The Effect of Alternative World Fertility Scenarios on the World Interest Rate, Net International Capital Flows and Living Standards

Transcription

1 6/09/2002 Te Effect of Alternative World Fertility Scenarios on te World Interest Rate, Net International Capital Flows and Living Standards Ross S. Guest Griffit University Australia Ian M. McDonald Te University of Melbourne Australia ABSTRACT Tis paper applies a multi-region Ramsey model to te question of te impact of alternative fertility scenarios on te world interest rate, net international capital flows and living standards. Te world economy is divided into nine regions consisting of te eigt regions in te United Nations long run demograpic proections (998 Revision) plus Japan as a separate region. Age-specific consumption demands and age-specific labour productivity levels are applied. Te model is simulated for te low, medium and ig fertility scenarios as proected for all regions by te United Nations. Population ageing in five of te nine regions Japan, Europe, Nort America, Cina and Oceania - over te next few decades causes a fall in te world interest rate, wic is greater te lower te fertility scenario. Te size of net international capital flows is smaller for te low fertility scenario and iger for te ig fertility scenario. Tis is due to te different effect on optimal saving of a cange in te world interest rate for borrowing and lending regions. For lending regions te income effect is positive but for borrowing regions it is negative. Tis causes lending regions to reduce teir lending, and borrowing regions to reduce teir borrowing, in response to lower fertility; te result is smaller current account deficits and surpluses. Higer fertility results in bigger current account deficits and surpluses. Lower fertility results in an initial boost in living standards followed by reduced living standards later on. Borrowing regions are better off, in terms of living standards, tan lending regions following a lower fertility sock.

2 . Introduction Tere is currently, in OECD countries in particular, a vigorous debate about te role tat policy makers can, and ougt to, play in influencing te fertility rate. Public concern is widespread tat lower fertility rates imply lower living standards, yet te teoretical and empirical evidence is ambiguous. Te transition to lower population growt yields a sort term dividend for living standards in te form of lower yout dependency and lower capital widening requirements. Tis dividend is offset later on by iger old age dependency. For a discussion along wit simulation results see Weil, 999; Elmendorf and Seiner, 2000; and Guest and McDonald, Tese models are, owever, single region models. In a multi-regional framework te effect on living standards depends on te effects of lower fertility rates on net international capital flows, troug te effect on saving and investment flows, wic in turn alters world interest rates. Te effect of canging world interest rates on living standards differs for borrowing and lending regions. For borrowers te income effect is negative wile for lenders it is positive. Tis paper quantifies te impact of alternative fertility rates on international capital flows, world interest rates and living standards by simulating a multi-regional Ramsey model. Te model incorporates te canging demograpic structure implied by UN population proections for te 2 st century under tree different fertility assumptions low, medium and ig. Tere are several classes of multi-regional macroeconomic models tat can be used to analyse te macroeconomic implications of population ageing. Te model in tis paper belongs to te class of multi-regional Cass-Ramsey-Solow models. Tis is a single good model. Altoug generations are not overlapping in tis framework, eterogeneity of workers and consumers is captured by weigting for age-specific

3 2 productivity and consumption needs, respectively. Oter examples of tis class of models is te OECD model in Borsc-Supan (996) and te two region open economy model in Cutler et al. (990). A variation on tis approac is te international overlapping generations models of Turner et al. (998) using te OECD s Minilink model, te models of te Ingenue Team (200) and Fougere, M. and Merette, M. (999), and te macroeconometric model of Masson and Tryon (990). A tird and newer class of models is te multi-good general equilibrium models suc as te G- cubed model (McKibbin and Wilcoxen, 995). However, we were unable to find any specific applications of tis class of models to te impact of population ageing. 2. Demograpic proections In tis model, te world economy is divided into nine regions consisting of te eigt regions in te United Nations (2000) long run demograpic proections plus Japan as a separate region. Te nine regions are: Africa, Asia (excluding Cina, India and Japan), Cina, Europe, India, Japan, Latin America, Nort America and Oceania. Tis is a larger number of regions tan as been adopted in oter models listed above. We coose for comparison te low, medium and ig fertility scenarios in te United Nations proections (up to 250). Te medium scenario assumes tat fertility in all maor areas stabilizes at replacement level around 2050; te low scenario assumes tat fertility is alf a cild lower tan in te medium scenario; and te ig scenario assumes tat fertility is alf a cild iger tan in te medium scenario. For eac of tese population scenarios, employment proections by age and sex are calculated from te International Labour Organisation (ILO, 200) database: Key Indicators of te Labour Market (KILM). Tese data provide labour force and population by age group and sex for eac country in te world for te latest year typically 999 or From

