international credit flows and pecuniary externalitiesย ยท loss spiral 1/{1 โ๐ ... due to...
TRANSCRIPT
Bru
nn
erm
eier
& S
ann
iko
v
International Credit Flows and
Pecuniary ExternalitiesMarkus K. Brunnermeier & Yuliy Sannikov
Princeton UniversityโInternational Credit Flows,โฆโ
Bank of International SettlementBasel, August 29th, 2014
Bru
nn
erm
eier
& S
ann
iko
v
Motivation
Old โWashington consensusโ in decline Free trade: flow of goods/services intratemporal Free finance: flow of capital intertemporal
When does full capital account liberalization reduce (capital controls/macropru regulation improve) welfare?
1. Liquidity mismatch can lead to sudden stop runs Technological illiquidity: irreversibility (adjustment costs)
Market illiquidity: redeployability/specificity โ not in this paper
Funding illiquidity: short-term debt, โhot moneyโ Type of capital flow matters: FDI, portfolio flows (equity), long-term debt
2. โTerms of trade hedgeโ (Cole-Obstfeld) can be undermined when Industryโs output is not easily substitutable. Consumers cannot easily
find substitutes No strong competitors in other countries
Natural resources: oil, copper for Chile, Hard drives in Thailand, Bananas in Ecuador
2
Bru
nn
erm
eier
& S
ann
iko
v
Motivation
Old โWashington consensusโ in decline Free trade: flow of goods/services intratemporal Free finance: flow of capital intertemporal
When does full capital account liberalization reduce (capital controls/macropru regulation improve) welfare?
1. Sudden stop including runs due to liquidity mismatch Technological illiquidity: irreversibility (adjustment costs)
Market illiquidity: redeployability/specificity โ not this paper
Funding illiquidity: short-term debt, โhot moneyโType of capital flow matters: FDI, portfolio flows (equity), long-term debt
2. โTerms of trade hedgeโ (Cole-Obstfeld) can be undermined when Industryโs output is not easily substitutable. Consumers cannot easily
find substitutes No strong competitors in other countries
Natural resources: oil, copper for Chile, Hard drives in Thailand, Bananas in Ecuador
3
Ass
et s
ide
Bru
nn
erm
eier
& S
ann
iko
v
Motivation
Old โWashington consensusโ in decline Free trade: flow of goods/services intratemporal Free finance: flow of capital intertemporal
When does full capital account liberalization reduce (capital controls/macropru regulation improve) welfare?
1. Sudden stop including runs due to liquidity mismatch Technological illiquidity: irreversibility (adjustment costs)
Market illiquidity: redeployability/specificity โ not this paper
Funding illiquidity: short-term debt, โhot moneyโ Type of capital flow matters: FDI, portfolio flows (equity), long-term debt
2. โTerms of trade hedgeโ (Cole-Obstfeld) can be undermined when Industryโs output is not easily substitutable. Consumers cannot easily
find substitutes No strong competitors in other countries
Natural resources: oil, copper for Chile, Hard drives in Thailand, Bananas in Ecuador
4
Ass
et s
ide
Liab
ility
Bru
nn
erm
eier
& S
ann
iko
v
Motivation
Old โWashington consensusโ in decline Free trade: flow of goods/services intratemporal Free finance: flow of capital intertemporal
When does full capital account liberalization reduce (capital controls/macropru regulation improve) welfare?
1. Sudden stop including runs due to liquidity mismatch Technological illiquidity: irreversibility (adjustment costs)
Market illiquidity: redeployability/specificity โ not in this paper
Funding illiquidity: short-term debt, โhot moneyโ Type of capital flow matters: FDI, portfolio flows (equity), long-term debt
2. โTerms of trade hedgeโ (Cole-Obstfeld) can be undermined when Industryโs output is not easily substitutable.
