Stress test stress test

How did you interpret the title of this Blog......? 

Can you read it in more than one way? In how many ways can you read it?

Still confused?

Enough questions to start this blog.

In short: the more ways you are able to read this title, the more successful you'll probably be in defining and executing 'Stress Tests" in practice.

Let's dive a little deeper to illustrate this important 'multi interpretation talent' you need, to make stress testing a success.

Although there are far more ways (please add some interesting suggestions as comments to this blog) to interpret the title 'stress test stress test', we'll analyze in this blog two interesting  interpretations that follow from the fact that 'stress test' can be interpreted as a noun or as a verb.

1. stress test [noun] stress test [noun]

Interpreting the title as two nouns could:

• make you aware of the importance of stress tests
• emphasize the importance of repetitive execution of stress tests
• illustrate the feeling of disinterest and apathy that occur when important words are repeated to often without enough plowing depth..
  
  

2. stress test [verb] stress test [noun]

This perhaps the most interesting interpretation.
How can you really stress test your stress test?  

The way we stress test at financial institutions like banks, insurance companies and pension funds, is basically more or less as follows:

• Project historical crisis crash-data into the future. Simulate what would happen and take a look at the consequences
  
  
• Test crash scenarios on basis of the question: What would happen if.... (prices go down, S&P 500 collapses, etc., etc.)
  
  
• Take several economic scenarios. Project them on your balance sheet and see what happens.

To stress-test a stress test we have to develop a different view on stress tests.

A view based on the answer of the next leading question:

How many sides has a coin?

Let's demonstrate this new crucial view on a stress test.

A Different View on Stress Tests
Some examples:

• Inverted Stress Test
  An interesting way of stress testing is 'working the other way around': Try to define financial situations where you never want to end up in (e.g. equity< -5%, etc.) and try to imagine scenarios that could lead to this unwanted financial situation  Paul Duijsens, ALM Principal Mercer Investment Consulting, mentions this approach).
   
• Idiot Proof Stress Test
   Andrea Enria, chairman of Europe’s new banking regulator stated recently:
  
  A stress test is only as good as the scenarios you plug into it
  
  Therefore, make stress tests 'idiot proof' as much as possible.
  Once a stress test is developed, don't present it to the board directly. Organize a 'second opinion' from a professional company that's undependable and critical enough to seriously test and analyze your stress test and its assumptions.
  
  Presenting a stress test without clear statements about the limits, vulnerabilities, constraints and shortcomings (every test has shortcomings) is like playing with fire and  offering your board 'the wrong end of the stick'.
  
  So we can add another conclusion quote:
  
  A good stress test transparently presents its weaknesses
  
  If nobody can find a weak spot spot in your stress test: ask a 5 year old child to ask some simple questions.....
  
  
• Compare Stress Tests
  Please  think about :
  
  'One stress test' = 'No stress test'
  
  Comparing successively executed stress tests on assumptions, methods and outcomes is essential for a correct understanding of the impact and consequences.
  
  Not only compare tests of your own company, compare also with tests of other financial institutions. Questions like: why do we as a [pension fund]  have different assumptions and methods than an international [bank], are key to a correct understanding of your own risks.
  
  
• Surpass Regulator Constructed Stress Tests
  Regulator Constructed Stress Tests(RCS-tests) seem relatively easy to implement.
  As a risk manager or board member you don't have to think about scenarios or methods. That's all been taken care of by the Regulator. Easy, isn't it?......... Wrong!
  
  Heedlessly implementing RCS-tests is risky. First of all, the regulator's principles are partly biased. As an example, take the risk of treasury bonds on your balance sheet.  Treasury bonds are commonly seen (and valued) as save (AAA-rating). However, some countries (Greece, Spain,Ireland etc.) have already been downgraded. Which countries will follow? Does the RCS-test includes this non-hypothetical risk? No? Does your own test includes this risk?
  

• By definition regulators have to act and communicate in a responsible and 'prudent' way about government financial issues. If they wouldn't, world wide financial chaos would be the inevitable result.
  
  
  The other side of this 'prudent coin' is that the actual risks are probably larger than can be concluded from the government (treasury) interest rates and interest spreads in a particular country. Here you'll have to develop your own risk model or - if data fail - formulate  your own risk approach (get out!).
  
  Key is that, given the general level of  systemic risk, all financial institutions must be able to withstand haircuts on all their own sovereign debt holdings.
  
  A 'third side' of this coin is the fact that regulators (in time) might decrease risks on certain asset categories that are not in line with your own risk view. Stay awake to prevent from becoming 'Supervisory Stress Compliant'.........
  
