aging bu he share of he elderly in Africa and he Middle Ea while rising, will remain relaively small. The world s populaion will coninue o age. The elderly will accoun for an increasing share of he populaion alhough he pace and iming of aging varies widely beween counries and regions and he median age of he world is expeced o increase by over 1 years during 2 5 o 37 years. The elderly dependency raio which shows he populaion aged 65 and older as a share of he working-age aged populaion is projeced o rise dramaically in Japan and Europe, wih lesser increases anicipaed in he Unied Saes. 8 Among he developing counry region aging is already under way in cenral and easern Europe, a process ha is expeced o accelerae from abou 215. Aging will also begin o accelerae in Asia and Lain America around his ime wih China experiencing paricularly rapid aging bu he share of he elderly in Africa and he Middle Ea while rising, will remain relaively small. The share of he working-age populaion will fall in advanced counrie bu increase in many developing counries. In Japan and some European counrie his decline has already sared and is projeced o accelerae. In he Unied Sae a high rae of immigraion and higher feriliy raes resul in a more modes projeced decline unil 225, afer which he share of he working-age populaion sabilizes. In developing counrie he share of he working-age populaion is projeced o increase unil 215, and hen sabilize a his higher level as a declining share of he young offses a rising share of he elderly. The demographic changes ha are projeced o ake place in he coming years are sriking. Clearly, cauion mus be exercised when using long-erm projecions of any kind, and demographic projecions do become much more uncerain he furher ino he fuure one goes. However, he basic rends oulined in his secion oward an increasing share of he elderly and a declining share of he young in he populaion are apparen in mos plausible scenario and he main issue herefore is he exen o which he global populaion will age over he nex 5 years and beyond. Agains his background, i is imporan o undersand how demographic change is likely o affec global economic and financial marke oucomes. 8 These elderly dependency raios are only an approximaion o he suppor needs of an elderly populaion. Some people coninue o work afer hey have reached 65, while no everyone in he age group is in employmen hey may be sill in school, unemployed, or ouside he labor force. Furher, in some counrie children younger han 15 are in full-ime employmen. I is possible o develop a measure of economic dependency ha adjuss for hese facor bu his alernaive measure is difficul o calculae, paricularly for developing counries.

11 - 9 - III. MODELING THE ECONOMIC IMPACT OF DEMOGRAPHIC CHANGE Numerous sudies in he lieraure examine he impac of demographic change on one region of he world by simulaing closed-economy muliple generaion models. Muliple generaion models of he kind used in hese sudies were firs proposed by Samuelson 1958 and Diamond In heir pures overlapping-generaions OLG form, hese innovae upon he Ramsey infiniely-lived represenaive-consumer hypohesis by inroducing, a each poin in ime, individuals of differen generaions see, for example, Auerbach and Kolikoff, Work by Blanchard 1985, Buier 1988, Weil 1989, and more recenly Faruqee 2, 23 suggesed simplified closed-economy alernaives o pure OLG models. Mulicounry exensions in he OLG radiion are relaively recen and include Buier 1981, Culer and ohers 199, Faruqee, Laxon, and Symansky 1997, Aanasio and Violane 2, INGENUE 21, Brooks 23, Börsch-Supan, Ludwig, and Winer 23, McKibbin, and Nguyen 23, Fehr, Jokisch, and Kolikoff 23, and Bryan and McKibbin 24 among ohers. Resuls from such models differ according o he way household consumpion, saving, and wealh accumulaion are reaed in each model, as well as wheher he models incorporae fricions in he adjusmen o equilibrium and wheher and how hey model fiscal including pension schemes and moneary policy arrangemens. The model used in his paper is a modified version of he MSG3 model. The MSG3 model is a hree-secor energy, nonenergy, and capial-producing version of he G-Cubed model developed by McKibbin and Wilcoxen The model in his paper is much smaller han he original model because of he incorporaion of demographic variables. In his version he world is divided ino four regions: Japan, he Unied Sae oher indusrial counries largely in Europe, and he res of he world in essence, he world s developing bloc. I combines he modern ineremporal opimizaion approach o modeling economic behavior as found in Blanchard and Fischer, 1989, and Obsfeld and Rogoff, 1996 wih shor-run rule-of-humb behavior. In doing hi i brings ogeher feaures of real business cycle models wih a fully ariculaed analysis of forward-looking producers and consumers and modern macroeconomeric models describing he effecs of demand downurns in he face of wage and price sickiness. The main feaures of he model are as follows. Demographics. The model includes demographic consideraion such ha economic agens in he model possess finie lifespans and heir income varies as hey age. Specifically, his draws heavily on Faruqee 23a, 23b who exended he Blanchard 1985 model of finiely lived agens o include aging consideraions. Following Bryan and McKibbin 24, we exend his approach o allow for children who are dependen on working parens for 17 years. These children hen become adul a a mauriy rae m. Each adul has age specific produciviy ha rises over heir life and hen gradually deerioraes owards zero as hey age. These workers

