Agent-Based Modelling of the Emergence of the UK Banking Sector Philip Garnett, Institute of Hazard Risk and Resilience, Durham University and Simon Mollan, University of Liverpool Management School Introduction Networks & Simulation for Business History • The vast majority of businesses leave few traces • • This is especially true of the British banking sector, despite many rich archives We cannot know what resources, capabilities, decisions, strategies, organizational cultures, personalities, or policies had an impact on those organizations Introduction • Business history therefore has: • a limited sample of organizations • an archival bias toward successful survivors • (a tendency toward?) post-hoc justifications of what makes a business ‘successful’ • (frequently) a dominant - broadly conceived - “strategic management” view of corporate behaviour, success and failure based on analyses of strategy, structure, ownership and performance, etc Baker, Mae and Collins, Michael (2007), ‘Methodological Approaches to the Study of British Banking History’, Revue Economique, 58, 1, 59-78. Discuss Holmes and Green’s history of the Midland Bank: The Midland’s history is rightly presented as a success story and a great deal of the “analysis” consists of the authors explaining how important strategic decisions (admittedly identified in hindsight) were made and how the bank grew in scale (whether in terms of number of accounts, capitalisation, balance sheet, number of offices, or number of employees (Baker and Collins, 2007: 61). Reflecting on Ackrill and Hannah (2001) the creation of Barclays in 1896 was strategic… …a defensive alliance amongst twelve familyrun banks against the intensifying threat of competition and possible take-over from the larger joint-stock banks such as Midland and the National Provincial (Baker and Collins, 2007: 62). Introduction • These are successful organizations • Information comes from Archives • & Corporate histories Holmes, A.R. and Green, Edwin (1986), Midland. 150 Years of Banking History. London: Batsford. Ackrill, Margaret and Hannah, Les (2001), Barclays: The Business of Banking, 1696-1996. Cambridge: Cambridge University Press. • They try to explain the growth and success of the banks in terms of the banks’ own operations • Can this be challenged? • What about the environment that the banks existed in? 1) Collected data on the changing demography of the UK banking sector from 1550s to the present day. 2) Stored the data in a network database so we can track individual banks through time. 3) Produce an agent-based simulation of the amalgamation period. The period of time from 1820s that shaped the banking sector that we see today. Relational Database Node: Child and Co. Node: Glyin Mills and Co. Relationship: Acquired Attributes: • Creation Date • End Date • … Extending the Database: • Client information • Interlocking Directorships • Debtor/Creditor relationships Relationship: Banks With L1 Relationship: Client Node: Some Co. L2 Network Database Data is basic ‘biographical’ details – no financial data Stores data in the form of Nodes, and Edges in a Neo4j Network Database. Nodes are banks, they have attributes. Bank Name. Creation Date. Edges store the relationship between the nodes. Powerful method of storing data that allows the relationships between data points to be analysed. Merged with/Acquired by. Network databases have a lot to offer real work complex systems research. It is possible to 'walk' the network. This allows us to find all the banks the came before a particular bank – the bank’s tree of antecedents. Changing bank population Between 1559 & 1810 the population of UK banks grows exponentially. Growth of ~2.7%. Around 1810 population starts decreasing exponentially (~1.2%). This looks very similar to ecological population collapse. Population increase far beyond the carrying capacity of the environment until it eventually collapses. Mergers to form bigger banks don't have a good biological equivalent. Banks are selected out of existence either through failure or merger/acquisition. • Sykes, J. (1926) The Amalgamation Movement in English Banking, 18251924. London: P.S. King and Son Ltd. • Prior to 1825 banks were limited in there size. For a century afterwards amalgamations/mergers in British banking occur. • Is the environment that the banks are operating more important than the individual agency of a bank and the people running that bank? Competitive selection and the growth of banks Growth of Bank Groups All bank “agglomerations” seem to be growing at approximately the same rate – as measured by the number of agglomeration events – these are also selection events. Hypothesis: In a limiting environment final group size could be a combination the of point in time where expansion started and the rate of growth. HSBC grows (as) quickly but starts late, and therefore can never catch-up. Suggests that all banks in the population expand by acquisition as quickly as they can. Bank group sizes flatten out at almost exactly the same time ~1925. No more banks left to merge with/acquire Change in regulatory environment – bank amalgamations