  

• Unmask Derivatives
  Market valuation with respect to derivatives is tricky business and probably only valid as long as there's a 'normal' market activity. Nobody is able to value derivatives under severe market conditions as is the case in stress tests. So, depending on the size and characteristics of a stress test, don't hesitate to to unmask your derivatives by applying a large discount on the value of your derivatives.
  
  

Conclusion
Stress testing is not for dummies, but for professionals.
It turns out that the more you're able to look different, critical and 'out of the box', the more stress testing will be successful.

Making your audience aware that a coin has three sides instead of two, is probably the essence of an actuary's or risk manager's profession. 

It has become clear that analyzing assumptions, models, outcomes,constraints and shortcomings of a stress test is no superfluous luxury. So stress test your stress test!

Related links:
- KAS BANK develops stress test for UK pension funds
- Concerns over latest EU bank stress tests
- EU bank stress tests, a joke (2011) 
- Lateral Thinking:  US Economy Stress test (2011)
- World Wide Interest Spread by Country (2011) 
- Government Bonds yields 10 Year Notes 
- How many sides has a coin? 
- Worst-Case Scenario Survival Card Game 

Zero Problems Risk Management

More and more we actuaries and risk managers become aware that our risk models can't just be based on numbers and statistics exclusively.

Some examples:

Systemic risk

The recent financial crisis made it clear that a 'mono risk approach' on a sole risk-object (mortgage, fund-investment) is insufficient.

Investments and loans are embedded in a worldwide sea of connected financial instruments and reinvestments. Systemic risk has to be included in our models.

Main challenge here is that systematic risk essentially depends on macroeconomic and (mostly) irrational factors. Further, systemic risk is related to the structure and dynamics of the market. More than numbers.....

Supervisory Herding Risk

In their effort to control and support financial institutions like banks, pension funds and insurance companies, country supervisors, regulators and 'accounting standards boards', defined a meticulously set of guidance rules (Basel I/II/II, Solvency I/II, Qis-I-V, IFRS, FAS, AIFMD, FTK, FIRM, etc.,etc.)

Financial institutions not only confirmed and adopted to those new rules, but - in their rush and driven by cost and time pressure - also implicitly (and often unintentionally) declared those same imposed rules and rationales as their own business 'Risk Appetite'. This way, most financial Institutions became so called: 'Supervisory Compliant'.

Instead of  expliciting their specific company-targets and successively developing their own correspondent risk appetite and risk framework, they incorporated the supervisor's risk philosophy. 

Without a sound own (board) risk vision that would undoubtedly have included some extra safety on 'company specific risk issues', financial institutions became - like a herd - all in the same way extremely vulnerable to (less defined) external risks.

Summarized:

Overregulation increases Herding Risk

Financial institutions all measure and respond to regulated risks in the same way. Supervisory Herding Risk is born.....

Too Much Focus Risk

As a consequence of pre-subscribed risk categories and ruling by law or (accounting) standards, there's the risk of 'too much focus' on specific risks while forgetting, denying or neglecting other important risks. Remember, the devil is in the (correlating) details....

Here's a useful, but not exhaustive, checklist to keep track on your risk models...

Average Premium Risk	Diversification Risk	Matching Risk
Commodity risk	Employer Continuity Risk	Operational risk
Compliance Risk	Environmental Risk	Outsourcing Risk
Compliance Risk	Equity Risk	Oversight Risk
Concentration Risk	Herding Risk	Price Inflation Risk
Counterparty Risk	Interest rate risk	Property Derivatives Risk
Coverage Ratio Risk	IT Risk	Reinsurance Risk
Credit Risk	Legal risk	Reinvestment risk
Culture Risk	Legislative Risk	Reputation risk
Currency Risk	Liability Risk	Sex Calculation risk
Default Risk	Liquidity risk	Strategic risk
Deflation risk	Longevity Risk	System Risk
Disaster Risk	Market Risk	Systemic Risk
Discount Risk	Matching Risk	Wage Cost Inflation Risk

ALM Simplifying Risk

Univariate models are killing and even multivariate models have proven to be too vulnerable and too limited in the recent crisis. It's not just about correlation and covariance matrices. What we need is an self-explaining model. A model  that predicts or generates expected values in an economic context, depending on exogenous economic variables like inflation rate, GDP-Level, etc. and that is based on the same structured historical economic data-set.

We need 'Asset Liability Modeling New Style' and not only Stress Testing or
advanced and excellent Crash Modelling as well explained by EMB.

Geopolitical Risk

With Europe and Japan as recent examples, it's clear that risks come from everywhere around the world.

