Aging in Europe: Reforms, international diversification and behavioral reactions

Transcription

1 Agng n Europe: Reforms, nternatonal dversfcaton and behavoral reactons Axel Börsch-Supan a, Klaus Härtl a and Alexander Ludwg a,b a Munch Research Insttute for the Economcs of Agng (MEA) at the Max Planck Insttute for Socal Law and Socal Polcy b CMR, Unversty of Cologne Ths draft verson: 31. December 2013 Abstract [exactly 100 words] The extent of demographc changes n Europe s much more drastc than n the US. Ths paper studes the effects of populaton agng on the nteractons between economc growth and lvng standards n Europe wth labor market and penson reform, behavoral adaptatons, and nternatonal captal flows. Our analyss s based on an overlappng generatons model wth behavoral reactons to reform whch s extended to the mult-country stuaton typcal for Europe. Whle the negatve effects of populaton agng on growth n Europe can n prncple be compensated by reforms and economc adaptaton mechansms, they may be partally offset by behavoral reactons. Keywords: agng; penson reform; labor market reform; labor supply reactons JEL classfcaton: J11, J21; D13; E27; H55; F16, F21 Emal: axel@boersch-supan.de, Haertl@mea.mpsoc.mpg.de, ludwg@cmr.un-koeln.de Acknowledgements: Ths paper was prepared for the Sesson Macroeconomc Impact of Populaton Agng at the 2014 ASSA Conference n Phladelpha, PA, n January We thank XXX for helpful comments. Fnancal support was provded by the Deutsche Forschungsgemenschaf the Max-Planck-Gesellschaf and the Land North Rhne Westphala. The usual dsclamer apples. [Shorten: body of paper max 9 pages 1.5 spaced ncl. references, tables and fgures]

2 Whle agng s global, there are marked nternatonal dfferences n the speed and the extent of the agng processes, notably also between the US and Europe. Ths paper focuses on the three largest Contnental European countres: France, Germany and Italy, and juxtaposes them to the US. These three European countres do not only feature a much larger declne n the support rato (number of workers per populaton) than the US but also have substantally larger publc budgets ncludng ther pay-as-you-go fnanced socal securty systems. Moreover, these countres have labor markets characterzed by low partcpaton rates of young women and ndvduals aged 55 and over. In spte of these structural problems, France, Germany and Italy have been remarkably resstant to labor market and penson reforms. Ths paper extends a lne of research based on mult-country overlappng generatons models to study the effects of populaton agng on the nteractons between economc growth and lvng standards wth relatvely mld labor marke penson and educatonal reforms, behavoral adaptatons, and nternatonal captal flows. Key queston s whch polcy mxes are sutable to mantan lvng standards despte the strong declne n the support rato. I. The model We extend the overlappng generatons model of the Auerbach and Kotlkoff (1987) type n several dmensons: we acknowledge the nternatonal trade and captal flows of European countres by a mult-country verson of the model (Börsch-Supan, Ludwg and Wnter, 2007); we model the large frctons n European labor markets by the dstncton between exogenous and endogenous labor supply components (Börsch-Supan and Ludwg, 2010); and, n ths paper, we add a model of an earnngs-related pay-as-you-go publc penson scheme typcal for France, Germany and Italy whch combnes aspects of a defned contrbuton system wth those of defned benefts. Households have preferences over consumpton and lesure. Total labor supply of a household of age j s the product of an exogenous component l j, and an endogenous component h j. The exogenous component l j can be thought of as the maxmum lfe-tme number of hours possble for a household, gven by restrctons n labor market entry (e.g., due to length of mandatory schoolng), restrctons durng man workng lfe (e.g., by the avalablty of day care facltes for famles, or the 35-hour week n France), and n older age by restrctons through mandatory retrement. Households have some ablty to choose ther preferred labor supply by choosng h j, the endogenous component of labor supply. They can be thought of hours wthn 1