4 3 tese data te aggregate labour force participation rates (LFPR) for eac of te nine regions, by age group, are calculated. Tese age and sex-specific LFPRs are assumed to be te age-specific employment to population (L/N) ratios. Employment proections for eac region are calculated by assuming tat te age-specific L/N ratios (for bot sexes) converge toward tose of Nort America according to te following formula: ( L N ) = ( L N ) ( L N ) ( L N ) NA a, a, a (), a γ were is te year from 200 to 250, NA is Nort America, a is te age group, γ is te convergence parameter set equal to Following Cutler et al. (990) and Elmendorf and Seiner (2000) te employment and population numbers are weigted to account for, respectively, agespecific differences in labour productivity and consumption needs. Te ageproductivity relation in Miles (999) is adopted, were te productivity weigt is a quadratic function of age: 0.05age age 2. Te consumption weigts are tose applied in Cutler et al (990); tat is 0.72 for 0-9 year olds,.0 for year olds and.27 for over 64 year olds. Bot productivity weigts and consumption weigts are non-gender specific. Te aggregate weigted L/N ratio for eac region is te support ratio (Cutler et al., 990). A decrease in te support ratio implies a diminised capacity to meet a given level of consumption needs per capita. An increase in te support ratio implies te opposite. Cart plots te support ratios for te nine regions. Te tree regions tat face imminent steep declines in teir support ratios are Japan, Europe, Cina and Nort America. Te first tree of tese are large net lenders to te rest of te world and it will be seen below tat it is optimal for tis to continue in te next few decades. Tis

5 4 will enable tem to build up wealt to provide for teir reduced capacity to meet future consumption needs. 3. Te model Te simulations are based on a multi-region Ramsey model of optimal saving, wit eterogeneous workers and consumers, abit formation in consumption, and a vintage production function. A social planner maximises, for eac region, a social welfare function (see Table for a list of variable names). ( ) ( ) ) ( ) ( ψ ρ ω β ρ ψ β = = N W N P C P C N V =,.., (2) were C 0 is exogenous and equal to actual consumption for te region. Equation (2) is maximised subect to te evolution of foreign debt Y r D C I D,.., ) ( = = (3) te vintage production function, L L I A Y Y,.., 2 ) ) ( ( ) ( = + = α α δ δ (4) and te definition of terminal wealt = + = = I K K were D K W 0 ) ( ) ( δ δ (6) = = 9, 9, i i i i I = S =,.., (7) Te social welfare function (2) includes a reference level of consumption, C -, wic captures abit formation were te memory in abit-formation lasts for only

6 5 one period. Furer (2000) finds empirical evidence to support suc a low degree of persistence in abit formation, using aggregate consumption data for te U.S. over te period 966 to 995. Te world interest rate is determined suc tat world investment equals world saving, condition (7). However, te pat of te world interest is exogenous to te social planner for eac region, wic implies tat no region as market power in te world capital market and tat indebted regions do not face risk premia on teir borrowings. Tis assumption ensures tat te solution is a global optimum. Te initial value of total factor productivity, A, in production function (4) is calibrated for all regions suc tat optimal investment is equal to actual investment in te initial year. Tereafter, total factor productivity growt is exogenous and constant at % per annum for all regions. Tis implies two tings first, tat tere is no influence of ageing on tecnical progress and, second, tat tere is no convergence of labour productivity among te nine regions. On te effect of ageing on tecnical progress bot te teoretical and empirical evidence is ambiguous, as discussed by Cutler et al. (990, p. 38). On te one and slower population growt makes innovation less profitable by reducing te gains from economies of scale troug te spreading of fixed costs; and a smaller yout sare of te population may reduce innovation troug a loss of dynamism. Also, in endogenous growt models wit uman capital, te effect of ageing on productivity growt is ambiguous. In te model in Steinman et al. (998) lower population growt results in less uman capital accumulation and terefore a lower growt rate of labour productivity. On te oter and, te model in Fougere and Merrete (998) sows tat ageing will increase uman capital formation, troug tax effects. Furtermore, slower labour force growt implies Te rationale for adopting a vintage production function in simulating te impact of population ageing,

