Monday, October 14, 2013

31000 Frameworks for Scenario Analysis

This week Causal Capital will be delivering a speech on how to 'stress test' an organisation for its resilience in managing catastrophes that originate from operational risk.

An associated draft white paper for this speech describes an entire framework for scenario analysis and can be found here LINK ].

Scenario Analysis
Scenario Analysis is often discussed among operational risk managers as the ideal way to scope and potentially quantify impacts from catastrophes. Yet, pulling all the parts together to ensure a scenario analysis framework meets its goal, is far from straight forward.

This paper outlines the actual steps required to build and then facilitate a successful Scenario Analysis Workshop from start to finish. The paper also covers the 'human' assessment methods for scenario analysis, as well as the modelling techniques required to make scenario analysis deeply useful.

Scenario Analysis Framework Whitepaper [ LINK ]


Please accept this is a draft working paper that is currently under review and amendments / revisions will be posted in due course as and when they are published.


Webinar Presentation
The Metricstream webinar will review this scenario analysis framework but will also describe other aspects of a leading best practice risk framework including:

[] Developing a robust matrix and taxonomy for monitoring losses.
[] Explain how to implement a set of scenarios for scoping the impacts from catastrophes.
[] Ensure that scenarios can be framed within a risk matrix so that they can be managed.

This is a free online webinar that is open to all people interested in risk management and more information on the program can be found here [ LINK ].

4 comments:

  1. Hello Martin,

    I enjoy your research and efforts to explain extremely complex data in an understandable format. I'm working in risk management from an accounting background and I am looking for coherent alternatives to risk assessment aside from probability x severity. I understand the weaknesses of that approach from articles by yourself and others, but what other tools are there to create normal distributions for assessing risks?

    I am not a 'quant' person, so I'm having trouble navigating through the whole risk assessment process without using wholly qualitative measures, especially when one is working in organizations that do not compile data.

    Also, if you happen to know what statistical courses are available online or elsewhere that you would recommend I take to be able to better understand the quantitative side of risk?

    Thanks for your help in advance and let me know how I can help you.

    Doug

    ReplyDelete
  2. Doug,

    Thank you for the positive comments and that is what the Causal Capital blog is attempting to achieve; explain specific risk management concepts and paradoxes in straight forward manner so that risk analysts can appreciate the various twists and turns required to take on a subject matter area in this vocational endeavor.

    There are a lot of blogs out there which simply don't go into enough detail to be anything other than a little innocuous when it comes to risk management. There is also a handful of blogs on the internet that perhaps dive too deep at times. What we are attempting to do here is to strike a happy medium.

    Probability x Severity is not a coherent measure of risk and it is deterministic as you rightly point out and we have debated this issue before on this site.

    Have a look at our Monte Carlo article as a solution to this problem http://causalcapital.blogspot.sg/2012/11/monte-carlo-example.html there is a spreadsheet that can be downloaded from that article which gives you a full working example on how MC can be used.

    In regards to getting up to speed with statistics, my recommendation is to acquire the knowledge in a practical manner. It is kind of frustrating learning the quant for the sake of the math yet not being able to practice the use of the skill. Here is an idea, why not try installing R-Project from this location http://cran.rstudio.com/ and go through one of the foundation manuals to learn R. In this way you will learn statistics but also how to apply the techniques you are reading about and in one of the best statistical tools around. What is also exciting is that R-Project is free !!!

    If you send through a request to this address https://sites.google.com/site/causalcapitalerm/contact-us, I should be able to share with you specific reading material to bring you up to speed with R-Project and the statistics it is designed for.

    On this point "I am not a 'quant' person" ---> Many risk people are not this and quants often don't handle the business requirements of risk management so well either. In my opinion, the best risk analysts are "all rounders", they know a bit about the statistics, a lot about risk, behavioral finance, valuation, they are able to pull together a risk report or design a database to hold risk variables. I suppose it is a never ending quest for knowledge with risk management but definitely an interesting road to walk.

    ReplyDelete
  3. Thanks for your ideas and recommendations Martin. I look forward to getting that reading material about R-Project and its statistics. Again let me know how I can help you.

    I'm currently working on a risk job in the public sector and was wondering if you have any reference material about public sector and risk.

    Thanks,

    ReplyDelete
  4. Thanks for your comment and your curiosity for reviewing risk management frameworks from the public sector is interesting. A lot of councils and governments haven't published their frameworks but North Sydney Council has done so and their draft risk management policy program can be reviewed here

    http://www.northsydney.nsw.gov.au/resources/documents/MS06_1.pdf

    It is an interesting document that explains the whole framework very well and specifically focuses on how the council assesses disruption and the associated recovery plan inline within the framework.

    ReplyDelete