3 the maxmum lfe-tme number of labor hours l j. Ths ablty, however, s lmted and asymmetrcal as h j may not exceed an upper lmt ħ. More formally, a household of age j at tme t n country derves utlty from consumpton c and lesure 1-l h where the household s per perod utlty functon s gven by 1 1 (1) c,1 h l ) c l h 1 u ( 1. 1 The maxmzaton problem of a cohort born n perod t at j=0 s gven by J j (2) max uct, 1 lt ht j0, where s the pure tme dscount factor. In addton to pure tme dscountng, households dscount future utlty wth ther uncondtonal survval probablty, j k 0 s tk, k,. Here, s t+k,k, denotes the probablty to survve from perod t+k, age k to perod t+k+1, age k+1 n country wth s 0, =1. Snce the tme of death s uncertan, we assume that accdental bequests resultng from premature death are taxed by the government at a confscatory rate and used for otherwse neutral government consumpton. Labor productvty changes over the lfe-cycle accordng to age-specfc productvty parameters ε j. Hence, the age-specfc wage s w t, j, wt,. j Denotng total assets by a, maxmzaton of the household s ntertemporal utlty s subject to a dynamc budget constrant gven by a 1 a 1 r l h w 1 1 p c,, (3) t, j1, j t j where =1 for j=0,,r and =0 for j>r and R s the exogenous retrement age. τ denotes the contrbuton rate to the pay-as-you-go fnanced publc penson system and p the penson ncome, see below. As ponted out above, maxmzaton s subject to the constrant that the endogenous component of labor supply ( hours worked wthn the lmt ) are postve and may not exceed the upper lmt ħ. Snce the model cannot dstngush between the lmt ħ and the exogenous labor supply componen we normalze ħ to one: (4) h

4 In those varants of our model n whch the labor supply s fully exogenous, we replace the constrant (4) wth the constrant that h t j,, =1 for all. The government organzes a prototypcal European earnngs-related pay-as-you-go fnanced penson system. Benefts are gven by (5) p t, w (1- ) s where ρ denotes the net replacement rate (generosty of the penson system) and the contrbuton rate of the penson system n country at tme t. The penson stock - whch captures the earnngs related component of the system and s denoted by s,j - accumulates over the lfe-cycle accordng to (6) s t1, j1, s jl h R h, where h R j j1 R R l j1 l h h N N Households thus earn one earnngs pont f they receve average wage ncome n a gven perod. Earnng ponts are normalzed by the length of the workng perod R. As we do not model ntra-generatonal heterogenety here, dfferences n wage ncome are nduced by age-specfc productvty only. Our model households understand the lnkage between contrbutons to the penson system and penson payments n old age. Therefore, relatve to a flat beneft penson system, labor supply dstortons are smaller but not zero because the rate of return on the captal market exceeds the mplct return of the penson system. The man polcy parameters of the penson system are ether the net replacement rate ρ or the contrbuton rate. The other parameter s determned endogenously snce the penson system s budget s assumed to always be balanced. If ρ s large, publc pensons substantally crowd out prvate savng through the households savng decson gven by (1) and (2). Europe does not lve n splendd solaton. There are trade and correspondng nternatonal captal flows. Savng and nvestment decsons are governed by a common global nterest rate whch, va nternatonal captal flows, equalzes the return to captal across countres. Assets held by households n country therefore do not necessarly equal the captal stock n country nor does savng necessarly equal nvestment n a sngle country. 3

