Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Bubbles, Crashes & the Financial Cycle:The Limits to Credit Growth
Sander van der Hoog and Herbert DawidChair for Economic Theory and Computational Economics
Bielefeld University
WEHIASophia Antipolis, 21-23 May 2015
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
The Big Questions
I Which micro- or macro-prudential banking regulations arebeneficial to financial stability?
I Prevention and mitigation policies:
I How to prevent severe downturns from occurring?
I How to mitigate the cumulative economic losses?
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
Activity Role Agent Agent Role Activity
E u r a c e @ U n i b i
Bank
InvGoodFirm
Household
asset demandsavings decision
FinancialMarket
(index bond)
ECBMonetarypolicy
Gov
Policymaker
Investor
ConsGoodFirm
Consumer
Cons.Goods Market
(local malls)
cgood demandconsumption choice
Producercgood supplyposted prices
labor supplyreservation wage
Employee labor demandwage schedule
LaborMarket(search & matching)
Creditor Debtorcredit supplyrank credit risk
credit demandrank interest
CreditMarket(credit
rationing)
Employer
InvestorProducerCapitalGoodsMarket
igood supplyvintage menuposted prices
igood demandvintage choices
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Mechanisms in the modelCapital Adequacy RequirementReserve Ratio Requirement
Mechanisms in the model
1. Probability of Default (PD): Internal Risk-Based approach (IRB)
2. Interest rate rule for commercial banks
3. Debt-equity transformation: Insolvency / Illiquidity
4. Dividend payout rule
5. Credit rationing rule
6. Capital Adequacy Requirement (CAR)
7. Central Bank Reserve Ratio Requirement (RRR)
8. Future research: Capital Conservation Buffers & Counter-Cyclical Capital Buffers:
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Mechanisms in the modelCapital Adequacy RequirementReserve Ratio Requirement
Probability of Default, Interest rate rule
1. Firm’s default probability
PDft = max{0.0003,1−e−νDf
t /E ft }, ν = 0.1
2. Interest rate offered by bank b to firm i
rbft = rECB
(1 + λ
B ·PDft + ε
bt
), ε
bt ∼ U[0,1]
rECB = 0.01
λ B = 3: penalty rate for high-risk firm, uniform across banksεb
t : bank’s ideosyncratic operating costs
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Mechanisms in the modelCapital Adequacy RequirementReserve Ratio Requirement
Capital Adequacy Requirement
1. Risk-exposure of credit request (Expected Loss at Default):
rwabit = PDit ·Lit . and RWAb
t =F
∑i=1
K (i)
∑k=0
PDkt ·Lkt , (1)
2. Constraint 6: Capital Adequacy Requirement (CAR)
RWAbt ≤ α ·Eb
t , α ≥ 0 (2)
3. Risk-exposure "budget" of the bank:
V bt := α ·Eb
t −RWAbt (3)
4. Risk-constrained loan demand:
¯̀bit =
Lit if PDit ·Lit ≤ V bt
0 if 0≤ V bt ≤ PDit ·Lit
0 if V bt < 0.
(4)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Mechanisms in the modelCapital Adequacy RequirementReserve Ratio Requirement
Reserve Ratio Requirement
I Constraint 7: Reserve Ratio Requirement (RRR)
Mbt ≥ β ·Depb
t , β ∈ [0,1] (5)
I Excess liquidity "budget" of the bank:
W bt := Mb
t −β ·Depbt (6)
I Loan granted: risk- and liquidity constrained credit request
`bi ,t =
¯̀bi ,t if W b
t ≥ ¯̀bi ,t
φ · ¯̀bi ,t if 0≤W b
t ≤ ¯̀bi ,t
0 if W bt < 0.
(7)
Possibility of credit rationing: {φ : W bt −φ · ¯̀b
i ,t = 0}→ φ = W bt /
¯̀bi ,t
I Illiquid banks stop lending to all firms (bank lending channel)
I Risky firms cannot get loans (borrower’s balance sheet channel)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Amplitude of recessionsPrevention and mitigation: The Limits to Credit Growth
Parameter sensitivity analysis
0 200 400 600 800 1000
2000
3000
4000
5000
Months
Eur
osta
t_ou
tput
alfa_20_gamma
1.0 2.0 4.0 8.0 16.0 32.0
α-sensitivity: Cap. Adq. Req.
