Liquidity saving mechanisms and bank behaviour
1 like362 views
The document discusses liquidity saving mechanisms (LSMs) and their impact on interbank payment systems, particularly the real-time gross settlement (RTGS) mode, highlighting the role of internal queues and liquidity/delay trade-offs. It presents an agent-based model evaluating how banks can optimize their strategies regarding payment urgency and liquidity management. The findings suggest that LSMs can significantly reduce liquidity costs and improve settlement speed, but their effectiveness may require coordination among banks.
![Model of a payment system
• N banks
• A ‘day’ of several ‘seconds’
• Random payment orders
random pairs ‘payer&payee’
for each payment a ‘urgency’ parameter u ~ U[0,1]
• Each bank sends payment orders in either of two ‘pipes’
(streams): RTGS or Queue](https://image.slidesharecdn.com/galbiatipresentationabm-baf09-091125174827-phpapp01/85/Liquidity-saving-mechanisms-and-bank-behaviour-6-320.jpg)
Liquidity saving mechanisms and bank behaviour
- 1. Liquidity saving mechanisms and bank behaviour Marco Galbiati – BoE Kimmo Soramäki – VerticeTree ABM-BaF Torino - 10 February 2009
- 2. Interbank payment systems Real Time Gross Settlement (RTGS) mode • Incentives to queue • ‘Games’ being played on a liquidity/delay tradeoff
- 3. Interbank payment systems Real Time Gross Settlement (RTGS) mode • Incentives to queue • ‘Games’ being played on a liquidity/delay tradeoff • Pool internal queues!
- 4. Interbank payment systems Real Time Gross Settlement (RTGS) mode • Incentives to queue • ‘Games’ being played on a liquidity/delay tradeoff • Pool internal queues! Liquidity Saving Mechanisms (LSMs)
- 5. Aims 1. Model a system with internal queues (LMM) 2. Look at how much liquidity/delay a LSM reduces in theory 3. Look at how banks would use a LSM An ‘agent-based’ model
- 6. Model of a payment system • N banks • A ‘day’ of several ‘seconds’ • Random payment orders random pairs ‘payer&payee’ for each payment a ‘urgency’ parameter u ~ U[0,1] • Each bank sends payment orders in either of two ‘pipes’ (streams): RTGS or Queue
- 7. A E B D C
- 8. A E B D C
- 9. A E B D C
- 10. Low urgency? RTGS A 1 E B D C
- 11. Low urgency? RTGS A 1 E B D C
- 12. High urgency? Queue A 1 E B D C
- 13. High urgency? Queue A E B D C
- 14. A E B D C
- 15. A E B D C
- 16. A E B D C
- 17. A E B D C
- 18. A E B D C
- 19. A E B high u D C
- 20. A E B low u D C
- 21. A E B D C
- 22. A E B D C
- 23. A E B D C
- 24. A E B D C
- 25. A E B D C
- 26. A E B D C
- 27. A E B D C
- 28. A E B D C
- 29. A E B D C End of day Cost = delay cost + liquidity costs = _ uk • (tk - tk’) + _ • a
- 30. Illustration of costs Very low liquidity cost Liquidity cost % routed to RTGS
- 31. Illustration of costs Very low liquidity Delay cost cost % routed to RTGS
- 32. Illustration of costs Very low liquidity Total cost cost % routed to RTGS
- 33. Illustration of costs Very low liquidity Very high liquidity Total cost Total cost cost % routed to RTGS % routed to RTGS
- 34. The game • Banks choose –opening liquidity balance: _ –urgency threshold to queue: _ • For each strategy profile (_1, _1), (_2, _2), (_3, _3)… (_N, _N) a payoff (cost) function • We look at Nash equilibrium for 2 cases: • ‘LMM’ - low urgency payments in internal queues • ‘LSM’ - low urgency payments in central queue
- 35. AB A E B D C
- 36. AB A E B D C
- 37. AB A E B AB D C
- 38. A E B D C
- 39. A E B D C
- 40. Agent-based modelling Liquidity flows very complex simulate the settlement process to compute costs, and hence equilibria
- 41. Agent-based modelling Liquidity flows very complex simulate the settlement process to compute costs, and hence equilibria We look at symmetric equilibria: {(_1, _1), (_2, _2), (_3, _3)… (_N, _N)} : (_i, _i) = (_j, _j) for each i, j
- 42. 1 Model with internal queues (LMM)
- 43. Delay costs with LMM Delay costs 1
- 44. Delay costs with LMM Increase “your” threshold Delay costs 1
- 45. Delay costs with LMM Increase “your” threshold Delay costs 2
- 46. Delay costs with LMM Increase “your” threshold Delay costs 3
