Demographic Trends and the Real InterestRate

Noemie Lisack, Rana Sajedi, and Gregory Thwaites

Discussion by Sebnem Kalemli-Ozcan

1 / 20

What does the paper do?

Quantifies the role of demographic change (aging) on thedecline in global (advanced country) real interest rates

Investigate the effect of aging on house prices, household debt

Main finding: demographics can explain a large part of thedecline in real rates and rise in house prices and householddebt since 1980s.

Projection: trends will persist given the slow movingpersistence process of demographics; global imbalances willget worse

2 / 20

General Impression

I like the paper.

I am very sympathetic to taking demographics seriously as oneof the key reasons behind the decline in real rates.

I am going to highlight some modeling and data issues which,upon dealing with, will help to clarify and strengthen thepaper.

3 / 20

POINT 1: Savings Rate or Wealth Accumulation?

I believe the authors want wealth but they have life cyclesavings

What are the issues with a savings focus?

4 / 20

Shifts in Demand and Supply for Funds

Figure 3.5. Real Interest Rate and Shifts in Demand forand Supply of Funds

Source: IMF staff illustration.

Realrate

(percent)

Funds(U.S. real dollars, bond market)

Supply

Supply'

Demand

Demand'

Source: 2014, IMF WEO

5 / 20

General framework for the global decline in real rates

Investment Decline: The decline in relative price ofinvestment—matches the timing of decline starting in 1980s

Saving Increase: Savings of China—post 2000 period

S = Spriv + Spub Low public saving/high public debt depressprivate saving (OLG with no RE)–role of FP

No role for aging since aging decreases saving in standardPIH/life cycle model so real rates rise

Monetary Policy Easing: Important role since 1980s both forshort and long term rates

Portfolio Shifts: Important role since 2000s given demand forsafe US assets

6 / 20

Investment

18

20

22

24

26

28

30

32

34

1980 85 90 95 2000 05 10 13

Figure 3.6. Investment-to-GDP Ratios(Percent of GDP)

Global nominal investment (saving)-to-GDP ratioAdvanced economy nominal investment-to-GDP ratioEmerging market economy nominal investment-to-GDP ratio

Sources: Haver Analytics; Organization for Economic Cooperation and Development; and IMF staff calculations.

Source: 2014, IMF WEO

7 / 20

Savings

0

5

10

15

20

25

30

35

40

1980 83 86 89 92 95 98 2001 04 07 10 13

10

15

20

25

30

35

40

1980 85 90 95 2000 05 10 13

Figure 3.8. Saving Shifts in Emerging Markets

Advanced economiesEMEs

1. Nominal Saving-to-GDP Ratios(percent of GDP)

2. Saving in Total GDP for Emerging Markets(1980–2013, percent)

EMEsChinaOil exportersOther EMEs

20

25

30

35

40

2001 03 05 07 09 11 1330

35

40

45

50

55

60

2001 03 05 07 09 11 13

Actual Predicted Counterfactual

3. Emerging Markets 4. China

Contribution of Higher Growth to Increased Saving(percent of GDP, 2001–13)

Sources: Organization for Economic Cooperation and Development; World Bank, World Development Indicators database; and IMF staff calculations.Note: EMEs = emerging market economies; Actual = actual saving-to-GDP ratio; Predicted = predicted saving-to-GDP ratio obtained by regressing the EME saving rate on its lagged value and EME real GDP growth; Counterfactual = conditional forecast of the saving rate assuming real GDP growth is constant at the average value of the late 1990s.

Source: 2014, IMF WEO

8 / 20

Savings: Role of Public Savings and AsiaAlfaro et al., 2014 JEEA

Sudan

Senegal

Kenya

Mali

Benin

Ethiopia

Madagascar

Niger

Mozambique

Chad

Mauritius

South Africa

Uganda

Tanzania

Guinea

GhanaRwanda

Malawi

Congo, Rep.

Zambia

Burkina FasoCameroon

Togo

Cote d’Ivoire

Sri Lanka

Philippines

Thailand

Papua New Guinea

Bangladesh

India

Malaysia

Pakistan

Nepal

Indonesia

China

Albania

Poland

Armenia

Romania

Turkey

Bulgaria

Ukraine

LithuaniaBelarus

Latvia

Kyrgyz Republic

Mexico

Honduras

Colombia

Paraguay

Argentina

Costa Rica

Venezuela, RB

Bolivia

Uruguay Chile

El SalvadorBrazil

Ecuador

JamaicaPanama

Peru

Dominican Republic

Guatemala

Egypt, Arab Rep.