I Default: α = 32 (3%)I Lower: amplitude of recessions
increases
0 200 400 600 800 100020
0030
0040
0050
00
Months
Eur
osta
t_ou
tput
beta_20_gamma_10_alfa
0.01 0.02 0.05 0.10 0.20 0.50
β -sensitivity: Reserve Req.
I Default: β = 0.05 (5%)I Higher: amplitude of recessions
decreases
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
0 20 40 60 80 100 120
6000
6500
7000
7500
8000
8500
9000
Recessions and expansions
Quarters
Out
put
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Amplitude of recessionsPrevention and mitigation: The Limits to Credit Growth
Parameter sensitivity analysis
α-sensitivity: Cap. Adq. Req.I Basel III: 4.5−10.5%
α = 22.2−9.5I Lower: amplitude of recessions
increases
−80
00−
6000
−40
00−
2000
Parameter 1
full_
ampl
itude
_rec
essi
on0.00 0.01 0.02 0.05 0.10 0.20 0.50 0.90 0.99 1.00
−80
00−
6000
−40
00−
2000
β -sensitivity: Reserve Req.I EU: β = 0.01, US: β = 0.10, CA:
β = 0I Higher: amplitude of recessions
decreases
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Amplitude of recessionsPrevention and mitigation: The Limits to Credit Growth
Parameter sensitivity analysis 2D-grid
alpha
beta
−1200
−1100
−1000
−900
−800
−700
4.0 6.0 8.0 10.0 12.0 16.0 20.0 24.0 28.0 32.0
0.00
0.02
0.10
0.25
0.35
0.45
0.90
1.00
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Amplitude of recessionsPrevention and mitigation: The Limits to Credit Growth
Prevention and mitigation policies: The Limits to Credit Growth
Proposed regulations to limit excesses in banking (eg. Admati & Hellwig, 2013):
A. Default regulation: Capital ratio 12.5%, Reserve ratio 10%.
B. Banning bank dividend payouts→ Increases bank equity capital
C. Using non-risk-weighted capital ratios→ Prevents abuse of risk-weights
("risk-weight management optimization")
D. Cutting-off funding to all financially unsound firms→ Prevents leverage
E. Cutting-off funding to Ponzi firms only→ Prevents further leverage
F. Combined effect of BCD→ Does it help to prevent bubbles?
G. Combined effect of BCE→ Does it help to prevent bubbles?
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Amplitude of recessionsPrevention and mitigation: The Limits to Credit Growth
Prevention and mitigation policies: The Limits to Credit Growth
Comparison across regulations A - G
−30
00−
2500
−20
00−
1500
−10
00−
500
Parameters
full_
ampl
itude
_rec
essi
on
A B C D E F G
−30
00−
2500
−20
00−
1500
−10
00−
500
amplitude of recessions(output lost)
−10
000
−80
00−
6000
−40
00−
2000
Parameters
full_
cum
m_l
oss_
rece
ssio
n
A B C D E F G
−10
000
−80
00−
6000
−40
00−
2000
cumulative loss of output(amplitude & duration)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Main Conclusions
I To prevent large cumulative losses that follow from recessions,it is required to cut-off funding to all financially unsound firms(speculative and Ponzi firms).
I Mere capital ratios, and increasing them incrementally, do nothelp to prevent credit bubbles.
I Imposing strict limits to growth on the excessive supply ofcredit seems to work best to mitigate the severity of economicdownturns.
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
Thank you for your attention!
Model documentation:
www.wiwi.uni-bielefeld.de/lehrbereiche/vwl/etace/Eurace_Unibi/Papers:
I S van der Hoog & H Dawid (2015):Bubbles, Crashes and the Financial Cycle, Working Paper Bielefeld University.
I H Dawid, S Gemkow, P Harting, S van der Hoog & M Neugart (2014):Agent-Based Macroeconomic Modeling and Policy Analysis: The Eurace@UnibiModel. In: S-H Chen, M Kaboudan (Eds), Handbook on ComputationalEconomics and Finance. Oxford University Press.