12 - 1 - are responsible for supporing heir financially dependen children. Deah occurs wih a fixed probabiliy. 9 Explici opimizaion. The model is based on explici ineremporal opimizaion by agens consumers and firms in each economy. Thu ime and dynamics are of fundamenal imporance in he MSG3 model, making is core heoreical srucure like ha of real business cycle models. Rule-of-humb agens. To rack he inerial dynamics of some key macroeconomic variable he behavior of agens is modified o allow for shor-run deviaions from opimal behavior, owing eiher o myopia or o resricions on he abiliy of households and firms o borrow a he risk-free bond rae on governmen deb. Cash-in-advance consrains. Holdings of financial asses including money are explicily modeled. In paricular, money is inroduced ino he model hrough a resricion ha households require money o purchase goods. Nominal rigidiies. The model allows for shor-run nominal wage rigidiy by differen degrees in differen counries and herefore allows for proraced periods of unemploymen depending on he labor marke insiuions in each counry. Two ypes of capial. The model disinguishes beween he sickiness of physical capial wihin secors and wihin counries and he flexibiliy of financial capial ha can flow immediaely where expeced reurns are highes. This disincion leads o a difference beween he quaniy of physical capial ha is available a any ime o produce goods and services and he valuaion of ha capial as a resul of decisions abou he geographical allocaion of financial capial. Esimaion/calibraion. Key parameers in he model such as he elasiciies of subsiuion in producion and consumpion decisions are esimaed, enhancing he model s abiliy o reproduce he dynamics of hisorical daa. As a resul, he model exhibis a rich dynamic behavior, driven on he one hand by asse accumulaion and, on he oher hand, by wage adjusmen o a neoclassical seady-sae. The core equaions of he model and a descripion of he daa used in he calibraion are available a We should sress a he ouse ha he fiscal secor in he model is ariculaed wihou a public pension scheme, and we herefore absrac from he fiscal side of he demographic debae. Throughou he analysis we assume ha governmen expendiures are held consan as a share of GDP and ha ax revenues flucuae wih changes in labor and corporae incomes. To he exen ha fiscal deficis or surpluses emerge as a resul of he demographic 9 See McKibbin and Nguyen 24 for a complee discussion of he heory.