Jordan

Yemen, Rep.Morocco

Iran, Islamic Rep.Tunisia

Sri Lanka

Philippines

Thailand

Papua New Guinea

Bangladesh

India

Malaysia

Pakistan

Nepal

Indonesia

China

Sudan

Senegal

Kenya

Mali

Benin

Ethiopia

Madagascar

Niger

Mozambique

Chad

Mauritius

South Africa

Uganda

Tanzania

Guinea

GhanaRwanda

Malawi

Congo, Rep.

Zambia

Burkina FasoCameroon

Togo

Cote d’Ivoire

−10

−5

05

10A

vera

ge G

over

nmen

t Sav

ings

/GD

P (%

), 1

990−

2004

−2 0 2 4 6 8Average per capita GDP Growth (%), 1990−2004

Legend:Red dash line − Developing Countries excluding ChinaRed letters − Developing Countries in AsiaGreen letters − Developing Countries in Africa

9 / 20

Savings: No Role for EM Private SavingAlfaro et al. 2014 JEEA

Sudan

Senegal

Kenya

Mali

Benin

Ethiopia

Madagascar

Niger

Mozambique

Chad

Mauritius

South Africa

Uganda

TanzaniaGuinea

GhanaRwanda

Malawi

Congo, Rep.

Zambia

Burkina FasoCameroon

TogoCote d’Ivoire

Sri LankaPhilippines

Thailand

Papua New Guinea

Bangladesh

IndiaMalaysiaPakistan

Nepal

Indonesia

China

Albania

PolandArmenia

Romania

Turkey

Bulgaria

Ukraine

Lithuania

Belarus

Latvia

Kyrgyz Republic

MexicoHonduras

Colombia

Paraguay

ArgentinaCosta Rica

Venezuela, RB

Bolivia

Uruguay

ChileEl Salvador

Brazil

EcuadorJamaica

PanamaPeru

Dominican RepublicGuatemala

Egypt, Arab Rep.Jordan

Yemen, Rep.MoroccoIran, Islamic Rep.

Tunisia

Sri LankaPhilippines

Thailand

Papua New Guinea

Bangladesh

IndiaMalaysiaPakistan

Nepal

Indonesia

China

Sudan

Senegal

Kenya

Mali

Benin

Ethiopia

Madagascar

Niger

Mozambique

Chad

Mauritius

South Africa

Uganda

TanzaniaGuinea

GhanaRwanda

Malawi

Congo, Rep.

Zambia

Burkina FasoCameroon

TogoCote d’Ivoire

010

2030

40A

vera

ge P

riva

te S

avin

gs/G

DP

(%),

199

0−20

04

−2 0 2 4 6 8Average per capita GDP Growth (%), 1990−2004

Legend:Red dash line − Developing Countries excluding ChinaRed letters − Developing Countries in AsiaGreen letters − Developing Countries in Africa

10 / 20

Public Savings in Asia and Demand for Safe Assets

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0

5

10

15

20

1990 96 2002 08 14

Figure 3.12. Portfolio Shifts and Relative Demand for Bonds versus Equity

Sources: Beltran and others (2013); and IMF staff calculations.Note: EMEs = emerging market economies.

Change in foreign exchange reserves (left scale)Gross saving(right scale)

1. Percent of Global GDP

0

1

2

3

4

5

6

1984 90 96 2002 08 11

ChinaOther EMEsTotal

2. Foreign Holdings of U.S.Government Securities(trillions of U.S. dollars)

0

1

2

3

4

5

6

1984 90 96 2002 08 11

OfficialTotal

3. Foreign Holdings of U.S.Government Securities(trillions of U.S. dollars)

0

1

2

3

4

5

1984 90 96 2002 08 11

Government securitiesPrivate securitiesTotal

4. Foreign Official Holdings ofU.S. Securities(trillions of U.S. dollars)

Source: 2014, IMF WEO

11 / 20

K/Y Ratio rather than Savings

In a life cycle model, easy to get a rise in K/Y ratio withdemographics and get an associated decline in real rates

Can we add to the model to make wealth accumulationcentral given increases in longevity?

