Sandia National Laboratories Performance and resilience to liquidity disruptions in interdependent RTGS payment systems Fabien Renault 1 Morten L. Bech 2 Walt Beyeler 3 Robert J. Glass 3 Kimmo Soramäki 4 1 Banque de France 2 Federal Reserve Bank of New York 3 Sandia National Laboratories 4 Helsinki University of Technology Joint Banque de France / European Central Bank conference on "Liquidity in interdependent transfer systems" Paris, 9 June 28 The views expressed in this presentation are those of the authors and do not necessarily reflect those of their respective institutions 1
Motivation The 21 Group of Ten Report on Consolidation in the Financial Sector (the Ferguson report) noted a possible increased interdependence between the different systems due to: The emergence of global institutions that participate to many systems The emergence of global service providers offering services to many systems The development of DvP procedures linking RTGS and SSS The development of CLS The report suggested that these trends might accentuate the role of payment and settlement systems in the transmission of disruptions across the financial system. To complement this previous work, the CPSS (Committee on Payment and Settlement Systems) commissioned a working group to: describe the different interdependencies existing among the payment and settlement systems of CPSS countries analyze the risk implications of the different interdependencies 2
Motivation Could a modeling approach provide any useful additional information to the regulators? So far, payment and settlement system modeling has been mainly limited to a single system, with a few exceptions We want to model the interactions between two payment systems and understand how interdependencies arise We wish to understand how disruptions in one system manifest in the other System-based Interdependencies Institution-based Interdependencies Environmental Interdependencies System System System System System System Dealt with in this presentation Financial Institution IT service provider 3
Coupled RTGS model Model description Two RTGS systems in two different currencies: $ and RTGS $ Extended core A 5 A 23 A i A 97 Both systems are similar in structure with 1 banks A2 A 6 A 1 A 2 A 3 Core E1 E 2 E 3 A 4 A22 A 21 Smaller $ local players Six global banks. Top three banks in each system have a presence also in the other system FX market The global banks make FX trades (at constant exchange rate) among each other E 2 A1 A 2 Core E 1 E 2 A 3 E4 E i E 97 E 23 E 22 Smaller local players All banks make local payments E 3 E 6 Extended core E 5 RTGS E 21 4
Coupled RTGS model Model description Payment instructions arrive according to a non-homogenous Poisson process intuition: customers who have received funds issue payments more frequently than bank customers who have already sent many payments FX trades arrival is similar as above, now taking into account balances in both currencies E.g. banks with high euro positions are likely to sell euro and vice versa Those two systems are linked Via the dual participation of some global banks that can make FX trades (institution-based interdependency) Via a possible PvP (Payment versus Payment) constraint on the FX trades (system-based interdependency), the alternative being a non-pvp settlement 5
Correlation dynamics Local $ payment orders High liquidity PvP or non-pvp FX trades $ legs of FX trades legs of FX trades Correlation:.22 (institution-based interdependency) Local payment orders Local $ payment orders $ legs Congestions and cascades Settled payments Low liquidity non-pvp FX trades legs Local payment orders Congestions and cascades Settled payments Correlation: -.2 (none) Low liquidity PvP Local $ payment orders $ legs FX trades legs Local payment orders Congestions and cascades Congestions and cascades Settled payments PvP link Settled payments Correlation:.83 (system-based interdependency) 6
Summary of main results PvP increases queues eliminates exposures Lower liquidity increases queues Increases exposures (in case of non-pvp) Liquidity differences in the two systems Reducing liquidity in one system increases queuing in the other (in case of PvP) Banks selling the more liquid currency face higher exposures (in case of non- PvP) Higher priority for FX trades Decreases queues in the more liquid system (in case of PvP) Does not affect queues when both systems have same liquidity Substantially reduces exposures (in case of non-pvp) 7
An operational disruption affects a significant local bank The affected bank does not participate in RTGS $, nor engage in FX transactions The affected bank is unable to submit its local payments for a certain duration The affected bank acts as a liquidity sink for RTGS To which extent will the disruption affect RTGS $? Four different cases: PvP or non-pvp High Liquidity or Low Liquidity (the same in both systems) What are the channels of propagation through which the crisis spreads from one RTGS to the other? 8
2 A B C D E F G H 18 Settlem ent Rate (Payments/Tim e) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity 9
2 A B C D E F G H 18 Settlem ent Rate (Payments/Tim e) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) 1
2 A B C D E F G H 18 Settlem ent Rate (Payments/Tim e) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots 11
2 18 A B C D E F G H Period A Steady state 16 Settlem ent Rate (Payments/Tim e) 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots 12
