Advanced International Macroeconomics Sessions 3-4

Transcription

1 Advanced International Macroeconomics Sessions 3-4 Nicolas Coeurdacier - nicolas.coeurdacier@sciencespo.fr Master in Economics - Spring 2018

2 Long-term capital ows and global imbalances

3 Reminder some important stylized facts on capital ows - Global imbalances. Since mid-80s, the US are running current account de cits. Quite large starting the 90s, amounting for ' 5% of GDP per year until recently. Other smaller countries show similar patterns (UK, Australia, NZL...) but less of a concern for global stability. - On the credit side, these de cits were largely nanced by emerging markets (Asia and Middle East), even though also some large creditors in developed countries (Japan, Germany, Switzerland). Relates to the allocation puzzle : emerging marketsn (e.g. Asia) have been growing at a fast pace over the period. - Net foreign asset positions of the US deteriorated over the last 30 years (moved from creditors to very large debtors). - Together with the emergence of global imbalances, fall in world long-term interest rates.

4 Motivation What is driving (persistent) low- What is driving global imbalances? frequency capital ows? Why do fast growing emerging markets lend to slow growing developed markets? Uphill ows. Focus on long-term capital movements Focus on net capital ows

5 New views on global imbalances Traditional models with e cient allocation of capital cannot account for the following type of imbalances: fast growing emerging markets lending to slow growing developed countries. Need new theories based on some market imperfections/frictions (heterogeneous across countries). New views rely on the Metzler approach. Remaining important empirical question: what is the relevant friction or heterogeneity across countries?

6 What would traditional models predict? If we assume that emerging markets have faster productivity growth. Following a permanent (expected) increase in productivity, standard models (think neoclassical growth model) would predict: - A rise in investment and a fall in savings in the fast growing country - A current account de cit in the fast growing country - A rise in world interest rates Think intertemporal approach of current account. Opposite of the data.

7 Metzler diagram and capital ows Suppose two countries. () and () () : high saving rate, low interest rate in autarky. Think of China (): low saving rate, high interest rate in autarky. Think of the US Exogenous for now but endogenous in models due to market imperfections. What happens if nancial integration? - convergence of interest rate (fall in US interest rates) - capital ows from (() = China) to (() = US)

8 R Metzler Diagram

9 Theories of global imbalances a quick summary Theories (mostly) generating low autarky interest rates in emerging markets (high in the US) due to: - Di erences in ability to capitalize output between emerging Asia and developed countries Caballero, Fahri and Gourinchas (2008) - Financial frictions at the corporate level Benhima and Bachetta (2011), Sandri (2010), Buera and Shin (2011), Song et al. (2011)

10 Theories of global imbalances a quick summary - Di erences in household credit constraints/development of credit markets Coeurdacier, Guibaud and Jin (2015) - Precautionary savings and uninsurable idiosyncratic risk in emerging markets Ranciere and Jeanne (2006), Jeanne and Caroll (2009) - Precautionary savings and di erences in nancial markets development ( - nancial markets incompleteness): Mendoza, Quadrini and Rios-Rull (2009)

11 1. Financial Frictions and Capital Flows

12 Caballero, Fahri and Gourinchas (2008) Motivation: 3 stylized facts - global imbalances: current account de cits in the US (and UK) nanced by surpluses in Asia and Europe - fall in world interest rates - rising share of US nancial assets

13 Current accounts by region

14 World interest rates

15 Share of US assets (% of world nancial assets or world output)

16 Caballero, Fahri and Gourinchas (2008) Main contribution Provide a tractable framework to analyze equilibrium in global nancial markets and the impact of regional macroeconomic shocks. Shed some light on the mechanisms that could be generating these important facts as an equilibrium

17 Basic elements of the model (closed economy) Continuous time OLG; birth rate = death rate = One good. output = X t grows at rate g Consume when dies: C = W (wealth = W ) One tree / no physical investment Capitalizable and non-capitalizable output: = share of output capitalized in nancial markets X t = X t + (1 )X t

18

19 Autarky interest rate r aut = + g - Countries with higher growth rates will have higher interest rates in autarky - Countries with higher supply of nancial assets will have higher interest rates in autarky Main novelty: asset supplies matter for interest rates.