Consumers cannot easily find substitutes No strong competitors in other countries
Natural resources: oil, copper for Chile, Hard drives in Thailand, Bananas in Ecuador
5
Bru
nn
erm
eier
& S
ann
iko
v
Model setup - symmetric
Preferences
๐ธ 0
โ
๐โ๐๐ก๐๐ก1โ๐พ
1 โ ๐พ๐๐ก
โข Same preference discount rate ๐ โ โsaving out of constraintโ
Two output goods ๐ฆ๐ and ๐ฆ๐ - imperfect substitutes
๐ฆ๐ก =1
2๐ฆ๐ก
๐๐ โ1๐ +
1
2๐ฆ๐ก
๐๐ โ1๐
๐ /(๐ โ1)
(Comparative) advantages:
12
Good ๐ Good ๐
Country A ๐๐๐ก ๐๐๐ก
Country B ๐๐๐ก ๐๐๐ก
Bru
nn
erm
eier
& S
ann
iko
v
Two country/sector model
World capital shares: ๐๐ก
๐ด๐ + ๐๐ก๐ด๐ + ๐๐ก
๐ต๐ + ๐๐ก๐ต๐ = 1
World supply of (output) goods:
๐๐ก๐ = ๐๐ก
๐ด๐๐ + ๐๐ก๐ต๐๐ ๐พ๐ก ๐๐ก
๐ = ๐๐ก๐ต๐๐ + ๐๐ก
๐ด๐๐ ๐พ๐ก
Price of output goods ๐ and ๐ in terms of price of ๐ฆ
๐๐ก๐ = 1
2
๐๐ก๐๐ก๐
1/๐
and ๐๐ก๐ = 1
2
๐๐ก
๐๐ก๐
1/๐
โข Terms of trade ๐๐ก๐/๐๐ก
๐
13
Bru
nn
erm
eier
& S
ann
iko
v
Two country/sector model
Capital evolution for
โข ๐๐๐ก = ฮฆ ๐๐ก โ ๐ฟ ๐๐ก๐๐ก + ๐๐ด๐๐ก๐๐๐ก๐ด in country ๐ด
โข ๐๐๐ก = ฮฆ ๐๐ก โ ๐ฟ ๐๐ก๐๐ก + ๐๐ต๐๐ก๐๐๐ก๐ต in country ๐ต
ฮฆ concavity โ technological illiquidity
Single type of capital
Investment in composite good
Shocks are โข Two dimensional
Affect global capital stock ๐๐๐ก๐ด + ๐๐๐ก
๐ต
Redistributive (initial shock + amplification) โ affects wealth share, ๐๐กโข Example: Apple vs. Samsung lawsuit
14
Bru
nn
erm
eier
& S
ann
iko
v
Market structures
15
Markets Output ๐ฆ๐, ๐ฆ๐
Physical capital ๐พ
Debt Equity
Complete MarketsFull integration/First Best
X X X X
Open credit account(equity home bias)
X X X
Closed credit accountX X
Trade Finance
intratemporal intertemporal
Add taxes/capital controls
Bru
nn
erm
eier
& S
ann
iko
v
Returns on physical capital
โข ๐๐๐ก/๐๐ก = ฮฆ ๐๐ก โ ๐ฟ ๐๐ก + ๐๐ด๐๐ก๐๐๐ก๐ด
Postulate
โข ๐๐๐ก/๐๐ก = ๐๐ก๐๐๐ก + ๐๐ก
๐๐ด๐๐๐ก
๐ด + ๐๐ก๐๐ต
๐๐๐ก๐ต
Returns from holding physical capital
โข ๐๐๐ก๐ด๐ =
๐๐๐ก๐โ๐๐ก
๐๐ก+ ๐๐ก
๐+ ฮฆ ๐๐ก โ ๐ฟ + ๐๐ด๐๐ก
๐๐ด๐๐ก +
+ ๐๐ด + ๐๐ก๐๐ด
๐๐๐ก๐ด + ๐๐ก
๐๐ต๐๐๐ก
๐ต
โข ๐๐๐ก๐ด๐ =
๐๐๐ก๐โ๐๐ก
๐๐ก+ ๐๐ก
๐+ ฮฆ ๐๐ก โ ๐ฟ + ๐๐ด๐๐ก
๐๐ด๐๐ก +
+ ๐๐ด + ๐๐ก๐๐ด
๐๐๐ก๐ด + ๐๐ก
๐๐ต๐๐๐ก
๐ต16
Ito product rule:๐ ๐๐ก๐๐ก = ๐๐๐ก๐๐ก + ๐๐ก๐๐๐ก + ๐๐๐๐๐๐ก
Bru
nn
erm
eier
& S
ann
iko
v
The 3 step solution procedure
1. Derive equilibrium conditionsโข Optimality and asset pricing conditions (from postulated processes)