5 The remander of the model s conventonal. Exogenous demography, determned by manstream fertlty, mortalty and mgraton assumptons (Eurostat 2013), determnes cohort szes and provdes the man exogenous drver of the model: populaton agng. Producton and wage settng n each country s neoclasscal. Equlbrum s acheved f supply equals demand n the natonal labor markets and n the global captal market. Further detals ncludng the numercal soluton and calbraton procedures are relegated to the techncal annex. II. Results A. results Our baselne s defned as status quo n terms of labor market and penson system. We assume constant age and gender-specfc labor force partcpaton rates and a constant replacement rate ρ, see equaton (5). Intally, we also assume equal productvty for all ages. All results refer to EU-3, the aggregate of France, Germany and Italy. As ther populatons age, the support rato declnes by 20 percent from 2005 untl As a consequence, GDP per capta would declne by 15 percent and consumpton per capta by about 10 percent relatve to a non-agng economy wth the same total factor productvty f polces and behavor were to reman at the current status quo, see the left panel of Fgure 1. FIGURE 1: results labor supply, GNP, GDP and consumpton per capta, detrended 105,0% Internatonal captal flows as share of GDP 3% 100,0% 2% 95,0% CperAN 1% 90,0% 85,0% GNPperAN YperAN Support 0% 1% 80,0% 2% 75,0% 3% Note: Authors computatons. Varables n the left panel are normalzed to 100% n 2005 and net of TFP trend. Captal flows are savng mnus domestc nvestment n Europe, relatve to GDP and normalzed to a balance of zero n B. Internatonal captal flows The declne n GDP per capta s smaller than the declne of the support rato because scarce labor due to populaton agng s partally substtuted by addtonal captal. Ths adaptaton occurs n response to rsng wages and fallng nterests rates. Snce the US s agng much less 4

6 than Europe, the return to captal would fall less (and wages ncrease less) than n Europe f these two regons were economcally solated. In an open economy settng, however, European households wll nvest n foregn captal dervng hgher returns untl a common nterest rate s acheved n equlbrum. From a lfe-cycle pont of vew, such behavor dffers accordng to age: eventually, households wll repatrate ther foregn savngs and, accordng to the lfe-cycle mechansm underlyng equatons (1) and (2), enjoy ther retrement consumpton. The aggregate effect depends on demography. The large cohort szes born n the 60s and 70s lead to frst rsng, then fallng net captal outflows, untl they turn negatve after about the year 2035, see the rght panel of Fgure 1. These nternatonal captal flows reach almost 2% of GDP and are substantal n the sense that consumpton per capta falls by about 5 percentage ponts less than GDP per capta n It also mples that de-trended GNP s substantally larger than de-trended GDP untl about 2055 (left panel of Fgure 1). C. Labor market reform Labor market reform s supposed to releve some of the current restrctons on European labor markets n order to ncrease labor supply and to offset the declne of the support rato n the course of populaton agng. In the language of our model, the exogenous component l j wll be ncreased. As a realstc example, our labor market reform scenaro ncludes four polcy changes: o ncrease n the retrement age by 2 years; o decrease n the job entry age by 2 years; o convergence of female labor force partcpaton to 90 percent of the rate for men; o reducton of unemployment to the NAIRU rate (Ball and Mankw, 2002). These reform steps are motvated by actual polcy proposals. The Mont-government n Italy has rased ts retrement age; n Germany, the statutory retrement age s gradually rasng from 65 to 67 years untl the year 2029; and n France, the mnmum pensonable age of 60 has been rased to 62. The change n the European hgh school and unversty system (the so called Bologna process) s expected to decrease duraton n schoolng by about 2 years. These reform steps wll be phased n lnearly between 2010 and If hours supply h j s exogenous, the economy ncreases ts capacty accordngly and the declne n the support rato s offset to about 94%, see the upper left panel n Fgure 2. In addton, savng and nvestment behavor reacts leadng to a small ncrease n the domestc captal stock relatve to the baselne scenaro of secton B (upper rght panel), thereby ncreasng 5

7 GDP per capta slghtly above the trend of the support rato (lower left panel). Furthermore, added savng flows abroad and eventually ncreases consumpton per capta stronger than per capta GDP (lower rght panel). As a resul economc lvng standards, here measured as per capta consumpton, can essentally be stablzed n spte of populaton agng n Europe. FIGURE 2: Labor market reform wth exogenous hours Support rato Captal stock 105.0% 240.0% 220.0% 95.0% 200.0% 180.0% 90.0% 160.0% 140.0% 85.0% 120.0% 80.0% GDP per capta Consumpton per capta 105.0% 105.0% 95.0% 95.0% 90.0% 90.0% 85.0% 85.0% 80.0% 80.0% Note: Authors computatons. All seres normalzed to 100% n GDP and consumpton per capta are net of TFP growth. D. Backlashes to labor market reform Overall, these reform steps do not appear to be overly radcal; n fac ther combnaton would lead n 2040 to labor force partcpaton rates farly smlar to those n Denmark, Sweden or Swtzerland today. Nevertheless, attempts to actually execute reforms wth those goals have faced stff opposton n France and Italy, and most recently durng the new grand coalton government s formaton also n Germany. Such backlash appears n our model n the substtuton between the endogenous component h j and the exogenous component l j. In the absence of constrants, the two components of labor supply are perfect substtutes such that the exogenous varaton of l j leaves the labor supply of the household unaffected: as the age-specfc employment l j s exogenously ncreased, the household endogenously decreases hours worked, h j. The exogenous varaton of l j affects total effectve labor supply, however, for those households for whom the tme endowment 6