20 Key mechanism if capital market integration - Europe has a low g, if open up to the US, capital out ows towards the US (high interest rate country). - Asia has a low, if open up to the US, capital out ows towards the US (high interest rate country). Thus if g in Asia not too high compared to the US. - US interest rates falls. Similar mechanism if fall in g in Europe (and markets integrated) and fall in in Asia (Asian crisis).

21 Small open economy Take world interest rate r as given, g < r < g +

22 Metzler diagram

23 Europe (E)-US (U) World Same but fall in growth g E in Europe r t = h x U t g + 1 x U t g E i +

24 - Fall in world interest rates: r t = h x U t g + 1 x U t g E i + = r U aut 1 x U t g E - In the short run V U V E raises. So does relative wealth W U W E (due to non-capitalized income). - Consumption raises in the US and falls in Europe. - Europe runs a current account surplus

25 A fall in ge

26 Asia (R)-US (U) World Fall in R in Asia (Asian crisis) with g R = g: R = R r t = g + x U t + 1 x U t R such that r R aut r t r U aut

27 A fall in R when g R = g

28 - Again fall in world interest rates: r t = r U aut 1 x U t R - Again V U V R raises as share of capitalized asset is falling in Asia. - Asia runs a surplus [supply of asset scarce in Asia and buy abundant US assets] Note: if g R > g, demand for assets can rise faster than supply. This implies that r t continues to fall, further expanding the asymptotic current account de cit in the US.

29 Three country world: Asia (R) Europe (E) US (U) Combination of both shocks. Consistent with the 3 broad facts presented above. Note if the crash in R takes place in a world where g E < g, then the asset appreciation is much larger in U than E. Capital ows mostly from R to U

30 A fall in ge followed by a fall in R

31 Some caveats Very stylized model. Not aiming to be quantitative. Asian surpluses did continue more than 10 years after the Asian crisis increasing in magnitude. Drop in supply of assets potentially important but data suggest that demand for assets (saving) are even more crucial.

32 Coeurdacier, Guibaud and Jin (2015) Two of the most striking trends in the past three decades. Financial integration of large emerging markets Fast growth in Emerging Asia

33 Motivation Macro facts 1. A fall in the world long-term interest rate. 2. An increase in private/household savings rate in Asia and a fall in private/household savings rate in Advanced OECD 3. Global imbalances and large current account surplus in Asia

34 Motivation On the top of these macro facts: current account mostly driven by savings (and particularly private savings) especially in the cross-section. Dispersion in savings accounts for 75% of the cross-section dispersion of current accounts (of which more than 50% on average due to private savings). Similar to ndings in Gourinchas and Jeanne (2013) = saving wedge accounts for allocation puzzle. Need a model where dispersion in savings rate across countries is crucial.

35 Motivation (New) Micro facts: US vs. China 1. Younger households decreased signi cantly their savings in the US, much less so in China. 2. Savings of middle-aged increased in both countries, but more signi cantly so in China. 3. Divergence of savings for retirees, but less of a contribution to GDP.

36 Private Saving Rates

37 Household Saving Rates

38

39 Theory Life cycle patterns matter for understanding household savings across countries. Incorporates household credit constraints (the extent of which is asymmetric across countries) into an open economy, general equilibrium OLG model. Analyzes the interaction between growth and credit constraints and its impact on the global equilibrium. Can match macro and micro level evidence.

40 Intuition for the results Country with tight constraints: high saving rate, low autarky rate. Opposite if loose constraints. Fall in world interest rate driven by nancial integration and fast growth in low autarky rate country, e.g. Asia Asymmetric response of saving rates to a fall in world interest rate leads to greater dispersion in saving rates. Asymmetric credit constraints translate into di erent weights placed on borrowers vs savers across economies.