13 shocks we assume ha a lump sum ax on households changes o finance he addiional ineres coss of any changes in governmen deb. Fiscal susainabiliy is imposed in his sudy so ha we can focus on he macroeconomic adjusmens o he demographic change. IV. HOW WILL DEMOGRAPHIC CHANGE AFFECT THE GLOBAL ECONOMY? The demographic changes projeced over he coming years are large, bu are hey likely o have an imporan effec on he economies of advanced and developing counries? This secion repors simulaions from he MSG3 model o assess his issue. Our modeling approach is as follows. We firs projec he world economy from 1985 o 21 assuming he UN mid-range demographic projecion convered ino he equivalen model variables given by he analyical framework. We also make assumpions abou produciviy growh by secor and counry using he approach oulined in Bagnoli, McKibbin, and Wilcoxen 1996 exended in McKibbin, Pearce, and Segman 24. This projecion exercise gives us a baseline simulaion conaining he expeced demographic ransiion. We hen rerun he model removing he demographic ransiion from he model projecion saring in 1985 Figure 3. Removing he demographic ransiion is defined as seing he birhraes of aduls and children equal o he raes assumed in he long-run seady-sae a 22. These raes are arbirary, bu no implausible. They assume a child birhrae of 1.9 percen and an adul mauriy rae of 1.5 percen plus a consan probabiliy of deah of aduls of 1 percen and of children of.46 percen. Due o he requiremen of he modeling approach ha here be a well defined seady-sae wih all counries growing a he same rae, we are forced o assume ha all counries will evenually have he same demographics in erms of growh raes in he seady-sae however, his is cenuries ino he fuure. The period of he curren demographic bulge, from 1985 unil he seady-sae, is inerpreed as a emporary deviaion from his long-run seady-sae. Thus removing he emporary deviaion also gives us a measure of he impac of he demographic ransiion over he recen pas and ino he fuure. The process of removing a shock of his naure is no a concepually easy exercise because he model assumes raional expecaions in a variey of markes. Thus he iniial condiions for 1985 i.e., he acual daa in he baseline have expecaions abou he demographic ransiion already embodied in sock variables such as physical capial sock ne asse posiions boh domesic and foreign, and human capial. We hus have a problem in he counerfacual exercise ha in 1985, when we remove he demographic shock, we are capuring boh he impac of he underlying demographics as well as he impac of he change in expecaions abou fuure demographic change. For a period afer he new informaion is realized, here will be a large adjusmen in asse socks which reflecs he expecaions revision. To ry and separae ou he expecaions revision from he underlying demographic change, we le he model run for 2 years o 25 so ha much of his iniial asse adjusmen is compleed and we are capuring more of he pure demographic effec and less of he revision o expecaions. Thus in he following analysis all resuls are presened from 25 o 28 afer he asse sock adjusmen and asse price volailiy o he change in informaion in 1985 have washed hrough he economy.

14 We conver he UN populaion projecions mid-case scenario ino he parameers of he model given he consan probabiliy of deah for aduls and a differen bu consan probabiliy of deah for children. This is done in a way which gives u as close as possible, he aggregae adul and child populaions over ime for each counry as projeced by he Unied Naions. I is no an exac represenaion of he UN projecions because he probabiliy of deah is changing over ime and a his sage of he research we are unable o incorporae his. Thus he resuls should be inerpreed as illusraive raher han as precise predicions abou he fuure. Figure 3. Removing he Demographics a Sylized Represenaion Developing counries Indusrial counries No demographics Baseline projecion Populaion growh Sources: Auhors' esimaes

15 Figure 4 presens he deviaion of he child birh- rae and adul mauriy raes from he long-run seady-sae raes. These are he basic shocks ha are removed from he baseline in he counerfacual simulaion. Several imporan poins can be seen in hese figures. In he Unied Saes and he developing counry bloc, he child birhraes and adul mauriy rae alhough declining, are well above he raes in heir seady-sae raes and populaions are rising. For Japan and he oher indusrial counrie child birhraes and adul mauriy raes are currenly below seady-sae level and populaions are falling. The pace a which child birhraes and adul mauriy raes are declining, however, are faser in developing counrie and in his imporan sense he demographic ransiion in developing counries is greaer han in he advanced regions. Resuls for he impac of demographic change on he global economy for he period from 25 o 28 are shown in Figure 5. Each figure conains resuls for each of he four regions showing he esimaed impac of demographic change ha i he difference beween he baseline which includes demographic change and an alernaive scenario in which he demographic variables are a heir seady-sae raes of growh. The resuls are expressed in such a way ha a posiive number is a variable which is higher because of he impac of changing demographics han i would oherwise have been. For example, he level of Japanese GDP in 25 is esimaed o be 3 percen lower han would have been he case wihou he impac of he demographic ransiion he decline in he labor force is only parly offse by a higher capial sock see below. On he oher hand, developing counry GDP is esimaed o be 6 percen higher by 25 as a resul of he srong increase in he labor force due o he demographic ransiion. I is imporan o sress ha resuls for all counries depend on he demographic change in each counry and he spillover beween counries. 1 In Figure 5 i is clear ha he demographic ransiion is imporan in affecing he GDP growh rae, wih growh in developing counries more han 2 percen higher by 22 han wihou he demographic change, and growh falling in Japan afer 21 o reach a low poin of 1.3 percen below he siuaion in he no demographic change scenario by 24. The Unied Saes and he oher indusrial counries are inermediae beween hese case alhough i should be sressed ha wihin Europe and wihin developing counries here are also large differences in demographic oucomes which are averaged away in he aggregaion we are using in his paper. 11 The picure per adul he proxy for per capia in he model is quie differen o ha for he real growh rae. The Japanese oucomes are a poin o noe. GDP per adul is projeced o be higher in Japan ou o 225 and sill be above wha would 1 The relaive impac of own and cross border demographic change is considered furher in McKibbin This is explored furher in McKibbin 25 where he oher indusrial counries and developing counry blocs are furher disaggregaed.