12 / 20

Wealth accumulation in a model of retirement

Life cycle savings is not a good model for wealthaccumulation

Better with retirees and uncertain survival (Carvalho, Ferrero,Nechio, 2016))

Here an OLG model with fixed retirement age, so if agingcomes from longevity increase then it is mechanical to get arise in K/Y

The literature shows that a bigger driver of aging and wealthaccumulation is reduced fertility. (See Weil, in Handbook ofPopulation and Family Econ)

The model should have endogenous fertility and retirement

Then the role of pensions, annuities become importanttogether with the role of baby-boomers: not every country haspay-as-you-go systems

Abel and Blanchard (1983); Cutler et al. (1990); Auerbachand Kotlikoff (1987); Lim and Weil (2003)

13 / 20

Wealth accumulation in a model of retirement

Life cycle savings is not a good model for wealthaccumulation

Better with retirees and uncertain survival (Carvalho, Ferrero,Nechio, 2016))

Here an OLG model with fixed retirement age, so if agingcomes from longevity increase then it is mechanical to get arise in K/Y

The literature shows that a bigger driver of aging and wealthaccumulation is reduced fertility. (See Weil, in Handbook ofPopulation and Family Econ)

The model should have endogenous fertility and retirement

Then the role of pensions, annuities become importanttogether with the role of baby-boomers: not every country haspay-as-you-go systems

Abel and Blanchard (1983); Cutler et al. (1990); Auerbachand Kotlikoff (1987); Lim and Weil (2003)

13 / 20

Wealth accumulation in a model of retirement

Life cycle savings is not a good model for wealthaccumulation

Better with retirees and uncertain survival (Carvalho, Ferrero,Nechio, 2016))

Here an OLG model with fixed retirement age, so if agingcomes from longevity increase then it is mechanical to get arise in K/Y

The literature shows that a bigger driver of aging and wealthaccumulation is reduced fertility. (See Weil, in Handbook ofPopulation and Family Econ)

The model should have endogenous fertility and retirement

Then the role of pensions, annuities become importanttogether with the role of baby-boomers: not every country haspay-as-you-go systems

Abel and Blanchard (1983); Cutler et al. (1990); Auerbachand Kotlikoff (1987); Lim and Weil (2003)

13 / 20

Wealth accumulation in a model of retirement

Life cycle savings is not a good model for wealthaccumulation

Better with retirees and uncertain survival (Carvalho, Ferrero,Nechio, 2016))

Here an OLG model with fixed retirement age, so if agingcomes from longevity increase then it is mechanical to get arise in K/Y

The literature shows that a bigger driver of aging and wealthaccumulation is reduced fertility. (See Weil, in Handbook ofPopulation and Family Econ)

The model should have endogenous fertility and retirement

Then the role of pensions, annuities become importanttogether with the role of baby-boomers: not every country haspay-as-you-go systems

Abel and Blanchard (1983); Cutler et al. (1990); Auerbachand Kotlikoff (1987); Lim and Weil (2003)

13 / 20

Wealth accumulation in a model of retirement

Life cycle savings is not a good model for wealthaccumulation

Better with retirees and uncertain survival (Carvalho, Ferrero,Nechio, 2016))

Here an OLG model with fixed retirement age, so if agingcomes from longevity increase then it is mechanical to get arise in K/Y

The literature shows that a bigger driver of aging and wealthaccumulation is reduced fertility. (See Weil, in Handbook ofPopulation and Family Econ)

The model should have endogenous fertility and retirement

Then the role of pensions, annuities become importanttogether with the role of baby-boomers: not every country haspay-as-you-go systems

Abel and Blanchard (1983); Cutler et al. (1990); Auerbachand Kotlikoff (1987); Lim and Weil (2003)

13 / 20

POINT 2: Evaluating Model’s Success

Authors are upfront on being not the only explanation butclaim a large fraction of the decline explained.

14 / 20

Calibration

15 / 20

Calibration

Mismatch on timing; demographics is slow moving, reductionin interest rate more sudden–calibrate to 5 year intervals?

Or use US decennial censuses so initial point can be 1940, useonly demographics, no financials and see how muchnon-targeted rates can be explained by the model?

Nothing is NOT targeted it seems–distributions of targetedand non-targeted variables will help

Figure 5 and 6 are all targeted so how do we evaluate modelsuccess?

Open economy dimension is important, maybe calibrationshould focus on that extension, instead of closed economy

Recent data shows global imbalances are narrowing, the modelimplies they will get worse via demographics, important toquantify this effect on real rates.

16 / 20

Calibration

Mismatch on timing; demographics is slow moving, reductionin interest rate more sudden–calibrate to 5 year intervals?

Or use US decennial censuses so initial point can be 1940, useonly demographics, no financials and see how muchnon-targeted rates can be explained by the model?

Nothing is NOT targeted it seems–distributions of targetedand non-targeted variables will help

Figure 5 and 6 are all targeted so how do we evaluate modelsuccess?

Open economy dimension is important, maybe calibrationshould focus on that extension, instead of closed economy

Recent data shows global imbalances are narrowing, the modelimplies they will get worse via demographics, important toquantify this effect on real rates.

16 / 20

Calibration

Mismatch on timing; demographics is slow moving, reductionin interest rate more sudden–calibrate to 5 year intervals?