41 Intuition for the results A fall in world interest rate causes the young to borrow more and the middle-aged to save more (income e ect). Di erent weights on borrowers vs savers lead to asymmetric responses of saving rates across countries. Micro evidence on savings across age groups for US and China broadly supportive of quantitative model predictions.

42 The Basic Model One-good model of n large open economies. No uncertainty. OLG structure with three-period lived agents = young borrowers, middle-aged savers, old retired. Borrowing constraints: the young can only borrow up to a fraction of their discounted future labor income. Asymmetry: tighter credit constraints in Asia

43 Production Output in country i Y i t = K i t h A i t e i t L i y;t + L i m;ti 1 ; e i t < 1: Wages and rental rates of capital w i m;t = (1 )A i t k i t w i y;t = e i tw i m;t; r i K;t = k i t 1 ; R i t = 1 + r i K;t : with capital-e ective-labor ratio k i t Ki t =fai t (ei t Li y;t + Li m;t )g and depreciation rate.

44 Households Lifetime utility of an agent born in period t in country i U i t = u(ci y;t ) + u(ci m;t+1 ) + 2 u(c i o;t+2 ): Isoelastic utility with i.e.s coe cient 1 u(c) = c1 1 1 : 1 1

45 Households budget constraints An agent born in period t faces the following sequence of budget constraints: c i y;t + a i y;t+1 = wi y;t; c i m;t+1 + ai m;t+2 = wi m;t+1 + Ri t+1 ai y;t+1 ; c i o;t+2 = Ri t+2 ai m;t+2 : The old decumulate all their assets (no bequests). Bequest motive later in quantitative exercise.

46 Credit constraints Young agents can only borrow up to a fraction i of the present value of their future labor income a i y;t+1 (lower! tighter credit conditions) iwi m;t+1 Rt+1 i : Constraint is binding if life income pro le is steep enough. We restrict our attention to parameter values for which the constraint is always binding in equilibrium.

47 Households asset holdings Binding credit constraints on the young imply: a i y;t+1 = iwi m;t+1 R i t+1 (< 0): FOC for the middle-aged gives: a i m;t+1 = (R i t+1 )1 (1 i )w i m;t:

48 Autarky equilibrium Capital market equilibrium: K i t+1 = Li y;t ai y;t+1 + Li m;t ai m;t+1 :,! di erence equation driving the dynamics of k i t. Autarky rate of return in steady-state (for = = 1) R i = (1 + g A )(1 + g L ) 1 + [1 + e(1 + g L )] + i (1 ) (1 ) 1 i : dr i di > 0, i.e., tighter constraints imply lower interest rate.

49 Integrated equilibrium Financial integration in period t implies: R i t+1 = R t+1; for all i: Steady state: g i A = g A, g i L = g L, e i = e. Let i A i;t(el i y;t +Li m;t ) Pj A j;t(el j y;t +Lj m;t ). P World steady state interest rate using i i (for = = 1): R = (1 + g A )(1 + g L ) 1 + i [1 + e(1 + g L )] + (1 ) (1 ) 1 : R falls as more constrained economies become larger.

50 Dynamics of k t : autarky versus integration

51 Net aggregate savings in steady-state S i Y i = g(1 ) i 1 + e(1 + g L ) R + where g (1 + g A )(1 + g L ) 1. g 1 1 i 1 + g 1 + e(1 + g L ) 1 + R 1 ; Under integration, saving rates di er across countries in the long run: saving rate higher in more constrained countries. Suppose we start from an integrated steady state and after an episode of high growth in the more constrained countries, the world reaches a new steady state. Lower! fall in R. Saving rates respond di erently across 2 (S=Y > 0! fall in R leads to more dispersion in saving rates

52 Life cycle savings empirical evidence using micro data Our model has implications for the evolution of saving rates by age groups. 1. the saving rate as function of age, in level and in change, has an inverted-u shape in both Advanced Economies and Emerging Asia; 2. the fall in the saving rate of the young dominates in Advanced Economies. The rise in the saving rate of the middle-aged dominates in Emerging Asia. We look at cohort-level data for China and the US to see if these predictions hold.

