Showing posts with label complexity. Show all posts
Showing posts with label complexity. Show all posts

Dec 6, 2010

Actuarial Simplicity

What is simplicity? What's the power of simplicity?

Goethe
It was Johann Wolfgang von Goethe ( listen), a German writer (poet), but also a polymath (!), who
stated
:


And indeed Goethe was right, in (actuarial) science and  practice it's the challenge of overcoming (transcending) this paradox of simplicity and complexity.

The art of actuarial mastership
As models become more and more complex, it takes the art of actuarial mastership to condense this complexity into an outlined, understandable and (for the audience) applicable outcome.

A 'best practice example' of condensing complexity into a powerful inspiring statement, is Einseins famous equation E=MC2 :

Like Paulo Coelho states in his blog about Einstein:
A man (actuary) should look for what is, and not for what he thinks should be. Any intelligent fool can make things bigger and more complex… 

It takes a touch of genius – and a lot of courage to move in the opposite direction.

Or, to quote Einstein:

Everything should be as simple as it is, but not simpler


How to cut through the actuarial cake?
Just three simple examples on how to cut through the complex actuarial cake. Examples that might help you to simplify complexity:

1. Think more simple

A perfect example of 'thinking more simple' is finding the solution of the next math problem (on the left), grabbed from an old high school math test.

Can you solve this problem within 10 seconds?

Found it? Now move your mouse over the picture or click it, to find the refreshing simple answer.....


Remember however not to oversimplify things. Sometimes problems need the eye of the actuarial master to identify important details...



2. Visual Results
Second example is to visualize the outcome of your models instead of power-point bullet conclusions or explaining how complex your model really is.

A nice example is the online dollar-bill-tracking project "Where's George?" from Research on Complex Systems, that measures the flow of dollars within the U.S. (over 11 millions bills, 3109 counties).
About 17 million passengers travel each week across long distances. However, including all means of transportation, 80% of all traffic occurs across distances less than 50 km.
One picture says it all and 'hides' the complex algorithms used, to get  stunning results.

On top of, George collects relevant data about 'human travel' that could be used for developing models of the spread of infectious diseases.

Just take look at the video presentation of George called Follow the Money to find out how to extract simple outcomes from complex models.

One of the simple results (by Brockmann) of this project is that the probability P(r) of traveling a distance (r, in km)  in a short period of time in days (max 14 days) can be expressed as a power law, i.e.:

P(r)= r -1.6

 3. Listen better
Every (actuarial) project outcome fails if there's no well defined goal at first.

Main problem is often, that the client isn't really capable of defining his goal (or problem) very precise and we - actuaries - start 'helping' the client.  In this 'helping' we are imposing our thoughts, beliefs and experiences onto others, by what we think 'is best' for the client. The outcome might often be an actuarial solution that fits the problem in our own actuarial head, but fails to meet the clients problem.

Main point is that we - as advisors - don't really listen well.
Of course that doesn't apply to you as an actuary personally, but it does apply to all other qualified actuaries, doesn't it?

Just to test if you're part of that small elite troop of 'well listening qualified actuaries' (WLQAs), just answer the next simple Client Problem:

Client: I'm confused about 'distances'. It turns out that measuring the distance between two points on earth is really complicated math, as the world is round and not flat.

But even in a 'flat world' I find measuring distance complicated. As an actuary, can you tell me:


What’s the shortest distance between two points in a flat world?

O.K. Now think for a moment.....

Have you got the answer to this complex client problem?

Now that you're ready with your answer, please click on the answer button to find out the one and only correct answer.
The answer is: the shortest distance between two points is zero
Hope you safely (without any mental damage) passed the above WLQA-test......

A Simple Application
A nice demonstration of actuarial simplicity is the well known 'compound interest doubling rule' that states that an investment with compound interest rate R, doubles itself in N≈72/R years.

So it'll take (p.e.) approximately N≈ 12 (=72/6) years to double your investment of $100 to $200 at an compound interest rate of 6% p.a.

