Indebted Demand

and Economic Policy in a Post-Covid World

Atif Mian, Princeton

Ludwig Straub, Harvard
Amir Sufi, Chicago Booth

Virtual Macro Seminar

April 2020

1
Rise in debt and decline in r∗ — especially relevant post-Covid!

Rising debt ... ... and falling rates

3 6

2.5
Debt to GDP ratio 4

Real interest rate

3
2
2

1
1.5
0

1 −1
1960 1970 1980 1990 2000 2010 2020 1980 1990 2000 2010 2020

• How did this happen? Do the two plots interact? What are the implications?

2
Rise in debt and decline in r∗ — especially relevant post-Covid!

• How did this happen? Do the two plots interact? What are the implications?

2
Rise in debt driven by households and government

2
household + gov debt

1.5

Ratio to GDP
1

corporate debt
.5

0
1960 1970 1980 1990 2000 2010 2020

3
The rich lend to the non-rich

Debt owed minus debt held as asset

Scaled by NI, relative to 1982

.4

.2

−.2
1960 1980 2000 2020
Top 1% Next 9% Bottom 90%

• “Saving glut of the rich and the rise in household debt”

4
Why might this matter? — Rich & wealthy save more

• Dynan Skinner Zeldes (2004): saving rates increase in current income

0.512
.5

saving rate by income percentile

.4 0.372

.3
0.236

.2 0.173

.1

0
60 to 80 top 20 top 5 top 1
From Dynan, et al, Table 3, column 2

5
Why might this matter? — Rich & wealthy save more

• Straub (2019): consumption has elasticity < 1 w.r.t. average income

log consumption residuals

1

−1

−2
−2 −1 0 1 2
average log income residuals
5
Why might this matter? — Rich & wealthy save more

• Fagereng Holm Moll (2019): saving rate across the wealth distribution

5
The indebted demand framework

• Introduce non-homothetic consumption-saving behavior into conventional

two-agent endowment economy
→ the rich have a higher saving rate

6
The indebted demand framework

• Introduce non-homothetic consumption-saving behavior into conventional

two-agent endowment economy
→ the rich have a higher saving rate

• Main insight: “Indebted demand”

≡ shifts & policies that stimulate demand today through debt creation, reduce
demand in the future by shifting resources from borrowers to savers

6
The indebted demand framework

• Introduce non-homothetic consumption-saving behavior into conventional

two-agent endowment economy
→ the rich have a higher saving rate

• Main insight: “Indebted demand”

≡ shifts & policies that stimulate demand today through debt creation, reduce
demand in the future by shifting resources from borrowers to savers

• Implications:
• rising inequality depresses r, amplified by rising debt levels
• monetary + fiscal policy have limited ammunition when they create debt
• economies can fall into a “debt trap” — liquidity trap driven by too much debt
• once in it, debt-financed stimulus deepens recession in the future
• redistributive policies help 6
At the center of our analysis is a simple diagram

interest rate
short-run supply
of savings

long-run supply
of savings
demand for debt
debt

7
Literature

1. Secular stagnation + theories: Summers (2013), Rachel Summers (2019), Eggertsson

Mehrotra Robbins (2019), Auclert Rognlie (2018), Caballero Farhi (2017), Straub (2019)
2. Non-homothetic preferences: Old idea (Böhm-Bawerk, Hobson, Fisher), old models
(Schlicht, Bourguignon). New: Uzawa (1968), Carroll (2000), Dynan Skinner Zeldes
(2004), De Nardi (2004), Straub (2019), Fagereng Holm Moll Natvik (2019), Benhabib
Bisin Luo (2019)
3. Inequality and debt (theory): Kumhof Ranciere Winant (2015), Cairo Sim (2018),
Illing Ono Schlegl (2018), Rannenberg (2019)
4. Inequality and debt (empirics): Cynamon Fazzari (2015), Mian Straub Sufi (2019)
5. Debt + demand: Dynan (2012), Mian Sufi (2015), Mian Sufi Verner (2017), Jorda
Schularick Taylor (2016), Bhutta and Keys (2016), Di Maggio et al (2017), Beraja Fuster
Hurst Vavra (2018), Di Maggio Kermani Palmer (2019), Cloyne Ferreira Surico (2019)
6. Deleveraging: Eggertsson Krugman (2012), Guerrieri Lorenzoni (2017) 8
Outline

