After last week's post about the meeting we had at the Club regarding the RBS glitch, I received an email from one attendee, Peter Miller. Well worth a read so I thought I would post it here as additional input to this challenging and important area in our group.
Last week’s Financial Services Club meeting, covering the RBS systems failure of last summer, stopped at the point that it might have got really interesting. It ran out of time.
The meeting’s chairman, Ralph Villa, led with the worthy points that;
- Banks are in business to make their customers’ lives both easier and possible, and thus that failure of the core systems (note that this is much more than their payment systems) makes customers’ lives harder and in some cases impossible.
- He also recognized that reputational risk was the key issue for the bank concerned, and that frequent and repeated apology, assurance that the bank would rectify the problem and that customers would be compensated for the inconvenience were the key things for the bank concerned to do loudly and publicly – note that RBS have behaved pretty well in this regard.
The real events remain shrouded in mystery and it was disappointing that Chris Dunne from Vocalink confined his comments to reporting the correspondence between Stephen Hester and Andrew Tyrie, good stuff, but old news. And he added real but slightly motherhood-and-apple pie wisdom on the problem being caused not necessarily or exclusively by technology, but by some combination of people, process (failure or lack of) and technology. Too many of his shareholders were in the audience!
The really interesting questions (for me) were either only briefly touched upon, or avoided.
Would the risk of failure be reduced by the adoption of new technology core systems by banks in general?
Roy Vella (ex-RBS) got on his new technology horse. His simple idea was that modern object-oriented technology, operating accounts in real time would solve the problems for any bank. RBS problems were caused by a “behemoth” of a banking system, and its operation, largely in batch mode. So, the problem was caused and/or enhanced by the old, creaking and diffuse nature of its technology.
This ignores the fact that some of the many operations of an accounting system must be – by definition – batch updates (for example, charges to be applied, interest rate application and indeed the transmissions to and from Bacs, the highest volume of the clearings). So, the question goes - would banking systems themselves be more reliable if they were modern real-time systems? Would this add to the security, and reliability of systems and would the absolute requirement for systems to recover to their point of failure without unsignalled transaction loss or duplication be thereby enhanced or reduced? Answers on a postcard, please.
A real problem was highlighted by a comment from the audience, which was not that old technology is inherently unreliable, but that the skills and old-fashioned disciplines required to fix, or change systems were being lost – to old age. (Perhaps, we should think about adding courses in Cobol back into the curriculum?)
But the key problem missed in making the assertion that new technology solves all problems is that life is as it is. If you asked Nick Clegg why good sound liberal principles seem to have been left at the door by the Lib Dems going into government, he would answer (and has) that dealing with government and the economy as it is leads you to different and less than idealistic solutions than you would like. Ditto banks and their technology.
Changing the core accounting system is akin to rebuilding a house by changing its foundations. Ask the tower owners in Pisa how easy that is! In making a system change, a bank has to continue to provide service for its customers and meet the demand for new requirements and regulatory change - and yes, test and implement them in a disciplined manner. (If the only value of the evening were to remind bank IT operators of the need to embrace good old-fashioned disciplines then it was worthwhile.)
As an example of core system replacement, Chris Dunne was asked what the catalyst was for changing the core Bacs system. (Bacs old technology was replaced by a modern combination of object-oriented, Oracle-based technology several years ago.) What he was not asked was how this had changed the functions offered to Bacs clearing members, where the answer would have been “not very much.” In replacing the old technology, the key requirement was that the new system must exactly replace the old one. This requirement is always the key problem – that because of the interconnected nature of a bank’s systems, the advantage of replacing old with new is limited. The new must do exactly what the old did, before you can move on.
The second question related to the interconnectedness of banks and their systems in the clearings.
What was the effect of the RBS glitsch on the UK clearings?
The wheels of our economy are greased by the movement of money between bank accounts across clearing systems – in the case of the UK, this is mostly CHAPS, Faster Payments, Bacs, the cheque clearing and LINK. All of these systems connect banks to other banks, so the failure of one bank to process, particularly one of the bigger ones, will tend to bring the movement of money and thereby the economy as a whole to a grinding halt. If RBS processes say 20% of market volumes they are the sending or the receiving bank in 40% of the transactions. The failure of RBS, Lloyds, Barclays or HSBC payment systems will not only starve their own customers of service, but will remove some of the liquidity necessary to make the money move across the clearings.
It was reasonably obvious that this failure caused no such disaster. Why not? How did other banks help? What lessons were learned?
Our clearing systems are essentially fragile, because they operate as a single interconnected entity of member bank accounting and payment systems, clearing infrastructures and settlement facilities offered by the Bank of England. The failure of one bank’s systems (and all of the UK banks will hold their hands up and recognize that RBS systems and operations are no more likely to fail than their own) causes a systemic problem not just a bank one.
In this case, disaster was avoided – and one suspects that many hands were called to the emergency pumps to limit any damage. How was real disaster avoided? How could the facilities in the middle of the market be improved to reduce the risk of failure of one bank’s systems? Is systemic failure more or less likely as a result of the RBS glitsch?