53 Life cycle savings empirical evidence Use CEX for US ( ) and UHS for China ( ). Micro surveys on consumption. Need to correct for biases: - under reporting bias in CEX - aggregation and selection bias in UHS due to multigenerational households

54

55 Evidence for China Common practice: examine savings at the household level. average saving rate of households with head of age x = average saving rate of individuals of age x: Two issues: Aggregation bias: multi-generational households. Selection bias: household heads might not be random.

56 Evidence for China

57

58

59 Evidence for China Main issue: we have data on individual income but not consumption (only household) Two alternative approaches to correct for biases. Method 1: keep only uni-generational households and reweight sample for selection according to observables Method 2: method disaggregation method, following Chesher (1997) = Projection

60

61 Quantitative model: Extended setup with J+1 periods and bequests Utility of agent born at date t in country i: U i t = JX j=0 () j u(c i j;z;t+j ) + ()J u(r i t+j+1 bi t+j ) Bequest motive captured by as in Abel (2001). Output Y i t = (Ki t ) h A i t P Jj=0 e i j;t Li t ji 1 : Wages wj;t i = ei j;t (1 )Ai t (ki t ) ; kt i Kt i A i P Jj=0 t e i : j;t Li t j Gross rate of return between t and t + 1 : R i t = 1 + (ki t ) 1 :

62 Budget constraints c i j;t+j + ai j;t+j = R i t+j ai j 1;t+j 1 + wi j;t + qi j;t ; j < J; c i J;t+J + bi t+j = R i t+j ai J 1;t+J 1 + wi J;t+J : Credit constraints a i j;t+j ihi j+1;t+j+1 Rt+j+1 i ; H i j;t wi j;t + P J j =1 w i j+;t+ Q s=1 R i t+s

63 Quantitative model: Simulations Financial integration in Demographic evolution to match population structure in China and the US. Productivity growth to match relative GDPs. Initial capital-labor ratios to match data in 1988 (Hall and Jones (1990)) Evolution of the relative e ciency of di erent age groups to match income pro le by age in China and the US. Credit constraints, bequests and preference parameters chosen to match cohort-level saving rates/aggregate savings in 1988.

64

65

66

67 Some caveats Not all countries with underdeveloped nancial markets are lenders. Deterministic environment. Perfect nancial markets for rms while imperfect for households.

68 Possible extensions Mortality and life expectancy. Endogenous fertility. Productivity growth biased towards some age-groups. Social security and welfare state. Risk and Precautionary Savings.

69 Financial frictions on the rms side an introduction Two important modi cations from standard models - Borrowing constraints for entrepreneurs: can only pledge a fraction of their output (or capital) to borrow - Heterogeneous rms in terms of productivity. Capital misallocation within countries (Hsieh and Klenow (2009)). Song, Storesletten and Zilibotti (2011), Martin and Ventura (2012), Buera and Shin (2017), Matsuyama (2011), Bachetta and Benhima (2015)

70 Financial frictions on the rms side an introduction Key insights High and low productivity entrepreneurs might coexist. Financial frictions prevent most productive entrepreneurs to operate at e cient scale. Frictions prevent equalization of MPK within a country. In autarky, rate of interest can be high or low depending on distribution of rms and level of nancial frictions.

71 Financial frictions on the rms side an introduction Key insights If large amount of capital in the hand of low productivity entrepreneurs, average returns are low. In this case, nancial integration provides higher returns to unproductive entrepreneurs. Save abroad and free up resources (labour) for the most productive entrepreneurs. Average productivity and returns go up and capital out ows. The opposite if average returns to capital are high under autarky (more developed nancial markets where low productive entrepreneurs are out of business). Capital in ows and fall in returns.

