Global Banks and Systemic Debt Crises Juan Martin Morelli Pablo - - PowerPoint PPT Presentation

Global Banks and Systemic Debt Crises

Juan Martin Morelli Pablo Ottonello Diego Perez New York University Michigan and NBER New York University September 24, 2019

Motivation

• Emerging market debt crises are global in nature

◮ Affect multiple borrowing economies in synchronized fashion ◮ Involve the stability of global financial intermediaries (GFIs)

• Policy circles: key role of GFIs shaping systemic debt crises
• Most theories abstract from explicit GFIs

This paper: Study role of GFIs in international lending

• Model of the global economy + New empirical evidence

What We Do

• 1. Model of world economy

◮ Key ingredients: heterogeneous risky borrowers + GFIs ◮ Role of GFIs depends on degree of financial frictions

• 2. Empirical evidence: Effects of GFIs’ net worth in bond prices

◮ Exploit bond-level variation in prices during Lehman episode ◮ Larger price drops in EM bonds held by more affected GFIs

• 3. Relevance of GFIs in global debt market

Quantitative analysis:

◮ Infer degree of financial frictions from empirical estimates

Main results:

◮ Lenders are key for fluctuations in EM spreads & consumption ◮ GFIs’ exposure to EM debt and debt distribution matter

Outline

• 1. Model
• 2. Empirical Analysis
• 3. Quantitative Analysis

The Global Economy

Emerging-Market Economies (EMs)

• Risk-averse/impatient
• Tradable endowment
• Risk-neutral/patient
• Tradable endowment +

production technology

Developed-Market Economies (DMs)

• Invest in risky securities
• Financed with risk-free

bonds + costly equity

• Save in risk-free bonds

Global Banks

• Borrow without commitment

Risky Lending Risky Lending Financing

EMs: Environment

• Preferences of EM household i: E0

∞

t=0 βt EMu(cit)

• Endowment: yEMt

systemic

+ zit

• idiosyncratic
• If repays (ιt = 1), EM can borrow and consumes

cit = yEMt + zit + qi

EMt(bit+1 − ξbit) − bit

• If defaults (ιt = 0), EM temporarily looses credit acces & consumes

cit = H(yEMt + zit) H(x) ≤ x

• Partial equilibrium ⇒ qi

EMt = E

• mt+1ιit+1
• 1 + ξqi

EMt

• ◮ This model: mt+1 depends on global banks and distribution of

borrowing in world economy

Recursive problem

DMs: Households & Non-Financial Firms

DM Households

• Linear preferences over consumption, inelastically supply labor
• Deposit rate Rdt =

1 βDM

DM Firms

• CRS technology using capital and labor, shocks to quality of capital
• Expected mg. return DM capital

EtRDM,t+1 = Etωt+1[α (Kt+1)α−1 + (1 − δ)]

Global Banks

• Objective function

max Et

∞

• s=0

βs−t

DMπjt+s

• Initial portfolio:
• ai

EMjt−1

• i∈It−1 , aDMjt−1, djt−1
• Net worth:

njt =

• i∈It−1

Ri

EMtqi EMt−1ai EMjt−1 di + RDMtqDMt−1aDMjt−1 − Rddjt−1

• Flow-of-funds constraint:
• i∈It

qi

EMtai EMjt di

• purchase EM securities

+ qDMtaDMjt

• purchase DM securities

+divjt = njt + djt

• net worth

+new deposits

Recursive problem

Global Banks

• Financial frictions:

◮ Limited liability djt ≤ κnjt ◮ Cost of raising equity: C(div, n) = φ −div

n

• if div < 0
• Exit with prob 1 − σ, replaced by other entrant with initial n
• Optimal portfolio choice

◮ Same expected required return in all risky assets Re

EMit = Re DMt ≡ Re t

where Re

EMit ≡ Et[vt+1Ri EMt+1], Re DMt ≡ Et[vt+1RDMt+1]

◮ Under excess returns, deposits given by borrowing const: djt = κnjt ◮ External finance: equity issuance increasing in required return −2φ divjt njt

• = βDMRe

t − 1

Equilibrium

Equilibrium in EM Debt Market

Re

EM

AEM

Supply

• f Funds

Demand

• f Funds

Supply of funds As

t(Re EMt, Nt)