- 47. Delay costs with LMM Increase “your” threshold Delay costs 4
- 48. Delay costs with LMM Increase “your” threshold Delay costs 5
- 49. Delay costs with LMM Increase “your” threshold Delay costs 6
- 50. Delay costs with LMM Increase “your” threshold Delay costs 7
- 51. Delay costs with LMM Increase “your” threshold Delay costs 8
- 52. Delay costs with LMM Increase “your” threshold Delay costs 9
- 53. Delay costs with LMM system level
- 54. Equilibria - LMM
- 55. Equilibria - LMM Increase liquidity price a
- 56. Equilibria - LMM Increase liquidity price a
- 57. Equilibria - LMM Increase liquidity price a
- 58. Equilibria - LMM Increase liquidity price a
- 59. Equilibria - LMM Increase liquidity price a
- 60. Equilibria - LMM Increase liquidity price a
- 61. Equilibria - LMM Increase liquidity price a
- 62. Equilibria - LMM Equil. * Planner Cost Liquidity Too little liquidity, too much queueing
- 63. 2 Potential savings from a LSM (‘LSM mechanics’)
- 64. Savings in liquidity RTGS + LMM RTGS + LSM
- 65. Savings in liquidity All in RTGS All queued RTGS + LMM RTGS + LSM
- 66. Savings in delay costs
- 67. 3 How would banks use the LSM ?
- 68. Delay costs with LSM
- 69. Delay costs with LSM Increasing “your” threshold
- 70. Delay costs with LSM Increasing “your” threshold UAD
- 71. Delay costs with LSM Increasing “your” threshold
- 72. Delay costs with LSM Increasing “your” threshold
- 73. Delay costs with LSM Increasing “your” threshold
- 74. Delay costs with LSM Increasing “your” threshold
- 75. Delay costs with LSM Increasing “your” threshold
- 76. Delay costs with LSM Increasing “your” threshold
- 77. Delay costs with LSM Increasing “your” threshold
- 78. _ - equilibria with LSM
- 79. _ - equilibria with LSM Increase liquidity price
- 80. _ - equilibria with LSM Increase liquidity price
- 81. _ - equilibria with LSM Increase liquidity price
- 82. _ - equilibria with LSM Increase liquidity price
- 83. _ - equilibria with LSM Increase liquidity price
- 84. _ - equilibria with LSM Increase liquidity price
- 85. _ - equilibria with LSM Increase liquidity price
- 86. _ - equilibria with LSM Increase liquidity price
- 87. _ - equilibria with LSM Increase liquidity price
- 88. _ - equilibria with LSM Increase liquidity price
- 89. _ - equilibria with LSM Increase liquidity price
- 90. _ - equilibria with LSM Increase liquidity price
- 91. _ - equilibria with LSM Increase liquidity price
- 92. _ - equilibria with LSM Increase liquidity price
- 93. _ - equilibria with LSM Increase liquidity price
- 94. _ - equilibria with LSM Increase liquidity price At high liquidity cost this is the only equilibrium
- 95. _ - equilibria with LSM Increase liquidity price
- 96. _ - equilibria with LSM Increase liquidity price
- 97. _ - equilibria with LSM Increase liquidity price
- 98. _ - equilibria with LSM Increase liquidity price At very high liquidity cost planner and banks choose the same
- 99. _ - equilibria with LSM Increase liquidity price
- 100. _ - equilibria with LSM Increase liquidity price
- 101. _ - equilibria with LSM Increase liquidity price
- 102. _ - equilibria with LSM Increase liquidity price
- 104. LMM vs LSM (good) equilibria costs liquidity thresh.
- 105. LMM vs LSM (good) equilibria costs liquidity thresh.
- 106. LMM vs LSM (good) equilibria costs
- 107. LMM vs LSM (good) equilibria costs cost ratio
- 108. LMM vs LSM (bad) equilibria costs liquidity thresh.
- 109. Conclusions In the hands of a ‘planner’, LSM can save substantial amounts of liquidity, and largely improve settlement speed
- 110. Conclusions In the hands of a ‘planner’, LSM can save substantial amounts of liquidity, and largely improve settlement speed (but may require radical choices) In the hands of individual banks, LSM may require some ‘coordination device’ to yield its potential benefits
- 111. Conclusions In the hands of a ‘planner’, LSM can save substantial amounts of liquidity, and largely improve settlement speed In the hands of individual banks, LSM may require some ‘coordination device’ to yield its potential benefits
- 112. Slide 112
- 113. Slide 113 (!) ~_~ Many thanks