Applications of BoF-PSS2 simulator and how to use it in agent based - - PowerPoint PPT Presentation

Applications of BoF-PSS2 simulator and how to use it in agent based models

Workshop on Agent based modeling for banking and finance Villa Gualino, Torino 10 February 2009 Matti Hellqvist Oversight of Market Infrastructure division Financial Markets and Statistics department Bank of Finland

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

ABM-BaF 10 February 2009 • 1

Outline

• BoF-PSS2 simulator basics
• Intended applications and examples
• How to use this simulator in agent based models

As usual, the views are those of the speaker and do not necessarily reflect Bank of Finlands position

ABM-BaF 10 February 2009 • 2

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Background of BoF-PSS2

Bank of Finland...

– National central bank and monetary authority – Member in European System of Central Banks – Long tradition as a research oriented CB – Statutory task to ”participate in maintaining the reliability and efficiency of the payment system and overall financial system and participate in their development”

... Payment system simulator

– BoF-PSS1 was built before Finland joined European Monetary Union (1.1.1999) – BoF-PSS2 was released in 2004 (beta 2003)

• Specialised tool for payment system experts
• Easy to use, versatile and modular
• Designed to be shared with others
• Available free of charge for research purposes

ABM-BaF 10 February 2009 • 3

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

BoF-PSS2 - principles

• Simulations of market infrastructures are often performed with one-
• ff tailored models or inflexible test environments
• BoF-PSS2 in contrast…

– offers a versatile toolbox with most of the common ”building blocks” needed in a model of a payment system – has modular design: allows combinations of different blocks – Is extendable: users have possibility to introduce new modules – Records various statistics and has ready made reports – It is designed as an analysis tool – Is based on widely used free tools – standard interfaces

Laboratory of payment and settlement systems

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

ABM-BaF 10 February 2009 • 4

From input data to output statistics by replicating the process logic of real payment system

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

ABM-BaF 10 February 2009 • 5

What goes in...

• Data

– Transactions, participants, balances, credit or bilateral limits – i.e. The account structure, bookings and constraints affecting the settlement process – Also optional data fields and user defined or meta data

• System definitions

– Type of system: RTGS, continuous netting, deferred netting – General parameters – Process logic, which is

• Divided into algorithms classes: Entry, queue release etc.
• Composed by selecting appropriate algorithm for each

needed class. – Multiple separate and interlinked systems are possible

ABM-BaF 10 February 2009 • 6

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

What comes out

Output database

• Large number of indicators

recorded during simulation

• Only selected tables are saved
• Has four levels of statistics:

1.General information of executed simulations 2.Statistics and indicators in the system level 3.Statsitics and indicators in the participant/ account level 4.Transaction level Reports:

• First view summaries
• Time series aggregation tool
• Comparison reports

ABM-BaF 10 February 2009 • 7

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Analysis tool aspect

• Data management

– Parallel data sets and projects are possible – Data files in CSV format and templates for data – Input or output database can be accessed directly

• Data imports, exports or modifications
• Special queries

– Easy standalone use

• Computational efficiency

– Minimised file i/o compared to real systems – Simplified process with only the necessary variables

• Based on platform independent and widely used technologies

– Java, MySQL (OEM licence needed) – Available as PC software for MS operating systems (currently)

• Capability for other analyses from the same data e.g. network

analysis

ABM-BaF 10 February 2009 • 8

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

User community

• Over 70 granted licences (January 2009) of which 61% in central banks, 24% in

academic institutions, 7% in automated clearing houses (ACH), 3 % in international organisations, 5% others

• Licenced users among others: University of Torino, Banca di Italia, Bank of

England, Banco de Mexico

ABM-BaF 10 February 2009 • 9

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Outline

• BoF-PSS2 simulator basics
• Intended applications and examples
• How to use this simulator in agent based models

ABM-BaF 10 February 2009 • 10

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Use of BoF-PSS2

• Basic idea:

Model what ever you want related to payment system. This tool allows you to replicate (closely) realistic settlement process in your analysis and record a variety of statistics from it

ABM-BaF 10 February 2009 • 11

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Purposes of simulations

Simulations are well suited for multiple CB tasks

(Deepening of) OVERSIGHT LVPS

• perations

Efficiency Scenario analysis Stress testing Assesment of contingency plans Testing & development Risk identification Risk quantification New policies & regulations Analysis of market structure and practicies Simulations

• f market

infrastructure

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

ABM-BaF 10 February 2009 • 12

Classification of analyses with BoF-PSS2 with publicly available results*

Network analysis ** Liquidity analysis Credit risk analysis Delay/queuin g analysis Oversight 2