= Nt(1 + κ)

• Net worth

+New Deposits

+ E(REMt, φ)Nt

• Equity issuance

− ADMt(REMt, α)

• Investment in DM

Demand of funds Ad

t (Re t) =

• i∈It

ιit+1 Re

EMt

• 1 + ξqi

EMt+1

• bit+1d

Low Costs of Equity Issuance

A Decrease in Global Banks’ Net Worth: Financial Frictions & Price Effects

High Costs of External Finance

Re

EM

AEM

Supply

• f Funds

Demand

• f Funds

∆N(1 + κ) R0

EM

R1

EM

Low Costs of External Finance

Re

EM

AEM

Supply

• f Funds

Demand

• f Funds

∆N(1 + κ) R0

EM

R1

EM

Outline

• 1. Model
• 2. Empirical Analysis
• 3. Quantitative Analysis

EM Spreads & US Banks Net Worth

• 2.5
• 2.0
• 1.5
• 1.0
• 0.5

0.0 0.5 1.0 1.5 2.0 1 3 5 7 9 11 13 Sep-96 Jun-97 Mar-98 Dec-98 Sep-99 Jun-00 Mar-01 Dec-01 Sep-02 Jun-03 Mar-04 Dec-04 Sep-05 Jun-06 Mar-07 Dec-07 Sep-08 Jun-09 Mar-10 Dec-10 Sep-11 Jun-12 Mar-13 Dec-13 Sep-14 Sovereign EM spreads (left axis) Corporate EM spreads U.S. banks' net Worth (right axis)

Russian/LTCM Crisis Beginning of the Great Recession Lehman Bankruptcy GreekDebt Haircut

• Key empirical challenge:

◮ DM shocks affecting GFIs correlated with drivers of EM default

Countries included By country Historical default rates

Empirical Strategy

• Exploit EM bond-level variation in prices in narrow window around

the Lehman episode

• Key idea:

◮ During Lehman’s episode GFIs experienced differential changes in their net worth primarily driven by DM factors ◮ Exposure of GFIs to EM debt small ◮ In narrow window: can exploit price differences for bonds with similar characteristics

⇒ Identification: Bonds w similar default risk but different holders

Connection Empirical Strategy & Model

• Empirical strategy exploits price variation across EM bonds
• In model: all EM bonds priced by aggregate net worth
• Model extension with segmented markets

Details

◮ Banks specialize in bond varieties (trading networks) ◮ Trade bonds with banks in the same network

• Model extension features same price variation as empirical strategy
• Elasticity arises due to both financial and trading frictions

Data

Data:

• 1. EM bond prices

◮ Bond characteristics (maturity, liquidity, amount, currency)

Selection criteria

• 2. Share of holdings by individual GFIs for each individual EM bond
• 3. GFIs’ stock prices around Lehman’s episode

Global Banks List

Key variables:

• 1. Bond yield-to-maturity yi
• 2. Change in GFIs’ avg net worth per bond around Lehman’s episode

∆ei = J

j=1 θij∆ej

EM Bond Yields during Lehman

• Avg. Change in Yields

Dispersion in Change in Yields∗

• .004
• .002

.002 .004 .006

• 60
• 40
• 20

20 40 60 80 100 120

Days (0=Lehman Bankruptcy) All bonds Only Sovereign

.005 .01 .015 .02 .025

• 60
• 40
• 20

20 40 60 80 100 120

Days (0=Lehman Bankruptcy) All bonds Only Sovereign

∗ Residualized std dev after controlling for bonds observed characteristics

Empirical Model

Dynamic effects of global banks’ net worth on bond prices: ∆hyi = αksh + αch + βh∆ei + γ′

hXi + εi

• where

◮ yi: yield-to-maturity of bond i from country-sector k & curr c ◮ ∆ei: average change in GFIs’ net worth ◮ αksh: country by sector fixed effects

Sector shares

◮ αch: currency fixed effects

• βh: elasticity of bond prices to changes in holders’ net worth
• Additional controls:

◮ Xi: maturity, liquidity, initial ytm, share of GFIs in holders

EM Bonds Held by More Distressed GFIs Experienced Larger Price Contractions

• .25
• .2
• .15
• .1
• .05

Beta

• 20

20 40 60

Days Post

∆hyikc = αkh + αch + βh∆ei + γhXi + εikc

Further Analysis

• Results robust to alternative specifications

Baseline Only Sov. Similar Maturity IV (1) (2) (3) (4) Impact Effect

• 0.013∗∗∗
• 0.006
• 0.008

0.0029 (0.004) (0.005) (0.018) (0.011) Peak Effect

• 0.136∗∗
• 0.054∗∗
• 0.252∗∗∗
• 0.195∗∗∗

(0.053) (0.024) (0.065) (0.053) N Observations 402 198 70 108

• Alt. windows

No pre-trends Only Banks

• Exc. Mkt Makers

Only Dollar Bonds

• Results not driven by selection

◮ No sorting on observables within country-sector

Sorting table By Country By Sector

• Results do not capture reverse causality

◮ Low exposure of GFIs to EM debt

GFIs Exposure

◮ IV estimates w/ share of bonds held by AIG as instrument

Outline

• 1. Model
• 2. Empirical Analysis
• 3. Quantitative Analysis

Solution and Calibration Strategy

Solution method:

• Agg+idiosync shocks → Krusell-Smith approach to approx debt

distribution

Detail

Calibration Strategy:

• 1. Set subset of parameters to predetermined values

Functional forms Parameter values

• 2. Calibrate subset of parameters to match key aggregates

◮ debt level (avg), default rate (avg), spread (avg, vol & cyclicality),

portfolio of global banks (avg), global banks net worth (vol)

Parameter values Calibrated moments Banks Balance Sheet

Mapping Model and Empirical Estimates

• Effect in EM yields of a shock to ω leading to a contraction in GFIs

net worth as in the Lehman episode

2 4 6 8 10 Cost of raising equity 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

EM

Baseline Model nbaseline × 0.75 nbaseline × 0.62 Data

• Validation in model with secondary market

Details

Untargeted Moments

Comovements in debt prices

Data Model Comovements in Debt Prices corr(SPEM, SPEM,i) 0.69 0.71 corr(SPEM, SPDM) 0.51 0.69 Comovement with Global Banks corr(log V (N), SPEM)

• 0.57
• 0.59

corr(log V (N), SPDM)

• 0.79
• 0.8

EM business cycle statistics

Target Description Data Model σ(Ci)/σ(Yi) Excess Volatility of Consumption 1.14 1.05 corr(Ci, Yi) Correlation Consumtpion Endowment 0.90 0.96 σ(TBi/Yi) Volatility of Trade Balance 0.04 0.02 corr(TBi, Yi) Correlation Trade Balance & Endowment

• 0.31
• 0.15

Global Financial Crisis: Drivers

What is the relevance of global banks in the global financial crisis? Global Bank’s Net Worth

2007 2008 2009 2010 2011 2.0 1.5 1.0 0.5 0.0 0.5 1.0 1.5 Model Data

EMs’ Income

2007 2008 2009 2010 2011 0.5 0.0 0.5 1.0 1.5 Model Data

Global Financial Crisis: EMs’ Spreads & Consumption

EM’s Spreads

2007 2008 2009 2010 2011 100 200 300 400 500 600 Model Data

EMs’ Consumption

2007 2008 2009 2010 2011 0.5 0.0 0.5 1.0 1.5 2.0 Model Data

Decomposing EMs’ Spreads Dynamics

During Global Financial Crisis

• Increase in spreads in GFC: 402bps in data vs 531bps in model
• Model decomposition:

◮ 1/4 EM shock only & 3/4 DM shock only

Details

Unconditional Decomposition

• Avg Spreads: 410bps in data vs 404bps in model
• Model decomposition:

◮ 2/3 default premium & 1/3 risk premium

Details

Systemic vs. Idiosyncratic Income Shocks: The Role of GFI’s Exposure to EM Debt

Current Exposure

1 2 3 4 5 6 7 450 500 550 600 650 700 750 800 850 YEM Shock ZEM Shock

High Exposure

1 2 3 4 5 6 7 500 600 700 800 YEM Shock ZEM Shock

Systemic vs. Idiosyncratic Income Shocks: The Role of Debt Distribution

Ergodic Dispersion of EM Debt

1 2 3 4 5 6 7 500 600 700 800 YEM Shock ZEM Shock

High Dispersion of EM Debt

1 2 3 4 5 6 7 500 600 700 800 900 YEM Shock ZEM Shock

Conclusions

Shift focus to role of global banks in debt crises

• 1. In the micro-data, shocks to global banks affect bond prices
• 2. In the aggregate, global banks relevant in the generation &

amplification of systemic debt crises

◮ Distribution of EM debt & asset portfolio of global banks determine role played by global banks

Appendix

Empirical Evidence

The Sample: Countries

Focus on countries in JPMorgan’s EMBI with data on bond prices and fundamentals for 10 years+ in 1994-2014

• Argentina, Brazil, Bulgaria, Chile, China, Colombia, Croatia,

Ecuador, El Salvador, Hungary, Indonesia, Jamaica, Latvia, Lithuania, Malaysia, Mexico, Morocco, Pakistan, Panama, Peru, Philippines, Poland, Russia, South Africa, Thailand, Turkey, Ukraine and Venezuela

Back spreads and net worth Back to empirical strategy

Comovement between EM Bond Spreads and Global Banks’ Net Worth (Market Value)

• 0.7
• 0.6
• 0.5
• 0.4
• 0.3
• 0.2
• 0.1

CRO THA RUS BRA HUN PER BUL COL PHI ECU TUR VEN CHN LAT ARG PAN MOR JAM POL MEX ZAF ELS UKR PAK MAL CHL LIT IND corr(SP,XLF) EMBI

Back

EM Sov Default and U.S. Banks’ Net Worth

5 10 15 20 25 30 35

• 3.0
• 2.0
• 1.0

0.0 1.0 2.0 3.0 4.0 1945 1947 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Net Worth (left axis) Default (right axis) Back

Change in Net Worth of Global Banks During Lehman’s Episode

Aegon NV GE Capital Northern Trust Allianz SE Genworth Financial PNC Allstate Goldman Sachs Principal Financial Group American International Group HSBC Prudential Financial Ameriprise Financial Hartford Raiffeisen Bank International AG Ares Management Huntington Bancshares Regions BNP Paribas Intesa Sanpaolo Royal Bank of Canada BNYM Invesco SEI Investments Co Banca Mediolanum JPMorgan Schroders Banco Bilbao Vizcaya Argentaria Janus Henderson Group Societe Generale Banco Santander KBC Group NV Standard Life Aberdeen Bank of America Legg Mason State Street Barclays Bank Loomis Sayles Sumitomo Mitsui Financial Group. BlackRock M&T Bank Sun Life Financial CIBC Merrill Lynch T Rowe Price Group Citigroup MetLife TD Bank Commonwealth Bank of Australia Mitsubishi UFJ U.S. Bancorp Credit Suisse Morgan Stanley UBS Daiwa Securities Group NN Group NV UniCredit Deutsche Bank Natixis Virtus Investment Partners Fidelity National Financial Nikko Asset Management Co Wells Fargo Franklin Resources Nomura Holdings GAM Holding AG Nordea Bank Abp Back to data Back to sumstats

EM Bonds Sector Shares

Sector Share YTM Maturity Bid–Ask Spread Government 53.9% 7.2% 385 0.42% Industrial 3.4% 13.6% 201 0.83% Financial 21.0% 9.8% 332 0.52% Utilities 3.6% 9.2% 238 0.56% Communications 6.0% 9.2% 345 0.51% Energy 4.7% 7.6% 290 0.42% Other 7.4% 8.7% 495 0.68% Average 15.4% 9.4% 298 0.54%

Other includes: consumer (68%), Basic material (35%), Diversified (7%), and Technology (0.5%)