7 6 a iger relative price of labour and terefore greater incentive to innovate troug capital investment. Also, diseconomies of iger population growt, troug congestion for example, can reduce labour productivity growt. Te empirical evidence on te effect of fertility on labour productivity is relatively scarce - see for example Galor et al. (997), Aituv (200) and Hondroyiannis and Papapetrou (999) - and somewat conflicting. Hence te assumption ere of zero net effect of ageing on total factor productivity growt seems to be a reasonable starting point. Wit respect to te zero convergence assumption, Barro and Salai-Martin (995, p.26) report tat te ypotesis of absolute convergence were poor countries catc-up wit ric countries in teir GDP per capita, witout allowing for any control variables as received mixed empirical support. 2 Neverteless, most multi-regional macroeconomic models adopt some form of productivity convergence. Using te OECDs multi-regional Minilink model, Turner et al (998) assume slow convergence in te rates of tecnical progress of teir five world regions, as distinct from convergence in teir productivity levels. Te Ingenue Team (200) assume extremely slow convergence in levels of productivity te gap between ric and poor countries appears to close by about 20% over 00 years. Dynamic intertemporal general equilibrium models, suc as te G-Cubed Model, also typically incorporate some form of tecnology catc-up (McKibbin, 999). In tis paper, zero convergence in total factor productivity is assumed because, in our initial simulations, we found tat all but a very small rate of convergence tended to swamp te effects of differences in fertility rates tat we were attempting to isolate. As a result of tis, and te uncertainty is discussed in Guest and McDonald, 2002(a) (available from te autors). 2 Te ypotesis of conditional convergence, wic controls for various caracteristics of economies, as received stronger empirical support. However, even te testing of tis weaker ypotesis faces some serious econometric problems. See Durlauf and Qua in Taylor and Woodford eds. (999) for a compreensive discussion of te teory and empirical tests of convergence ypoteses.

8 7 in te empirical literature about productivity convergence, we felt tat zero convergence was a reasonable assumption for our purposes. Te parameter ω is set suc tat te ratio of terminal wealt to consumption is equal to te actual ratio of wealt to consumption in Tis prevents a rundown of wealt in te simulations. Te simulation lengt,, is equal to 50 years wic is lengt of te United Nations demograpic proections and also long enoug tat optimal values in te first 00 years te period of interest - would not be canged significantly by increasing. Te rate of time preference, ρ, is set equal to for eac region, wic is approximately te rate suc tat in a steady state te sare of consumption in output approaces a constant value. Table. Definition of variables year of te planning period tat runs from to i region, from to 9 Y real GDP C aggregate consumption W wealt=capital stock plus overseas assets I aggregate investment D overseas debt K capital stock r world interest rate δ rate of depreciation (5%) ρ rate of time preference (2.5%) β reciprocal of intertemporal elasticity of substitution (2.0) ω weigt attaced to bequest of terminal wealt in te social welfare function L aggregate employment in relative efficiency units terminal period of te maximisation problem (50 years) N population P population in effective persons A efficiency parameter te initial value is set suc tat te optimal level of investment is equal to te actual level of investment. Tereafter A grows at % p.a. National accounts data for te initial year are from IMF International Financial Statistics Yearbook (2000). Te capital stock to GDP ratio is assumed to be uncanged

9 8 from teir latest values available from te Penn World Tables (Summers and Heston, 993). 4. Te world rate of interest and net international capital flows Figure 2 sows te impact of te future patterns of demograpic cange on te world rate of interest. Te base case is te medium fertility scenario. Under all fertility scenarios, te world interest rate falls. Te regions tat face imminent population ageing and ence falls in teir support ratios Japan, Europe, Cina, Nort America and Oceania find it optimal to increase saving and reduce investment. Higer saving allows tese regions to smoot consumption and terefore smoot te burden of ageing over time. At te same time teir lower employment growt reduces bot te marginal product of new capital and capital widening requirements, implying lower optimal investment to output ratios. Te relative size of te regions facing tese forces puts downward pressure on te world interest rate. Te lower is te fertility rate te greater are tese forces, and ence te bigger te drop in te world interest rate. Te effect of alternative fertility scenarios on net international capital flows works troug te effects on investment, saving and ence current account balances. Te direction of tese effects is ambiguous in a model in wic te world interest rate is endogenous. As noted above, lower fertility reduces te optimal investment ratio. However, tis is a first round effect. Tere is a second round effect tat works in te opposite direction, increasing investment troug te lower world interest rates tat result from lower fertility. Our simulations sow tat tese effects approximately cancel out for at least several decades after wic te negative employment dominates. Tis is illustrated in Figure 3. Tere are minor differences in te magnitude of tese effects between regions.

