endogenous derivatives networks. - banque de france fileendogenous derivatives networks. guillaume...
TRANSCRIPT
![Page 1: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/1.jpg)
Endogenous Derivatives Networks.
Guillaume Vuillemey 1,2 Regis Breton 2
1Sciences-Po, Department of Economics
2Banque de France
ACPR - Banque de France - SoFIE, July 2014
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 2: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/2.jpg)
Disclamer
Any opinions expressed are only the presenter's own and should not
be regarded as opinions of the Bank of France or the Eurosystem.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 3: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/3.jpg)
This paper
Question: How is the network for OTC derivatives likelyto change as a consequence of the OTC derivativesmarket reform?
Central clearing set to become mandatory (Dodd-Frank,EMIR).First channel: Increased demand for safe collateral assets tomeet increased margin requirements.Neglected channel: Changes in network size and structure.
The model
OTC derivatives network formation model.Both prices and quantities are negociated.Both idiosyncratic counterparty risk and (exogenous)collateralization and clearing requirements are accountedfor in banks' decision to trade.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 4: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/4.jpg)
Motivation
Same network of net CDS exposures.
No central clearing (left chart), full central clearing (rightchart).
0
1
2
3
4
5
67
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
3334
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104105
106
0
1
2
3
4
5
6
78
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
3435
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105106
107
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 5: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/5.jpg)
CCP clearing and collateral requirements
Wide range of estimates for required additional marginrequirements.
ISDA (2011): 1000 bn USD for IR swaps.IMF (2012): 100-200 bn USD for all OTC derivatives.Heller and Vause (2012): 720 bn USD for IR swaps and CDS.Due, Scheicher and Vuillemey (2014): Estimates for severalmarket structures.
Main limitation of previous studies: Neglect of the fact
that the network of exposure is endogenous to the
collateralization and clearing requirements.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 6: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/6.jpg)
CCP clearing and collateral requirements
Focus instead on changes in the market structure.
The net benet of opening a derivative exposure depends onthe relative weighting of costs and benets of collateral andclearing.Costs: Exposures that have the lower net expected marginalbenet may disappear if increased collateral requirements areimposed ceteris paribus.Benets: Or, alternatively, increased collateral requirementsmay induce some counterparties to trade more, as counterpartyrisk is then mitigated.
Results underpinned by bank trading behaviour
What if counterparty risk increases?What if the safe asset becomes scarcer?
Rough assessment of collateral demand mis-estimation when
network exogeneity is assumed.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 7: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/7.jpg)
Contribution
New model of OTC derivatives network formation.
Accounts for features of derivative networks (upfront payment,o-balance sheet exposure).Collateralization and clearing requirements enter banks'objective functions.Counterparty credit risk as a key ingredient.Both prices and quantities negociated.
Investigate the eect of three regulatory schemes.
Increased collateralization of bilateral exposures.Rehypothecation or not of collateral received.Increased scope for central clearing.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 8: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/8.jpg)
Preview of the results
Main results.
Market liquidity / intermediary activity decreases with highercollateral requirements.Some end-users are no longer able to obtain credit riskprotection.Rehypothecation does not change the ability of end-users toobtain credit risk protection, but increases market liquidity.Overall: Redistributive eects.
Additional results.
Increased counterparty risk reduces market size to a largeextent. Collateralization can mitigate the market shrinkage.A drop in the safe asset supply has a rst-order eect onmarket outcomes. Market changes due collateralizationrequirements are second-order in this context.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 9: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/9.jpg)
Related literature
Endogenous network formationNetwork models: Acemoglu et al. (2013), Battiston et al.(2012), Bluhm et al. (2013), Babus (2013).
Reviews by Dutta and Jackson (2003), Jackson and Zenou(2013).
OTC markets: Due et al. (2005, 2007).
Central clearing
Theoretical: Acharya and Bisin (2013), Biais et al. (2012,2013), Koeppl and Monnet (2010), Due and Zhu (2011).Empirical: Heller and Vause (2012), Sidanius and Zikes(2012), Singh (2013), Due et al. (2014).
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 10: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/10.jpg)
Model - Notations
Set Ω = 1, ..., n of nancial institutions indexed by i .