Back

EM Bond Yields during Lehman

• Avg. Yields

Dispersion in Yields∗

.01 .02 .03 .04 .05

Beta

10 20 30 40 50 60

Days Post

.01 .02 .03 .04 .05 .06

• 60
• 40
• 20

20 40 60 80 100 120

Days (0=Lehman Bankruptcy) All bonds Only Sovereign

∗Residualized std dev after controlling for bonds observed characteristics Back

Alternative Windows around Lehman

Different Start Date Different End Date

• .3
• .2
• .1

Beta

10 20 30 40 50 60

Days Post

• .3
• .2
• .1

Beta

10 20 30 40 50 60

Days Post

∆hyikc = αkh + αch + βh∆Ei + γhXi + εikc

Back

Baseline Results for Extended Sample

• .3
• .2
• .1

.1

Beta

• 50

50 100 150

Days Post

∆hyikc = αkh + αch + βh∆Ei + γhXi + εikc

Back

Robustness Analysis: Alternative Specifications

Baseline Only Dollar Only Banks

• Exc. Mkt. Makers

(1) (2) (3) (4) Impact Effect

• 0.013∗∗∗
• 0.012∗∗∗
• 0.024
• 0.013∗∗

(0.004) (0.004) (0.015) (0.006) Peak Effect

• 0.135∗∗
• 0.134∗∗
• 0.235∗∗∗
• 0.143∗∗∗

(0.053) (0.052) (0.090) (0.049) N Observations 402 305 356 397

Back

Sorting By Country

.05 .1 .15 .2 Mexico Brazil Kazakhstan Argentina India Philippines Turkey Colombia South Africa Venezuela Poland Indonesia Panama Greece Ukraine Thailand Uruguay Peru Lebanon Croatia Russia Jamaica Pakistan Costa Rica

Δe < Avg Δe Δe > Avg Δe

Back

Sorting By Country

Sector All bonds ∆ei < ∆ei ∆ei > ∆ei Government 53.9% 66.7% 45.6% Industrial 3.4% 4.0% 3.0% Financial 21.0% 11.3% 27.4% Utilities 3.6% 4.0% 3.3% Communications 6.0% 4.5% 7.0% Energy 4.7% 4.5% 4.8% Other 7.4% 5.1% 8.9%

Back

EM Bonds’ Characteristics by Holders’ Change in Net Worth

No Fixed Effects Country by Sector FE ∆ei < ∆ei ∆ei > ∆ei ∆ei < ∆ei ∆ei > ∆ei Residual Maturity 361 365

• 28.9

18.97 [265] [286] [237.7] [214.5] Bid–Ask Spread 0.43% 0.51%

• 0.02%

0.02% [0.02%] [0.01%] [0.01%] [0.02%] Yield (Pre-Lehman) 7.9% 8.4%

• 0.18%

0.12% [0.25%] [0.25%] [0.16%] [0.14%] Amount Issued 20.56 20.25

• 0.00

0.000 [0.079] [0.080] [0.063] [0.057] where: maturity expressed in days, amount in log-difference mn dollars

Back

EMs Debt in Global Banks’ Balance-Sheets

Financial Institution Ratio of Sovereign Ratio of Sovereign Non-U.S. Ratio of Risky Non-U.S. Debt to Assets Debt to Risky Assets Assets to Equity Aegon 0.044 0.32 1.9 Allianz 0.072 0.17 7.8 American International Group 0.008 0.28 0.3 Ameriprise 0.001 0.02 1.1 Banco Santander 0.054 0.41 2.3 Bank of America 0.022 0.06 3.8 Barclays 0.052 0.43 5.5 CIBC 0.005 0.01 9.9 Citigroup 0.058 0.14 6.7 Deutche Bank 0.005 0.04 4.1 Goldman Sachs 0.021 0.4 1.2 Hartford 0.002 0.21 0.3 HSBC 0.08 0.13 9.8 Intesa 0.053 0.67 1.2 JPMorgan 0.048 0.15 3.8 Merrill Lynch 0.026 0.39 1.4 MetLife 0.024 0.11 3.3 Mitsubishi 0.021 0.08 4.8 Morgan Stanley 0.025 0.51 1.6 PNC 0.001 0.003 3 Principal Financial Group 0.006 0.23 0.5 UBS 0.03 0.21 6.2 Wells Fargo 0.014 0.03 4.9 Average 0.048 0.139 5.5

Back to empirical analysis Back to calibration

Model

Global Banks: Recursive Problem

v(s, n) = max

{aEM,(b,z)≥0}, aDM≥0,d,div

(1 − σ)n + σ (div(1 + Idiv<0C(div, n) + v(s′, n′)) subject to

(b,z):g+(b,z)>0

qEM,(b,z)(s)aEM,(b,z)dbdz + qDM(s)aDM = n + d − div, d ≤ κn n′ =

(b,z):g+(b,z)>0

ιEM,(b,z)(s′)aEM,(b,z)dbdz + ω′ αAα−1

DM + 1 − δ

• aDM − Rdd.