I H Dawid, S Gemkow, P Harting, S van der Hoog & M Neugart (2012):The Eurace@Unibi Model: An Agent-Based Macroeconomic Model for EconomicPolicy Analysis. Working Paper University Bielefeld.
I H Dawid, S Gemkow, P Harting, S van der Hoog & M Neugart (2011):Eurace@Unibi Model v1.0 User Manual. Working Paper Bielefeld University.
I H Dawid & P Harting (2012): Capturing Firm Behavior in Agent-Based Models ofIndustry Evolution and Macroeconomic Dynamics, in: G. Bünstorf (Ed), AppliedEvolutionary Economics, Behavior and Organizations. Edward Elgar, pp.103-130.
I H Dawid & M Neugart (2011): Agent-based Models for Economic Policy Design,Eastern Economic Journal 37, 44-50.
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Outlook & Future research
I Macroprudential regulation
I Systemic risk (SIFIs, SIBs)
I Bank-firm networksI size effectsI balance sheet contagion
I Empirically-grounded bank behavior
I Credit quotasI Credit rationing of SMEsI Tighter integration of Basel III regulation
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Scenario: Capital Adequacy Requirement
Output
0 100 200 300 400 500
1500
2000
2500
3000
Months
Eur
osta
t_ou
tput
alfa
2.0 8.0
Bank activity (α = 2)
0 100 200 300 400 500
05
1015
20
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Firm activity (α = 2)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Scenario: Minimum Reserve Requirement
Output
0 100 200 300 400 500
2000
2200
2400
2600
2800
Months
Eur
osta
t_ou
tput
min_cash_reserve_ratio
0.10 0.50
Bank activity (β = 0.50)
0 100 200 300 400 500
05
1015
20
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Firm activity (β = 0.50)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
Scenario: Capital Adequacy RequirementOutput
0 100 200 300 400 500
1500
2000
2500
3000
Months
Eur
osta
t_ou
tput
alfa
2.0 8.0
Bank activity (α = 2)
0 100 200 300 400 5000
510
1520
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Firm activity (α = 2)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
0 100 200 300 400 500
050
0010
000
1500
020
000
Months
Ban
k_eq
uity
alfa
2.0 8.0
Bank equity
0 100 200 300 400 500
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Months
F_E
AR
atio
alfa
2.0 8.0
Firm fragility
0 100 200 300 400 5000.
040
0.04
50.
050
0.05
50.
060
0.06
50.
070
Months
Fir
m_m
ean_
inte
rest
alfa
2.0 8.0
Mean interest
Scenario: Minimum Reserve RequirementOutput
0 100 200 300 400 500
2000
2200
2400
2600
2800
Months
Eur
osta
t_ou
tput
min_cash_reserve_ratio
0.10 0.50
Bank activity (β = 0.50)
0 100 200 300 400 5000
510
1520
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Firm activity (β = 0.50)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
0 100 200 300 400 500
5000
1000
015
000
2000
0
Months
Ban
k_eq
uity
min_cash_reserve_ratio
0.10 0.50
Bank equity
0 100 200 300 400 500
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Months
F_E
AR
atio
min_cash_reserve_ratio
0.10 0.50
Firm fragility
0 100 200 300 400 5000.
045
0.05
00.
055
0.06
00.