USA, UK

20

FIN, UK, SWE, CAN, AUT, JPN, SUI, HOL, HUN, DEN, POL, NOR, BIS

16

ISL, UK, CAN, FIN, AUT, JAP, SUI, HOL, DEN, POL, BIS

14

DEN, FIN, UK, SWE, CAN, AUT, JAP, HOL, HUN, POL

Operational development issues 7

USA, KOR, FIN, CAN, UK

4

ISL, CAN, UK, FIN

5

USA, DEN, CAN, UK, FIN

Policy concerns 2

USA, UK

15

FIN, KOR, DEN, SWE, CAN, USA, UK, POL, NOR

10

FIN, ISL, DEN, CAN, USA, UK, POL

13

FIN, DEN, SWE, CAN, USA, UK, POL

Basic research 5

USA, FIN, JAP, BIS

4

USA, FIN, JAP, BIS

4

USA, DEN, FIN

The number of unpublished studies is likely much larger

* See extra slides for more details ** studies performed with same network module which is (currently being) integrated in BoF-PSS2

ABM-BaF 10 February 2009 • 13

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Typical oversight studies

• Network analysis

– Which are the most important participants and connections? – Do connections vary during the day and from day to day? – Which are the end-of-day and overnight patterns? – How would the network change in outages?

• Liquidity analysis

– Which are the current liquidity levels and effects of general drainage? – What is the impact of large participants’ liquidity problems or stopping? – Are some participants working with too low liquidity levels? – How large shocks can the current liquidity supply sustain?

• Credit risk analysis

– How large are the current credit risk levels and systemic risk probability? – How much will credit risks grow in crisis situations? – To which extent can participants sustain current credit limits? – What impact will more stringent credit risk requirement have?

• Delay/queuing analysis

– Which effects will large participants’ and system stop have on delays – Does some participants delay payments more than others?

ABM-BaF 10 February 2009 • 14

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Typical policy studies

• Network analysis

– What would be the impact of new participant access criteria? – What is the current level of tiering and the impact of possible changes? – What is the network characteristic of prioritised transactions?

• Liquidity analysis

– What is the impact of new liquidity regimes – What would the effects be of new liquidity pricing schemes? – What would be the impact of different prioritising regimes? – What would be the most efficient liquidity/delay levels?

• Credit risk analysis

– What would the impact be of new credit risk rules? – What would be the most efficient credit risk/delay levels?

• Delay/queuing analysis

– What could be the effects of new rush-hour pricing policies? – Which gridlock resolution algorithms would be most efficient for given payment flows? – Which are the effects of bypassing FIFO or other processing order rules? – Which are the effects of new open hour rules?

ABM-BaF 10 February 2009 • 15

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Typical operational studies

• Network analysis

– Which are the current capacity profiles and bottlenecks? – How would different shocks change the capacity profiles? – How much capacity does different prioritising schemes for different system outage scenarios require

• Liquidity analysis

– What effects would different kinds of marked-based syncronisation and timing rules have? – What impact would shifting volumes between different systems have? – Which are the effects of different types of liquidity bridges? – How fast should liquidity expansion operations be?

• Credit risk analysis

– How fast need credit limit change procedures to be?

• Delay/queuing analysis

– Which would be effects of introducing prioritising, queueing and gridlock resolution features? – How efficiently are current prioritising, queuing and gridlock resolution features used? – How would algorithms’ parameter changes affect overall queuing?

ABM-BaF 10 February 2009 • 16

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Typical basic research studies

• Network analysis

– Which are current network structures in different systems? – Are there typical trends in network structure changes? – Are there differences in marked-driven and society-optimal network structures?

• Liquidity analysis

– Benchmarking liquidity levels across systems and countries? – Are there typical trends in liquidity provision regime changes? – Which are the differences between public and private liquidity regimes?

• Credit risk analysis

– Are there general warning signals for increased systemic risk? – How could systemic risk probabilities be best measured? – Which are the historic correlations between systemic risks and participants credit risks? – Which are the efficiency and risk concerns of private vs public credit risks?

• Delay/queuing analysis

– Can general optimality among gridlock algorithms be established? – Are there differences among service providers’ and customers’ preferences?

ABM-BaF 10 February 2009 • 17

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Outline

• BoF-PSS2 simulator basics
• Intended applications and examples
• How to use this simulator in agent based models

ABM-BaF 10 February 2009 • 18

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

• Common approach with BoF-PSS2 so far has been

deterministic event based simulations

• Alternatives were mentioned

already in 2003

• 1. Integrate with external

dynamic model

• 2. Modify the algorithms

and include e.g. Agent based modeling elements

Integration in agent based models

ABM-BaF 10 February 2009 • 19

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Solution 1, external dynamic model

• Independent agent based model, which creates input

data or parameters: small economy, artificial exchange...

