The GIRO Report
by Louise Francis, Chairperson, Committee on the Theory of Risk
Editors Note: In addition to chairing the Committee on the Theory of Risk, Louise Francis participates on the International Research Committee. This is part of that committee's continuing series of articles on international research of potential interest to CAS members.
The 2003 GIRO Convention was held October 14-17 in Cardiff, Wales. (For those who are geography buffs, Cardiff is the capital of Wales). GIRO is a subset of the UK and Irish actuarial organization, the Institute of Actuaries and GIRO is an acronym for General Insurance Research Organization. General Insurance is the UK term for non-life insurance (which is mostly property-casualty insurance, but not limited to that).
Several CAS members, including Gail Ross and Raji Bhagavatula were on the program. Gail talked about the key insurance issues of concern to actuaries on this "side of the pond" and Raji reported on the asbestos and environmental liability issues.
Like the CAS, GIRO tends to have an applied orientation. That means many of the papers presented at GIRO are accessible to the typical CAS member, though some require more advanced technical knowledge.
Some highlights of the conference:
- Reserve cycle research
Research by a GIRO working party indicates that use of chain ladder methods for reserving appears to be associated with cycles in the adequacy of loss reserves. The research found that when a chain ladder reserving method is used and mechanical procedures are used to select factors (thereby removing subjectivity from the process) cycles in reserving occurred which tracked fairly well the underwriting cycle. That means reserves were less adequate in soft markets and more adequate in hard markets). The finding applied to both incurred and paid loss chain ladder methods. The working party did not know why this happens but they had some suggestions. The issue will be studied further in 2004 by a working party.
- Extreme value theory
Most financial data follows a heavy - tailed probability distribution. Hence, the normal and lognormal distributions do a poor job of approximating the data. Two papers presented at the GIRO conference addressed this issue. The first, "Extreme Events", by David Saunders was subtitled "The Overthrow of Modern Financial Theory" because Financial Theory is heavily dependent on an assumption of normally or lognormally distributed data and does a poor job of approximating the behavior of real financial variables when it comes to extreme events. The regular occurrence of events assumed to be extremely uncommon was cataloged by the author. Then, alternatives, such as catastrophe theory and chaos theory were introduced. The second paper "The Importance of Being Normal" by Stephen Carlin and Andrew Smith presents a specific alternative to the normal/lognormal model: the conic moment or Normal Inverse Gaussian distribution.
- Correlation research
The topic of dependencies between lines of business was tackled by Papachristou and Grenon. The authors presented methods of modeling dependencies based on copulas. Four copulas, the Clayton, Student-t, Gumbel and Bivariate Normal were covered. They also discussed using extreme value theory to model tail dependencies. In addition, they addressed approaches which could be used in the absence of adequate data.
- Enterprise Risk Management
The working party on Operational Risk produced a paper on the topic which discusses and incorporates much of what has occurred to date in this field. Operational risk addresses such risks as lapses in internal controls and systems and the impact of some external events on company profitability. There is little data available to measure such risks (although the paper indicates a database for measuring bank operational risk has been assembled) and such risks are considered difficult to quantify. The Working Party reported on the results of a survey sent to managements of Insurance Companies, Regulatory Agencies and Risk Management firms. The working party also produced a case study (using fictitious data) to help illustrate concepts on how to quantify operational risk.
- Solvency Monitoring
- The Risk and Capital Assessment and Supervision in Financial Firms produced a paper on Solvency Monitoring. The paper summarizes and critiques the work going on in this area. The paper discusses many of the risks that are relevant to solvency and discusses methods for modeling and managing the risks.
- A loss modeling team from Lloyds presented their paper ""The Overhaul of Lloyd's Realistic Disaster Scenario", where they attempt to present improved methods for modeling and quantifying the impact of catastrophes (including those from natural catastrophes and liability disasters).
- Relationships with Underwriters
On a more practical level the paper "Developing Relationships with underwriters", discusses ways to improve communications between underwriters and actuaries, a perennial issue for both UK and US actuaries. The working party authoring the paper presented an entertaining and humorous video at a GIRO plenary session.
The papers presented at the conference are available on the institute's web site.
The GIRO actuaries have a great sense of humor and know how to have a good time. This was manifested in some of the videos that were played at the conference which were nearly as good as some of those famous BBC comedies that can be viewed in the US on some of the Public Television channels. The conference also featured a tradition that I rather liked: the awarding of a bottle of champagne to the person who asked the best question at each plenary session. The banquet held the second day of the conference featured entertainment by a harpist and by a Welsh men's choir. The Welsh take their men's choirs very seriously, as each town has a choir and a national competition between the choirs is held each year.