1 Model

2 Equilibria & indebted demand

3 Inequality & financial liberalization

4 Fiscal & monetary policy

5 Debt trap

6 Indebted demand post-Covid

7 Extensions & conclusion

9
Model
Model of indebted demand

• Deterministic ∞-horizon endowment economy with real assets (“trees”)

• Populated by two separate dynasties

• Same preferences, but different endowments of trees

• mass 1 of borrowers i = b: endowment ω b
• mass 1 of savers i = s: endowment ω s > ω b
• total endowment ω b + ω s = 1

• Trees are nontradable, dynasties trade debt contracts

• Agents within a dynasty die at rate δ > 0, wealth inherited by offspring

10
Preferences

• Dynasty i consumes cit , owns wealth ait .

11
Preferences

• Dynasty i consumes cit , owns wealth ait . Preferences:

Z ∞  
δ
e−(ρ+δ)t log cit + · v(ait ) dt
0 ρ

• Budget constraint
cit + ȧit ≤ rt ait

11
Preferences

• Dynasty i consumes cit , owns wealth ait . Preferences:

Z ∞  
δ
e−(ρ+δ)t log cit + · v(ait ) dt
0 ρ

• Budget constraint
cit + ȧit ≤ rt ait

• v(a) = utility from bequest [future consumption, “status” benefits from wealth,
artwork, gifts (to relatives or charities), adjustment frictions in illiquid accounts]

• Key object: η(a) ≡ av0 (a) — marginal utility of v(a) relative to log
• homothetic model: η(a) = const ⇒ v(a) ∝ log a
• non-homothetic model: η(a) increases in a
11
Borrowing constraint & asset market

• Total wealth = real asset wealth net of debt

ait = ω i pt − dit

where pt = price of a Lucas tree: rt pt = 1 + ṗt

12
Borrowing constraint & asset market

• Total wealth = real asset wealth net of debt

ait = ω i pt − dit
where pt = price of a Lucas tree: rt pt = 1 + ṗt

• Agents can pledge ` trees each to borrow dit

dit ≤ pt `

12
Borrowing constraint & asset market

• Total wealth = real asset wealth net of debt

ait = ω i pt − dit
where pt = price of a Lucas tree: rt pt = 1 + ṗt

• Agents can pledge ` trees each to borrow dit (λ ≡ bond “decay rate”)
ḋit + λdit ≤ λpt `
| {z }
new debt issuance

12
Borrowing constraint & asset market

• Total wealth = real asset wealth net of debt

ait = ω i pt − dit
where pt = price of a Lucas tree: rt pt = 1 + ṗt

• Agents can pledge ` trees each to borrow dit (λ ≡ bond “decay rate”)
ḋit + λdit ≤ λpt `
| {z }
new debt issuance

• steady state: di ≤ p` [paper: generalize to ` = `({rs }s≥t )]

12
Borrowing constraint & asset market

• Total wealth = real asset wealth net of debt

ait = ω i pt − dit
where pt = price of a Lucas tree: rt pt = 1 + ṗt

• Agents can pledge ` trees each to borrow dit (λ ≡ bond “decay rate”)
ḋit + λdit ≤ λpt `
| {z }
new debt issuance

• steady state: di ≤ p` [paper: generalize to ` = `({rs }s≥t )]

• Market clearing dst + dbt = 0 pins down interest rate rt

• Focus on debt of borrowers: dt ≡ dbt (state variable) 12

Scale invariance

• Non-homothetic model is typically not scale invariant in aggregate

• economic growth ⇒ $28’000 today is like $200’000 around 1900
• so . . . someone with $28’000 today should save a ton?!

13
Scale invariance

• Non-homothetic model is typically not scale invariant in aggregate

• economic growth ⇒ $28’000 today is like $200’000 around 1900
• so . . . someone with $28’000 today should save a ton?!