Consumptionwith log-utility: ๐๐ก = ๐๐๐ก (no precautionary savings)
Asset pricing (from above)with log-utility: Sharpe Ratio of asset = volatility of net worth
Internal investment rate ๐๐ก: ๐ฮฆโฒ ๐๐ก โ 1 = 0
โข Market clearing conditions
2. Derive evolution of state variable ๐๐ก =๐๐ก
๐๐ก๐พ๐ก
3. Express in terms of ODE
โข All ๐postolated and ๐postulated are expressed in terms of ๐โฒ(๐), ๐โฒโฒ(๐), โฆ
17
For
any
mar
ket
stru
ctu
rem
arke
t st
ruct
ure
sp
ecif
ic
Bru
nn
erm
eier
& S
ann
iko
v
Market structures
22
Markets Output ๐ฆ๐, ๐ฆ๐
Physical capital ๐พ
Debt Equity
Complete MarketsFull integration/First Best
X X X X
Open credit account(equity home bias)
X X X
Closed credit accountX X
Trade Finance
intratemporal intertemporal
Add taxes/capital controls
Bru
nn
erm
eier
& S
ann
iko
v
Market structures
1. Complete markets โ First best
2. Incomplete markets (equity home bias)โข Levered short-term debt financing
โข Sudden stops: (varying technological illiquidity)
Amplification
Runs due to sunspots
3. Closed capital account: capital controls (no equity, no debt)
4. Welfare analysis
23
Bru
nn
erm
eier
& S
ann
iko
v
1. Complete markets: First Best Remarks
Perfect capital allocation + perfect risk sharing
Prices are constant and independent of shocks
Economy shrinks/expands with (multiplicative) shocks
Elasticity of substitution, ๐ , has no impact on prices
27
Bru
nn
erm
eier
& S
ann
iko
v
Market structures
1. Complete markets โ First best
2. Incomplete markets (equity home bias)โข Levered (short-term) debt financing
โข Sudden stops: (varying technological illiquidity, irreversibility)
Amplification
Runs due to sunspots
3. Closed capital account: capital controls (no equity, no debt)
4. Welfare analysis
33
Bru
nn
erm
eier
& S
ann
iko
v
2. Equilibrium characterization: state variable
Equilibrium is a mapHistories of shocks prices allocation{๐๐
๐ด, ๐๐ ๐ต , ๐ โค ๐ก} ๐๐ก, ๐๐ก
๐ด๐โฆ, ๐๐ก๐ด, ๐๐ก
๐ต, ๐๐ก๐ด, ๐๐ก
๐ต
wealth distribution
๐๐ก =๐๐ก
๐๐ก๐พ๐กโ 0,1 Aโs wealth share
๐๐ก๐ด๐ + ๐๐ก
๐ด๐ + ๐๐ก๐ต๐ + ๐๐ก
๐ต๐ = 1 and ๐ถ๐ก๐ด + ๐ถ๐ก
๐ต = ๐๐ก โ ๐๐ก๐พ๐ก
Portfolio weights: ๐๐ก
๐ด๐
๐๐ก,๐๐ก
๐ด๐
๐๐ก, 1 โ
๐๐ก๐ด๐+๐๐ก
๐ด๐
๐๐ก
Consumption rates: ๐๐ก๐ด = ๐ถ๐ก
๐ด/๐๐ก ๐๐ก๐ต = ๐ถ๐ก