Back

EMs: Repayment Choice

Value at repayment stage V (b, z, s−) = max

ι

ι V r(b, z, s+)

• value of repayment

+(1 − ι) V d(z, s+)

• value of default

s.t. s+ = Γ+(s−,˜ ι(ˆ b, ˆ z, s−)). Value of default V d(z, s+) = max

b′

u(c) + βE

• φV r(0, z′, s′

+) + (1 − φ)V d(z′, s′ +)

• s.t. c = H(yEM + z),

s′

+ = Γ+(s′ −,˜

ι(ˆ b, ˆ z, s′

−)),

s′

− = Γ−(s+, s′ x, ˜

ADM(s+), ˜ D(s+),˜ b′(ˆ b, ˆ z, s+)).

Back

Equilibrium

Definition

• i. Allocations for DM households, non-financial firms, global banks,

and EM households

• ii. Prices {
• qi

EMt

• i∈(0,µEM) , qDMt, wt}

such that

• Allocations solve agents problems at the equilibrium prices,
• Assets and labor markets clear.
• Returns of securities

◮ EMs: Ri

EMt+1 = ιit+1(1+ξqi

EMt+1)

qi

EMt

◮ DM: RDMt+1 = ωt+1[αAα−1

DMt+(1−δ)]

qDMt

Back

Functional Forms (I)

• Period utility EM households

u(c) = c1−γ 1 − γ

• Output net of default costs

H(y) = y

• 1 − d0yd1

[Arellano (2008), Chatterjee and Eyigungor (2012)]

Back

Functional Forms (II)

• EM endowment processes

ln yEMt = ρyEM ln yEMt−1 + σyEM ǫEMt, ǫEMt ∼ N(0, 1), ln zit = ρzi ln zit−1 + σzEM ǫit, ǫi,t ∼ N(0, 1).

• DM productivity process

ln ωt = ρω ln ωt−1 + σωǫωt, ǫωt ∼ N(0, 1).

Back

Fixed Parameter Values

Parameter Description Value γ Risk aversion 2.00 θ Reentry probability 0.25 ρEM Systemic endowment, autocorrelation 0.68 σEM Systemic endowment, shock volatility 0.03 βDM Discount rate of DM 0.98 α Share of capital 0.35 δ Depreciation 0.15 κ Debt-to-asset ratio 4.50

Back

Calibrated Parameter Values

Parameter Description Value βEM Discount rate of EMs 0.90 d0 Default cost — level 0.03 d1 Default cost — curvature 14.0 σ Bank survival rate 0.71 φ Marginal cost of raising equity 4.00 ηEM Mass of EM economies 2.16 σDM Volatility of DM shock 0.065 ¯ n Net worth of new entrants 0.40 ρDM,EM Correlation of exogenous shocks 0.55

Back

Calibrated Moments

Target Description Data Model E[Di/Yi] Average EM debt 15.0% 14.0% P[DFi] EM default frequency 1.5% 1.8% E[SPi] Average EM-bond spreads 410bp 404bp σ(SPi) Volatility EM-bond spreads 173bp 166bp corr(SPi, log Yi) Correlation EM-bond spreads & endowment −31% −75% σ(log V (N)) Volatility global banks’ net worth (NW) 0.28 0.25 corr(log V (N), log YEM) Correlation banks’ NW & systemic EM endowment 35% 31% E[AEM/(AEM + ADM)] Global banks’ exposure to EMs 10% 10% ηEM,N Elasticity EM spreads to banks’ NW 0.058 0.06

