Aging and the Regional Economy: Simulation Results from the Chicago CGE Model

Transcription

1 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model Seryoung Park* and Geoffrey J.D. Hewings** *Deparmen of Economics and Regional Economics Applicaions Laboraory, **Deparmens of Geography, Economics and Urban and Regional Planning, Regional Economics Applicaions Laboraory and Insiue of Governmen and Public Affairs, Universiy of Illinois a Urbana-Champaign, 607 S. Mahews #220, Urbana, IL , USA spark2@uiuc.edu; hewings@uiuc.edu Inroducion For many decades, he relaionship beween demographic ransiions and economic growh has remained a subjec of debaes and generaed a large volume of academic research. In earlier years, he main focus was on he impac of populaion size on economic growh. Debae cenered on wheher populaion growh resrics, promoes, or is independen of economic growh. Populaion pessimiss (Coale and Hoover, 958; Ehrlich, 968) believed ha rapid populaion growh was a hrea o limied resources because a large par of invesmen had o be used o suppor he growing populaion. On he conrary, populaion opimiss (such as Kuznes, 967 and Simon, 98) believed ha a large populaion faciliaed he realizaion of economies of scale and promoes echnological innovaion. However, populaion neuraliss (Kelly, 988) empirically showed ha here was lile cross-counry evidence ha populaion growh had eiher a significan posiive or negaive effecs on economic growh. They argued ha he correlaion disappeared once oher major influences, such as counry size, openness, educaional aainmen, and poliical insiuions, were aken ino accoun. All hese heories, however, paid lile aenion o he age srucure of populaion and is influences of he economic growh. In recen years, economiss have ended o focus on populaion dynamics beyond populaion growh iself. More sudies have decomposed he populaion growh in erms of feriliy and moraliy componens and, more imporanly, examined impacs on age disribuion and economic growh. Since individuals behave differenly a differen sages of life,

2 2 Seryoung Park and Geoffrey J.D. Hewings changes in he age srucure in he economy can have an immediae effec on economies. For example, a counry wih a high proporion of he elderly migh experience slower economic growh, because a large share of resources has o be allocaed o suppor he less producive populaion. This poenially negaive impac is he main problem associaed wih an aging populaion and will become a phenomenon experienced in many developed economies wihin a nex few decades. Fuure demographic changes, arising from an aging populaion, have araced a lo of aenion because of heir sizable and rapid ransiions. In paricular, considering heir significan economic impacs, here has been an explosion in research on many aspecs of he aging populaion. However, mos of hese sudies on economic consequences of aging populaion have been conduced a he naional level, so ha heir resuls fail o capure he local effecs of demographic changes due o he exisence of he regional differences in erms of economic and demographic srucures. The nex quesion is closely relaed o he region-specific demographic changes which is an imporan bu widely ignored influence of he aging populaion. Alhough an aging populaion is a naion-wide phenomenon ha will affec all regions, i could significanly aler he inerregional demographic srucure hrough significan asymmeric migraion flows among regions. For example, he migraion of reirees will accelerae under an aging populaion, resuling in fundamenal changes in he demographic srucures of boh regions of origin and desinaion. This issue is he leas well undersood demographic change which very few previous sudies have addressed. For his sudy, he Chicago region is seleced for a reference region since i has long been boh a leading immigraion place while i is expeced o face a significan demographic change wih increasing reiremen ou-migraion as he populaion ages over he nex wo decades. Over he las hree decades, Illinois has los is place as one of he leading saes in he naional economy. Tradiionally, Illinois had been one of he highes per capia income saes, ranking among he op en saes wih he income much above he naional average. However, in 2005, Illinois per capia income fell o jus slighly above he naional average, ranking fifeenh highes among he saes. This decline has been more serious when economic performance has been viewed in erms of oal income raher han in per capia erms. Abou hiry years ago, he Illinois economy accouned for 6 percen of naional income; by 2005, is share had fallen o 4.5 percen. Much of his economic downurn is linked o he meropolian Chicago economy, which plays an imporan role as a developmen engine for he sae, accouning for abou 65 percen of sae populaion and over 70 percen of sae producion. There can be lile doub ha he lower level of relaive economic performance of boh Chicago and Illinois parly resuled from he recessions in he manufacuring secor saring from he early 980s. Since 990 alone, he sae has los 23,000 manufacuring jobs a a rae ha is almos wice as high as ha for he Midwes as a whole. However, he seady decline in economic performance over The Chicago area is he MSA, comprising he counies of Cook, Will, DuPage, McHenry, Lake, and Kane.