16 Figure 4. Child Birh and Adul Mauriy Rae Relaive o Seady Sae Unied Saes Japan Oher indusrial counries Developing counries Child Birh Rae Percenage deviaion from seady sae Adul Mauriy Rae Percenage deviaion from seady sae Sources: Auhors' esimaes

17 Figure 5. Impac of Global Demographic Transiion - 4 Region MSG3 Model Percen deviaion from baseline unless oherwise noed Real GDP Adul Populaion Growh Rae Unied Saes Japan Real GDP Growh Rae Invesmen/GDP 1 Toal Savings/GDP 1 25 Capial/Oupu Raio Oher indusrial counries Developing counries Real GDP per Adul Privae Consumpion/GDP Figure 5. Impac of Global Demographic Transiion - 4 Region MSG3 Model concluded Unied Saes Japan Curren Accoun/GDP 1 Trade Balance/GDP 1 Real Effecive Exchange Raes Shor Real Ineres Rae Raio of Impors o GDP Oher indusrial counries Developing counries Change in Ne Foreign Asses/GDP Real 1 Year Ineres Rae Tobin's Q Source: Auhors' esimaes 1Percen poin deviaion from baseline.

18 have been experienced unil 25. This demonsraes ha in a world wih falling labor forces i is imporan o focus on per capia variables raher han aggregae variables in hinking hrough he welfare impacs of demographic change. As expeced, he changing populaion and labor force have imporan impacs boh on privae invesmen hrough changes in he expeced marginal produc of capial as well as on consumpion and saving decisions. The increase in he labor force in developing counries raises he marginal produc of capial and simulaes higher invesmen, wih he invesmen o GDP raio 4 percen higher by 225. There is also iniially an increase in invesmen in Japan as a resul of he need o subsiue capial for a diminishing supply of workers over ime, alhough evenually invesmen begins o fall as he declining labor force needs less capial o equip i. The subsiuion of workers wih capial is clear in he change in he capial oupu raio in Japan. The response of households o boh a change in youh dependency and an expeced change in elderly dependency can also be seen. In Japan, privae saving has been boosed in recen years as he working age populaion move hrough heir high saving year bu is projeced o fall over ime as he increasing number of elderly disserve o supplemen heir falling employmen income. The raio of saving o GDP is 4 percenage poins higher by 25 due o he demographic ransiion, bu by 27 is projeced o be 11 percenage poins lower. In he Unied Sae he saving-o-gdp raio in 25 is 3 percenage poins lower due o demographic change bu is projeced o be 5 percenage poins higher by 24 as an increasing number of he populaion move ino heir high saving years. Aggregae privae consumpion reflecs hese rends wih consumpion rising in developing counries and o a lesser exen in he Unied Saes and falling in Japan. Per capia consumpion no shown, however, acually rises iniially in Japan and is only slighly lower by I is clear ha he demographic ransiion has imporan effecs on saving and invesmen behavior and consequenly curren accoun balances. As a resul of he demographic ransiion occurring boh in Japan and he res of he world, he Japanese curren accoun in 25 is 2.6 percen of GDP in surplus relaive o he case of no demographic change. Over ime, however, as saving falls in Japan, and here is a rise, hen fall, in invesmen, he curren accoun surplus narrows. The curren accoun posiion also deerioraes albei more modesly in he oher indusrial counries while he Unied Saes and developing counries see an improving curren accoun posiion. These changes resul in a subsanial reallocaion of global capial, as Japan and he oher indusrial counries repariae capial o mainain consumpion in he face of falling labor incomes. These changes in curren accoun posiions are consisen wih he resuls presened in IMF 24 from simulaions of he INGENUE model 21 and economeric esimaes. I is also worh noing here ha he changes in saving behavior in he simulaions here are no dissimilar o 12 See McKibbin and Nguyen 24 for deailed discussion of he per capia behavior.