10 9 Te effect of lower fertility on saving is also ambiguous. Te iger saving described above arising from lower fertility is also a first round effect only. A second round effect occurs troug te effect on saving of a canging world interest rate. Here te effect is different for borrowing and lending regions. Lending regions face a positive income effect (lower income) and a negative substitution effect from te lower interest rate tat results from lower fertility. Bot of tese effects tend to lower saving in response to lower fertility and terefore work in te opposite direction to te first round effect on saving. Simulation sowed tat te second round effect dominates te first round effect for lending regions leading to lower saving. See Figure 3. Hence lending regions face an initial small increase in investment and a larger decrease in saving. Te result is lower current account surpluses for lending regions as a result of lower fertility. Te opposite occurs for iger fertility. Tat is, te iger interest rate leads to iger saving and an initial small fall in investment te net result being an increase in te current account surpluses for lending regions. For borrowing regions, te second round effect on saving is different to tat for lending regions as sown in Figure 3. Te sign of te income effect is negative for borrowing regions because tey gain income from lower interest rates. Hence te income and substitution effects work in te opposite direction on saving. Te net of te income and substitution effects is small enoug tat te first round effect on saving dominates te second round effect of te lower interest rate. Te result is iger saving for borrowing regions in response to lower fertility. Combining wit te small initial effect on investment and large negative effect later on - te outcome is a reduction in current account deficits for borrowing regions. Again, iger fertility as te opposite effect an increase in current account deficits for borrowing regions.

11 0 To summarise, lower fertility leads to lower current account surpluses for lending regions and lower current account deficits for borrowing regions. Hence net international capital flows are lower under te lower fertility scenario. Te opposite occurs for iger fertility an increase in net international capital flows. Tis is illustrated in Figure 4 wic sows te size of net international capital flows for te tree alternative fertility scenarios over te next 00 years. 5. Living standards Living standards are defined as aggregate consumption per effective person, C/P. In considering te impact of fertility scenarios on living standards te analysis above suggests te need to distinguis between borrowing and lending regions. In addition, we compare te case were te wole world experiences lower fertility wit te case were only one region experiences lower fertility. Te effects of population ageing on optimal consumption are discussed in (Elmendorf and Seiner, 2000). Te new steady state level of consumption in response to an ageing sock is lower as a result of te net of two effects: te dependency effect and te Solow effect. Higer overall dependency (te combination of yout and old age dependency) implies fewer workers per person wic lowers consumption possibilities. Te Solow effect refers to te gains from te lower capital widening requirements tat come wit lower employment growt. Te pat to tis lower level of consumption is determined by te canges in te return to saving during te demograpic transition. In te closed economy tis is determined by te effect of canges in te capital labour ratio on te marginal product of capital. In te open economy it is determined by canges in te interest rate at wic te economy borrows and lends from overseas.

12 Witin an overall scenario of population ageing, te effect of lower fertility on living standards in a single economy Ramsey model as been simulated by Guest and McDonald (2002b, 2002c). 3 Tey sow tat, in te case were te interest rate is constant, lower fertility is likely to result in an improvement in living standards trougout te planning period. Te reason is tat tere are two consumption dividends tat accrue as soon as fertility drops. Tese are te dividend from lower yout dependency and from lower capital widening requirements. Tese are eventually offset by iger old age dependency. Te net effect is a gain wic can be smooted out over te planning period resulting in iger living standards trougout. Were te interest rate is endogenous te result is somewat different in tat optimal consumption is initially iger, but subsequently lower, under lower fertility (Guest and McDonald, 2002c). However, for a range of reasonable discount rates and parameter values tere remains a small net discounted gain from lower fertility. Te difference in te effect of te fertility rate on living standards between te endogenous and exogenous world interest rate models arises troug te second round effects of a cange in te interest rate on te return to saving and on te cost of capital. Two cases are simulated ere. One were a different rate of fertility is experienced by all regions in te world and anoter case were only one region experiences a different fertility scenario. Te latter case is described as te own region effect of lower fertility. Te own region effect is muc smaller because te effect on te world interest rate is muc smaller. Figure 5 illustrates te effect of fertility on living standards for bot a representative lending region (Europe) and a representative borrowing region (Africa). Taking te lending region first, lower fertility results in an initial rise in living 3 Te first reference assumes a constant exogenous interest rate and te second reference allows for a