• In reality, savings preferences probably closer to v(a/A) or v(a/Y)

• We work with v(a/Y), where so far Y = 1 (total endowment = 1)

13
Equilibria & indebted demand
Saving supply curves

• Savers’ Euler equation

ċst cs
s = rt − ρ − δ + δ t s · η(ast )
ct ρat

14
Saving supply curves

• Savers’ Euler equation

ċst cs
s = rt − ρ − δ + δ t s · η(ast )
ct ρat

• Setting ċ = 0 in Euler and use cs = ras ⇒

1 + ρ/δ
r =ρ·
1 + ρ/δ · η(as )

14
Saving supply curves

• Savers’ Euler equation

ċst cs
s = rt − ρ − δ + δ t s · η(ast )
ct ρat

• Setting ċ = 0 in Euler and use cs = ras ⇒

1 + ρ/δ
r =ρ·
1 + ρ/δ · η(as )

• This is a long-run saving supply curve:

• r necessary for which saver keeps wealth constant at as

• η(as ) determines the shape of the saving supply curve

14
Long-run saving supply curves

r
η(a) ↓ in a (saving is necessity)

η(a) = const (homothetic)

η(a) ↑ in a (saving is luxury)

15
Long-run saving supply curves

r
η(a) ↓ in a (saving is necessity)

η(a) = const (homothetic)

η(a) ↑ in a (saving is luxury)

• If η(as ) increasing: larger wealth as requires lower return on wealth r for

saver to be indifferent about saving! 15
Steady state equilibria

• Steady state: intersect long-run supply curve with debt demand curve
1 + ρ/δ `
r =ρ· d=
1 + ρ/δ · η(ω s /r + d) r

16
Steady state equilibria

• Steady state: intersect long-run supply curve with debt demand curve
1 + ρ/δ `
r =ρ· d=
1 + ρ/δ · η(ω s /r + d) r

supply
demand
d

16
Steady state equilibria

• Steady state: intersect long-run supply curve with debt demand curve
1 + ρ/δ `
r =ρ· d=
1 + ρ/δ · η(ω s /r + d) r

supply
demand
d

16
Indebted demand

• Start from a steady state & raise debt service costs by some dx

• What is response of aggregate spending? (partial equilibrium, r fixed)

17
Indebted demand

• Start from a steady state & raise debt service costs by some dx

• What is response of aggregate spending? (partial equilibrium, r fixed)

s   0 !
ρ + δ 1 r η (a)a
dC = dcs + dcb = − 1− 1−4 1− dx
r 2 ρ+δ η(a)

⇒ Thus increase in debt service costs weighs on aggregate demand

• dC < 0 if η 0 > 0

17
Indebted demand

• Start from a steady state & raise debt service costs by some dx

• What is response of aggregate spending? (partial equilibrium, r fixed)

s   0 !
ρ + δ 1 r η (a)a
dC = dcs + dcb = − 1− 1−4 1− dx
r 2 ρ+δ η(a)

⇒ Thus increase in debt service costs weighs on aggregate demand

• dC < 0 if η 0 > 0

• Call this phenomenon “indebted demand”

17
Equilibrium transitions

supply
I II demand
d

18
The indebted demand diagram

long-run supply
demand
d

• Saving supply curve = how low does r have to be given % resources

controlled by savers
• Debt demand = how much do borrowers want to borrow given r
19
The indebted demand diagram

r
short-run supply

long-run supply
demand
d

• Saving supply curve = how low does r have to be given % resources

controlled by savers
• Debt demand = how much do borrowers want to borrow given r
19
Inequality & financial liberalization
Rising inequality ω s ↑: lowers r and raises debt plot

Homothetic model
r
Old and new steady state

20
Rising inequality ω s ↑: lowers r and raises debt plot

Homothetic model Non-homothetic model

r r
Old and new steady state

Old steady state

New steady state

d d

20
Rising inequality ω s ↑: lowers r and raises debt plot

Homothetic model Non-homothetic model

r r
Old and new steady state

Old steady state

New steady state

d d

• Effects of rising inequality ω s ↑ in non-homothetic model:

1. inequality ↑ ⇒ more saving by the rich ⇒ r ↓ ⇒ debt ↑
2. debt ↑ first raises demand, pushing against decline in r
3. high debt eventually lowers demand, aggravating decline in r 20
Inequality and debt across 14 advanced economies

200 16

Household & government debt to GDP (%)

Top 1% income share (%)

14
150

12

100
10

50 8
1960 1970 1980 1990 2000 2010 2020

Household & government debt to GDP (%)

Top 1% income share (%)

21
Financial liberalization: raising pledgability `

Homothetic model
r

22
Financial liberalization: raising pledgability `

Homothetic model Non-homothetic model

r r

d d

22
Financial liberalization: raising pledgability `

Homothetic model Non-homothetic model

r r

d d

• Mechanism in non-homothetic model:

1. raises debt & demand, pushing r up (short-run saving supply slopes up)
2. ultimately high debt weighs on demand, lowering r, stimulating further debt!
→ resolves puzzle in literature [e.g. Justiniano Primiceri Tambalotti] 22
Fiscal & monetary policy
Fiscal policy implications

• Gov’t spends Gt , has debt Bt , raises income taxes τts , τtb , subject to

Gt + rt Bt ≤ Ḃt + τts ω s + τtb ω b

• Total demand for debt now dt + Bt

23
Fiscal policy implications

• Gov’t spends Gt , has debt Bt , raises income taxes τts , τtb , subject to

Gt + rt Bt ≤ Ḃt + τts ω s + τtb ω b

• Total demand for debt now dt + Bt

• Result: In the long run

1. larger gov’t debt B ↑: depresses interest rate r ↓, crowds in household debt d ↑
2. tax-financed spending G ↑: raises r ↑, crowds out d ↓
3. fiscal redistribution τ s ↑, τ b ↓: raises r ↑, crowds out d ↓

• With homothetic preferences none of these policies change r or d !

23
Deficit-financed fiscal policy plot r−g

24
Deficit-financed fiscal policy plot r−g

d
• Caveat: this assumed gov’t pays same interest rate r
• In many advanced economies, gov’t actually pays a lower rate
• e.g. when investors derive other benefits from their debt (safety, convenience)

24
Deficit-financed fiscal policy plot r−g

d
• Caveat: this assumed gov’t pays same interest rate r
• In many advanced economies, gov’t actually pays a lower rate
• e.g. when investors derive other benefits from their debt (safety, convenience)
• In that case, what matters is how those benefits affect savers’ investments
24
→ paper: natural case where things are unchanged
“Japanification” — how high public debt makes r less likely to rise

Imagine inequality falls exogenously. How much does the interest rate rise?

Low B High B

25
“Japanification” — how high public debt makes r less likely to rise

Imagine inequality falls exogenously. How much does the interest rate rise?

Low B High B
r
Strong recovery of r
with low gov’t debt

d+B

25
“Japanification” — how high public debt makes r less likely to rise

Imagine inequality falls exogenously. How much does the interest rate rise?

Low B High B
r r
Strong recovery of r
Little recovery of r
with low gov’t debt
with high gov’t debt

d+B d+B

25
“Japanification” — how high public debt makes r less likely to rise

Imagine inequality falls exogenously. How much does the interest rate rise?

Low B High B
r r
Strong recovery of r
Little recovery of r
with low gov’t debt
with high gov’t debt

d+B d+B

With higher B, any given increase in r weighs down more on aggregate demand
25
Monetary policy has limited ammunition when it raises debt

• Can extend our setup to include nominal rigidities (see paper)

• Monetary policy sets path of interest rates {rt }, output is endogenous

Main result:
r
monetary possible natural
intervention interest rate paths
r

more debt,
greater η
full ammunition

reduced
ammo.
t
T
26
Debt trap
Introducing the lower bound

• Consider lower bound r on interest rate r

• r > 0 if r is return on wealth (e.g. r ≈ 3.5% during recent US ZLB)

27
Introducing the lower bound

• Consider lower bound r on interest rate r

• r > 0 if r is return on wealth (e.g. r ≈ 3.5% during recent US ZLB)

• What happens if the steady state natural rate falls below r ?

d
27
The debt trap (≡ a debt-driven liquidity trap)

• Result: if natural rate < r, get stable liquidity trap steady state: “debt trap”

→ Output persistently below potential

   
r −1 ρ
Ŷ = Y · η (1 + ρ/δ) − ρ/δ − B <Y
(1 − τ s )ω s + ` r

• Liquidity trap more likely if

• income inequality ω s is high, low taxes on savers τ s
• pledgability ` high, gov. debt B high