19 hose from he INGENUE model alhough in he laer model he decline in saving in Europe is larger due o he explici inclusion of a pension sysem in he model. 13 The global asse adjusmen is refleced in real exchange raes as a resul of demographic change. There are a number of facors driving real exchange raes in his model. 14 A key facor is he assumpion ha all goods ener ino consumers consumpion bundles idenified by counry of origin. Thu if he supply of goods from one counry shrink he relaive price of hose goods will end o rise, aking all oher facors as given. Thu as he labor force falls in Japan and less Japanese goods are produced, he relaive price of all Japanese goods will end o rise. This will be refleced in an appreciaion of he real exchange rae of Japan. By 25 he Japanese real exchange rae is appreciaed by close o 6 percen. Similarly, he real exchange rae of developing counries is depreciaed by close o 6 percen because of he large increase in supply of developing counry goods on world markes. In addiion o hese underlying facor he adjusmen of global capial flows will also end o reinforce his effec. Wih income being repariaed back ino counries wih aging populaions from counries wih expanding populaion he real exchange rae of he aging economies will also end o appreciae. Finally, real ineres raes in developing counries are 1.5 percen 15 basis poins above wha hey would have been because of demographic change, reflecing he rising marginal produc of capial as he labor force rises in hese economies. This is no arbiraged away because of adjusmen coss in physical capial accumulaion ha preven large differences in marginal producs of capial from being exploied because developing counries canno quickly absorb large amouns of capial. This differenial is reduced over subsequen decades. In Japan, he changing capial-labor raios as a resul of demographic change is esimaed o reduce real ineres raes by up o 1.5 percen by 25. Long-erm ineres raes fall much sooner han his given he expecaions in he model. A he same ime, equiy markes summarized by Tobin s q lower righ-hand corner are expeced o rise especially in developing counrie again reflecing he changing marginal produc of capial over ime. In Japan and he oher indusrial counries as he demographic ransiion maure equiy markes fall below wha hey would have been by 22 in Japan and 24 in he oher indusrial counries. 15 Overall, hese resuls sugges ha he demographic ransiion is already, and will increasingly, impac global growh and he disribuion of his growh across regions. Saving, invesmen, and curren accoun balances in differen regions will also be affeced. Clearly, 13 Social securiy conribuions have o be raised o finance he large increase in pension expendiures in Europe. This effecively ransfers resources from he working-age populaion which has a higher propensiy o save o he older generaions who have a lower propensiy o save in he model. 14 See Bryan and de Fleurieu 25 for a deailed analysis of he exchange rae sensiiviy o a wide variey of assumpions. 15 See IMF 24 for an overview of how populaion aging may affec financial markes.

