Accountants, Prices and Upsidedown Elastic...

I am sure I am not the only one who has had to suffer the boredom of a economics lecture on price elasticity, but my interest in this old topic was sparked by an article by Tony Jackson in the FT on Monday, providing a very simple and clear explanation of how mark-to-market accounting (see earlier post) can conspire with leverage to turn price elasticity on its head, so the more something goes up in price, the more in demand it becomes...perhaps I should have paid more (or less?) attention to what the dusty prof was saying...

CEP - Part of the technology furniture?

The CEP market is apparently maturing - don't miss this post "CEP: LaserDisc or DVD?" by Adam Honoré at Aite Group with an interesting view of the future of CEP technology.

The Value in Product Control

Good post from Robert Peston on the BBC website on part that the Product Control Group did (or rather didn't?...) play in the problems at Lehman's, according to the official US bankruptcy report on Lehman's by Anton Valukas.The post highlights the report's findings that the Product Control Group did not have the quant experience to keep up with CDO trading desk.

Interesting findings on Lehman's, but variants on this theme seem to be elsewhere too. A contact who knew Merrill's New York trading operation in the run up to the crisis recently asked me how many quants did I think used to work on the CDO trading desk. The surprising (?) answer was not one...

Data models are not what they used to be...

AIM have released the results from their 2009 survey on reference data management which is worth a look, particularly given the 2008 results are also shown for comparison. Seems like Mike Atkin and the EDM Council have their work cut out in getting the Semantics Repository adopted if the survey is anything to go by, with the number of institutions using standards-based data models having dropped significantly when comparing 2009 to 2008. What is going on there in these heady days of the finance industry sorting out its data problem through adopting standards? - In cash starved times, maybe it costs more to conform to a standard? - Is the survey data not broad enough? Any ideas appreciated!

Risk, Data Transparency and the MBS Market

I spent the morning yesterday over at the FIMA USA event in New York, and caught the panel discussion chaired by Neil Edelstein of GoldenSource. Stand out speakers were Amy Hawkins of BNY Mellon and John Bottega of the Federal Reserve.

Neil started the panel by asking the panel for their thoughts on the current drive to improve "data management for risk". Transparency and quality were mentioned a lot unsurprisingly, with John Bottega adding that he was aware that a lot of banks were now focussed on the data that in the past had been "not available" for risk management, not just the quality of data that is readily accessible. All panelists focussed on the need to manage risk across the whole institution, not just by product silo.

On the topic of data standards and transparency, John referred the audience to testimony on the Mortgage Backed Securities (MBS) market presented to the US Government by the XBRL group. Apparently the filing process for mortgages allows free format filing and so is of little use from an automated processing point of view. John also pointed out that a key piece of data in assessing risk is that the "first time buyer" flag was found to be present in only 15% of the filings.

John also mentioned that if loans and mortgages could be given standard identifiers, then this would enable new levels of risk management - for instance it should be able to extract those obligations against a specific region that for example is experiencing economic recession. These would be the benefits of getting data standards in place.

As was later expanded upon in a later talk by Kay Vicino of Northern Trust, there was a lot of panel discussion on organisational data governance and the management structures needed to achieve it. On the governance side of things then whilst it is not an exciting topic, it is obviously vital - main point seems to be establishing data ownership and responsibilities which brings me back to the point that a lot of (most?) data management issues are down to managing people and organisational politics, not just down to good technology (although it helps!).

Overall a reasonable panel, and the XBRL testimony looks worth a more detailed read (if the testimony link doesn't work then go to the www.xbrl.org site and search for a report called "Using Standards for Transparency")

How not to do marketing #1

I ran into this very funny post on the rebranding of Fortis into "ageas". Worth reading (and learning from it)! Also don't miss some of the comments posted for how other banks in the news could be renamed - join the debate and enter your suggestions too!  

One man's speculation is another man's insurance...