13 2 standards as consumption is brougt forward by te lower interest rate. Tis implies lower consumption later, wic is reinforced by te income effect of lower interest rates (implying lower income). Te outcome is muc lower living standards later in te planning period. Te reverse story applies in te case of iger fertility. Te iger interest rate causes consumption to be postponed early on, relative to te base case, but te effect of iger income enoyed by te lending country eventually allows for iger living standards. For te borrowing country, te pattern is somewat different. For lower fertility, te initial rise in living standards as consumption is brougt forward is reinforced by te income effect of lower interest rates (implying iger income) wic allows iger living standards. Hence te initial rise in living standards is iger and tere is no subsequent fall in living standards for te borrowing country. By te same token, te borrowing country suffers lower income from iger world interest rates under iger fertility. Hence living standards are lower under te ig fertility scenario. Te own region effect remains relatively small for bot te borrowing and te lending country. Table sows, for eac of te 9 regions, te average annual loss in living standards between te year 2000 and 200 from experiencing lower fertility trougout tat time period. Te average losses are all less tan 0.5% wic can be considered to be very small. Table. Te effect on living standards of lower fertility Average annual % loss Japan 0.09 Europe 0.2 Nort America 0.06 Oceania 0.07 Asia 0.02 Cina 0.4 risk premium in te interest rate.

14 3 Africa -0.0 India 0.06 Latin America 0.06 Finally, we found tat te effect of alternative fertility scenarios on living standards is almost entirely accounted for by te dependency effect. Tis is illustrated in Figure 6 for te case were Europe, as an example, faces a lower fertility scenario. Te gap between te total effect and te dependency effect represents te net of two effects described above te Solow effect and te capital intensity effect. Tese effects are clearly dominated by te dependency effect wic is itself te net of a yout dependency effect and an old age dependency effect. 6. Conclusion From te simulations of a global model of optimal consumption, investment and saving in tis paper, tree maor conclusions stand out. First, te lower te fertility rate te lower te world rate of interest. Tis is due to bot lower demand for new capital implied by slower employment growt, and iger saving implied by smooting of te reduced consumption possibilities. To equilibrate te world capital market, te equality of world saving wit world investment requires a lower equilibrium rate of interest. Second, lower fertility tends to reduce te size of international capital flows. Lending regions lend less and borrowing regions borrow less. Tis is largely due to te fact tat te income effects of lower interest rates are of opposite sign for borrowers and lenders. Tird, lower fertility reduces te average growt in living standards over time. Tis loss is greater for lenders as a result of te lower income tat tey earn on overseas assets as a result of lower interest rates.

15 4 Tere are several simplifying assumptions tat ave been adopted in tis analysis in order to quarantine te effect of fertility on te outcomes. For example, all regions experience in te 2 st century te same rate of growt of total factor productivity and tere are no risk premia faced by borrowers. Tese assumptions could - peraps ougt - to be relaxed in future work.

16 5 APPENDIX Te first order conditions for te maximisation problem for eac region, described by equations (2) to (6) are: V C V C = ( + r ) k= k for =,..,- (A) dy+ = di for =,,- (A2) k ( δ) k= + k (+ r ) τ= τ Te terminal first order condition is V V = C W (A3)

17 6 References Aituv, A. (200), Be Fruitful or Multiply: On te Interplay Between Fertility and Economic Development, Journal of Population Economics, 4, 5-7. Barro, R. and Sala-I-Martin, X. (995), Economic Growt, McGraw-Hill, New York. Borsc-Supan, A. (996), Te Impact of Population Ageing on Savings, Investment and Growt in te OECD Area, in Future Global Capital Sortages, OECD, Paris, Cutler, D.M., Poterba, J.M., Seiner, L.M. and Summers, L.H. (990) "An Aging Society: Opportunity or Callenge?" Brookings Papers on Economic Activity, (), pp.-74. Elmendorf, D.W. and Seiner, L.M. (2000), Sould America Save for its Old Age? Fiscal Policy, Population Ageing and National Saving, Journal of Economic Perspectives, 4, 3, Fougere, M. and Merette, M. (998), Population Ageing, Intergenerational Equity and Growt: Analysis wit an Endogenous Growt, Overlapping Generations Model, International Conference on Using Dynamic Computable General Equilibrium Models for Policy Analysis, Denmark, June 4-7 Fougere, M., and Merette, M. (999) Population ageing and economic growt in seven OECD countries. Economic Modelling, 6, Furer, J.C. (2000), Habit Formation in Consumption and Its Implications for Monetary-Policy Models, Te American Economic Review, 90, 3, Galor, O. and Hyoungsoo, Z. (997), Fertility, Income Distribution, and Economic Growt: Teory and Cross-Country Evidence, Japan and te World Economy, 9,