8 constrant ħ s bndng. As a consequence, the exogenous employment varaton of l j has a postve effect on labor supply but the overall effect s substantally smaller than n the prevous secton where labor supply wafully exogenousthe nterplay between l j and ħ s most obvous n the household context where female and male workng hours are jontly determned. In our model, the resultng substtuton s extreme for households n the mddle ages of ther lfe-cycle who are unconstraned. In the real world we would probably expect a less than one for one reducton of male hours when female hours ncrease. FIGURE 3: results Support rato Captal stock 110.0% 105.0% 240.0% 220.0% 95.0% 200.0% 90.0% BACKLASH 180.0% BACKLASH 85.0% 160.0% 80.0% 140.0% 75.0% 120.0% 70.0% GDP per capta Consumpton per capta 105.0% 105.0% 95.0% 95.0% BACKLASH BACKLASH 90.0% 90.0% 85.0% 85.0% 80.0% 80.0% Note: Authors computatons. All seres normalzed to 100% n GDP and consumpton per capta are net of TFP growth. E. Penson reform The publc penson systems n France, Germany and Italy are pay-as-you-go fnanced. They used to be defned beneft (DB) systems n whch the contrbuton rates were rased to mantan a poltcally determned relatvely hgh replacement rate. Recently, reforms have ntated a process whch ntroduces elements of notonal defned contrbuton (NDC) systems n these DB systems. Ths transton s strongest n Italy, where a formal NDC system of the Swedsh type has been ntroduced for new entrants, and has recently been accelerated by the Mont government. France and Germany have earnngs-related penson systems n whch the labor 7

9 market rsk to penson benefts s borne by workers, an mportant element of a defned contrbuton (DC) system. In addton, Germany ntroduced a sustanablty factor whch adjusts penson benefts not only to productvty ncreases, but also to the rato of pensoners to workers, effectvely transtng to a mxed system wth DB and DC features. Our model captures the earnngs-related nature of the penson systems n France, Germany and Italy. In addton, we smulate two polces to set the replacement and the contrbuton rate whch bracket the current mxture of DB and DC: o o the replacement rates are constant and roughly correspond to the 2010 levels (OECD, 2013). Ths corresponds to a DB system; the contrbuton rate adjusts accordngly to mantan a balanced budget; the contrbuton rates are frozen at ther 2010 levels correspondng to a DC system; the replacement rate adjusts accordngly to mantan a balanced budget. Fgure 4 shows the resultng ncrease n the European captal stock, expressed n relaton to GDP and for the case of exogenous labor supply. The base case s a DB system n whch the replacement rate s fxed. In a DC system, the declnng replacement rate nduces workers to save more for ther retremen resultng n a larger domestc captal stock. Moreover, also nternatonal captal flows ncrease substantally to about 3.5 percent of GDP, more than offsettng the reversal after 2035 n the baselne scenaro snce the new young cohorts keep buldng up assets to fnance ther retrement consumpton. FIGURE 4: Penson reform 6,50 Increase n domestc captal stock 4% Internatonal captal flows DC 3% 5,50 DB 2% 4,50 3,50 1% 0% 1% DC DB 2% 2,50 Note: Authors computatons. Captal stock as multple of GDP. Captal flows are savng mnus domestc nvestment n Europe, normalzed to a balance of zero n % The hgher captal stock plus the larger foregn assets lead to hgher consumpton per capta, see the left panel of Fgure 5. FIGURE 5: Consumpton per capta after penson reform Exogenous hours Endogenous hours 8