20 here are a myriad of assumpions underlying he model used in his analysis. The following secion pus he magniudes of hese resuls in he conex of some oher assumpions ha migh impac he variables ha have been explored in his secion. V. SENSITIVITY ANALYSIS In his secion, we change some assumpions boh o see how imporan hese migh be for he analysis above, as well as o give a benchmark o compare he size of he adjusmens we have presened in he previous secion. The firs assumpion we explore is wha happens if produciviy growh ouside he Unied Saes grows faser for a decade from 25. The second assumpion we explore is wha happens if here is a fall in he counry risk premium facing developing counries which is benchmarked in he model o ha in A. Faser Global Growh hrough More Rapid Technological Convergence In his secion we explore he implicaions of more rapid echnological convergence han in he underlying baseline projecion. This is modeled as a rise in produciviy growh defined as labor-augmening echnical change of 1 percen per year for 5 years from 25 o 254 in he developing counry region, a rise of.1 percen per year over he same ime in Japan and oher indusrial counries, and no change in underlying produciviy growh from baseline in he Unied Saes. Resuls are presened in Figure 6 for he same se of variables as for he demographic shock. As expeced, faser echnical change leads o higher real GDP over ime. In developing counries his growh begins immediaely, alhough for he res of he world GDP growh is acually slighly lower as resources are iniially channeled ino developing counries. This can be seen in he deerioraion of he developing counry curren accoun. Afer a decade, however, real GDP growh rises in he res of he world, wih Japan and he oher indusrial counries experiencing heir own growh from domesic produciviy as well as income growh from sronger aciviy in developing counries. The surge in growh iniially reduces invesmen in all counries. This may seem surprising, bu i reflecs he rise in consumpion by households in anicipaion of higher fuure income which reduces savings and raises real ineres raes hereby crowding ou privae invesmen. I akes a decade before invesmen raes rise above he baseline in developing counries. This rise of consumpion in anicipaion of higher incomes and he iniial fall in invesmen resuls in a fall in capial-oupu raios in all counries which is consisen wih higher real ineres raes. Higher real ineres raes persis for five decades while produciviy growh is high, bu hen fall below baseline when he growh surge concludes. This overshooing of ineres raes reflecs he overshooing of capial accumulaion which can be seen in he sharp invesmen reversal afer 25. Despie he presence of raional consumers and firms in he model, here is also considerable persisence hrough he presence of backward-looking households and firms.

21 Figure 6. Implicaions of Faser Produciviy Growh for 5 Years - 4 Region MSG3 Model Percen deviaion from baseline unless oherwise noed Figure 6. Implicaions of Faser Produciviy Growh for 5 Years - 4 Region MSG3 Model concluded Unied Saes Japan Oher indusrial counries Developing counries Unied Saes Japan Oher indusrial counries Developing counries 45 Real GDP Real GDP Growh Rae Curren Accoun 2 Trade Balance Adul Populaion Growh Rae Real GDP per Adul Real Effecive Exchange Raes Change in Ne Foreign 3 Asses Invesmen/GDP 1 Toal Savings/GDP Shor Real Ineres Rae1 Real 1 Year Ineres Rae Capial/Oupu Raio Privae Consumpion/GDP Raio of Impors o GDP 1 Tobin's Q Source: Auhors' esimaes 1Percen poin deviaion from baseline. 2Change as percen of baseline GDP. 3Percen of baseline GDP.

22 - 2 - In comparison wih he resuls for he demographic ransiion, we can make several observaions. The firs is ha higher produciviy growh in developing and modesly higher produciviy growh in developed economies can offse he impac of demographic change on aggregae GDP. Higher produciviy growh of.1 per year in Japan and he oher indusrial counries offses half of he fall in GDP caused by demographic change. For developing counrie he higher growh rae almos removes he expeced fuure decline in growh raes caused by heir aging afer 23. This is also rue for he aggregae consumpion oucomes. I doe however, place more upward pressure on real ineres raes in he shor run. B. Improving Capial Marke Access for Developing Counries Reforms o improve he invesmen climae are assumed o resul in a reducion in he risk premium associaed wih invesing in developing counry asses. The risk premium is assumed o be 1 percen lower forever. 16 Figure 7 shows he resuls for he same se of variables as in he earlier figures. The reducion in he risk premium encourages more capial o flow ino developing counrie and his reduces real ineres raes and has a considerable impac on real GDP. Domesic saving increases for several decades in he developing counries as he rae of reurn on domesic capial improves as a resul of he reform furher reinforcing he posiive effec on growh. The curren accoun posiion of developing counries deeriorae while he advanced counry regions he suppliers of he capial o he developing counries experience improvemens in heir exernal balances. The change in risk premia relocaes producion from he indusrial economies o he developing counries causing he capial oupu raio o rise in developing counries and fall in indusrial counries. In comparison wih he growh shock, he ransmission of growh from developing counries produciviy is more posiive for indusrialized economies han he fall in developing counry risk which leads o a reallocaion of global producion and higher global income, bu all of he gains are capured by consumers in developing counries. VI. CONCLUSIONS AND POLICY IMPLICATIONS The impac of ongoing demographic shifs across he globe will be wide-ranging. The modeling in his paper suggess ha in advanced counrie populaion aging will likely reduce per capia growh raes in he fuure, while in developing counries increases in he relaive size of he working-age populaion could lead o sronger per capia growh provided he addiional labor resources are effecively uilized. Inernaional capial flows could also be subsanially affeced. The resuls presened in his paper sugges ha large changes in saving, invesmen, and curren accoun balances could ake place over he nex 8 years as a resul of demographic change. 16 The issue of changes in counry and equiy risk premia using he same model bu wihou demographics is considered in McKibbin and Vines 23.

