The Point About Stress Tests
This morning I could be back at school or University, reading this; "City traders are braced for turmoil in UK financial sector stocks this week, when the Federal Reserve releases results of health checks on 19 US banks on Thursday. At least two big banks are believed to have failed the "stress tests"." from here: the Herald.
But it's just not the point! The article paints a picture of not only the financial markets but also the risk departments of the banks themselves hanging around behind the bike sheds in short pants waiting to get punished as the government, like the headmaster in the shape of the US Treasury reads out the results of stress tests (like examinations) which we know that two of the pupils have failed. It's ridiculous, the British Bankers Association yesterday warned the UK Treasury Select committee to stop criticising bank management! Until managers in banks do a little growing up and conduct their own stress tests to produce their own internal ratings of risk and their own estimates of economic capital then they are going to be treated globally not only by the authority of the state but also by the body politic as recalcitrant teenagers.
Why don't they realise (bank management I mean) that they have run out of road. They have been hanging around on the sidelines of the pitch rubbing tearful eyes complaining that statistics was "just too hard" for them (awwwwwwh) and they could not do it. They have been doing this for at least 6 or 7 years, Basel II made it crystal clear that they had to participate in a modern supervisory framework which required banks to conduct internal stress testing to quantify economic capital which supervisors would review and agree. They did not do it. Now (as we have blogged elsewhere on this site) a recent conference in Chicago, sponsored by REvolution Computing and MICROSOFT HPC has published detailed techniques on how to do stress testing in a Free (yes FREE) Open Source development environmnent called R. Still the banks are looking for a way NOT to do this! And they complain at the same time that the state is doing it for them. After they have created a financial and economic catastrophe the likes of which only my grandfather could remember!
I give up! Clearly the academics and some of the software industry is taking its social responsibilities seriously but the banks are just tripping out 5-bullet lists of so-called good practice, the academics and the software engineers are stepping up to the plate, they can do no more, it seems to me, if the management incumbent in the banks will just not take its legal and social duties seriously and engage with the rest of society. There is no room for relcacitrant teenagers in senior positions in society anymore, they hold us all hostage to their pecadillos as they dance angels on pinheads avoiding what is required of them. Until bank management get a grip, accept that what they have been asked to do by Basel II is the only way forward and that expertise and technology exists NOW and is available to them and that they must deploy it then disruption to market activity and banking generally by state interference is inevitable with the consequnet constraints upon the liberty of us all and disruption to free market pricing of credit and liquidity.
Rant over!
Are the key legislative pillars such as Basel II & III, UCITS IV and Solvency II forcing you to re-examine how you identify, measure and manage risk and capital?
Is the goal of your website to sell services or products, educate, or collect data?
Comments
Stress testing credit risk: a survey of authorities' approaches
by Antonella Foglia, Banca d'Italia, Banking and Financial Supervision
This paper reviews the quantitative methods developed at selected authorities for stress testing credit risk, focusing in particular on the methods used to link macroeconomic drivers of stress with bank-specific measures of credit risk (macro stress test). Authorities with a mandate for financial stability are particularly interested in quantifying the macro-to-micro linkages and have developed specific modeling expertise in this field. Stress testing credit risk is also an essential element of the Basel II Framework. The paper highlights recent developments in macro stress testing and details a number of methodological challenges that may be useful for supervisors in their review process of banks' models as required by Basel II. It also contributes to the on-going macroprudential research efforts to integrate macroeconomic oversight and prudential supervision, for early detection of key vulnerabilities and assessment of macro-financial linkages.
http://www.bancaditalia.it/pubblicazioni/econo/quest_ecofin_2/qf_37/QEF_37.pdf
Stress Testing: What is Required and How to do it
I have split the references between 3 domains;
1) Requirement Definition,
2) Predictive Techniques in R and
3) Solution Architecture considerations.
As a final piece I am adding a section on my current interest which is the integration of an external structural model (such as the European Commission DG ECFIN’s AMECO) with your own Stress Testing platform (probably implemented in the Factor Model & optimally implemented in REvolution R). You are then left with the question of ‘What is my data management platform?’
All of the references below are given as Hyperlinks to help you navigate around the content on this site in a structured way.
The final item on the list (in the Structural Models Integration section, which is by far the most difficult of an extremely stretching body of work, let’s face it) is a brand new toy for all you quants out there to play with!
It’s a new ‘windows-like’ user interface to the Dynamic Stochastic General Equilibrium (DGSE) Structural model of the European economy and each of the member states, developed and maintained by the European Commission Directorate General Economic and Monetary affairs (DG ECFIN). The model is called AMECO and it allows you to download all of the time series in the model as input to your Stress Test Factor Model. It is straight off the production line, hot-off the presses, having been released by DG ECFIN at the Brussels Economic Forum last week.