3 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 3 he long erm could be convincingly demonsraed by he slow populaion growh and changing srucure of populaion in his region. Alhough inernaional (legal and illegal) immigraion is an increasingly imporan componen of naional populaion change, regional demographic srucure is deermined by he combinaion of naural increase (birhs - deahs), and wo ypes of migraion, inernaional and inerregional. However, as regional feriliy and moraliy have become more uniform hroughou he Unied Saes, migraion has become by far he more imporan facor in changing regional populaions. Hence, par of he reason for he slower pace of populaion growh in Chicago migh be explained by he oucome of he ou-migraion of reirees, because Chicago is he second larges loser, nex o New York, in reiremen ou-migraion. Moreover, over he nex couple of decades, reiree migraion may be expeced o have a dramaic impac on he Chicago economy, because of he rapid ransiion o a saus where he aging populaion will comprise a larger share of oal populaion han a he presen ime. Considering hese issues, his chaper ses wo main objecives. The firs goal in his sudy is o improve our undersanding abou economic and inergeneraional welfare condiions given he fuure demographic changes, wih paricular focus a he regional level. The second goal is o provide he sensiiviy of hese oucomes o seleced policy reforms and sugges he policy implicaions o boh local and federal governmen. This sudy would be he firs aemp o rigorously quanify he dynamic effec of demographic changes on he regional economies. This chaper is srucured as follows. In he nex secion, aenion will be direced o a discussion of he link beween demographic changes and heir economic impacs, wih some empirical analysis derived from he Chicago economy. Thereafer, a review of he received heory and some prior empirical work ha has explored formal demographic-economic ineracions will be provided. The model is presened nex and hen he resuls focus on graphical presenaion of he resuls of wo scenarios, one in which no aging process is assumed and he oher in which he expeced paern of aging is modeled. The analysis is cenered on a wo-region model of he Chicago economy (Chicago and he Res of he US) alhough he resuls in his chaper ignore he influence of migraion. A brief summary complees he chaper. Demographic Changes and Economic Impacs Aging Populaion Over he nex couple of decades, mos indusrialized counries including he Unied Saes are expeced o face significan aging populaion problems as a large proporion of populaion moves from middle age o reiremen. Alhough he raio of elderly populaion is already rising due o increased longeviy and lower birh raes, he effecs of aging populaion are expeced o become more pronounced

4 4 Seryoung Park and Geoffrey J.D. Hewings wih he aging of he baby-boom generaions. In he case of he U.S., all 77 million baby boomers will have reired wihin 30 years, leaving wice he number of elderly. The significan rise in he proporion of he elderly populaion will lead o dramaic increase in he old age dependency raio, which is defined as he raio of he dependen old age populaions (hose 65) o he populaion in he working age groups (beween 5 and 64). According o a Unied Naions projecion of he evoluion of he share of he populaion, he rise in old age dependency raio in developed counries will be gradual unil 200, bu increases a a much faser rae hereafer. For he Unied Saes, he dependency raio is prediced o climb from 9 percen o 33 percen beween 2000 and Such an unprecedened demographic ransiion aribuable o a sizable increase in elderly populaion should have a significan impac on economic performance. Especially, a significan loss in he working-age populaions who usually exhibi higher produciviy could generae a negaive impac on he growh of per capia income unless sufficien echnological progress has been made o compensae for he loss in labor resources. In addiion o he shorage of labor supply, he aging populaion could have powerful impacs on savings and capial accumulaion. Based on he classical life cycle heory of saving, proposed by Modigliani and Brumberg (954), saving rises wih income from labor supply during he middle ages, and declines and urns negaive over he reiremen period. As a resul of his uneven paern of savings over an individual s lifeime, a significan change in age srucure under aging populaion may primarily aler he aggregae savings in he economy, and hus he economic growh. I is eviden ha an aging populaion could have a negaive impac on aggregae savings because i increases he elderly populaion who usually dissave by consuming ou of heir savings. However, posiive effecs may also arise as a resul of precauionary savings moives in anicipaion of an exended lifespan. Here, i is assumed ha individuals save more hrough privae savings in order o provide for increased old-age consumpion. Therefore, he overall effec on aggregae savings migh be quie differen according o he age srucure of he populaion and an individual s opimal saving behavior. The differen predicion for he impac of aging on saving will also generae various responses on capial sock, wages, and economic growh associaed wih he aging populaion. In addiion o imporan macroeconomic effecs, he aging populaion also pus remendous financial pressures on curren pension sysems such as pay-as-you-go (PAYG) schemes. Wih a rapidly aging populaion, he pension conribuions may no be sufficien o cover he oal pension benefis which are expeced o grow quie subsanially. In order o finance he defici of pension sysem, conribuion rae may have o be raised, in some cases quie significanly. The implicaions here could be profound; however, hey are no addressed in his chaper. A furher imporan consideraion of he impac of an aging populaion is on migraion. According o U.S census, he oal esimaed number of reiremen migrans, which are defined as migrans over he age of 60 who moved across he sae lines, increased o over 2 million during he 995~2000 period, which was 9 percen increase over he 985~990 period. In case of Illinois sae, as a dominan migraion origin, i has experienced heavy loss in elderly popula-

5 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 5 ions due o he reiremen migraion. Since he 980s, more han 00,000 people age 60 and older in Illinois, which accouns for abou 5 percen of sae s cohor of populaion over 60 years old, ou-migrae every 5 years. Considering he inmigrans, in Illinois has experienced a ne loss of 70,000 reirees, who were 60 and older, ranking he second larges among all he saes. Furher, he Chicago area accouns for over 80 percen of he older migrans who leave Illinois, i.e. Chicago region lose abou 0.9 percen of old populaion, or abou 2,000 reirees, every year. As he resul, Illinois has los more han one billion dollars a year in ne income hrough reiremen migraion since he 980s (Longino, 995). The implicaions for he regional economy have been described in Park (2007). 2. Economic Implicaions Mos economic models, especially CGE models, devoe relaively modes aenion o he demographic side of he economy. In many cases, a represenaive household approach is adoped; hence, i is difficul, under hese circumsances, o learn much abou he dynamics of an aging populaion. In a precursor o he curren chaper, Yoon and Hewings (2008) explored he implicaions of an aging populaion on he Chicago region by adoping a modified linear expendiure approach wih several age cohors (a furher disincion was also made by income). Since he model employed was an economeric-inpu-oupu sysem, i was possible o race he impacs of aging on consumpion in oal and o explore he ramificaions of a changing mix of age cohors on he way in which expendiures would be made across a se of goods and services. Table provides a sample of he resuls obained for he allocaion of income across a selecion of goods and services by differen age groups for seleced years hrough There are some obvious changes across age groups and over ime; for example, all age groups can expec o allocae a larger share of oal expendiures on healh care bu some of he larges percenage increases occur in he and age groups. The decline in food expendiures is compensaed by an increase in he Oher caegory wherein can be found resaurans; hence, i is anicipaed ha less food will be purchased for consumpion a home. Fig. shows how hese differen age-cohor households consumpion will be expeced o grow over ime, in comparison o he growh of he represenaive household. The spread in growh raes is significan (confirmed by formal ess) and given he combinaion of changing numerical composiion of hese households in he region, he inersecion of differen growh raes and allocaions across goods and services generaes differen impacs on he region s economy han he consumpion effecs from a represenaive household. The combinaion of an aging populaion and he changes in consumpion from income generae imporan growh paerns in he middle and higher age cohors while he younger cohors exhibi a declining rae of growh. However, hese resuls are presened for an economy in which only modes migraion effecs have been considered. As Wakabayshi and Hewings (2007) demonsraed for Japan, he impacs of aging in he absence of significan in-migraion are very differen