23 Figure 7. Implicaions of One Percen Poin Decline in Developing Counry Risk - 4 Region MSG3 Model Percen deviaion from baseline unless oherwise noed Figure 7. Implicaions of One Percen Poin Decline in Developing Counry Risk - 4 Region MSG3 Model concluded Unied Saes Japan Oher indusrial counries Developing counries Unied Saes Japan Oher indusrial counries Developing counries Real GDP Real GDP Growh Rae Curren Accoun Trade Balance Adul Populaion Growh Rae Real GDP per Adul Real Effecive Exchange Raes Change in Ne Foreign Asses Invesmen/GDP Capial/Oupu Raio Toal Savings/GDP Privae Consumpion/GDP Shor Real Ineres Rae 1 Real 1 Year Ineres Rae Raio of Impors o GDP 1 Tobin's Q Source: Auhors' esimaes 1Percen poin deviaion from baseline. 2Change as percen of baseline GDP. 3Percen of baseline GDP.

24 There are, however, considerable uncerainie and our undersanding of how demographic change will affec economic performance is far from complee. For example, he impac of demographic change on exernal balances and capial flows will criically depend on he reacion of privae saving, bu i remains unclear o wha exen households will adjus heir behavior as he demographic ransiion unfolds. Furher, he size of he poenial changes need o be considered in he conex of he scale of a wide range of oher shocks ha migh occur in coming decades such as changes in produciviy growh in various counries. I is clear ha subsanially more research is required o explore hese issues.

25 APPENDIX Appendix: The Analyical Approach This appendix summarizes he analyical approach followed in incorporaing demographics ino he model. A summary of he key feaures of he MSG3 model was oulined in secion III. More deailed documenaion can be found in McKibbin and Wilcoxen 1998 and McKibbin and Nguyen 24, as well as a Adul Populaion In each period, a cohor of children maures and joins he adul populaion. The size of he newly maured cohor, a ime wih respec o he exising adul populaion, Ns is referred o as he mauriy rae, bs. The mauriy rae and is relaionship o he populaion of children will be addressed in anoher secion, below. Following Blanchard, we make he simplifying assumpion ha a any ime all aduls in he economy face he same moraliy rae, 17 p, defined here as he probabiliy of any given agen dying before he nex period. The number of aduls who maured a a previous ime who are sill alive a a subsequen ime is given by: p s n = b s N s e 1 The adul populaion size can hen be deermined for any ime by summing he number of living aduls from all of he cohors ha have ever maured: N = = n ds b s N s e p s ds 2 where N represens he adul populaion size, a ime. Taking he derivaive wih respec o ime yields an equaion governing he evoluion of he adul populaion size over ime: N& N = b p 3 The above equaion has a simple inerpreaion: he adul populaion grows a a rae deermined by he mauriy rae less he moraliy rae. 17 Blanchard 1985 noes ha he assumpion of a common moraliy rae is a reasonable approximaion for aduls wihin he ages of 2 o 4. The fac ha children and reiree whose behavior is of ineres in sudies of populaion aging, fall ouside of his age bracke cerainly indicaes ha he issue requires furher aenion.

