We are ALL funding an industry’s “prediction addiction”
Wednesday, 9 January, 2013 Leave a comment
We “know” (well, understand) that we cannot predict the future – which should be pretty worrying for the financial sector, whose success or failure relies upon the frequency and cost of a variety of events that haven’t yet and may never happen. Except that, the expiry date is approaching, for relying upon a steady stream of (mis)information, discredited economic theories, spending vast amounts on personnel and technology that convey the impression of knowledge.
In the absence of “special powers”, insurers have to rely upon what they ‘know’ i.e. what they have learnt from the impact and frequency of past events, that happened to OTHER, similar, risks!
We now have vast quantities of data, accumulated over many years, from a wide variety of sources. The type of information with which Statisticians, Actuaries, Economists (and Carol Vorderman) can have hours and hours of fun, aided by tried and tested techniques, using the most sophisticated technology in our history. But it doesn’t change the basic fact that we cannot predict the future although we must learn from past events.
The invaluable lessons for our man-made world are, that:
- non-linear [real world] interactions CANNOT be modelled – concatenated probabilities are still linear
- we should question what we think we know – you know what they say about assumptions!
- we cannot manage risk – we CAN influence what is within the scope of our control
- conventional tools CANNOT identify, map or measure complexity
- resilience is a function of complexity
- resilience (or, per Nassim Taleb “anti-fragility”) should be our primary concern
- we can learn many more lessons from nature
Insurers continue to “dither”, despite recognition of complexity as a source of risk; the effect of inter-connectedness upon probability; the glaring limitations of conventional rating; lack of reliable current financial (risk) data; potentially misleading analyses of pre-digital era stats. Too many people are clinging dearly to techniques they have grown up with…because, well, it is what they know(!) even though the pace and demands of inter-connected business systems has outstripped conventional capabilities.
The insurance industry is geared for “known knowns” i.e. known probability [frequency] known financial impact [cost], but as its policyholders are ill-equipped for the impact of uncertainty i.e. high impact, low probability events, insurers are unable to accurately rate for such events so are ill-prepared to absorb associated costs…and not all such events fall into the “Acts of God” category!
We can overlook or ignore risk but it doesn’t dissipate, it is amplified and comes back through external interactions [feedback loops] as uncertainty: the butterfly effect. FRAGILE
Interrogate the data, measure complexity, and uncertainty can be reduced whilst, previously unseen, sources of risk are identified: extending the “risk horizon”.
All parties benefit from a culture of loss prevention: RESILIENT
Think about it this way. If you go to your Doctor and are diagnosed with a life-threatening illness, would you rather a team of specialists:
set to work calculating the likelihood of your survival based upon your symptoms and historic data about individuals with a similar health profile to your own?
conduct further investigations to identify and assess the nature and extent of the illness, to tailor treatment to your specific needs?
told you about how your lifestyle [risk culture] has been a major contributory factor and sell you on the notion that a homeopathic remedy (i.e. without any scientific basis) may, somehow, cure you?
Complexity is a measure of the total amount of structured information (which is measured in bits) that is contained within a system and reflects many of its fundamental properties, such as:
- Potential – the ability to evolve, survive
- Functionality – the set of distinct functions the system is able to perform
- Robustness – the ability to function correctly in the presence of endogenous/exogenous uncertainties
In biology, the above can be combined in one single property known as fitness.
Like any mathematically sound metric our complexity metric is bounded (metrics that can attain infinite values are generally not so useful). The upper bound, which is of great interest, is called critical complexity and tells us how far the system can go with its current structure.
Because of the existence of critical complexity, complexity itself is a relative measure. This means that all statements, such as, “this system is very complex, that one is not”, are without value until you refer complexity to its corresponding bounds.
Each system in the Universe has its own complexity bounds, in addition to its current value. Because of this a small company can, in effect, be relatively more complex than a large one, precisely because it operates closer to its own complexity limit.