The current finanical crisis in Greece has prompted an outburst of entertaining discussion at the FT about CDS contracts, initiated by a feature article by Wolfgang Munchau who advocates that naked CDS contracts should be banned. The main argument used is that you should not be able to insure against a risk that you do not face e.g. buying insurance on somebody else's house then arranging to have the house burnt down. In support of Mr Munchau, one reader letter points out that insurance without interest in the insured item has been illegal since 1746, which on the face of it seems a long enough time to be a credible point in the discussion.

However, in using this argument then Mr Munchau seems be to attacking the whole of the derivatives industry not just CDS, for example the same argument could be used to ban the use of naked index puts to hedge equity market risk. I guess he is also helping some of the politicians in the EU direct attention away from Greece's financial mismanagement more towards the "evils" of the derivatives markets and hedge funds.

Some good letters in response, for instance this one with a good illustration of what hedging would be like without intermediaries to buy and sell risks that they do not own, plus another more direct one from the Association of Corporate Treasurers.

Whilst talking of Greece and credit, the FT Alphaville team also poked some fun at Anatole Kaletsky, the economist of the London Times Newspaper, who has recently done some interesting articles in the paper concerning his predictions about the stresses being suffered by Greece and the Euro. From their post, it would seem that Mr Kaletsky also runs a credit related fund, so it is implied that some of his newspaper views need to "calibrated" against his own vested interests...

Data Management Panel

Thomson Reuters held a panel event on data management at their London offices on Tuesday last week, with speakers from Barcap, LCH.Clearnet, DB, Mizuho and Citi. This event was held in follow up to their recent report "Beyond Golden Copy". Below are some of my notes on the summary points the panelists made:

• The Value of Data - Kris Bhattacharjee of Barcap said that there were currently two main drivers behind the perceived business value of data; i) Regulators are expecting more information, adding additional requirements and conducting more adhoc reporting requests. ii) Business users/decision makers want more granular understanding of trading and risk management data, in order to decide how best to allocate scarce capital to what trading positions.
• Data Metrics - Kris said that the metrics were many but timeliness of data was becoming a key metric - over the past two years regulators have moved from allowing say 2 months as a reporting timeline down to 10 days recently. Additionally timeliness is again vital as regulators demand adhoc reporting in response to market events.
• Accuracy/Completeness - Again regulators are driving this, with the "bad numbers in, bad numbers out" as the main motivation. Unsurprisingly, counterparty data is also being required at a new level of detail and accuracy down to a portfolio level in light of the crisis.
• Granularity of Data - Deeper granularity of data being driven by scarce capital and the need to understand how efficiently it is being used. Basel II has also driven greater granularity over Basel I. Reflecting what I have heard from some our clients, Kris added that the data associated with securitised products had increased greatly as people need to understand exposure/risk and pricing in more detail (rather than assume blanket statistical behaviour for a whole basket of assets).
• Stress Scenarios - Kris again mentioned the understanding of counterparty exposure driving the need for new data sets, as had the initiative of banks having "living wills" to allow a bank to be wound down in an orderly manner.
• Everybody has Left the Building! - Martin Taylor of LCH.Clearnet was a great speaker and said that the biggest new problem that the collapse of Lehman's created was that ordinarily there are people around to help with extracting from systems what the exposure is to the various counterparties. In the Lehman's case there was nobody around to help, making the process very difficult and leading to the need for changes to address this problem.
• Mandating Data Integrity - Martin added that data security, integrity and auditabiliy were vital, and in particular put emphasis on the people that are running the systems that they have their own form of integrity so that an institution knows that the people can trusted but is also capable to deal with a situation where the people are not around to help. Martin felt that this level of data management should be mandated on the industry and that there was an awful lot that finance could learn from industries such as Pharmaceuticals in terms of product approval and management/robustness of data.
• Data with No Cost or Value - Neil Fletcher of DB was another good speaker who started his talk by saying that pre-crisis people thought of data as project based, otherwise dealt with it on an adhoc basis and considered data as having no cost or value. Institutions had a spaghetti approach to data, with systems/projects being process not data based i.e. the systems get only the isolated data sets they need only when they need it.
• Quality is Now the Data Driver - Neil said that 18 months on from the crisis, then whilst ROI is still important for data projects then quality of data is the key driver.
• Sponsorship and Ownership of Data - Neil added that quality data is an asset as are the systems that produce data quality, and to ensure success data management projects needed high level business sponsorship, but also ongoing and clearly defined ownership of all data sets and their quality.
• Enterprise Data Virtualisation - Neil said that DB were embarking on a long term project to ensure that all systems get data from the same logical place on a global basis, and that they were investing heavily in data virtualisation technology as a key means of achieving this goal. DB are starting with reference data, moving to transactional/positional data and on to other data types. For each type/category of data ownership would be clearly defined across all systems and would enable real-time transformation of the data into whatever format it is needed in.
• Enterprise Data Model - Neil said that as a result of this virtualisation approach then you have to invest in putting together an enterprise data model for all data used in an institution. From my point of view this could be interpreted as a move back to "big EDM" (with all the project risk that implies) but I guess it is being approach on a more staged manner.
• Lip Service to Data has Ended - Neil summarised by saying that lip service to data management has ended with the start of the crisis and that 18 months on the enthusiasm for dealing with the data problem has not diminished.
• Publish/Validate/Subscribe - Simon Tweddle of Mizuho echoed a lot of what Neil said in approach to global data management and ownership, but added that he believed that the model of publish/subscribe needs to change to publish/validate/subscribe to ensure data quality.