6 6 Seryoung Park and Geoffrey J.D. Hewings from cases in which in-migraion migh provide a source of younger, acive enrans ino he labor force. Table. Consumpion by Age: Seleced Goods and Services by Year, Chicago Consumpion Types Toal Under Over 64 Food Housing Clohing Transporaion Healh care Ohers HH Noe: HH is he single represenaive household Fig.. Growh in Consumpion by Age-Cohor Households, Chicago

7 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 7 Received Theory and Prior Empirical Analyses A large number of sudies on he economic impacs of demographic changes already have been carried ou embracing boh empirical analysis and heoreical research. 2 The empirical sudies primarily infer he economic effecs of demographic ransiions using cross-secional survey daa or microeconomic and macroeconomic ime series daa. However, he empirical lieraure has presened a mixed oucome of he impac of populaion aging on privae savings and economic growh, in ha he relaionship beween savings and he dependency raio changes significanly, depending on he ypes of daa and analyses used. Basically, cross secional sudies suggesed a srong negaive effec of he dependency raio on he savings rae, while ime series sudies for individual counries and sudies using micro survey daa yielded weak, or even posiive, demographic effecs on savings. For insance, using he pooled ime series daa for 6 counries, Masson e al. (998) obained highly significan esimae suggesing ha he dependency raio causes a negaive impac on he savings rae. However, analyzing he consumer expendiure survey daa, Poerba (994) and Canari (994) found no empirical evidence ha suppors he srong effec of he dependency raio on savings raes. The main drawback of hese empirical sudies is ha hey do no provide he explici mechanisms ha link he changes in individual opimal behavior associaed wih demographic ransiions o he changes in macro-economic variables. In general, demographic changes affec he economy hrough he several complicaed ineracions and feedback mechanisms beween economic agens. Thus, he complee economic implicaions of demographic changes canno be adequaely derived from he parial equilibrium approaches ha characerize many of he empirical analyses. Considering his limiaion, mos recen papers on he effecs of demographic changes have been carried ou wih a general equilibrium framework ha conains deailed informaion abou all commodiies and facor markes ogeher wih each agen s decision making process. Much of he general equilibrium model is based on he sandard overlapping generaions (OLG) model, originally inroduced by Samuelson (958) and Diamond (965). Unlike he single represenaive model, OLG model includes inrageneraional heerogeneiy and a much more deailed specificaion of demographic srucure by incorporaing a more realisic paern of birhs and lengh of life. However, in earlier sudies, he obsacle o overcome in working wih overlapping generaions models was he heerogeneiy implied by he age srucure of he populaion. Auerbach and Kolikoff (987) offered a breakhrough in overcoming his obsacle by developing a racable framework wih a reasonable demographic srucure. Since hen, heir model has been widely applied o he analysis of economic quesions, arising from demographic changes, policy reforms, and endogenous growh. More recenly, several conribuions have been made o modify and improve 2 A survey on aging populaion and is economic impac can be found in Bosworh, e al. (2004).