The consequence of earthquakes (Japan, Australia), a possible  country default (Ireland, Greece, Portugal, ..), political instability (Libya, Ivory Coast, ..), war threat (Vietnam,Iraq, ..), financial easing (US, Europe,...), on our economic system, prices and financial institutions seems substantial and - moreover- predictable.

More than just trying to catch and capitalize these kind of risks in our risk models, we need to develop (financial) mechanisms and products that can cope as best as possible with these kind of risks.

The Riskmap 2011, Managing Risk | Maximising Opportunity, offers a good description of the actual risks that influence our lives and risk models.

A nice example is the recent (unexpected) leading role that France took in action against Libya.  'Riskmap 2011' mentions the 'Arabic Poll 2010' that clearly shows (despite the lack of sympathy for president Sarkozy) the trust and sympathy for France. France clearly outperforms the US and president Obama  unfortunately has lost the trust of the Arabic world... Take a look at the next slide summary (or the original complete pdf):  

Arab Public Opinion Poll 2010 Summary

Arab Public Opinion Poll 20... by Jos Berkemeijer

The Arabic poll shows that the prime minister of Turkey, Erdogan, has clearly gained  the confidence and trust of the  Arabic countries. With Ergodan, Turkey - at the cross road between East and West - takes a leading role in the 'World Risk Management Process'.  Ergodan's Risk Philosophy, invented  by the Turkish Foreign Minister Ahmet Davutoglu,   is 'Zero Problems'.....

Perhaps that should be the philosophy of actuaries too...

Zero Problems

Conclusion

From now on 'Modeling Risk' is more than just a financial exercise.
It's building scenario's, mechanisms and products that can cope with this risky world.  Success as actuary or risk manager!

Related Links:

- Committee (behaviour) assessment tool
- Control Risks:Riskmap 2011
- Arabic Poll 2010
- Supervisory Compliant
- Maplecroft Risk Maps 
- EMB: How to Model a Crash (REVO) 

Fukushima Risk Management

With all due respect for the Japanese risk managers, engineers and the Japanese people in general: The latest 2011 M=8.9 Earthquake in Japan made clear that the current Risk Models, as well as Risk Management itself, have tragically failed.

Shortcomings
Shortcomings in headlines:

• Underestimation of the probability of a M=8.9 (M>8) event
• Risk Modeling mainly focused on 'direct' earthquake effects
• Underestimation of cumulative correlated devastating effects that occurred as a consequence of a combination of Earthquakes, Tsunamis, Nuclear Disasters and Physical Concentration (4 reactors in a row!).
  
  

Frequency?
It's true, the worldwide annual frequency of a serious earthquake with a magnitude of 8 or more (M>8) seems small....

Yet - no matter how low the probability- the consequences (loss) is unacceptable if the damage in case of an event (like an earthquake M>8) is not manageable, exceeds a country's financial recuperation power, or devastates thousands of lives...

End of Time?

Lots of worrywarts think the end of time has come, as severe earthquakes are increasing.

Simple statistics show that this is not (yet ?) the case:

The expected average Magnitude value of all earthquakes with M>=7 over the period 1950-2011 (March) is M_a=7.6 with a standard deviation of SD= 0.4. The average value of the last 10 years turns out M_a=7.4 with the same SD (0.4).

Nevertheless, when we examine the graphic more closely, there seems a light (significant?) increase in large magnitude earthquakes in recent decades.....


Back to the year 0
For what it's worth, the average magnitude for M>=7 earthquakes over a longer period from 0 to 2011 (March) , turns out in line with M_a=7.6 and a SD=0.44. Or in graphics:

History Earthquakes in GoogeMaps
Finally, take a look at the history of all earthquakes with a magnitude of 7 or higher on GoogleMaps... (take look at your country!)


Earthquakes, Magnitude> 7, years 0-2011 weergeven op een grotere kaart

Let's hope and pray for the Japanese people that that the brave engineers who are fighting the current meltdown at Fukushima will succeed and survive.....

Related links
- Spreadsheet (Excel) Earthquakes 0-2011 (March), M>=7
- Crash Course Earthquake calculation by Professor Nicholas Pinter
- Effect of Seismic Risk Measures on Japanese Housing Prices
- Earthquake Software
- Download Earthquake data from USGS
- Earthquake Safety of Nuclear Power Plants
- PPT:Japan’s Nuclear Energy Program(2003-2004)
- Design criteria Japanes Nuclear Plants  
- Prediction of ground motion
- Seismic Risk Evaluation (2004)
- Earthquake Facts 
- Earthquake FAQS 
- GeoBatch: Earthquakes (history M>=7)
- GoogleMaps: Earthquakes (history M>=7)