Most of the panelists agreed that bringing in experience from external industries (Pharma, Oil & Gas, Internet Search etc) would be beneficial since we should not assume that the financial market has the expertise to get data management right first time (take a look at this article from the FT for a related idea). Martin of LCH.Clearnet was convinced that mandated data management would come and would be beneficial, which some of other panelists did not agree with and suggested that the industry needs to get ahead of the regulators to head this possibility off. Simon said that the focus on complex data/products was wrong given that the basics (what is our exposure to this counterparty?) were not being done (not sure I agree with this totally, both are needed given the losses from CDOs etc). Overall it was good panel with some interesting debate and speakers.

Beyond Golden Copy?

Interesting reading in a survey put together by Lepus and Thomson Reuters and publicised on Finextra this week. Summary findings:

• Data management budgets are increasing, with 77% of firms intending to increase spend on data quality and consistency and 32% saying spend would increase significantly.
• Tearing down data silos is a key initiative, 70% of firms are looking to revise data management solutions as a result of the crisis, and 31% of firms cited data quality and consistency as the most important driver.
• Data management for risk is the top concern, with 87.25% of firms looking to integrate data repositories in risk, and 62.5% saying that they were close/very close.

This seems to be consistent with another article on Finextra this week, with Deloitte predicting a much greater spend on risk management projects. Putting the marketing aspects aside for a moment, I don't think it is abundantly clear from the actual content of the Lepus survey as to why the title includes the phrase "...Beyond Golden Copy" other than the type of data management they refer to seems to have more emphasis on global/firm-wide data integration than your traditional EDM golden copy data warehouse approach.

It is also interesting to hear so much about consistent data across the entire enterprise (driven by risk and regulation) which seems to echo the "big EDM" projects of old that did prove that successful, and to some degree is at odds with what the likes of Golden Source and Asset Control are currently saying about choosing smaller projects to bite off on rather than the enterprise approach. I would suggest however that there is no issue in having smaller projects in mind so long as they are compatible with the overall goal.

The integration and consistentency of data across front, middle and back office was also interesting, and in particular the front office integration echos some of the things I have been saying about the need for analytics management and the management of front office data as part of the data management process, not something to be ignored in the hope it sorts itself out.