• BoF-PSS2 as a model of the settlement system

– Integration with the external model with data files in specified locations – BoF-PSS2 has command line interface ⇒automated simulation runs can be started from the external model

• Benefits

– No limitation for the tool used in the external model

• Disadvantages

– Currently only file based batch run approach is supported

ABM-BaF 10 February 2009 • 20

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Solution 2: user module changes

Recall that

• Process logic is splitted

into algorithm classes

• Existing algorithms are
• pen source code
• Users can easily*

include own modules

Logical place for an agent based model is in the submission algorithm: ”What happens next in the simulated setup”

*See the following:

• Algorithm descriptions and user module development guide
• Source code for all algorithms can be found in the simulators home pages or installation folder
• Step by step demo in the simulator extranet (only for licenced users though)

Output DB

Simulator engine: User interface, process control, all common data

Input DB

System X

Entry Booking queue Transaction queue

Settlement Update balances& bookings

Split Injection QUE MNS Bil.Ofs. PNS Trans- action? End Add to queue Update queue Account & bilateral balances

System Y

Entry Booking queue Transaction queue

Settlement Update balances& bookings

Split Injection QUE MNS Bil.Ofs. PNS Trans- action? End Add to queue Update queue Account & bilateral balances

Submission algorithm

ABM-BaF 10 February 2009 • 21

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

User modules solution continued

• Benefits

– Allows the ABM to react on internal state variables of the settlement system on transaction level – Use of Java based toolboxes possible

• Repast, Jawa Swarm…

– Succesfull new algorithms can be distributed to wide user community – BoF is open for joint project proposals – good track record from sponsorship projects for enlarging the capabilities of BoF-PSS2

• Disadvantages

– BoF-PSS2 is not fully open source; engine, interfaces and main data structures can’t be changed by the user

ABM-BaF 10 February 2009 • 22

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Summary

• BoF-PSS2 is

– Designed to be an analysis tool – Flexible and versatile description of payment system process logics – Widely used for payment system related oversight or policy analyses, operational development and research – Available free of charge for these purposes – Also compatible with agent based approach

• It is not

– Limited to existing set of algorithms – Limited to deterministic simulations with historical data

Be the first one to build an agent based model into this tool!

ABM-BaF 10 February 2009 • 23

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Thank you!

Contact details: matti.hellqvist(at)bof.fi bof-pss(at)bof.fi Simulator home pages www.bof.fi/sc/bof-pss

– Documentation – Ordering info – List of published studies – Call for papers for the 7th simulator seminar in Aug 09 is open

ABM-BaF 10 February 2009 • 24

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

ABM-BaF 10 February 2009 • 25

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Extras

• Publicly reported studies which have utilised BoF-PSS2

classified into four cathegories (see table on slide 13)

– Oversight – Policy studies – Operational studies – Basic research

• Works which share strongly aspects from several

cathegories have been listed under each relevant cathegory

• For more details see the BoF-PSS2 home pages for

– List of published studies – List of previous simulator seminars and workshops and their presentations

ABM-BaF 10 February 2009 • 26

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Oversight studies I

Finnish BoF-RTGS - assess liquidity effects of introduction of TARGET and the shift to a greater use of RTGS settlement – results published 1997 in BoF E:14 Iceland’s Sedlabanki - netting vs. real-time gross settlement - setting credit limits for the system, 1999 Bech and Soramäki (2002) ‘Liquidity, Gridlocks and Bank Failures in Large Value Payment systems’, E-Money and Payment systems Review (Bech-Soramäki) Paul Bedford - Stephen Millard - Jing Yang, BoE: Analysing the impact of operational incidents in large-value payment systems: A simulation approach Björn Segendorff, Sveriges Riksbank: Liquidity levels and delays in RIX Darcey McVanel, Bank of Canada: The impacts of unanticipated failures in Canada's Large Value Transfer System Matti Hellqvist, Bank of Finland: Stress testing securities settlement systems using simulations Claus Puhr, Stefan W. Schmitz / Oesterreichische Nationalbank: Risk Concentration and Operational Risk in Payment Systems – A Simulation Approach Kei Imakubo, Yutaka Soejima / Bank of Japan: Intraday Settlement Activities in the BOJ-NET RTGS

ABM-BaF 10 February 2009 • 27

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Oversight studies II

Ana Lasaosa, Merxe Tudela / Bank of England: Risks and efficiency gains of a tiered structure in large-value payments: a simulation approach Matti Hellqvist / Bank of Finland: Stress testing liquidity needs in Finnish retail securities settlement system Martina Glaser, Philipp Haene / Swiss National Bank: Simulation of participant-level