8 8 Seryoung Park and Geoffrey J.D. Hewings he earlier version of he model by incorporaing many rich componens such as uncerainy wih respec o income and lifeime, liquidiy consrains, endogeneiy of labor marke, and openness. Uncerainy wihou perfec annuiy marke exers a major influence on he paern of savings because of he iming of deah or unemploymen, and he limied opions agains unforeseen coningencies. Liquidiy consrains will also change he implicaion for saving behavior by limiing saving agains expeced fuure income. As a pioneer of hese ypes of model, Imrohoroglu e al. (995, 998) and Conesa and Krueger (999) analyzed he quaniaive role of uninsured idiosyncraic uncerainy wih respec o individual labor produciviy and lifespan in an overlapping generaions model o invesigae opimal pension reforms. 3 Gerler (997) made subsanial progress oward developing a more realisic overlapping generaions framework where agens move sochasically ino reiremen, and once in reiremen, face a consan moraliy risk. This exension wih life ime uncerainy, pioneered by Blanchard (985), is no only more realisic in erms of is demographic srucure and life-cycle feaures, bu also i opens up new applicaions ha consider boh aging populaion and fiscal policy implicaions. Morrow and Roeger (2004) applied his framework o invesigae he effecs of demographic ransiions and pension reforms in Europe. Keuschnigg and Keuschnigg (2004) incorporae endogenous labor supply combined wih employmen search o capure he poenial labor marke disorions of pension reforms. In hese sudies, however, demographic ransiions have been analyzed in closed economy models ha ignore immigraion and accordingly inernaional capial mobiliy. More recen sudies have begun o address hese issues. For example, Soresleen (2000) invesigaed he fiscal effec of immigraion policies in he U.S. adoping an open economy feaure in an OLG framework. Ingenue (200) divided he world ino six regions which have differen demographics o examine he effec of global demographic ransiion on inernaional capial flows. Fehr e al. (2004) developed a hree-counry version of he model wih inergeneraional earnings heerogeneiy, bequess moives, and capial adjusmen coss o quanify he macroeconomic and welfare effecs of skill-specific immigraion policy during he demographic ransiion. However, even wih hese significan advances in he inernaional models, very few sudies have conduced boh heoreical and numerical evaluaions of he relaionship beween demographic changes and regional economies. One of he bigges obsacles in dealing wih regional models is he complicaions arising from greaer openness. 4 For example, labor is more mobile among regions wihin a naion han beween counries, demanding a quie differen reamen of labor mobiliy. Also, he influence of savings by residens in a region can spread o invesmen in oher regions, generaing differences beween he region of facor employmen and he region of expendiure of facor income. In addiion, complex 3 Imrohoroglu e al. (995) found ha he maximum benefi of social securiy occurs a a replacemen rae in he neighborhood of 30 percen. And Conesa and Krueger (998) argued ha a majoriy of curren voers would lose along he ransiion and herefore favor he saus quo. 4 For deails, see Parridge and Rickman (998)

9 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 9 ineracions beween federal and local governmens add complicaions for regional modeling. Reflecing hese complexiies, he firs aemp o analyze he inerregional consequences of an aging populaion in he conex of overlapping generaion framework was very recenly conduced by Fougere e al. (2004). They seup a six region OLG model for Canada o evaluae he effecs of raising he proporion of immigrans and found ha he effecs on real-per capia oupu vary subsanially depending on he regions. However, heir model does no consider he labor mobiliy since i reas labor as compleely immobile beween regions; hus, i excludes all he efficiency issues associaed wih he regional movemen of labor. Moreover, i basically adoped he sandard overlapping generaions framework neglecing he uncerainy issues which are closely relaed o saving moives agains idiosyncraic risk. Moving he ineres o regional demographic changes under an aging populaion, here appears o be a dearh of previous sudies. In paricular, here appear o be no sudies relaing he impacs of reiremen migraion under aging populaion on he regional economy in he conex of general equilibrium framework. Insead, mos of he sudies have analyzed he specific economic impacs based on empirical evidence such as cross-secional surveys and macroeconomic ime series. Serow and Hass (992) used he survey daa colleced from reirees who have recenly moved ino he Norh Carolina and macroeconomic daa a he couny level. They find ha migrans of reiremen age make subsanial conribuion o he local economy in erms of income, job creaion, and local governmen revenues. More recenly, Day and Barle (2000) examined he associaion beween elderly in-migraion and regional economic growh for Texas Hill Counry counies. They found ha counies ha have become more reiree-oriened have experienced significanly more economic growh. The Model The model o be presened has been developed for use in analyses of he impacs of aging, migraion and he fiscal issues associaed wih he opimal funding of social securiy and reiremen programs. However, in his chaper, only he aging issues will be feaured bu he complee model srucure will be described o provide a sense of he ways in which he CGE framework has been expanded. This model is represened by he wo-region dynamic general equilibrium model wih an overlapping generaions framework, whose naional version was originally developed by Auerbach and Kolikoff (987). Like he original version, his model incorporaes individual earnings heerogeneiy, demographic ransiions, and social securiy sysem. However, here are some novel feaures ha are differeniaed from former overlapping generaions framework in wo ways. Firs, his model newly inroduces consideraion of regional feaures ha are omied in he naional overlapping generaions framework. In his model, each region is inerlinked wih each oher by migraion, rade, and he social securiy sysem. Secondly, unlike Fougere e al. (2004), his model feaures age-specific moraliy

10 0 Seryoung Park and Geoffrey J.D. Hewings and borrowing consrains which is criically imporan o generae realisic implicaions of he effecs of demographic changes. Deailed feaures are presened in he following secion. Feaures of he Model. Regional Seup The model economy is composed of wo regions, Chicago (HOME) and res of he US (ROUS), bu he basic srucure of his regional model is closely relaed o is naional counerpars. Households 5 maximize heir uiliy by choosing a profile of consumpion over he lifecycle and firms demand facors following from profi maximizaion, responding o differences in goods and facor prices. Prices adjus in boh goods and facor markes o clear he excess demand. However unlike he prooype of OLG model, his model has a complicaed srucure, even more han inernaional rade models. This model adds various componens and linkages ino is naional version o capure he regional feaures. Firs, labor is assumed o be parially mobile in domesic regions, while inernaionally immobile, aking ino accoun people s preference for saying in he region where hey originally reside. 6 This locaional preference is represened by he wage elasiciy of labor migraion. Wih parial mobiliy of he labor, wage differenials beween regions ake muliple periods o be adjused because of he lagged responses of labor marke. However, capial is assumed o be immobile inerregionally. 7 This resuls in he reurn on capial being differen across he regions. Secondly, he nesing srucure is necessary o complee he household s decision process, since boh regions rade in goods and each individual considers producs from differen regions as imperfec subsiues following he familiar Armingon assumpion. Under Armingon assumpion, a good produced in one region is reaed qualiaively differen from good produced in oher regions. Thus, Armingon assumpion ensures ha consumers demand all he goods produced in boh regions. The hierarchy in nesing srucure of his model consiss of he following wo seps. In he firs sep, each agen deermines he aggregaed consumpion pah over ime, maximizing a ime-separable uiliy funcion subjec o lifeime income. Time separabiliy allows a separaion beween ineremporal and inraperiod decision-making in he nesing srucure. Once opimal condiions governing he aggregae consumpion levels are esablished, he nex sep is o al- 5 Since each household consiss of one agen in his model, household and individual agen will be used inerchangeably. 6 According o Jones and Whalley(986), perfec labor mobiliy is no useful in analyzing he region specific effec of governmen policies because under perfec mobiliy, he policy effec migh be underesimaed wih complee labor movemen beween regions. 7 The reamen of capial mobiliy is imporan when assessing he regional invesmen policies.