72 Financial frictions on the rms side an introduction Key insights If nancial reforms under integration (= relaxation of borrowing constraints in a country), more productive entrepreneurs borrow more and investment in more productive rms increases. Tends to generate capital in ows. But world interest rate goes up and less productive entrepreneurs are downscaling. Free up resources (capital and labour) and net saving increases. Leads to capital out ows. Which one of the two forces dominate depend on the level of the nancial friction and the distribution of productivities. Application: Song et al. (2011), SOE versus private rms in China.

73 2. Demographics, Capital Flows and the World Interest Rate

74 Motivation Aging World. Low interest rates. Highly persistent net capital ows. Renewed interest for the importance of demographics to understand international allocation of capital and the world rate of interest. Capital ows: Domeij and Floden (2006), Ferrero (2010), Backus, Cooley, and Henriksen (2014), Barany, Coeurdacier and Guibaud (2017) Secular stagnation and low interest rates: Eggertsson, Mehrotra and Summers (2016), Carvalho et al. (2016), Lisack et al. (2017)

75 Aging across the world The world is aging. Drop in fertility. Fall in mortality rates; rising life expectancy. Convergence of demographic patterns across countries. Emerging countries have become more similar to developed countries.

76 Aging across the world (cont d) The world is aging. But there are variations in aging across countries. Across emerging countries: speed and timing of convergence varies across countries. Across developed countries: Old Europe and Japan are aging faster than others.

77

78

79 Barany, Coeurdacier and Guibaud (2017) Investigates the impact on international capital ows of Global aging Country-speci c aging patterns. Develops a multi-country lifecycle model of savings, which incorporates Common and country-speci c demographic trends Cross-country heterogeneity in access to credit and social security.

80 Theoretical results Global aging and capital ows. Global aging depresses the world interest rate. Generates capital ows due to the di erent response of savings across countries. Can trigger uphill capital ows.

81 Theoretical results Country-speci c aging and capital ows. Countries aging faster than the rest of the world are more likely to export capital. Capitals ows downhill towards countries aging slowly. The impact of country-speci c aging tends to be stronger for less developed countries.

82 Theory Agents live for at most three periods (y; m; o). - Young: do not work, face credit constraints. - Middle aged: work, contribute to social security and save for retirement. - Old: consume out of accumulated assets and social security bene ts, no bequest. Demographics - Life expectancy: a middle-aged individual in period t reaches retirement with probability p t. - Fertility: L y;t = n t L y;t 1.

83 Production Output Y t = K t At L m;t 1 ; where productivity evolves as Wage rate A t+1 = (1 + A;t+1 )A t : w t = (1 )A t k t ; with k t K t =(A t L m;t ) capital-e ective-labor ratio Rate of return between periods t 1 and t (full depreciation). R t = k 1 t :

84 Social security Pay-as-you-go system. Contribution rate t ; replacement rate t. Balanced budget condition L m;t t w t = L o;t t w t 1 ) t = p t 1 w t 1 t : n t 1 w t

85 Household s decisions An agent born in a given country in period t maximizes u(c y;t ) + u(c m;t+1 ) + 2 p t+1 u(c o;t+2 ); where u(c) = c1 1=!,! 1, subject to budget constraints 1 1=! and credit constraint c y;t + a y;t = 0; c m;t+1 + a m;t+1 = (1 t+1 ) w t+1 + R t+1 a y;t ; c o;t+2 = R t+2a m;t+1 p t+1 + t+2 w t+1 : a y;t t w t+1 R t+1 :

86 Savings decisions Assume that the credit constraint is always binding a y;t = t w t+1 R t+1 : The optimal savings of the middle aged is then a m;t = p t(1 t t 1 ) p t +! R 1! t+1 p t! Rt+1 1! w t+1 w t t p t +! Rt+1 1! : R t+1 Credit constraints and social security a ect individual savings.