Settlem ent Rate (Payments/Tim e) 2 18 16 14 12 1 8 6 4 A B C D E F G H Period A Steady state Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots 13
Settlem ent Rate (Payments/Tim e) 2 18 16 14 12 1 8 6 4 A B C D E F G H Period A Steady state Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity Period C RTGS $ down to local activity 2 9 1 11 12 13 14 15 PvP, High Liquidity PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots 14
Settlem ent Rate (Payments/Tim e) A B C D E F G H 2 18 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP High Liquidity PvP, High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots Period A Steady state Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity Period C RTGS $ down to local activity Period D Because of the queuing of FX trades (PvP), customers have lower $ funds and make fewer $ local payments Period E-F Queued local payments settle Queued FX trades settle Period G Return to equilibrium generates extra trades 15
Settlem ent Rate (Payments/Tim e) A B C D E F G H 2 18 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP High Liquidity PvP, High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots Period A Steady state Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity Period C RTGS $ down to local activity Period D Because of the queuing of FX trades (PvP), customers have lower $ funds and make fewer $ local payments Period E-F Queued local payments settle Queued FX trades settle Period G Return to equilibrium generates extra trades 16
Settlem ent Rate (Payments/Tim e) A B C D E F G H 2 18 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP High Liquidity PvP, High Liquidity Outage: settlement rate in RTGS $ decreases (-5 %) Recovery: settlement rate in RTGS $ overshoots Period A Steady state Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity Period C RTGS $ down to local activity Period D Because of the queuing of FX trades (PvP), customers have lower $ funds and make fewer $ local payments Period E-F Queued local payments settle Queued FX trades settle Period G Return to equilibrium generates extra trades 17
2 A B C D E 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, Low Liquidity PvP Low Liquidity 18
2 A B C D E 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, Low Liquidity PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-65 %) 19
2 A B C D E 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 PvP, Low Liquidity PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-65 %) Recovery: settlement rate in RTGS $ overshoots (reaches maximum rate) 2
Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 A B C D E 9 1 11 12 13 14 15 PvP, Low Liquidity PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-65 %) Recovery: settlement rate in RTGS $ overshoots (reaches maximum rate) Period B balances vanish local payments are queued Both legs of FX trades are queued, RTGS $ deprived of FX activity Period C The queuing of FX trades decreases $ deposits. Agents are uncertain about their $ position, fewer $ local payments emitted The distribution of $ deposits is brought out of equilibrium because of the disruption. In this low liquidity context, this causes $ local payments to be queued Period D Queued local payments settle Queued FX trades settle Queued $ local payments settle Period E 21 Return to equilibrium marginally affects settlement rate
2 A B C D E F G 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity 22
2 A B C D E F G 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-17 %) Recovery: no overshoot in RTGS $ 23
2 A B C D E F G 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-17 %) Recovery: no overshoot in RTGS $ 24
Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 A B C D E F G 9 1 11 12 13 14 15 Non-PvP High Liquidity Outage: settlement rate in RTGS $ decreases (-17 %) Recovery: no overshoot in RTGS $ non-pvp, High Liquidity Period B balances vanish local payments are queued leg of FX trades are queued $ leg of FX trades still settle Period C The queuing of local payments decreases deposits. Agents are uncertain about their position, fewer FX trades emitted. RTGS $ is deprived from FX activity Period D Only local activity in RTGS $ Period E Queued local payments settle Queued leg of FX trades settle Period F Return to equilibrium generates extra trades 25
A B C D E F 2 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, Low Liquidity Non-PvP Low Liquidity 26
A B C D E F 2 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, Low Liquidity Non-PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-25 %) 27
A B C D E F 2 18 Settlement Rate (Payments/Time) 16 14 12 1 8 6 4 2 9 1 11 12 13 14 15 non-pvp, Low Liquidity Non-PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-25 %) Recovery: settlement rate in RTGS $ overshoots 28
Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 A B C D E F 9 1 11 12 13 14 15 non-pvp, Low Liquidity Non-PvP Low Liquidity Outage: settlement rate in RTGS $ decreases (-25 %) Recovery: settlement rate in RTGS $ overshoots Period B balances vanish local payments are queued leg of FX trades are queued $ leg of FX trades still settle Period C The queuing of local payments decreases deposits. Agents are uncertain about their position, fewer FX trades emitted. RTGS $ is deprived from FX activity Period D Only local activity in RTGS $ The distribution of $ deposits is brought out of equilibrium because of the disruption. In this low liquidity context, this causes $ local payments to be queued Period E Queued local payments settle Queued FX trades settle 29 Queued $ local payments settle