11 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model locae hese expendiure levels among differeniaed good in erms of geographic origin, i.e. home produced good and impored good from res of he U.S. In his sep, subsiuion elasiciies play an imporan role in deermining each agen s opimal choice, hus, he values of elasiciies beween wo regions are very imporan o influence he magniude of he regional effecs. For example, even if he aging populaion changes he age srucure in a similar paern across he naion, he effec on regional economies will depend on elasiciy..2. Dynamic Overlapping Generaions Framework To measure he effecs of he demographic change on he behavior of differen generaions, i is necessary for he model o be disaggregaed by he age cohors as well as dynamic processes, describing he pah of consumpion and savings behavior of each age cohor over ime. The dynamic overlapping generaion framework saisfies hese crieria. This model is consruced based on he dynamic OLG framework developed by Auerbach and Kolikoff (987). There are hree ypes of agens in each region: households, firms, and governmen. Each secor represened by hese agens has sylized componens, bu heir ineracions can be quie complex. By solving for he economy s general equilibrium ransiion pah, he model akes ino accoun all relevan feedbacks among hese agens according o demographic changes and relaing governmen policies. In his model, each region is populaed by individual agens who live up o age 85. This limied age does no appear o be crucial since, under his assumpion, only less han 3 percen of U.S. populaion is no considered. 8 The individual agen eners he labor marke a he age of 2 and reires mandaorily a he age of 65. Since all he individuals beween ages 0 and 20 are considered no o perform economic aciviies, reflecing hey are suppored by heir parens, his model deals wih only he individual agens above age 2. Lifeime uncerainy is considered in his model, i.e., each individual faces a differen probabiliy of deah in every period, which becomes higher as hey age. Therefore, in every period, some fracion of people dies earlier han age 85, and leaves accidenal bequess since annuiy markes are assumed o be missing. 9 Toal accidenal bequess are disribued evenly over all he agens alive in he nex period. Moreover, each individual is assumed o face borrowing consrains. Under borrowing consrains, social securiy could furher disor he ineremporal consumpion allocaion by levying he higher payroll ax on younger generaions who are binding in borrowing consrains. Individuals are endowed wih one uni of ime and supply he labor inelasically. Since all agens in he same age cohor are idenical in erms of preferences, individual heerogeneiy is presen only across age cohors wih respec o labor 8 Of course, all of hese sylized facs can be changed and par of he research agenda will be o consider changes. 9 Wih perfec annuiy markes, each individual does no leave uninended bequess. However, he social securiy sysem subsiues parially for he missing annuiy sysem and reduces uninended bequess.

12 2 Seryoung Park and Geoffrey J.D. Hewings produciviy and wage income depends on he individual s produciviy, which is assumed o be idenical across regions. However, wage income migh differ across regions because he wage rae per uni of effecive labor is region specific due o he parial labor mobiliy. Because of wage differences by age, he individual life cycle of an individual is described by a hump shaped income profile. The individual agen sars o work a age 2 and receives he highes wage income during his/her middle age. Reiremen erminaes he flow of wage income and eniles he individual o pension benefis. As a resul of he uneven paern of wage raes over heir working lifeime and borrowing consrains, individuals save during middle aged working periods and dissave in reiremen, which resuls in uneven disribuion of wealh by age cohors. 2 Technical Descripion of Model Srucure 2. Household Secor Each individual makes lifeime decisions abou consumpion and savings a he beginning of his/her adul life, leaving no volunary bequess and receiving no inheriances. Since each agen is represened as forward looking and having perfec foresigh, he evoluion of consumpion and savings depend on all fuure ineres raes and afer ax wages. Represenaive agens of each age cohors maximize a ime-separable expeced lifeime uiliy funcion (U) ha depends on sreams of aggregae consumpion goods (C). Moraliy risk is represened by he condiional probabiliy (s). is hen he uncondiional probabiliy of being alive a age k. j =Π s k k 65 j γ ( C, ) j i j Ui = ( ) [ sk] j= + ρ Π (.) k = γ where C i,j is he aggregae consumpion of an individual of age j in i h generaion, ρ is he subjecive discoun rae, γ is he inverse of he ineremporal elasiciy of j j subsiuion. Thus, he effecive discoun rae is expressed as ( ) sk + ρ Π, k = γ meaning ha wih moraliy risk, he uiliy of fuure consumpion is more heavily discouned. A every period, each individual faces he budge consrains described as follows: C ( + τc) PCi, j + ai, j = ( τw τ p) wej + ( + ( τr) ra ) i, j + peni, j +Φ where a i,j is he asse of generaion i a age j, τ w, τ c, and τ r are ax raes on labor income, consumpion, and capial income respecively, τ p is he social securiy ax rae, i.e. pension conribuion rae, w is he wage, r is he ineres rae, and P C is