Three dates t = 0, 1, 2.Stylized balance sheet of i ∈ Ω.
Riskless securities ci .
Gross yield r0 > 1 at t = 2 with probability one.
Risky credit exposures ai .
Gross yield R > r0 > 1 at t = 2 with probability (1− δ1).Gross yield 0 with probability δ1.δ1 observed only at t = 1.Imperfect signal δ0 = E [δ1] at t = 0.
The portfolio ai , ci is exogenously given at t = 0.
Vector of idiosyncratic bank probabilities of defaultΛ = λ1, ..., λn.
Provides a rationale for collateralizationAssumption: Independence of any λi and λj .The joint probability of i and j failing at t = 2 is λiλj .
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 11: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/11.jpg)
Model - CDS contracts
Assumption: The risky asset a is illiquid and in xed supply.
Thus, instead of trading a for c, c is traded against creditderivatives written on a.
Consider CDS contracts on reference entity a.
OTC contract between i and j dened by the tuple ωij , pij.ωij ∈ R: Gross notional amount.
ωij > 0 in case i sells the contract.ωij = −ωji
pij > 0: Premium per unit of notional
pij (xed leg) paid at t = 0 by the buyer.
Proceeds invested in the safe asset, yielding ωjipij r0 at t = 2.
In case of default at t = 2 (w. p. δ1), the seller j pays −ωijR .
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 12: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/12.jpg)
Model - Timing
Initial period: t = 0
Each bank i observes its portfolio ai , ci.δ0 ∈ [0; 1] is observed and the p.d.f. of δ1, denoted π (δ1)(with c.d.f. Π), is common knowledge.Banks are matched and credit derivatives are traded over alarge number of subperiods s = 1, ...,S.Initial margins are posted (with the riskless asset c).
Interim period: t = 1
True asset probability of default δ1 is observed.Change from δ0 to δ1 gives ground for variation marginsposting.
Final period: t = 2
All assets mature.In case a defaults, CDS payments take place.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 13: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/13.jpg)
Model - Trade without collateral
Banks mawimize VNM expected utility V = E [U].Utility of a bank when it fails normalized to U(0).Expected utility for i when being oered a contract ωij , pijwith j :
Vi (ωij , pij) =
∫ 1
0
π (δ1)[λiU (0) + (1− λi)
[(1− δ1)U (Ci + aiR)
+δ1[(1− λj)U (Ci − ωijR)
+λjU (Ci −max 0;ωijR)]]dδ1
With Ci ≡ (ci + ωijpij) r0.Both asset default probability and counterparty defaultprobability are accounted for.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 14: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/14.jpg)
Model - Trade without collateral
Both ωij and pij are negociated.
Any bank i can thus be both buyer and seller.
Objective: Nash product with equal bargaining power:
ωij , pij ∈ arg max (Vi (ωij , p)− Vi (0, .)) (Vj (ωji , p)− Vj (0, .))
Subject to budget and incentive-compatibility constraints :−pijωij ≤ ci(cj + pijωji ) r0 ≥ ωjiR
Vi (ωij , pij)− Vi (0, .) ≥ 0
Vj (ωji , pij)− Vj (0, .) ≥ 0
Note: Naked CDS buying is allowed.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 15: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/15.jpg)
Model - Matching
Period t = 0 is divided into a large number S of sub-periods
indexed by s.
At each s, a randomly chosen pair of banks is matched.
Preferential attachment is modeled and parameterized.
Fixed cost κ of creating a rst link with a counterparty.
Limited rationality: A bank considers each match as the last
trading opportunity before S .
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 16: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/16.jpg)
Model - Collateral requirements
Rationale for collateralization: idiosyncratic probabilities of
default Λ.
Collateralization requirements are exogenously given.Initial margins
Meant to cover potential future exposure.
χIMji = τ IM f IM · σ (δ1)max 0;ωji
τ IM ∈ [0; 1]: Collateralization level.σ (δ1): Standard deviation of δ1.f IM : Constant.
Variation marginsMeant to cover current exposure.
χVMji = τVM f VM · (δ1 − δ0)max 0;ωji
f VM : Constant.