Parameter values Parameters and moments Back

Model with Secondary Market

Back

Model with Secondary Markets

• 1. EM Borrowers

◮ Issue continumm of bond “varieties” w same promised payoffs in primary markets

• 2. Global Banks

◮ Specialize in a variety: can buy all risky securities within that variety ◮ Choose asset portfolio & finance in primary & secondary markets

• 3. Secondary markets

◮ Trading networks: partition of set of banks according to varieties ◮ Trading frictions: Banks trade outstading securities w others in same trading network

⇒ Net worth at the trading network relevant for pricing

• 4. Primary markets

◮ New securities are issued by borrowers

⇒ Net worth at the aggregate level relevant for pricing

Back

Model w Secondary Mkts: Quantitative Analysis

• Same parametrization as baseline model
• Recreate Lehman as joint shock to ω + iid shock to borrowing

const κ at trading network

• Cross-sectional elasticity in model −0.031 vs −0.058 in data

Raw Model Simulated Data

0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 Change in Net Worth 0.02 0.01 0.00 0.01 0.02 0.03 Change in YTM

Demeaned w Country FE

0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 Change in Net Worth 0.015 0.010 0.005 0.000 0.005 0.010 Change in YTM

Back

Decomposing EMs’ Spreads Dynamics

During Global Financial Crisis ∆YEM ∆NW ∆ Spread ∆ C Data

• 2.14
• 3.72

402

• 1.72

Model Joint Shocks

• 2.14
• 3.52

531

• 2.59

EM Shock Only

• 2.14
• 0.09

121

• 1.55

DM Shock Only 0.0

• 3.36

351

• 0.92

Back

Decomposing EMs’ Spreads Dynamics

Unconditional Decomposition Average Std Dev Data 410 173 Model 404 166 Default Premium 295 137 Risk Premium 109 91

Back

Appendix Solution Method

Back

Solution Method

• Heterogeneity + aggregate uncertainty

⇒ ∆ state variable in agent’s individual problems

• Solve for the equilibrium of the model numerically using set of

statistics that summarize this distribution

Solution Method

Considerations:

• 1. Problems of individual EMs involve a default choice without

commitment ⇒ global methods in the solution of these problems

• 2. With default risk, degree of aggregate uncertainty significantly

affects EMs’ debt price schedules and policy functions ⇒ Include statistics summarizing ∆ as states in the agents’ problems

• 3. Debt price schedules faced by EMs depend on ˜

ADM(s+) ⇒ Use auxiliary aggregate variable ˆ ADM as a state

Solution Method: Algorithm

• 1. Specify initial forecasting rules: Fj

A(.) and Fj m(.) for j = 0

• 2. Solve individual agents’ problems given Fj

A(.) and Fj m(.) for j = 0

• 3. Simulate data from the model using the policy functions from (2)

for a given sequence ˜ sx ≡ {sx,t}T

t=1

◮ Estimate parameters of the forecasting rule with model simulated data ⇒ new forecasting rules Fj+1

A

(.), Fj+1

m

(.) ◮ compute the distance δj+1 ≡ || ˜ Fj+1(˜ sx) − ˜ Fj(˜ sx)||. where ˜ Fj(˜ sx): sequence of forecasts under Fj

i (.) and ˜

sx

• 4. Update forecasting rules and iterate in steps (2) and (3) for

j = 1, 2, 3, ..., until distances δj+1 is sufficiently small

Predicting ˆ ADM without using updated values

“fundamental accuracy plot,” Den Haan (2009)

Actual vs. Predicted Log-Residuals

Density of Log-Residuals of Predicted ˆ ADM, without using updated values

Den Haan (2009) method

Predicting ˆ ADM, without using updated values

Accuracy Tests, Den Haan (2009)

Target YEM below YEM above Max log-residual ZDM below 0.0219 0.039 ZDM above 0.0231 0.040 Min log-residual ZDM below

• 0.213
• 0.235

ZDM above

• 0.245
• 0.192

R2 ZDM below 0.979 0.980 ZDM above 0.989 0.989

Predicting ˆ ADM one-step-ahead

Actual vs. Predicted Log-Residuals

Density of Log-Residuals of Predicted ˆ ADM,

• ne-step-ahead