While the precise equation of the doubling time is quite complex to handle, it's approximating equivalent, N≈72/R, is simple applicable and will do fine for small size compound interest rates.


It's our actuarial duty and challenge to develop simple rules of thumbs for board members we advice. We actuaries have to master the power of simplicity. Let's keep doing so!

Related links:
- The Complexity of Simplicity
- Where's George?: Wikipedia
- The scaling laws of human travel (2006)

Dec 11, 2009

Systemic Risk

In an excellent paper called 'Defining and Measuring Systemic Risk', professor Sylvester Eijffinger of the Tilburg University discusses actual developments around one of the most interesting risk topics of this moment: systemic risk (not to be confused with systematic risk).

Just a short warming up to actually download and read this excellent article:

ESRB
Main target of the 2010 launch of the European Systematic Risk Board (ESRB) is trying to identify and avoid future financial crises before they start. This implies that ESRB's main issue is 'how to detect systemic risks '. All this -of course - under the lead of the European Central Bank (ECB).

First of all the ECB does not have a clear concept of systemic risk, nor in the academia there exists a generally accepted definition. However, the G10 definition provides a good starting point:

Systemic risk
Systemic risk is the risk that an event will trigger a loss of economic value or confidence in, and attendant increases in uncertainty about, a substantial portion of the financial system that is serious enough to quite probably have significant adverse effects on the real economy


This still sounds pretty complex, and it is.
To get the right feeling, take a look at the next diagram illustrating a network of Credit Default Swaps (CDS) contracts:

In his blog 'complexity is our enemy' Steve Hsu, Professor of physics at the University of Oregon, explains in short and in simple words the principles and problems of the Credit Default Swap Market.

Hsu perfectly illustrates why some financial institutions are 'too connected to fail', as opposed to 'too BIG to fail'. Systemic risk is all about complexity.

New early warning models

There are several new models that can predict a financial crisis. Key challenge is to find a model with an indicator that predicts a potential crisis (just in time) with high probability, while at the same time minimizing errors of type I errors (missing crises) and type II false alarm).

One indicator can be qualified as the best current performing indicator: 'The global private credit gap', by Alessi and Detken (2009). This method predicts 82% of the crises correctly and has a 32% share of false alarms. 95% of the crises (price boom/bust cycles) are signaled in at least one of the 6 preceding quarters and the difference in the conditional and unconditional probability of a boom following a signal is 16%

Individual Institutions’ Contribution to Systemic Risk
For measuring risks of individual banks, a measure called CoVaR was developed by Adrian and Brunnermeier. The CoVar model measures the marginal expected shortfall (MES) as used in Value at Risk (VaR) as well as the systemic expected shortfall (SES).
Eijffinger's Conclusion
Finding new early warning instruments that are effective, easy to use, and independent of the interest-rate instrument seems to be an impossible task. And yet there is a solution according to Sylvester Eijffinger: "Central banks should give the growth of (broad) money supply more prominence in their monetary policy strategies."

The ECB with its often criticized monetary pillar may have a head start. Important central banks, such as the Bank of England and the United States Federal Reserve, kept their key interest rates too low for too long leading to a long period of double-digit growth in money supply.

The ECB was more cautious. To be sure, the fall of he risk premium on financial markets, the development of all kinds of exotic derivatives, and these derivatives’ subsequent misuse sowed the seeds for this crisis, but those factors could not have caused the crisis without the plentiful rainfall that allowed those seeds to grow.

Finally
What can pension funds and insurers learn from this?
The answer is simple:
  • Make Risk Management top priority nr. 1
  • Develop and implement in advance - cross financial institutions - early warning models.
  • Insist upon regulators to create a world wide central registration data base that registers and reports all possible derivate transactions in the financial market. Every financial institution has to report every transaction in a preformatted form.
  • New financial products are subject to approval ('no objection') by the regulator before market launch.
This way regulators will have a complete transparent view cross financial institutions. Systemic problem solved.

Sources
- Eijffinger:Defining and Measuring Systemic Risk
- The global private credit gap
- CoVaR modelM
- Steve Hsu: complexity is our enemy