• perational disruption in Swiss Interbank Clearing

Ronald Heijimans / De Nederlandsche Bank: Stress simulations: A Dutch case Ágnes Lublóy, Eszter Tanai / Magyar Nemzeti Bank: Operational Disruption and the Hungarian RTGS system VIBER Neville Arjani and Lana Embree, Bank of Canada: Consolidation in Canada's LVTS: A Simulation Study Kristian Sparre Andersen and Irene Madsen, Danmarks Nationalbank: A quantitative assessment of international best practices in relation to business continuity arrangements in payment systems Matti Hellqvist, Bank of Finland: Implicit intraday counterparty limits in large value payment systems

ABM-BaF 10 February 2009 • 28

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Oversight studies III

Jenni Koskinen, Bank of Finland: The liquidity impact of merging bond and equity settlement systems Agnieszka Grat-Osinska - Miroslaw Pawliszyn: Liquidity Levels and Settlement Delays in the Sorbnet System-Simulation-Based Approach With the Application of the BOF-PSS2 Payment Asbjorn Enge and Frode Overli: Intraday liquidity and the settlement of large-value payments: a simulation-based analysis (Economic Bulletin 1/2006, Norges Bank) Elisabeth Ledrut / Bank for International Settlements: Tit for Tat in Payment Systems

ABM-BaF 10 February 2009 • 29

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Policy studies I

Finnish BoF-RTGS - assess liquidity effects of introduction of TARGET and the shift to a greater use of RTGS settlement – results published 1997 in BoF E:14 Iceland’s Sedlabanki - netting vs. real-time gross settlement - setting credit limits for the system, 1999 Danmarks Nationalbank 2000 – main focus on gridlock resolution – results published in BoF discussion paper series 9/2001 and DNB Monetary Review 4/2001 Bank of Korea – alternative liquidity provision, optimisation methods, 2002 Leinonen and Soramäki (1999) ‘Optimizing Liquidity and Settlement Speed in Payment Systems’, Discussion Paper, Bank of Finland Soramäki (2000) ‘Alternative Liquidity Management Features in an RTGS Environment’, Financial Markets Department Working Paper Bech and Soramäki (2002) ‘Liquidity, Gridlocks and Bank Failures in Large Value Payment systems’, E-Money and Payment systems Review (Bech-Soramäki) Björn Segendorff, Sveriges Riksbank: Liquidity levels and delays in RIX Neville Arjani, Bank of Canada: Examining the Balance between Risk and Efficiency in Canada’s LVTS: A Simulation Approach

ABM-BaF 10 February 2009 • 30

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Policy studies II

Matti Hellqvist, Bank of Finland: Stress testing securities settlement systems using simulations Morten Bech and Kurt Johnson, Federal Reserve Bank of New York: BoF-PSS 2.0.0 A tool for policy analysis Kemal Ercevik, John Jackson / Bank of England: Simulating the impact of hybrid functionality on CHAPS banks Kristian Sparre Andersen and Irene Madsen, Danmarks Nationalbank: A quantitative assessment of international best practices in relation to business continuity arrangements in payment systems Jenni Koskinen, Bank of Finland: The liquidity impact of merging bond and equity settlement systems Agnieszka Grat-Osinska - Miroslaw Pawliszyn: Liquidity Levels and Settlement Delays in the Sorbnet System- Simulation-Based Approach With the Application of the BOF-PSS2 Payment System Simulator (SSRN, Financial Markets and institutions, May 2007) Neville Arjani: Examining the Trade-Off between Settlement Delay and Intraday Liquidity in Canada's LVTS: A Simulation Approach (Working Paper 2006-20, Bank of Canada) Asbjorn Enge and Frode Overli: Intraday liquidity and the settlement of large-value payments: a simulation- based analysis (Economic Bulletin 1/2006, Norges Bank)

ABM-BaF 10 February 2009 • 31

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR

Operational studies

Iceland’s Sedlabanki - netting vs. real-time gross settlement - setting credit limits for the system, 1999 FRB New York – alternative queuing/liquidity concepts – a ‘Receipt Reactive Gross Settlement’ queue, 2000 Danmarks Nationalbank 2000 – main focus on gridlock resolution – results published in BoF discussion paper series 9/2001 and DNB Monetary Review 4/2001 Bank of Korea – alternative liquidity provision, optimisation methods, 2002 Soramäki (2000) ‘Alternative Liquidity Management Features in an RTGS Environment’, Financial Markets Department Working Paper Neville Arjani, Bank of Canada: Examining the Balance between Risk and Efficiency in Canada’s LVTS: A Simulation Approach Kemal Ercevik, John Jackson / Bank of England: Simulating the impact of hybrid functionality on CHAPS banks Neville Arjani: Examining the Trade-Off between Settlement Delay and Intraday Liquidity in Canada's LVTS: A Simulation Approach (Working Paper 2006-20, Bank of Canada)

ABM-BaF 10 February 2009 • 32

Bank of Finland PAYMENT AND SETTLEMENT SYSTEM SIMULATOR