87 Net asset demand over GDP L y;t a y;t + L m;t a m;t K t+1 = (1 ) p t(1 t t 1 ) Y t p t +! Rt+1 1! (1 ) p t! Rt+1 1! t+1 p t +! R 1! Impact of aging, n t # and/or p t " t t+1 R t+1 nt (1 + A;t+1 ) k t R t+1! 1 1 : - Direct e ects: increases net asset demand. Indirect e ect via social security adjustment. - Aging leads to a greater increase in net asset demand in less developed countries (low, low ).

88 Net asset demand over GDP L y;t a y;t + L m;t a m;t K t+1 = (1 ) p t(1 t t 1 ) Y t p t +! Rt+1 1! (1 ) p t! Rt+1 1! t+1 p t +! R 1! t t+1 R t+1 nt (1 + A;t+1 ) k t R t+1! 1 1 : Impact of a drop in R t+1. - Overall impact is ambiguous. - A drop in R t+1 leads to a larger fall in net asset demand in more developed countries (high, high ).

89 Capital market equilibrium Under nancial autarky K t+1 = L y;t a y;t + L m;t a m;t : Under nancial integration X L i y;t a i y;t + L i m;ta i m;t : i K i t+1 = X i Financial integration in period t implies R i t+1 = R t+1 for all i:

90 Steady-state autarky interest rate Steady-state interest rate under nancial autarky satis es R = n(1 + A) p(1 ) p 1 + +! R 1! With log utility (! = 1) R = n(1 + A) p(1 ) (1 + p) : Autarky interest rate increasing in and. Aging depresses the interest rate: n # and/or p " ) R #. Even more so if! low and/or low.

91 World interest rate at integrated steady state Symmetric aging (n i t = n, pi t = p) and productivity growth (i A;t = A) De ne = P i i i and = P i i i, where i = A i;tl i m;t P j A j;tl j. m;t The steady-state world interest rate satis es R = n(1 + A) p(1 ) p 1 + +! R 1! The world interest rate is increasing in and. Global aging depresses the world interest rate. Even more so if! low and/or low.

92 Global Aging and Capital ows Net foreign asset position of country i In the integrated steady state NF A i t Y i t NF A i t = Li y;t ai y;t + Li m;ta i m;t K i t+1 : =(1 ) + " p p +! R 1! + n(1 + A) R p(1 ) p +! R 1! " 1 + n(1 + A) p! R! # # ( i ) ( i ): Countries with tighter credit constraints than the world average ( i < ) and/or with lower social security than the world average ( i < ) tend to export capital.

93 Global aging and capital ows Steady-state comparative statics NF A L t Y L t NF A H t Y H t = ( H L )(1 ) + ( H L p(1 ) ) p +! R 1! " p p +! R 1! + n(1 + A) R " 1 + n(1 + A) p! R! # : # Low vs high- countries: A global fall in fertility triggers larger dispersion of NFA positions if the drop in R is large enough. A rise in longevity always triggers a larger dispersion of NFAs. Low vs high- countries: Global aging triggers larger dispersion of NFAs if the drop in R is large enough, or if social security systems adjust mostly through higher contribution rates (rather than lower replacement rates).

94 Global aging and capital ows. Heterogenous credit constraints

95 Global aging and capital ows. Heterogenous social security

96 Country-speci c aging and capital ows Illustration for an emerging small open economy: soe < row, soe < row

97 Country-speci c aging and capital ows (cont d) Experiment with slow demographic convergence

98 Quantitative model Agents live for at most J + 1 periods. Age j = 0; : : : ; J Conditional survival probability p i j;t Lifetime utility of agent born in period t in country i U i t = 0 1 JX jy p ì ;t+` j=0 `=0 with isoelastic preferences u(c) = c1 1! 1 1 1! A j u(c i j;t+j );

99 Production Output Y i t = (K i t) 2 4A i t 3 X J 1 e i j;t Li 5 j;t j=0 : Wages w i j;t = ei j;t (1 )Ai t(k i t) ; k i t A i t Kt i PJ j=0 ei j;t Li j;t : Gross rate of return between t and t + 1 R i t+1 = 1 + (ki t+1 ) 1 :

100 Credit constraints with a i j;t H i j;t wi j;t + J Xj =1 p i i j;t Hi j+1;t+1 t Rt+1 i Q 1 s=0 j+s;t+s pi w i j+;t+ Q s=1 Rt+s i : ; Social security Contribution and replacement rates must satisfy the balanced budget condition i t J X j=0 L i j;t wi j;t = i t JX j=j+1 L i j;t wi J;t+J j :

101 Unintentional bequests Unintentional bequests left at the end of period t J Q i 1 t X (1 p i j;t )Li j;t ai j;t : j=0 Redistributed to surviving agents as lump sum transfers.