๐ต/(๐๐ก๐พ๐ก โ ๐๐ก)
35
Bru
nn
erm
eier
& S
ann
iko
v
2. State variable: 3 regions
Wealth share ๐โข Three regions
โข Symmetric
39
๐
Full specialization
Full specialization
๐ด produces ๐ ๐ ๐, ๐
๐ต produces ๐, ๐ ๐ ๐
0 11/2
๐๐ก๐ด๐ = ๐๐ก
๐๐ก๐ต๐ = 1 โ ๐๐ก
๐๐ก๐ต๐ = ๐๐ก
๐ด๐ = 0
Bru
nn
erm
eier
& S
ann
iko
v
2. Capital share, terms of trade, price of capital
Numerical: ๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%
Three different elasticities of substitution: ๐ = {.5,1,โ} 40
wealth share
Bru
nn
erm
eier
& S
ann
iko
v
TOT: Supply vs. demand shock
Supply versus demand shock
TOT improve for ๐ด as ๐๐ก declines for ๐๐ก โ ๐, . 5can be due to
โข ๐๐๐ด < 0: Negative supply shock World recession
โข ๐๐๐ต > 0: Positive demand shock World boom
TOT: Output price
โฆbut fire-sale of (physical) capital stock ๐๐ก
41
Bru
nn
erm
eier
& S
ann
iko
v
Stationary distribution drift volatility
Three different elasticities of substitution: ๐ = {.5,1,โ}
Difference to Cole & Obstfeld 1994: persistence of capital, ๐ฟ < โ
2. Stability, Phoenix Miracle for different ๐
42
Phoenix miracle
Masspointat {0,1}
wealth share
Bru
nn
erm
eier
& S
ann
iko
v
Overview
1. Complete markets โ First best
2. Incomplete markets (equity home bias)โข Levered short-term debt financing
โข Sudden stops: (varying technological illiquidity)
Amplification
Runs due to sunspots
3. Closed capital account: capital controls (no equity, no debt)
4. Welfare analysis
44
Bru
nn
erm
eier
& S
ann
iko
v
2. Amplification
46
๐๐ก๐๐ด
=
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
Leverage effect ๐๐ก๐ด๐/๐๐ก
Loss spiral 1/{1 โ ๐๐ก๐ด๐ โ ๐๐ก
๐โฒ ๐๐ก๐ ๐๐ก
} (infinite sum)
Technological illiquidity (๐ , ๐ฟ) โ market illiquidity ๐โฒ ๐(dis)investment adjustment cost
Bru
nn
erm
eier
& S
ann
iko
v
2. Amplification
47
๐๐ก๐๐ด
=
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
Leverage effect ๐๐ก๐ด๐/๐๐ก
Loss spiral 1/{1 โ ๐๐ก๐ด๐ โ ๐๐ก
๐โฒ ๐๐ก๐ ๐๐ก
} (infinite sum)
Technological illiquidity (๐ , ๐ฟ) โ market illiquidity ๐โฒ ๐(dis)investment adjustment cost
leverage
Bru
nn
erm
eier
& S
ann
iko
v
2. Amplification
48
๐๐ก๐๐ด
=
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
Leverage effect ๐๐ก๐ด๐/๐๐ก
Loss spiral 1/{1 โ ๐๐ก๐ด๐ โ ๐๐ก
๐โฒ ๐๐ก๐ ๐๐ก
} (infinite sum)