28
How does an economy fall into the debt trap? (i) Rising inequality

Household debt / GDP Interest rate Output gap

100 % 7% 0%

95 % 6% −0.5 %

90 % 5% −1 %

85 % 4% −1.5 %

80 % 3% −2 %
0 10 20 30 0 10 20 30 0 10 20 30
years years years

Without ZLB ZLB at r = 3.5%

• Anticipation of the liquidity trap pulls the economy in even faster

29
How does an economy fall into the debt trap? (ii) Credit boom-bust cycle

Borrowing capacity (% of s.s.) Household debt / GDP Output gap

10 % 100 % 6%

8% 98 %
4%
6% 96 %
2%
4% 94 %
0%
2% 92 %

0% 90 % −2 %
0 10 20 30 0 10 20 30 0 10 20 30
years years years

30
Fighting debt with debt? Deficit financing in the liquidity trap

Gov. spending Interest rate Output gap

4% 8% 0%
7%
3%
−1 %
6%
2%
5%
−2 %
1%
4%

0% 3% −3 %
0 2 4 0 2 4 0 2 4
years years years

31
Fighting debt with debt? Deficit financing in the liquidity trap

Gov. spending Interest rate Output gap

4% 8% 0%
7%
3%
−1 %
6%
2%
5%
−2 %
1%
4%

0% 3% −3 %
0 2 4 0 2 4 0 2 4
years years years

• Here, deficit financing is only temporary remedy against a chronic disease

• lessons for Covid crisis?

31
Indebted demand post-Covid
Covid shock set to further raise debt

32
Modeling Covid in our framework

• Assume agents work in two sectors, “social” and “distant”

• Assume borrowers are over-represented in “social”

[Dingel-Neiman, Mongey-Weinberg, Leibovici et al]

• Shock:
• potential output falls Y ↓ and inequality rises ω s ↑, ω b ↓
• assume this induces negative demand shock in “distant” sectors
[Guerrieri-Lorenzoni-Straub-Werning]

33
Covid in the indebted demand diagram

Effective lower bound

34
Covid in the indebted demand diagram

Effective lower bound

Induced demand shock

34
Covid in the indebted demand diagram

r
Reduced borrowing
capacity

Effective lower bound

Induced demand shock

34
Covid in the indebted demand diagram

r
Reduced borrowing
capacity
Covid shock:
r ↓, debt ↑

Effective lower bound

Induced demand shock

34
Three “archetypes” of policies in response to Covid shock

(A) Stimulating (non-productive) private debt to buffer the shock

• e.g. Fed’s lending facilities via SPV’s

→ model as increase in credit limit

(B) Government funds transfers using public debt, paid for by all taxpayers
• e.g. stimulus checks, UI, grants to businesses

→ model as increase in government debt

(C) Government funds transfers by taxing (now or later) very progressively

• e.g. Landais-Saez-Zucman, Greenwood-Thesmar

→ model as saver-financed increase in government debt

35
Three “archetypes” of policies in response to Covid shock

(A) Stimulating (non-productive) private debt to buffer the shock

• e.g. Fed’s lending facilities via SPV’s

→ model as increase in credit limit

(B) Government funds transfers using public debt, paid for by all taxpayers
• e.g. stimulus checks, UI, grants to businesses

→ model as increase in government debt

(C) Government funds transfers by taxing (now or later) very progressively

• e.g. Landais-Saez-Zucman, Greenwood-Thesmar

→ model as saver-financed increase in government debt

Different across (A), (B), (C): whether there is a transfer from savers to borrowers 35
Policies in the indebted demand diagram

Covid shock:
r ↓, debt ↑

Effective lower bound

36
Policies in the indebted demand diagram

Covid shock + (A):

r ↓↓, debt ↑↑

Effective lower bound

d
Policy (A) — Stagnation post-Covid

36
Policies in the indebted demand diagram

Covid shock + (B):

r ↓, debt ↑

Effective lower bound

d
Policy (B) — Softer stagnation post-Covid

36
Policies in the indebted demand diagram

r
Covid shock + (C):
r ↑, debt ↑

Effective lower bound

d
Policy (C) — No stagnation!

36
Policies in the indebted demand diagram

B
A

Effective lower bound

Bottom line: Transfers > Debt

(long term → address any structural problems leading to greater inequality)
36
Extensions & conclusion
Extensions

• Redistribution (e.g. wealth tax) = Pareto improvement in debt trap

• Investment can help, especially if it complements borrowers’ labor

• Similar results when there is gov’t bond pay lower rate

• Intergenerational mobility helps

• Sufficient statistic exercise

In paper:
• Open economy model

• Uzawa preferences, relative wealth preferences 37

Takeaway

Indebted Demand:

Demand decreases in r × debt

Particularly relevant post-Covid!

38
Takeaway

Indebted Demand:

Demand decreases in r × debt

Particularly relevant post-Covid!