102 Baseline calibration Sample of countries Focus on largest countries: 100 most populated in Drop main oil/gas producers and countries su ering very deadly con icts on their soil. If Oil&Gas Net Exports > 15% of GDP. If Death/Capita in a con ict > 1=1000. Drop countries missing necessary inputs for simulations. Sample of 70 countries. 91% of world GDP and 85% of world population.

103 Baseline calibration Parameters common across countries Adult life lasts for at most 16 periods (one period = 5 years). Retirement at age 65. Standard preferences/technology parameters. = 0:96 (annual),! = 1=2, = 0:3, = 0:075 (annual). Hump shape of age-income pro les (e i j;t ) calibrated on the US. Constant world TFP growth: A = 1%. Isolate the role of demographics. Integration date in Focus on

104 Baseline calibration Country-speci c parameters Demographics (United Nations World Population Prospects ( )). Fertility calibrated to match the size of the youngest age-group at each date. Survival probabilities from life-tables at each date. Country size Population and TFP level A i in each country to match countries GDPs in 2000.

105 Baseline calibration Country-speci c parameters Credit constraints parameters f i g. Calibrate i = US to match household debt/gdp in each country relative to the US in Household debt data for about 50 countries ( ). Predict household debt/gdp for remaining countries using data on private credit/gdp and nancial inclusion in 2011 (% of above 15 with a mortgage, % of above 15 having borrowed from a nancial institution). Calibrate US to match US aggregate savings/gdp in 2005 ) US = 9%.

106 Baseline calibration Country-speci c parameters E ective replacement rates f i g: i = i ~ i. Coverage ratio f i g in Data on social security coverage from ILO for a large sample of countries. Measures % of contributors to social security. Predict coverage for (few) remaining countries using demographics and GDP per capita. O cial replacement rate f~ i g in Hand-collected data from Social Security Administration. - Schedule of o cial replacement rates by years of contribution. Proxy number of years of contribution to back-out f~ i g.

107 Model versus Data Empirical strategy Regress capital ows data on model predictions. ca d i;t = cam i;t + t + i;t Unbalanced panel of 70 countries over country/period observations. ca d i;t = Current Account/GDP (average over 5y period). t = period xed e ect.

108 Model versus Data Empirical strategy Equivalently: demean Current Account/GDP by period. ca ^ d i;t = cam ^ i;t + i;t Pooled data. Std. errors clustered at the country level. Cross-section by period. Between & Within. Cross-section of changes over

109 Model versus Data Baseline results

110 Model versus Data Baseline results with a low eis

111 Model versus Data Between scatter plot

112 The world interest rate Fall relative to 1980, in %.

113 Model versus Data Selected countries/regions over time

114 Model versus Data Selected countries/regions over time

115 Inspecting the mechanisms Between regressions

116 Shutting down di erences in and

117 Common world demographics and uphill capital ows

118 Inspecting the mechanisms Uphill and downhill ows Cross-country di erences in demographics are crucial to generate downhill capital ows. Towards younger countries or countries not expected to age much in the near future. Credit constraints and di erences in social security are crucial to generate the right amount of downhill ows. Di erences in credit constraints/social security are essential to generate some uphill ows. Also important to account for di erences in capital ows within the group of emerging or developed countries.

119 Conclusion and direction for further research Government debt and the supply of assets. Implications for risky asset prices and gross positions Productivity growth versus aging. A solution to the allocation puzzle?