chnological illiquidity (๐ , ๐ฟ) โ market illiquidity ๐โฒ ๐(dis)investment adjustment cost
leverage
Market illiquidity(price impact)
Bru
nn
erm
eier
& S
ann
iko
v
2. Amplification
49
๐๐ก๐๐ด
=
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
Leverage effect ๐๐ก๐ด๐/๐๐ก
Loss spiral 1/{1 โ ๐๐ก๐ด๐ โ ๐๐ก
๐โฒ ๐๐ก๐ ๐๐ก
} (infinite sum)
Technological illiquidity (๐ , ๐ฟ) โ market illiquidity ๐โฒ ๐โข (dis)investment adjustment cost
leverage
Market illiquidity(price impact)
Bru
nn
erm
eier
& S
ann
iko
v
Quadratic adjustment cost
Investment rate of ฮฆ +1
๐ ฮฆ2 generates new capital
at rate ฮฆ
ฮฆ ๐ =1
๐ 1 + 2๐ ๐ โ 1
Three casesโข ๐ = 0 โ ๐ = 1
โข ๐ = 2
โข ๐ ๐<0 = 100 and ๐ ๐>0 = 2
2. Technological (๐ , ๐ฟ) โ market illiquidity ๐โฒ ๐
52
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stops: amplification & runs
Sudden stopโข Adverse fundamental triggers %-decline in debt that exceeds
%-decline in net worth; ๐(๐๐ด๐โ๐)
๐๐
๐
๐๐ด๐โ๐> 1
๐๐๐ด๐
๐๐>
๐๐ด๐
๐
pro-cyclical leverage
54
hyperbola
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stops: amplification & runs
Sudden stopโข Adverse fundamental triggers %-decline in debt that exceeds
%-decline in net worth; ๐(๐๐ด๐โ๐)
๐๐
๐
๐๐ด๐โ๐> 1
๐๐๐ด๐
๐๐>
๐๐ด๐
๐
pro-cyclical leverage
55
hyperbola
Slope of tangent vs. secant
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stops: amplification & runs
Sudden stopโข Adverse fundamental triggers %-decline in debt that exceeds
%-decline in net worth; ๐(๐๐ด๐โ๐)
๐๐
๐
๐๐ด๐โ๐> 1
๐๐๐ด๐
๐๐>
๐๐ด๐
๐
pro-cyclical leverage
๐๐ก๐ท๐๐๐ก๐ด
= 1 +๐ ๐๐ก
๐ด๐ โ ๐๐ก /๐๐๐ก ๐๐ก
๐๐ก๐ด๐ โ ๐๐ก
+
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
โข An unanticipated sunspot triggers a sudden capital price drop from ๐ to ๐, accompanied by a drop in ๐ to ๐.
๐ ๐ = max ๐๐ + ๐๐ด๐ ๐ โ ๐ , 0
58
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stops: amplification & runs
Sudden stopโข Adverse fundamental triggers %-decline in debt that exceeds
%-decline in net worth; ๐(๐๐ด๐โ๐)
๐๐
๐
๐๐ด๐โ๐> 1
๐๐๐ด๐
๐๐>
๐๐ด๐
๐
pro-cyclical leverage
๐๐ก๐ท๐๐๐ก๐ด
= 1 +๐ ๐๐ก
๐ด๐ โ ๐๐ก /๐๐๐ก ๐๐ก
๐๐ก๐ด๐ โ ๐๐ก
+
๐๐ก๐ด๐
๐๐ก(1โ๐๐ก)
1 โ [๐๐ก๐ด๐ โ ๐๐ก]
๐โฒ(๐๐ก)๐ ๐๐ก
๐๐ด
โข An unanticipated sunspot triggers a sudden capital price drop from ๐ to ๐, accompanied by a drop in ๐ to ๐.