regulated by the Bank Amalgamations Committee The emergence of the cartel – the Big 5 of large banks This is also the time at which the UK bank population stabilises Simulation • Can a simple agent-based simulation of mergers, creations and failures of banks produce the a simulated bank population that mirrors our real data? • Simulating the period from 1810 forward, from the maximal bank population (1100 banks). • Banks can (with some probability/rate): • Fail and leave the population altogether. • Merge with another bank. • New banks can be created. • Simulation timestep is one month. “Agglomerations” Size Distribution “Agglomerations” Size Distribution Simulation Mergers Network • Simulator stores results in a Network Database. • Resulting network structure can be compared with real network. • Changes in the assumptions in the simulation change the network structure but not the population level outputs. • New result, not analysed so far…. Isomorphic behaviour? • Institutional isomorphism (DiMaggio and Powell, 1983): • Units in a population come to resemble each other as a result of environmental factors • Coercive – government, regulation • Mimetic – imitating ‘successful’ behaviour • Both seem present in the case of UK banks Each of the Big Five has grown partly by amalgamation and partly by the opening of new branches. In the latter half of the nineteenth century local banks in the various parts of England began to coagulate into small branch systems. The process was continued by the merger of small branch systems into larger branch systems, finally resulting in the emergence of a few large banks whose head offices are in London. (Wernette, 1935: 369). When one bank made a bold stroke and captured a business possessing a number of branches, its rivals were stimulated also to seek expansion in order to protect the gains they had made previously. (Sykes, 1926: 116) • Very much like the mimetic institutional isomorphism as discussed by DiMaggio and Powell (1983) Conclusions • There is a complexity to the organizational past which is revealed by using different quantitative techniques. • It looks as if – as viewed from a population perspective – that the ‘successful’ banks were created as a result of a growth pattern which was at least in part based on mimicry as other banks were selected out of existence (failed or merged). • Growth occurs in this way until this system reaches saturation/maturity. • Are therefore the changes in the bank population a result of environmentally limited emergent behaviour? • Although human agency is present in the system, perhaps the environment is driving the trajectory of the population as a whole. Conclusions • We are (unfortunately?) unable to disprove our hypothesis. The simulation/date suggests that there are random fluctuations (noise) in the number of banks. These could be misinterpreted by regulators. Or blur the effect of regulation over the short term. Regulators could attach significance to a random event and alter the environment... post-hoc justification leads to reactive policy change, which could have unforeseen consequences. The natural trajectory of the system was environmentally driven. Limited agency of individual banks? Regulatory change set the banks down a path which their agency could not alter? If the year was 1870 would anyone believe this model had predictive capacity? How would we know if it did? The 1844 Bank Charter Act 1844 1866 It is tempting to say that we can see this and other crisis related events in the data (1866), but they don’t shift the population from its path. Blips might be nothing to do with the events. Questions? The Tipping Points Project Durham University http://www.dur.ac.uk/ihrr/tippingpoints/ http://tippingpointsproject.org/ philip.garnett@durham.ac.uk http://prgarnett.net References Ackrill, Margaret and Hannah, Les (2001), Barclays: The Business of Banking, 1696-1996. Cambridge: Cambridge University Press. Baker, Mae and Collins, Michael (2007), ‘Methodological Approaches to the Study of British Banking History’, Revue Economique, 58, 1, 59-78. Cameron, Rondo (1967), Banking in the early stages of industrialization: a study in comparative economic history. Oxford: Oxford University Press. Carroll, Glenn, and Hannan, Michael (2000) , The Demography of Corporations and Industries. Princeton, NJ: Princeton U.P. DiMaggio, Paul and Powell, Walter. (1983). ‘The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields’. American Sociological Review, 48, 147-160. Hannan, Michael and Freeman, John. (1977) . ‘The Population Ecology of Organizations’. American Journal of Sociology, 82, 929-964. Holmes, A.R. and Green, Edwin (1986), Midland. 150 Years of Banking History. London: Batsford. Steele, F.E. (1896), ‘Bank Amalgamations’, The Economic Journal 6, 24, 535-641. Sykes, J. (1925), ‘The Effect of English Bank Amalgamations on Working Expenses and Profits, The Economic Journal 35, 140, 583-589. Sykes, J. (1926) The Amalgamation Movement in English Banking, 1825-1924. London: P.S. King and Son Ltd. Wernette, J. Philip (1935), ‘The English Banking System’, Harvard Business Review, 13, 3, 366-380.