Exposure (Dollars or Euros Owed) 3 25 2 15 1 5 Dollars Owed Euros Owed (maximum at 11.) 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity, FX exposures 3
Exposure (Dollars or Euros Owed) 3 25 2 15 1 5 Dollars Owed Euros Owed (maximum at 11.) 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity, FX exposures Outage: huge increase in owed (1 times normal exposures) 31
Exposure (Dollars or Euros Owed) 3 25 2 15 1 5 Dollars Owed Euros Owed (maximum at 11.) 9 1 11 12 13 14 15 non-pvp, High Liquidity Non-PvP High Liquidity, FX exposures Outage: huge increase in owed (1 times normal exposures) Recovery: large increase in $ owed (15 time normal exposure) 32
Cross-currency channels of disruption propagation Channel 1: Low balances at the CB prevent settlement of leg of FX transactions (PvP) or create FX exposures (non-pvp) 33
Cross-currency channels of disruption propagation Channel 1: Low balances at the CB prevent settlement of leg of FX transactions (PvP) or create FX exposures (non-pvp) Settlem ent Rate (Payments/Tim e) 2 18 16 14 12 1 8 6 4 PvP: All FX settlement activity stops, RTGS $ is deprived from FX settlement activity 2 9 1 11 12 13 14 15 PvP, High Liquidity 34
Cross-currency channels of disruption propagation Channel 1: Low balances at the CB prevent settlement of leg of FX transactions (PvP) or create FX exposures (non-pvp) Settlem ent Rate (Payments/Tim e) 2 18 16 14 12 1 8 6 4 PvP: All FX settlement activity stops, RTGS $ is deprived from FX settlement activity Because of the queuing of FX trades (PvP), customers have lower $ funds and make fewer $ local payments 2 9 1 11 12 13 14 15 PvP, High Liquidity 35
Cross-currency channels of disruption propagation Channel 1: Low balances at the CB prevent settlement of leg of FX transactions (PvP) or create FX exposures (non-pvp) Exposure (Dollars or Euros Owed) 3 Dollars Owed Euros Owed (maximum at 11.) 25 2 15 1 5 9 1 11 12 13 14 15 non-pvp, High Liquidity PvP: All FX settlement activity stops, RTGS $ is deprived from FX settlement activity Because of the queuing of FX trades (PvP), customers have lower $ funds and make fewer $ local payments Non-PvP: Very high exposures (unsettled legs) during crisis 36
Cross-currency channels of disruption propagation Channel 2: Low customer funds lead to fewer emitted FX trades Banks customers liquidity is trapped within queued payments. Therefore, customers emit fewer FX trades 37
Cross-currency channels of disruption propagation Channel 2: Low customer funds lead to fewer emitted FX trades Banks customers liquidity is trapped within queued payments. Therefore, customers emit fewer FX trades Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 This leads to RTGS $ being eventually deprived of FX activity, even in the non-pvp case 9 1 11 12 13 14 15 non-pvp, High Liquidity 38
Cross-currency channels of disruption propagation Channel 3: As not all banks are similarly affected, the system becomes unbalanced The FX banks for which the disrupted bank is an important counterparty see their level of customer funds decrease more rapidly. These banks become net buyers ($ sellers) on the FX market. RTGS $ becomes unbalanced. 39
Cross-currency channels of disruption propagation Channel 3: As not all banks are similarly affected, the system becomes unbalanced The FX banks for which the disrupted bank is an important counterparty see their level of customer funds decrease more rapidly. These banks become net buyers ($ sellers) on the FX market. RTGS $ becomes unbalanced. Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 Low Liquidity: this leads to the queuing of several $ local payments, even in the non-pvp case 9 1 11 12 13 14 15 non-pvp, Low Liquidity 4
Cross-currency channels of disruption propagation Channel 3: As not all banks are similarly affected, the system becomes unbalanced The FX banks for which the disrupted bank is an important counterparty see their level of customer funds decrease more rapidly. These banks become net buyers ($ sellers) on the FX market. RTGS $ becomes unbalanced. Settlement Rate (Payments/Time) 2 18 16 14 12 1 8 6 4 2 Low Liquidity: this leads to the queuing of several $ local payments, even in the non-pvp case And to an overshoot at recovery, even in non- PvP case 9 1 11 12 13 14 15 non-pvp, Low Liquidity 41
Cross-currency channels of disruption propagation Channel 3: As not all banks are similarly affected, the system becomes unbalanced The FX banks for which the disrupted bank is an important counterparty see their level of customer funds decrease more rapidly. These banks become net buyers ($ sellers) on the FX market. RTGS $ becomes unbalanced. Exposure (Dollars or Euros Owed) 3 Dollars Owed Euros Owed (maximum at 11.) 25 2 15 1 5 9 1 11 12 13 14 15 non-pvp, High Liquidity Low Liquidity: this leads to the queuing of several $ local payments And to an overshoot at recovery, even in non- PvP case Non-PvP: This creates a peak in $ owed exposures at recovery, event at high 42 liquidity
Conclusions A simple model of interconnected RTGS was developed During normal operation, the two RTGS are shown to be interdependent When a liquidity crisis affects one RTGS, the crisis propagates to second RTGS in all considered cases PvP: sharp decrease in activity (local and FX) in second RTGS Non-PvP: Decrease in activity in second RTGS due to fewer FX trades emitted At low liquidity, local payments in second RTGS are also affected Large increase of FX exposures during crisis and recovery 43