๐ =max ๐๐ + ๐๐ด๐ ๐ โ ๐ , 0
๐59
hyperbola
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stop due to sunspot
60
Bru
nn
erm
eier
& S
ann
iko
v
Sudden stop due to sunspot: Zoomed in
61
Bru
nn
erm
eier
& S
ann
iko
v
Overview
1. Complete markets โ First best
2. Incomplete markets (equity home bias)โข Levered short-term debt financing
โข Sudden stops: (varying technological illiquidity)
Amplification
Runs due to sunspots
3. Closed capital account: capital controls (no equity, no debt)
4. Welfare analysis
62
Bru
nn
erm
eier
& S
ann
iko
v
Market structures
64
Markets Output ๐ฆ๐, ๐ฆ๐
Physical capital ๐พ
Debt Equity
Complete MarketsFull integration/First Best
X X X X
Open credit account(equity home bias)
X X X
Closed credit accountX X
Trade Finance
intratemporal intertemporal
Add taxes/capital controls
Bru
nn
erm
eier
& S
ann
iko
v
3. Credit account: open vs. closed
๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%, s = 1
65
Capital price lowerโข Lower input priceโข Destabilizes
balance sheet
Bru
nn
erm
eier
& S
ann
iko
v
3. Credit account: open vs. closed
๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%, s = 1
66
Stability
More stabilityLess growth
Phoenix miracleslightly smaller
Bru
nn
erm
eier
& S
ann
iko
v
Overview
1. Complete markets โ First best
2. Incomplete markets (equity home bias)
3. Closed capital account: capital controls (no equity, no debt)
4. Welfare analysisโข Pecuniary externalities
โข Welfare calculations + Pareto improving redistributions
68
Bru
nn
erm
eier
& S
ann
iko
v
4. When are credit flows excessive?
Constrained inefficiency (in incomplete market setting)due to pecuniary externalities
โข Price of capital: fire sale externality if leverage is high
โข Price of output good: โterms of trade hedgeโ restrained competition
Price taking behavior undermined this hedge
69
shockautomatic
hedge
Price taking behavior
Bru
nn
erm
eier
& S
ann
iko
v
4. When are credit flows excessive?
Constrained inefficiency (in incomplete market setting)due to pecuniary externalities
โข Price of capital: fire sale externality if leverage is high
โข Price of output good: โterms of trade hedgeโ restrained competition
Price taking behavior undermined this hedge
70
shock
Complete market insurance
Bru
nn
erm
eier
& S
ann
iko
v
4. Welfare comparison
๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%,
75
Full specialization
Bru
nn
erm
eier
& S
ann
iko
v
4. Welfare comparison
๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%,
76
Full specialization
Inefficiency at the extremes:Role for redistributive Policydefault/bail-out/debt-relief
Pareto improving
Intuition:Other countryโs output price is high
Bru
nn
erm
eier
& S
ann
iko
v
4. Welfare comparison
๐ = 5%, ๐ = 14%, ๐ = 4%, ๐ฟ = 5%, ๐ = 2, ๐๐ด = ๐๐ต = 10%,
77
Full specialization
Bru
nn
erm
eier
& S
ann
iko
v
4. Welfare comparison
Any monotone transformation of ๐ would be equally good state variable
Normalization: take CDF of ๐
โข Uniform stationary distribution!
79
Bru
nn
erm
eier
& S
ann
iko
v
Conclusion
Sudden stopsโข Amplification of fundamental shock โข Runs due to sunspots โ vulnerability region
Phoenix miracle Tradeoff between capital allocation & risk sharing
โข โTerms of trade hedgeโ
When are short-term credit flows excessive?โข When can capital controls (financial liberalization) be welfare
enhancing (reducing)? โข Pecuniary externality
Price of physical capital fire-sales externality โ technological illiquidity Price of output goods: โterms of trade hedgeโ externality
Bailout/RestructuringRedistributive policy can be Pareto improving if one country is sufficiently balance sheet impaired
โข Reduces output good price
83
po
siti
ven
orm
ativ
e