38
Extra slides
Inequality and debt back

Top 1% income share Interest rate

14 %
7%
13 %
6%
12 %
5%
11 %
4%
10 %
0 10 20 30 40 50 0 10 20 30 40 50
years years

Household debt / GDP Debt service / GDP

4.5 %
80 %

4%
70 %

3.5 %
60 %
3%
50 %
2.5 %
0 10 20 30 40 50 0 10 20 30 40 50
years years

Homothetic model Non-homothetic model

39
Deficit spending causes indebted (government) demand back

Gov. debt / GDP Interest rate Household debt / GDP

20 %
6% 90 %

15 % 5.5 %
85 %
5%
10 %
4.5 %
80 %
5% 4%
3.5 % 75 %
0%
0 10 20 0 10 20 0 10 20
years years years

40
But ... what about r < g? (here: g normalized to zero) back

• Our r is return on wealth so always r > g. But what if gov’t pays rB < g?

41
But ... what about r < g? (here: g normalized to zero) back

• Our r is return on wealth so always r > g. But what if gov’t pays rB < g?

• Our model points to two objects that matter (see paper for details)

41
But ... what about r < g? (here: g normalized to zero) back

• Our r is return on wealth so always r > g. But what if gov’t pays rB < g?

• Our model points to two objects that matter (see paper for details)

1. Derivative of debt service cost of (rB − g)B w.r.t. B

∂(rB − g)B ∂rB ?
= rB − g + ≷ 0
∂B ∂B
| {z } |{z}
<0
>0

41
But ... what about r < g? (here: g normalized to zero) back

• Our r is return on wealth so always r > g. But what if gov’t pays rB < g?

• Our model points to two objects that matter (see paper for details)

1. Derivative of debt service cost of (rB − g)B w.r.t. B

∂(rB − g)B ∂rB ?
= rB − g + ≷ 0
∂B ∂B
| {z } |{z}
<0
>0

2. Where does the spread r − rB come from? Investors really like B!

• B is not negative for savers just because (rB − g)B < 0
• B ↑ still makes savers wealthier, as ↑, lowering required return on wealth r

41
Redistribution and welfare back

• What policy mitigates a debt trap? → redistribution

• Example: wealth tax of τ a > 0 on saver’s wealth, redistributed to borrowers

• Saver’s budget constraint becomes

cst + ȧst = (rt − τ a ) ast
→ Wealth tax reduces return on wealth at ZLB to r − τ a , raising Ŷ

• What about welfare?

• borrower clearly benefits: lower r + wealth tax transfers + higher incomes
• saver also benefits: greater incomes (& asset prices) more than compensate
for tax!

• Thus: Redistribution mitigates debt trap, at no welfare cost! 42

Introducing investment back

• Assume goods are now produced from capital and both agents’ labor

Y = F(K, Lb , Ls )
• F is net-of-depreciation production, K pinned down by FK = r
• σ ≡ (Allen) elasticity of substitution between K and Lb

43
Introducing investment back

• Assume goods are now produced from capital and both agents’ labor

Y = F(K, Lb , Ls )
• F is net-of-depreciation production, K pinned down by FK = r
• σ ≡ (Allen) elasticity of substitution between K and Lb

• Key: savers’ income share ω s = ω s (r) now a function of r!

FK K FLs Ls F b Lb
ω s (r) ≡ + =1− L
F F F

43
Introducing investment back

• Assume goods are now produced from capital and both agents’ labor

Y = F(K, Lb , Ls )
• F is net-of-depreciation production, K pinned down by FK = r
• σ ≡ (Allen) elasticity of substitution between K and Lb

• Key: savers’ income share ω s = ω s (r) now a function of r!

FK K FLs Ls F b Lb
ω s (r) ≡ + =1− L
F F F
• ω s (r) independent of r if σ = 1 [e.g. Cobb-Douglas]
• ω s (r) ↑ as r ↓ iff σ > 1 [e.g. capital-skill complementarity, robots]

43
Indebted demand and investment back

• Main result: Our results are unchanged if σ = 1. Amplified if σ > 1.

σ<1

σ=1
σ>1
d

44
Indebted demand and investment back

• Main result: Our results are unchanged if σ = 1. Amplified if σ > 1.

σ<1

σ=1
σ>1
d

• Related Q: Can corporate debt also cause indebted demand?