10 102,0% 102,0% 100,0% 100,0% 98,0% 98,0% 96,0% 96,0% 94,0% 94,0% 92,0% DC 92,0% DC DC END 90,0% 90,0% 88,0% 88,0% Note: Authors computatons. All seres normalzed to 100% n 2005 and net of TFP growth. Such penson reform steps wll not only generate reacton n savng behavor but also nfluence labor supply. However, compared to the reactons to labor market reform, the negatve behavoral responses are relatvely modest as can be seen n the rght panel of Fgure 5. III. Dscusson Our paper shows that drect quantty and ndrect behavoral effects of populaton agng on the economy are large. Drect quantty effects come from releasng restrctons on European labor markets. Ths ncreases labor supply, nduces more retrement savng and also a hgher domestc captal stock. Moreover, addtonal savng s nvested abroad and generates nternatonal captal flows. They go frst from the old to the young countres untl the savngs are repatrated to ncrease retrement consumpton. Indrect behavoral effects strengthen savng further but substantally weaken labor supply. Both effects sgnfcantly affect economc growth and lvng standards. Due to the nteracton effects between penson system and labor markets, a smart combnaton of penson and labor market polces can do more than each of such polces n solaton. These nteracton effects are shown n Fgure 6. 9

11 FIGURE 6: Consumpton per capta when polces are combned 102,0% 100,0% DC pensons exogenous hours 98,0% 96,0% 94,0% DB pensons exogenous hours DC pensons endogenous hours 92,0% 90,0% DB pensons endogenous hours 88,0% 86,0% 84,0% Note: Authors computatons. All seres are normalzed to 100 percent n 2005 and net of TFP growth. A combnaton of a relatvely moderate labor market reform wth a penson polcy to keep the contrbuton rate at the current level wll reduce the baselne declne of consumpton per capta (red crosses n Fgure 6) to about a half (orange trangles n the fgure). The bggest obstacle to further reductons are behavoral effects generated by the hgh preference for lesure and other ncentves not explctly modeled here but mplctly captured by the nterplay between the hours lmt ħ and the exogenous labor volume l j (dfference between orange trangles and purple crcles n Fgure 6). The man message of the paper s therefore twofold. Frs t s msleadng that Europe could resolve all agng related problems by moblzng the employment pool. Such polcy proposals target at the extensve margn of labor supply and gnore behavoral reactons at the ntensve margn. In the same ven, budgetary projectons based on employment scenaros are severely based because they mplctly assume that hours worked per person are kept constant. Hence, responses to demographc change do not only requre structural reforms of labor markets and penson systems but also changes n the atttudes towards reform. Second, our analyss suggests that such reform proposals can only be effectve when households are constraned n ther hours choce by nsttutons. Ye we do nether explctly model these nsttutons nor the complex household jont labor supply decsons. We leave an extenson of our model along these dmensons for future research. The strong substtuton between 10

12 the endogenous hours h j and the exogenous labor volume l j n our model may generate too pessmstc an estmate of the reform backlash and ts effects on economc growth and lvng standards. The fac however, that all three new governments n France, Germany and Italy are currently revertng penson and labor market reforms enacted by ther predecessors shows how real the effects are of these governments fear of resstance to reform. References Auerbach, A.J. and L.J. Kotlkoff, Dynamc Fscal Polcy. Cambrdge, MA: Cambrdge Unversty Press. Ball, L. and N. G. Mankw, The NAIRU n Theory and Practce, Journal of Economc Perspectves 16, 4, Börsch-Supan, A., A. Ludwg, and J. Wnter, Agng, Penson Reform, and Captal Flows: A Mult-Country Smulaton Model, Economca, 73, Börsch-Supan, A. und A. Ludwg,, Old Europe s Agng: Reforms and Reform Backlashes, In: Shoven, J. (ed.): Demography and the Economy, Unversty of Chcago Press, Chcago, Eurosta 2013, OECD, 2013, Pensons at a Glance, OECD, Pars. 11