IBS Journal February 2015


Many banks have ended up with a mish-mash of solutions, some of which are real-time, others that are quasi real-time and the rest that are batch. The proportion of systems in each category is likely to be largely determined by the technical architecture and the majority of banks still rely on end-of-day crunching of numbers. The complexity has come in waves.


From setting limits on the same basis as traditional lending. There was then a move to add-on methodologies for some, such as Monte Carlo simulations. Regulatory capital has gone through a similar sequence of added complexity, with more sophisticated internal models being created as the requirements mounted, through Basel II, bringing a need for new systems, processes and quantitative skills. The financial crisis heralded more regulation, rapidly accelerating the shift of OTC derivatives to central clearing and giving prominence to Credit Valuation Adjustment (CVA), whereby the value of a derivative is influenced by the associated counterparty risk. Collateral management moved firmly into the spotlight, with this typically shifting from the operations department to the front office. CVA is used to incorporate counterparty risk into the pricing of deals and for inter-desk charging within institutions.


‘No one is standing up and defining


what real-time risk management is,’ says Mikael Sorboen, head of risk systems at BNP Paribas. Within the bank itself, the first step has been to try to build consensus and this alone has taken six to eight months, he says. When it comes to the regulators, the Federal Reserve moves at one speed, European regulators move at other speeds, and a lot of the requests from them are piecemeal. At BNP Paribas, it is recognised that


there will be a need to shift from end-of- day to real-time risk management, but it is relatively early in the journey. At least, unlike most banks, it has a risk platform that covers all lines of business and both counterparty and market risk. It is also the only part of the bank that has an IT devel- opment team within the group function, reflecting the fact that risk management is deemed a special case. The bank uses Monte Carlo models


for calculating regulatory capital, with this now seen as important for the manage- ment of the bank, not just a regulatory


‘There is no point having out of date information


when trying to make dealing decisions. But there is also no point having real-time data if it cannot


be consumed in real-time.’ John Macdonald, IBM


task. It has automatic monitoring of around 900,000 limits. It has multi-di- mensional drill-down capabilities for risk analysis and reporting and stress testing for both market and counterparty risk. On the input side, it has a lot of source sys- tems and siloes, market data, calibration, static data (‘absolutely crucial, you have to be a lot more attentive and there are more and more descriptive regulations about how you classify counterparties’, says Sorboen) and risk indicators (typically more than 100,000 feed files per day, 800 million rows per day, and more than ten source systems). IBM’s Macdonald says ‘there is a lot


of collective demand away from pure financial assumptions towards more end-to-end processes and understanding the data that is used for calculations’. The emphasis is on ‘prove it, don’t just report it’. The position of ‘chief data officer’ or sim- ilar is now increasingly prominent. It is no longer merely about acquiring data, there needs to be an understanding of where it is from, whether it is fit for purpose and whether it is up to date. The crux, he says, is ‘can I trust what I am seeing?’ There will not always be an absolute answer, he suggests, with the need to ‘socialise it’, checking with colleagues and peers, un- derstanding the details and nuances, and building up a sense of whether something is being done correctly. On the issue of timeliness, Macdonald


feels some clients will ‘absolutely need real-time decision support’ but another way to class the need could be ‘right-


© IBS Intelligence 2015


time intelligence’ – in other words, the appropriate data at the moment that it is needed. ‘There is no point having out of date information when trying to make dealing decisions. But there is also no point having real-time data if it cannot be consumed in real-time.’ For instance, a corporate loan officer might need to do two or three days analysis on a prospec- tive loan and will need data throughout the process. Similarly, with ever more predictive requirements, there might be a number of moments in the day when stress tests need to be run, at which point the right data needs to be available. As an example of where much greater


governance is needed than in the past, a prime area is around financial models. For an aspect such as CVA, there could be ‘lay- ers and layers of models’, says Macdonald. There might be 30, 40 or 50 contributing to a single number. But are those models fit for purpose? The lifecycle of the models becomes important, including who built it (there have been plenty of ’rogue models’ built by one person on a spreadsheet), who has changed it, how has it been changed, what is the release procedure and where is it used? In terms of BNP Paribas’ infrastructure


to support its broad risk requirements, it has a three-layer database (Sybase IQ) in its Paris data centre, with three days of data in the first layer, 50 days in the second and five years of history in the third, amounting to around 40 terabytes of data. Input data is handled by a Java/ Python-based in-house developed script-


www.ibsintelligence.com 43


analysis: risk management


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56