065
0.07
0
Months
Fir
m_m
ean_
inte
rest
min_cash_reserve_ratio
0.10 0.50
Mean interest
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Firm activity
Number of illiquid firms
No constraint
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
Capital constraint (α = 2)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
LFirm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
Liquidity constraint (β = 0.50)
0 100 200 300 400 500
05
1015
20
Months
Fir
m_i
nsol
venc
y_S
L
Firm_insolvency_SFirm_insolvency_L
Firm_illiquidity_SFirm_illiquidity_L
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Bank activity
Number of active banks (unconstrained + constrained by equity/liquidity constraint)
No constraint
0 100 200 300 400 500
05
1015
20
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Capital constraint (α = 2)
0 100 200 300 400 500
05
1015
20
Months
Ban
k_ac
tive_
mul
tiBank_active_none Bank_active_exposure Bank_active_liquidity
Liquidity constraint (β = 0.5)
0 100 200 300 400 500
05
1015
20
Months
Ban
k_ac
tive_
mul
ti
Bank_active_none Bank_active_exposure Bank_active_liquidity
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Scenarios: Firm Fragility
Firm E/A-ratio = 1/leverage
Capital constraint
0 100 200 300 400 500
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Months
F_E
AR
atio
alfa
2.0 8.0
Liquidity constraint
0 100 200 300 400 500
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Months
F_E
AR
atio
min_cash_reserve_ratio
0.10 0.50
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Prevention and mitigation - Bank dividend payout−
3000
−25
00−
2000
−15
00−
1000
−50
0
Parameters
full_
ampl
itude
_rec
essi
on
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
−30
00−
2500
−20
00−
1500
−10
00−
500
amplitude of recessions
−10
000
−80
00−
6000
−40
00−
2000
Parameters
full_
cum
m_l
oss_
rece
ssio
n
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90
−10
000
−80
00−
6000
−40
00−
2000
cumulative loss
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Bank accounting
1. Bank profit
πbt = rb
i Lbi − rb(∑h Mb
h +∑i Mbi ) + rECB(Mb
t −Dbt )
2. Bank cash and reserves
Mbt+1 = Mb
t + ∆Mbh + ∆Mb
i + (1− τ)max[0,πbt ]−db(1− τ)max[0,πb
t ]
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Debt-equity transformation
3a. Insolvency bankruptcy
Debt renegotiation is addressed by re-scaling the total debt Dft with a debt rescaling
parameter ϕ.Target debt is given by:
D∗ = ϕAft with 0≤ ϕ ≤ 1 . (8)
After debt restructuring, the equity of the firm is now positive:
E∗ = (1−ϕ)Aft > 0. (9)
The new debt/equity-ratio is given by the constant D∗/E∗ = ϕ/(1−ϕ) < 1.
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Debt-equity transformation
3b. Illiquidity bankruptcy
Debt-renegotiation is not necessary per se, rescaling of the debt is either based on thelevel of total assets or on the level of the original debt:
D∗ =
{ϕAf
t if ϕAft ≤ Df
tϕDf
t if ϕAft > Df
t .with 0≤ ϕ ≤ 1 . (10)
The new debt/equity-ratio is given by the following piece-wise function:
D∗/E∗ =
{ϕ/(1−ϕ) if ϕAf
t ≤ Dft
ϕ/(A/D−ϕ) if ϕAft > Df
t .(11)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Dividend payout rule
I 〈Rf 〉nR : average revenues over previous nR months (nR = 3,6,12)
I 〈Πf 〉nE : average net earnings (after-tax profits) over the last nE months
〈Rf 〉nR =1
nR
nR−1
∑i=0
Rft−i (12)
〈Πf 〉nE =1
nE
nE−1
∑i=0
Πft−i (13)
I Prevent liquidity hoarding by firms: Liquidity Buffer Stock
4. Dividend payout rule:
Div f =
{d · 〈Πf 〉4 if M f
t ≤ µ · 〈Rf 〉6〈Πf 〉4 if M f
t > µ · 〈Rf 〉6.d = 0.7,µ = 0.5 (14)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle
The Big QuestionsEurace@Unibi Model
Simulation ResultsConclusions
Exogenous Credit Rationing
5a. Full/Partial credit rationing is based on the (exogenously prescribed, ex ante)constraints of the bank (CAR, CRR).
I Full rationing for CAR constraint:
¯̀bit =
Lit if PDit ·Lit ≤ V bt
0 if 0≤ V bt ≤ PDit ·Lit
0 if V bt < 0.
(15)
I Partial rationing ("filling up to constraint") for CAR constraint:
¯̀bit =
Lit if PDit ·Lit ≤ V bt
V bt /PDit if 0≤ V b
t ≤ PDit ·Lit0 if V b
t < 0.(16)
Sander van der Hoog Bubbles, Crashes & the Financial Cycle