13 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 3 he price of aggregae consumpion good. pen i,j sands for he pension benefi of generaion i a age j, and Φ is he ransfer from accidenal bequess. An individual s labor produciviy is assumed o be an exogenous funcion of his/her age. This produciviy difference by age is capured by e j, which changes wih age j in a hump shape way (Miles, 999). For simpliciy, produciviy age profile ( e j ) is consan in erms of ime and region. e j j 2 j = λ+ λ2 λ3 (.2) Wih he maximizaion procedure, he following sandard firs-order condiions can be derived, concerning consumpion per period. Equaion (.3) implies ha he marginal rae of subsiuion beween consuming now and consuming laer equals he relaive price of consuming laer insead of now. C C P (, ) γ + r γ + τ c i, j= C C i, j + ρ P ( + τ c, ) (.3) 65 A = 65 j+, j+, j j= C N C A where C is he aggregae consumpion a ime, N j, measures he number of people in age cohor j a ime. The following wealh accumulaion equaion can be obained wih he maximizaion procedure where A is he aggregae asse a ime. a = a { + ( τ ) r} + ( τ τ ) we + pen ( + τ ) P C +Φ (.4) C i, j i, j r, w, p, j i, j c, i, j 65 = 65 j+, j+, j j= A N a Once hese opimal condiions governing he aggregae consumpion levels a each period are esablished, he nex sep is o disribue he opimal consumpion of is purchases in erms of regional geographic disribuion. The represenaive agen of each age group minimizes oal expendiure, wih an aggregae level of consumpion being a CES composie of wo regional goods. min PC = p c + p c (.5) C HOME HOME ROUS ROUS s i, j, s s i, j, s s i, j, s subjec o : ( ) ( ) /,, ( ) ϕ s s HOME ϕ s s ROUS s i j s s ( i, j, s ) ( s) ϕ ϕ = β + β ( i, j, s ) ϕ C c c HOME( ROUS ) where c i, j, s is he consumpion of generaion i a age group j for a Chicago (HOME) or res of he US (ROUS) produced good a region s, β s is he consump-

14 4 Seryoung Park and Geoffrey J.D. Hewings ion share parameer for good produced in region s, and φ s is he parameer ha conrols ase for variey. Opimal consumpion of differeniaed good beween impors and domesic good akes he following forms: C σ s HOME Ps i, j, s = β s C HOME i, j, s ps c (.6) C σ s ROUS Ps i, j, s = ( βs) C ROUS i, j, s ps c where σ s is he Armingon elasiciy of subsiuion for consumpion in regions s beween home made good and impored good 0. Equaion (.6) implies ha he demand by an individual of region s for a good produced in each region is funcion of he price of ha good relaive o he price of aggregae good and of he quaniy of aggregae good he individual wans o buy. Combining equaion (.6) wih equaion (.5) yields he aggregae price ( P s ). { ( ) } /( s HOME σ ) s ROUS σ σ s β β s s P s = p p s + s (.7) 2.2 Producion secor In each region, here is a single represenaive firm, which specializes in he producion of a unique regional good. Producion in every period akes place wih a consan reurn o scale of a Cobb-Douglas producion echnology using capial sock insalled a he beginning of he period in he region and he full regional labor force. Y = AK L (.8) α α e, where Y is he oupu, A and α sand for scale parameer and capial income share respecively, and K and L e represen he capial sock and effecive labor force respecively. The curren cash flow of he firm π is deermined by equaion (.9). π = PY wl PI (.9) C C where I = K + ( δ ) K is he invesmen and δ sands for depreciaion rae of capial. The firm maximizes is value which is expressed as fuure cash flow discouned by gross ineres rae R: 0 σ is equal o / ( ϕ ). Thus, as long as φ is sufficienly less han, which implies σ is finie, consumers regard each good produced by differen origin as imperfec subsiues and prefer variey.

15 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 5 max ( ) R π (.0) = 0 s= 0 The firs order condiions o his problem yield he facor demand condiions: re C P = α AK L (.) α α e, w C P = ( α) AK L (.2) α α e, R P = + (.3) C + + ( re+ ( δ )) C P where re is renal reurn of capial. Equaion (.3) implies ha he unique gross ineres rae is increased by renal reurn of capial and capial gains. 2.3 Governmen secor The role of he governmen in his economy is simply o levy he axes and adminiser he social securiy programs. The governmen has hree ypes of axes: wage income ax, consumpion ax, and capial income ax. Since his economy ignores he public deb, he governmen balances he budge consrain spending ax revenues wihou issuing governmen bonds. The governmen decides ax raes according o budge consrains o balance for each period. The governmen budge consrain is defined as follows: 65 C C N j, ( τw, we j, + τc, P Cj, + τr, ra j, ) = P G (.4) j = where G is he governmen expendiures a ime. The governmen also manages he public pension sysem, which is iniially modelled as a pay-as-you-go (PAYG) scheme for he benchmark economy. Under PAYG sysem, he governmen grans a fixed pension benefi o he reired generaions while pension conribuions are compleely financed by he curren working generaions. The pension benefis are deermined as a fracion of he lifeime average wage earnings from age 2 hrough he previous age of reiremen. The fracion is given by he replacemen rae (ψ), which is assumed o be idenical across region. Aggregae pension benefi is represened by equaion (.5). PB N [ ( ( w e ))] j, ψ j+ k k j= k= = (.5)