18 7 Guest, R. and McDonald, I.M., (2002a), Vintage Versus Homogeneous Capital in Simulations of Population Ageing: Does it Matter, mimeo. Guest, R. and McDonald, I.M., (2002b), Would a Decrease in Fertility be a Treat to Living Standards in Australia, Australian Economic Review, 35,, Guest, R. and McDonald, I.M., (2002c), Population Ageing, Capital mobility and Optimal Saving, seminar paper presented at Queensland University of Tecnology, August, Hondroyiannis, G. and Papapetrou, E. (999), Fertility Coice and Economic Growt: Empirical Evidence from te U.S., I.A.E.R., 5,, International Labour Organisation (ILO)(200), Key Indicators of te Labour Market , ILO, Geneva. International Monetary Fund (IMF) (2000), Internationial Financial Statistics Yearbook 2000, IMF, Wasington DC. Masson, P.R. and Tryon, R.W. (990), Macroeconomic Effects of Proected Population Ageing in Industrial Countries, IMF Staff Papers, 37, 3. McKibbin, W. and Wilcoxen, P. (995), Te Teoretical and Empirical Structure of te G-Cubed Model, Brookings Discussion Paper in International Economics, #8, Te Brookings Institution, Wasington DC. McKibbin, W. (999), Forecasting te World Economy Using Dynamic Intertemporal General Equilibrium Multi-Country Models, Brookings Discussion Paper in International Economics, #45, Te Brookings Institution, Wasington DC. Miles, D., (999) Modelling te impact of demograpic cange on te economy Economic Journal, 09, pp.-36. Steinman, G., Prskawetz, A. and Feictinger, G. (998), A Model of Escape From te Maltusian Trap, Journal of Population Economics,,

19 8 Summers, L.H. and Heston, A. (993), Penn World Tables, available at ttp:// Turner, D., Giorno, C. DeSerres, A., Vourc, A. and Ricardson, P. (998), Te Macroeconomic Implications of Ageing in a Global Context, OECD Economic Department Working Papers, 93. Taylor, J. B. and Woodford, M. (Eds.) (999), Handbook of Macroeconomics Volume A, Elsevier, Amsterdam. United Nations (2000), Long Range World Population Proections. Based on te 998 Revision, United Nations, New York 200. Weil, D. (999), Wy as fertility Fallen Below Replacement in Industrial Nations, and Will it Last?, American Economic Review, 89, 2,

20 9 Figure. Support ratios to 250. Base fertility. Ageing regions Cina Japan Europe Nort America Oceania Oter regions Africa Rest of Asia India Latin America Figure 2. World Interest rate to ig base low

21 20 Figure 3. Saving, investment and current account balances S/Y lending regions, % of world GDP S/Y borrowing regions, % of world GDP Base Low Base Low I/Y lending regions, % of world GDP I/Y borrowing regions, % of world GDP Base Low Base Low CAB/Y lending regions, % of world GDP CAB/Y borrowing regions, % of world GDP Base Low - -2 Low Base

22 2 0. Figure 4. Size of net international capital flows for alternative fertility scenarios Calculated as te sum of te absolute value of CABs as a ratio to w orld GDP Hig Base Low

23 22 Figure 5. Impact of fertility rate on living standards for a LENDING region : EUROPE 0 05 C/P[low/base]* Eur-low;ROW-low Eur-low;ROW-base Eur-ig;ROW-base Eur-ig;ROW-ig Impact of fertility rate on living standards for a BORROWING region : AFRICA 05 C/P[low/base]* Afr-low;ROW-low Afr-low;ROW-base Afr-ig;ROW-ig Afr-ig;ROW-base

24 23 Figure 6. Dependency effect of low fertility : Europe Low fertility scenario for all world regions Percent of base C/P Dep effect Total