Failures from collateral shortage (inability to post variation
margins) are ruled out through an additional constraint.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 17: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/17.jpg)
Model - Sequential trading problem
Rene notations:ωij : CDS contract currently being negociated between i and j .ωsij : sum of all (adding up or osetting) agreed-upon trades between i and j
before instant s.
Expected utility for i at date s when matched with j :
Vi (ωij , pij) =
∫ 1
0
π (δ1)
[λiU (0) + (1− λi)
[(1− δ1)U
(Ci + aiR + χi
)
+δ1
[ ∑Φ⊆Ω\i
∏k∈Φ
λk∏k /∈Φ
(1− λk) · U
(Ci
+∑k∈Φ
[−max 0; ωsik+ χki ]−
∑k /∈Φ
[ωsikR + χik ]
)]]]dδ1
Some simplifying notations have been introduced.More on notations.
Budget constraints are also updated with the trading process s.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 18: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/18.jpg)
Calibration
Banks
n = 5 banks (main dealers in one area).U: CRRA utility with elasticity of substitution γ = 0.3.Total bank assets drawn from a Power law distribution.Random assignement of shares β and (1− β) for ai and ci .Bank probabilities of default drawn from a beta distribution(support [0; 1]) calibrated from credit ratings.
Assets
r0 = 1.029 (10-Year U.S. Treasury Constant Maturity Rate)R = 1.158 (Long-term arithmetic average of S&P 500 return).
Collateral
f IM = 10 (Potential future exposure)f VM = 2 (Current exposure)
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 19: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/19.jpg)
Outcome metrics
Restrict attention to key metrics.
Gross notional amount: Size of the market.
Multilateral net notional: Exposure at default
Net over gross ratio: Extent to which trading activity
generates exposure at default. Proxies for market liquidity.
Total collateral and collateral over net notional ratio:Counterparty risk mitigation.
Average volume-weighted price: Pricing of the contracts.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 20: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/20.jpg)
Results - Bilateral clearing, varying τ
Outcome metrics for changing collateralization levels on
bilateral trades.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 21: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/21.jpg)
Results - Bilateral clearing, varying τ
Gross market size shrinks by 21%, but net exposure decreasesto a lower extent (13%).
The net-over-gross ratio increases: Market liquidity /intermediary activity are lower.Some end-users no longer get credit risk protection.Thus, sizeable eects for both intermediaries and end-users.
Increased collateral requirements are priced: higher spreads
paid by CDS buyers.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 22: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/22.jpg)
Results - Rehypothecation
Rehypothecation: re-use of collateral.
Common practice on OTC derivative markets (Singh 2010)
Assume a fraction ρ ∈ [0; 1] of collateral received can be
rehypothecated.
At t = 0, rehypothecation of initial margins.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 23: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/23.jpg)
Results - Rehypothecation
Outcome metrics with bilateral trades and changing
rehypothecation, τ IM = τVM = 1.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 24: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/24.jpg)
Results - Rehypothecation
Rehypothecation increases gross market size by about 13.7%,but net notional remains unchanged.
The ability to rehypothecate does not increase the ability ofend-users to get credit risk protection.But intermediary activities increase to a large extent.
Collateral over net ratio drops (from 0.21 to 0.12).Policy debates on rehypothecation should weigh the costs andbenets of extensive collateralization and of well-developedintermediary activities.Trading volumes by end-users are not a rst-order concern.
Prices decrease slightly as rehypothecation increases.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 25: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/25.jpg)
Results - CCP clearing
Bilateral exposures satisfying a size criterion are novatedto a CCP.
All exposures above T have to be centrally cleared.Size criterion T can be interpreted as a proxy forstandardization.
Example: Novation of ωij < 0 to a CCP.
The CCP buys ωCCP,j = ωij from j .The CCP sells −ωCCP,j to i .
Multilateral netting of exposures novated to the CCP.
Collateral required by the CCP:
Full collateralization : τ IM = τVM = 1The CCP does not post initial margins.Default fund contribution: fraction ψ of initial margins.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 26: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/26.jpg)
Results - CCP clearing
Outcome metrics when increasing CCP clearing (reducing T )
Uncleared bilateral trades uncollateralized, τ IM = τVM = 0.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 27: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/27.jpg)
Results - CCP clearing
Eects very similar to those of increased bilateralcollateralization.