16 6 Seryoung Park and Geoffrey J.D. Hewings Aggregae pension conribuion, equaion (.6), is deermined by he produc of he populaion of working group ( N j ), social securiy ax rae ( τ p ), and labor income. 44 j, τ p, j j = PC = N w e (.6) Since he pension budge consrain is balanced every period, PB = PC, he model can calculae he pah of social securiy ax from he curren working generaion, which is endogenously deermined. 2.4 Migraion Working populaion in model age group from o 44 is assumed o be parially mobile across domesic regions. The ne ou-migraion of labor is deermined by he wage elasiciy of labor migraion. M HOME w η = POP ( ) (.7) w w w ROUS w w where M denoes he number of ne ou-migraion of labor a ime, POP is HOME ROUS he aggregae sock of labor given a he beginning of ime, w and w are he wage raes in Chicago (HOME) and res of he U.S. (ROUS), and η refers o he wage elasiciy of labor migraion. w The sock of effecive labor L e is defined as he number of ne workers ( N ) imes heir corresponding produciviy level (e j ). 44 w w e, = ( j, j, ) j j= L POP M e 44 = N e j= w j, j (.8) Reirees aged over 65 are assumed o migrae from one region o he oher region wih exogenously given rae ε where M R R and POP are he number of reiree migrans and oal reirees populaion a ime, respecively. M R = ε POP (.9) R 2.5 Marke Clearing Condiions There are wo equilibrium condiions ha close he model. Firs, he equilibrium condiions for good marke mus hold, which saes ha domesic oupu is equal o oal demand from household (C ), governmen (G ), and firms (I ).

17 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 7 D = C + G + I 65 = [ ( N C + N C )] + G + K ( δ ) K j= HOME HOME ROUS ROUS j, j, j, j, (.20) The second condiion is equilibrium in he financial marke. Financial marke equilibrium condiion assures ha he sock of asses accumulaed by he all individuals mus be equal o he sum of he capial sock used in boh regions. A + A = K + K (.2) HOME ROUS HOME ROUS 3 Daa and Calibraion One of he key issues in compuable general equilibrium modeling is calibraion, which is he process of selecing values of exogenous parameers o ensure ha he soluion is consisen wih wha is observed in he daa. The calibraion of he model is basically conduced o replicae he equilibrium condiions in he base year, 2005 in his model. Since naional values are easily obained from he accessible naional daa se like NIPA and previous sudies (Brown, e al., 992; Kouparisas, 998) he following mainly describes he choice of regional parameers. Seady sae condiions and microconsisen daa se for Chicago region are mosly obained from he Chicago Social Accouning Marix (SAM) consruced by MIG (997), Illinois inpu-oupu mulipliers and Chicago inpu-oupu ables prepared by REAL (Regional Economics Applicaions Laboraory) in he Universiy of Illinois. Furher, a compuable general equilibrium model for Chicago region under single represenaive household has been compleed and many of he parameers for his model are used in his wo-region sysem. Some regional parameers which appear in he uiliy and producion funcions are obained from he corresponding naional counerpars since he model assumes same ype of household preferences and producion funcion across regions. For example, coefficien of relaive risk aversion is chosen by γ =.9 following he esimaes esablished by Hurd (989) and Imrohoroglu e al. (999). The subjecive discoun facor is chosen by / ( + ρ ) =.0 following he suggesion of Imrohoroglu e al. (999) o reproduce a reasonable wealh-oupu raio. Boh preference parameers generae he wealh-oupu raio of 2.89, which is slighly lower han he empirical measuremen of 3.5 by Lainer (992). The producion parameers are calibraed along he line suggesed by previous sudies. The depreciaion rae (δ) and he echnology parameer (A) for boh Chicago region and res of he US are se a and.005, respecively. The labor share of oupu (α) for Chicago region is calibraed using Chicago SAM, yielding a value of 0.66, insead 0.69 for res of he US. Mehra and Presco (985) sugges ha he coefficien of relaive risk aversion is beween and 2.

18 8 Seryoung Park and Geoffrey J.D. Hewings For he demographic daa se, populaion change by age cohors unil 2050 is obained from he projecions provided by U.N. and Illinois Deparmen of Commerce and Economic Opporuniy. The condiional survival probabiliies (s) are aken from Faber (982). This implies a dependency raio resuls in a dependency raio of 7.7 percen in he base year, which is close o a raio of 7.8 percen based on he U.S. census daa for year And, over he demographic ransiion periods, he dependency raio calibraed in he model closely approximaes o he one from he U.N. projecion. The labor earning s profile is aken from Hansen (99) and Miles (999). Also, he price elasiciies in inerregional rade are assumed o be he same as hose in inernaional rade following he suggesion by Jones and Whalley (989). The labor migraion elasiciy is specified a 0.37, reflecing he pas sudies on inerregional migraion (Plau, 98; Seung and Kraybill, 200). Replacemen rae for he base year is aken o be 50 percen of he average wage income, which maches is empirical counerpar. 2 IMPACTS OF AGING POPULATION In his secion, aging populaion is addressed as one of he main sources of local demographic changes. Firs, he simulaed resuls wih no aging are described as he Baseline Scenario, and hen, hese resuls are compared wih he simulaion resuls under aging populaion.. Baseline Scenario The baseline simulaion basically assumes ha here is no change in populaion srucure. I runs wih he age disribuion and populaion growh rae unchanged forever from heir 2005 levels, keeping oher parameers a heir base levels. Also, i is assumed ha here are no changes in public policies on immigraion and pension sysems. Figs. 2 hrough 7 summarize he simulaed opimal decisions by he age cohors, represened as an individual agen living in Chicago region. The following describes he simulaion resuls on asse holdings, income, saving, and consumpion in urn. The classical life-cycle heory predics ha each individual who has o provide for old age consumpion from his privae savings accumulaes a cerain amoun of asses during he working age. Fig. 2 shows ha asses increase from he mid 20s unil around reiremen, and hen declines seadily o zero by deah as classical life-cycle predics. Fig. 3 shows he income profiles by age. Toal income, which includes wages, capial income from savings, and pension benefis, coninues o rise unil he age of mid 50s, when wage income reaches is peak, and 2 Conesa and Krueger (998) compue he replacemen rae of 50 percen using social securiy payroll ax and OASI (Old-Age Survivors Insurance) daa for he U.S.