• yes, if σ > 1! but always weaker indebted demand than household debt
• why? corporate debt productive, raising Y, easier to repay 44
Government yield spread back

• Allow for benefits from gov’t bonds [cf Krishnamurthy Vissing-Jorgensen (2012)]
δ
log (cst + ξBt ) + · v (ast + ξBt /r)
ρ
• Implies fixed spread ξ > 0
rB = r − ξ

45
Government yield spread back

• Allow for benefits from gov’t bonds [cf Krishnamurthy Vissing-Jorgensen (2012)]
δ
log (cst + ξBt ) + · v (ast + ξBt /r)
ρ
• Implies fixed spread ξ > 0
rB = r − ξ

• Define effective wealth as including benefits ξBt from bonds. In steady state:
ωs rB B ξB
aeff ≡ +d+ +
r | r {z r}
=B
• Savings supply curve unchanged in effective wealth
1 + ρ/δ
r=ρ
1 + ρ/δ · η(aeff )
45
Intergenerational mobility back

• With probability q > 0, savers turn into borrowers and vice versa

• Saver-turned-borrowers consume down their wealth instantly

• Borrower-turned-savers get transfer from other savers to raise wealth

46
Intergenerational mobility back

• With probability q > 0, savers turn into borrowers and vice versa

• Saver-turned-borrowers consume down their wealth instantly

• Borrower-turned-savers get transfer from other savers to raise wealth

• Saving supply curve becomes flatter with q

1 + δ/ρ δ/ρ · η(a)
r=ρ + qγδ
1 + δ/ρ · η(a) 1 + δ/ρ · η(a)
| {z }
contribution of mobility

• q ↑ thus mitigates indebted demand, especially if high income inequality γ

ωb − `
γ ≡1−
ωs + ` 46
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

c(r(a), a) = r(a)a

47
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

cr c dr dr
c(r(a), a) = r(a)a ⇒ + ca = +r
c a d log a |{z} d log a
MPCcap. gains
|{z}
semi-elast. r wrt r

47
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

dr MPCcap. gains − r
c(r(a), a) = r(a)a ⇒ =
d log a 1 − r ac

47
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

dr MPCcap. gains − r
c(r(a), a) = r(a)a ⇒ =
d log a 1 − r ac

• Standard PIH model: MPCcap. gains = r log preferences: r = 0

47
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

dr MPCcap. gains − r
c(r(a), a) = r(a)a ⇒ =
d log a 1 − r ac

• Standard PIH model: MPCcap. gains = r log preferences: r = 0

• Assume r = 0, r ≈ 0.06, MPCcap. gains ≈ 0.025

[Farhi-Gourio, Di Maggio-Kermani-Majluf, Baker-Nagel-Wurgler, Chodorow-Reich Nenov Simsek]

47
Is this first order? What is the slope of savings supply in the data? back

• Consumption function of rich c(r, a). Along curve:

dr MPCcap. gains − r
c(r(a), a) = r(a)a ⇒ =
d log a 1 − r ac

• Standard PIH model: MPCcap. gains = r log preferences: r = 0

• Assume r = 0, r ≈ 0.06, MPCcap. gains ≈ 0.025

[Farhi-Gourio, Di Maggio-Kermani-Majluf, Baker-Nagel-Wurgler, Chodorow-Reich Nenov Simsek]

dr
= −0.035
d log a

• In words: if wealth ↑ by 10%, required r ↓ by 35bps

47
Bottom 90% did not accumulate assets

Bottom 90% reduced saving

.02

Relative to 63−82
0

−.02

−.04

−.06

63−82 83−97 98−07 08−15

Contributions into housing (Nh)

Contributions into non−housing (Nnh)
Change in debt (∆D)
Saving (Θ)

48
How indebted is US demand? back

• Thought experiment: How large is dC implied by current levels of household

& government debt, had interest rates not come down?

49
How indebted is US demand? back

• Thought experiment: How large is dC implied by current levels of household

& government debt, had interest rates not come down?
• Counterfactual debt service burden, holding r constant:
Household Government
20 20

15
Percent

Percent
15
10

10
0
1980 1990 2000 2010 2020 1980 1990 2000 2010 2020

Household DSR Government DSR

Household DSR (counterfactual) Government DSR (counterfactual) 49
How indebted is US demand? back

• Thought experiment: How large is dC implied by current levels of household

& government debt, had interest rates not come down?

• Counterfactual debt service burden, holding r constant:

MPCcap. gains
dC ≈ −15% + · 15% = −8%
| {z } r {z
borrower debt service
| }
partial offset by savers

49