Think of it as a reward for labouring through all of this content, in Quant-land, this new tool is like someone delivering a new E-type for you to play with or an Overfinch conversion, whatever floats your boat.
1) Stress Testing - Requirement Definition
Credit Economic Capital & Financial Predictive Analytics – White Paper 2009.
European Central Bank: Integrated Economic Capital Models
Optimization Heuristics for Determining Internal Rating Grading Scales
Papers in April (2009) from the Bank for International Settlements
Financial Sector Pro-Cyclicality Lessons From The Crisis
Academic & Central Bank Papers on the topic of Risk Management
Economic (Risk) Capital - How To - References
Comparison and Pooling of MacroEconomic Forecasts - 2 interesting papers
The Asymptotic Single-Risk-Factor (ASRF) model Specification and Calibration Errors
The Case for Fully Integrated Models of Economic Capital
How to find plausible, severe, and useful stress scenarios - Oesterreichische Nationalbank
Firm Default and Aggregate Fluctuations
Loss, Default, and Loss Given Default Modeling
INFINITE-DIMENSIONAL VARS & Factor Models
Stress testing credit risk: a survey of authorities' approaches
2) Predictive Techniques in R
Economic Capital & Bank Stress Testing - The Optimal R Object Set
Presentations from the R in Finance 2009 Conference
Fitting mixed models
Analysis of Integrated and Cointegrated Time Series with R
Optimize your R development experience
How to get IT to accept, and love, R
Statistical programming with R
Portfolio optimization in R
Nonlinear Regression with R
Computational Finance with R
3) Solution Architecture Considerations
SAP White Paper on a Solution Architecture for Economic Capital Quantitative Analytics, pre Credit Crunch
Important References – Risk Management and Financial Predictive Analytics – A discussion
High Performance Backtesting - REvolution R with Vhayu
The Mega-Vendor road-map for Data Management consistent with High Performance Predictive Analytics
Bringing Open Source Best Practices into Corporations Using a Software Forge
ZOA an Open Source Framework for SAP ABAP
TOGAF™ Version 9
Enterprise Service Bus in Banking SOA Architecture
The MICROSOFT Enterprise Data Warehouse
REvolution R Enterprise with Parallel Processing Now Available for 64-bit Windows
4) Integration of an external Structural Model with an internal Factor Model for Stress Testing
How to traction the European Commission Structural Economic Model of the EU for Stress Testing
DSGE Model-Based Forecasting of Non-modelled Variables
A method to incorporate information from Dynamic Stochastic General Equilibrium (DSGE) models into Dynamic Factor Analysis
Testing a DSGE Model of the EU Using Indirect Inference
Macroeconomic Models of the Euro area at the European Central Bank
DG ECFIN’s macro-economic data series on the internet (AMECO)
AMECO ONLINE
CONCLUSION
Stress Testing is difficult! My personal experience of that is all the evidence I need to understand that proposition. Alongside a unique group of people we published the first road-map on how to do it properly at SAP in April 2006 but the resistance of the banking industry to our proposals was immense, the criticism was personal. All the Credit Crisis has proved is that we were right all along. The fact that most of the people who made decisions in this space at that time are no longer in position is testimony to that. But the way in which we at asymptotix monitor the legislative and supervisory landscape is telling us that in 2009 / 2010 you will require to implement a Solution Architecture along the lines presented in that 2006 White Paper. Asymptotix can be pretty Euro focused in its research but REvolution keeps us aware of developments in the US. Euroland and Obama’s US are genuinely very interlocked now under the auspices of the IMF, they are both dealing with the same key issue at government level related to Stress Testing which is fair valuation of that enormous stock of ‘impaired assets’.
But its not supervisory enforcement which will drive you towards implementation of this kind of Solution Architecture alone, there is a business driver and that is the re-opening of the wholesale credit markets. Neither Europe nor the US can meet the ‘demand for money’ of society without a wholesale credit market which channels savings directly from surplus institutions to credit (banking) institutions and leaves the equity market to do its job with risk capital. What the ebb tide of the crisis has revealed is that the banking system simply does not have the “Balance Sheet Width”. A wholesale credit market will never again look like it used to do, it will require to have personnel with at least one eye on their stewardship role to manage it and the manifestation of that stewardship, the ‘disclosure’ of it is that Stress Testing engines which compute forward looking ‘economic’ quantities of risk are in place. In short; Financial Predictive Analytics. Given the transparency ethos of life going forward from here, there is no logic which I can think to counter the proposition that will have to be ‘Open Source’.
So, you will have to DIY! Once granny state has done enough Stress Testing for you, you will have to go the garden centre or the hardware store and buy the tools to renovate your risk management platforms! Right? This above has been a wee exposition, a flyer if you like to give you a picture of what you might need to be buying and why, a plug a play guide to a very complex requirement for this new world in which we live, it will not always remain like it is now but we have all to do our bit to get ourselves out of it. This is an attempt to help start that process.