19 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model 9 hen drops abruply a he ime of reiremen. 3 Afer reiremen, income declines gradually due o he decrease in asse holdings. When we look a he income by age and is composiion, before reiremen, wage income accouns for more han 70 percen of he oal income, while capial income capures a small slice of oal income. However, afer reiremen, pension benefis replace he small fracion of wage income. During he early reiremen period, capial income and pension benefis compose he oal income wih approximaely equal shares. However, approaching he oldes age, capial income falls relaively fas as he sock of asses declines, while he pension benefi is consan. Thus, he pension benefi is significanly responsible for he oal income of he oldes generaions Fig. 2 Baseline simulaion resuls (no aging): Asses 3 If he reiremen decision was endogenous and hours of work could be gradually adjused afer reiremen, hen average income would no change abruply a reiremen.

20 20 Seryoung Park and Geoffrey J.D. Hewings oal income wage pension Fig. 3 Baseline simulaion resuls (no aging): Income Fig. 4 Baseline simulaion resuls (no aging): Saving Fig. 4 depics he opimal paern of lifeime saving. Individuals save asses beween heir mid 20s and early 60s and hen begin o conver hem for consumpion during reiremen. Especially, he oldes show srongly negaive saving moives because hey expec heir remaining lifeime o exend wih lower uncerainy, i.e., older reirees have a higher marginal propensiy o consume ou of heir lifeime savings han workers and younger reirees. This saving paern over he lifeime fundamenally coincides wih he consumpion smoohing moive, i.e., individuals end o save more early in heir lifeime and pospone consumpion laer when he reiremen reduces he income. Similarly, he saving rae (fig. 5), defined as household savings as a percenage of disposable income, urns from a posiive o a negaive value around reiremen.

21 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model Fig. 5 Baseline simulaion resuls (no aging): Savings rae Fig. 6 Baseline simulaion resuls (no aging): Consumpion

22 22 Seryoung Park and Geoffrey J.D. Hewings Fig. 7 Baseline simulaion resuls (no aging): Consumpion propensiy Figs. 6 and 7 show he age profiles of consumpion and is propensiy. The consumpion profile by age is smooher han he income profile. The level of consumpion sars o rise from he younges age, and keeps rising around he reiremen age. Afer reiremen, consumpion sars o decline because reiremen is a greaer negaive shock o income resources. However, reirees keep he consumpion a he level of he mid 30s, hough his gradually decreases o he end of life. As a resul, he consumpion propensiy, defined as household consumpion as a percenage of disposable income, shows a reversed lifeime paern in comparison o he level of consumpion. I moves around 0.7~0.8, which implies 70~80 percen of disposable income is spen for consumpion, unil middle age and hen increases o.2~.8 during reiremen. 4 This resul yields a very imporan implicaions for he economic condiions under aging populaion. I implies ha once a large porion of baby-boomers reire around he year 2030, hey will accoun for significan par of aggregae demand in he economy. Moreover, his resul suggess ha heavy losses in he elderly populaion hrough reiremen ou-migraion, he experience of some larger meropolian regions like New York and Chicago, could seriously hur he local economic growh over he nex couple of decades since he elderly ou-migraion is expeced o subsanially increase wih aging populaion. 4 The paern of his simulaed age-consumpion propensiy is very similar o he empirical counerpar repored by Yoon and Hewings (2008), who used CREIM (Chicago Region Economeric Inpu-oupu Model). However, he simulaed profiles for reired persons appear o have relaively higher propensiies. This can be aribued o he fac ha he beques moive does no exis, so, accumulaed wealh has o be compleely exhaused before deah.

23 Aging and he Regional Economy: Simulaion Resuls from he Chicago CGE Model Aging Populaion This secion invesigaes he effecs of aging populaion primarily on he economy of he Chicago region, reflecing he fac ha he overall effec of aging populaion on he res of he U.S. is basically similar o he Chicago region. Demographic changes associaed wih aging populaion are inroduced by replicaing he U.N. projeced dependency raio. The replacemen rae is held consan a 50 percen over he enire horizon. I is assumed ha he aging populaion is no expeced in he base year, which is 2005, when he economy is in he iniial seady sae asses(no aging) asses(aging) Fig. 8 Aging simulaion resuls: Asses Figs. 8 hrough 3 compare he projeced resuls from he simulaions wih and wihou aging populaion. For asse holdings, aging populaion appears o pu upward pressure on he asse holdings (see fig. 8). This happens basically because individuals in his economy who save hrough privae saving in order o provide for old-age consumpion, are moivaed o accumulae addiional asses in response o increasing life expecancy. However, he increase in asses is no large enough, especially during mos working age periods; hence, accumulaed asses do no change significanly before and afer aging populaion.