26 APPENDIX Child Populaion In every period, a cohor of children is born. If we hink of he adul populaion as represening he se of poenial paren hen i follows ha he size of a newly born cohor will depend upon he curren adul populaion size and he birhrae, b m. The expression for he number of children born a ime s who are sill alive a a laer ime, is hus given by: p s m = b s N s e 4 m The aggregae number of children, M, can be calculaed by summing he number of surviving children, who were born recenly enough ha hey have no ye reached adulhood. If we le represen he fixed number of years from when a child is born o when i reaches adulhood, i.e. he period of childhood, 18 hen: Differeniaing wih respec o ime: = = M m ds 5 m s p M b s N s e ds 6 & p = pm + bm N bm N e 7 M Noe ha in he final exponenial, p refers o he period of childhood muliplied by he moraliy rae, i does no represen a change in p. Relaionship beween he Birhrae and he Mauriy Rae Of he children who were born a ime -, hose who survive will maure a ime, a which ime hey are added o he adul populaion. Thu he mauriy rae a ime is dependen on he birhrae, and adul populaion size, of years pas; as well as he moraliy rae. p b N = b N e 8 m 18 In he simulaions ha follow, he period of childhood is defined as he firs 16 years of an agen s life.

27 APPENDIX Now, we know ha: N = N e p = N e b s p ds b s ds 9 so given he birhrae of years ago, and he mauriy raes over he las year he curren mauriy rae can be deermined: b s ds = b m e b 1 Since he mauriy raes over he las years will be dependen on previous values of he birhrae, i is clear ha he rae of mauriy is predeermined by any given series of birhraes. Adul Consumpion Aduls aemp o maximize he expeced uiliy derived from heir lifeime consumpion. Aduls mus ake ino accoun he uncerainy of heir life spans and hus hey discoun heir planned fuure consumpion by he probabiliy ha hey may no survive hrough o fuure periods. Assuming a logarihmic uiliy funcion, each agen will maximize he following: max θ + p v ln c v e dv 11 subjec o he budge consrain: w& = [ r + p] w + y c 12 where c is he consumpion, a ime, of an adul who maured a ime θ is he rae of ime preference, w is he financial wealh ha an adul who maured a ime s holds a ime ; and r is he ineres rae earned on financial wealh. Financial wealh includes domesic bond equiy, and foreign asses from all counries in he models. In addiion o ineres paymen aduls also earn a rae of p on heir holdings of financial wealh, due o he assumpion of a life insurance marke, as in Blanchard Children do no play a par in he life insurance marke, nor do hey earn inere as hey are assumed o hold no financial wealh. In he full model he cv funcion also includes a complee se of all goods from all counries in he models. Wihin each period, once aggregae consumpion for ha period is deermined, he allocaion across goods from differen counries is deermined based on relaive prices of goods disinguished by place of producion.

28 APPENDIX The opimal consumpion pah for an adul can be shown o be: [ w h ] c = θ + p + 13 where c is he consumpion, a ime, of an adul who maured a ime and h represens he human wealh of he adul. An adul s human wealh is defined as he presen value of he adul s expeced income over he remainder of his or her lifeime: v r i p di + h = e y v dv 14 A any ime, he sum of financial wealh and human wealh w and h represens an adul s oal wealh: he means by which he adul can pay for his or her fuure consumpion. Aduls consume a proporion of heir oal wealh each period, he proporion being deermined by heir rae of ime preference, and heir likelihood of perishing before he nex period. Aggregae adul consumpion, aggregae financial wealh, and aggregae human wealh are simply he sums of he consumpion, financial wealh, and human wealh for all aduls in he economy. N = C c n ds 15 W = w n ds 16 H = h n ds 17 where C N represens aggregae adul consumpion, W is aggregae financial wealh, and H is aggregae human wealh. The adul aggregae consumpion funcion can be shown o be given by: Labor Supply and Demographic Consideraions [ W H ] C N = θ + p + 18 Empirically, one of he key economic characerisics ha changes wih age is he income ha a person receives. This is usually summarized in an age-earnings profile of a counry. We inroduce age-earnings profile ino he model, such ha an agen s income is deermined by his or her age. We hen aggregae cohors over ime. Furher, we assume ha only aduls earn labor income and ha children are compleely dependen upon aduls. Faruqee 2 uilizes hump-shaped age-earnings profiles for adul fied o Japanese daa using nonlinear leas squares NLS. Inuiively, he hump-shaped profile of age-earnings