Gross market size decreases by 34.1%.Some end-users trade less in net termsNet over gross ratio increases from 0.52 to 0.66 (lower marketliquidity).Collateral demand increases to represent 18.2% of netexposures.Increased collateral requirements are priced.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 28: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/28.jpg)
Results - Calibrated example
Run an experiment with a calibration close to pre-reformparameter values.
Pre-reform case:
No central clearing.No bilateral initial margins.Variation margins τVM = 0.88 (ISDA 2013).Rehypothecation ratio ρ = 0.75 (ISDA 2013).
Post-reform case:
Full CCP clearing: T = 0.Default fund contribution ψ = 0.15 (Ra 2012).
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 29: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/29.jpg)
Results - Calibrated example
Not accounting for dynamic eects may lead to an
over-estimation of system-wide collateral demand between
22% and 47%.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 30: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/30.jpg)
Extensions - Probabilities of default
Role of bank probabilitities of default : What if all bankprobabilities of default are driven up by an idiosyncraticfactor?
Benets of collateral posting are larger ceteris paribus.
Low impact of bank PD if full collateralization.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 31: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/31.jpg)
Extensions - Probabilities of default
Benets of central clearing are also larger ceteris paribus.Dierences due to counterparty risk almost vanish whencentral clearing is widespread.
Lower incentives to screen counterparties.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 32: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/32.jpg)
Extensions - Safe asset supply
Keep each bank size constant, reduce the proportion of safeassets.
Akin to the downgrade of safe assets (≈ safe assetshortage).Blue: base case, red: ÷2, green: ÷4.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 33: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/33.jpg)
Extensions - Safe asset supply
Large consequences on market size, in particular netexposures.
Portfolio rebalancing: Traders less willing to take on morecredit risk when risky assets are in larger supply.
Much higher net over gross notional ratio when the
collateralization level is high enough.
However, the collateral over net ratio decreases.
First-order eect on CDS prices, which increase
signicantly.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 34: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/34.jpg)
Conclusion
Dynamic model of OTC derivatives network formation.
Network is endogenous to collateralization and clearingrequirements.Idiosyncratic counterparty risk accounted for.Both prices and quantities are negociated.
Dynamic eects of collateral and clearing on OTCderivatives market size and structure are large.
Reduction in both market liquidity and access to the marketfor end-users.Between 22% and 47% over-estimation of collateral demand incase dynamic eects are not accounted for.Part of the costs of collateral and clearing are passed ontobuyers through higher prices.Sizeable consequences of a change in safe asset supply.
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 35: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/35.jpg)
Further research
Model with multiple assets
Asset default correlation likely to play a role if initial andvariation margins are properly modeled.Benets from diversication
Robustness of simulated networks to contagion
Contagion when accounting for collateral and clearing.Quantify the risk-mitigating role of collateral.
Normative (welfare) analysis
Increased collateralization needs not be welfare-improving.Nature of CDS contracts that are no longer traded: truehedging needs or other trades?
Vuillemey and Breton (2014) Endogenous Derivatives Networks
![Page 36: Endogenous Derivatives Networks. - Banque de France fileEndogenous Derivatives Networks. Guillaume Vuillemey 1, 2 Regis Breton 2 1 Sciences-Po, Department of Economics 2 Banque de](https://reader033.vdocuments.net/reader033/viewer/2022050718/5e17ce397825df72153e0614/html5/thumbnails/36.jpg)
Appendix - Notations in the sequential problem
Notations introduced in setting up the sequential problem.
Ci ≡
(ci − Ki − χi + ωijpij +
∑k
ωsikpik
)r0
Ki ≡ κ∑k
1Ξs(i ,k)=1
χi ≡ χIMi
(ωij +
∑k
ωsik
)+ χVM
i
(ωij +
∑k
ωsik , δ1
)χij ≡ χIM
ij + χVMij
Return.
Vuillemey and Breton (2014) Endogenous Derivatives Networks