Or use US decennial censuses so initial point can be 1940, useonly demographics, no financials and see how muchnon-targeted rates can be explained by the model?

Nothing is NOT targeted it seems–distributions of targetedand non-targeted variables will help

Figure 5 and 6 are all targeted so how do we evaluate modelsuccess?

Open economy dimension is important, maybe calibrationshould focus on that extension, instead of closed economy

Recent data shows global imbalances are narrowing, the modelimplies they will get worse via demographics, important toquantify this effect on real rates.

16 / 20

Calibration

Mismatch on timing; demographics is slow moving, reductionin interest rate more sudden–calibrate to 5 year intervals?

Or use US decennial censuses so initial point can be 1940, useonly demographics, no financials and see how muchnon-targeted rates can be explained by the model?

Nothing is NOT targeted it seems–distributions of targetedand non-targeted variables will help

Figure 5 and 6 are all targeted so how do we evaluate modelsuccess?

Open economy dimension is important, maybe calibrationshould focus on that extension, instead of closed economy

Recent data shows global imbalances are narrowing, the modelimplies they will get worse via demographics, important toquantify this effect on real rates.

16 / 20

Calibration

Mismatch on timing; demographics is slow moving, reductionin interest rate more sudden–calibrate to 5 year intervals?

Or use US decennial censuses so initial point can be 1940, useonly demographics, no financials and see how muchnon-targeted rates can be explained by the model?

Nothing is NOT targeted it seems–distributions of targetedand non-targeted variables will help

Figure 5 and 6 are all targeted so how do we evaluate modelsuccess?

Open economy dimension is important, maybe calibrationshould focus on that extension, instead of closed economy

Recent data shows global imbalances are narrowing, the modelimplies they will get worse via demographics, important toquantify this effect on real rates.

16 / 20

POINT 3: Measurement—What are the real rates?

17 / 20

What are the real rates?

King and Low (2014) for real rates or Rachel and Smith(2015) for natural interest rate? (not equal unless monetarypolicy is neutral)

Directly observable real rates: yields on inflation-indexedbonds, only available for a handful of countries (countries hereare ok probably)

Approximate real rates: Difference between nominal rates andinflation expectations

At the end what we want is cost of capital

18 / 20

Real Rates and Cost of Capital

–8

–6

–4

–2

0

2

4

68

10

1970 75 80 85 90 95 2000 05 10 12

Figure 3.3. Real Interest Rates, Real Returns on Equity, andCost of Capital(Percent a year)

Three-month real rateTen-year real rateTerm spread

1. Short- and Long-Term Global Real Interest Rates

0

1

2

3

4

5

6

7

8

9

1973 78 83 88 93 98 2003 08 13

2. Expected Real Returns on Equity

United States United Kingdom

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.54.0

4.5

1991–2000 2001–07 2008–13

3. Global Real Interest Rates and Cost of Capital

Global real interest rateGlobal cost of capital

Sources: Bloomberg, L.P.; Haver Analytics; IMF, International Financial Statistics database; Organization for Economic Cooperation and Development; World Bank, World Development Indicators database; and IMF staff calculations.Note: Term spread is defined as the difference between short- and long-term real rates.

19 / 20

Conclusion

This is a great paper!

It made me really think about these issues, a must read forthose who work on these questions

For the next draft, it will help to make the point sharper ifauthors can clarify these points:

Endogenous retirement: fixed retirement age not veryreasonable when aging is due to longevity

Maybe adding endogenous fertility if the aim is to explain thedecline in real rates via wealth accumulation which goes viadeclining births (more important than housing)

Improve the calibration on non targeted moments

Clarify measurement issues

20 / 20

Conclusion

This is a great paper!

It made me really think about these issues, a must read forthose who work on these questions

For the next draft, it will help to make the point sharper ifauthors can clarify these points:

Endogenous retirement: fixed retirement age not veryreasonable when aging is due to longevity

Maybe adding endogenous fertility if the aim is to explain thedecline in real rates via wealth accumulation which goes viadeclining births (more important than housing)

Improve the calibration on non targeted moments

Clarify measurement issues

20 / 20

Conclusion

This is a great paper!

It made me really think about these issues, a must read forthose who work on these questions

For the next draft, it will help to make the point sharper ifauthors can clarify these points:

Endogenous retirement: fixed retirement age not veryreasonable when aging is due to longevity

Maybe adding endogenous fertility if the aim is to explain thedecline in real rates via wealth accumulation which goes viadeclining births (more important than housing)

Improve the calibration on non targeted moments

Clarify measurement issues

20 / 20