Nov 10, 2011

How to catch Risk?

Desperate attempts to catch Risk in new regulatory standards like Basel (II/III) for banks and Solvency (II) for insurers seem a dead end street....

What is happening?

That's the question we're about to answer in this blog!

Here are some observations:
  • Risk Weighting
    All new risk valuating standards are based on Risk Weighting. Some assets (or liabilities) are assumed to be more risky than others. In practice, every asset class that has been identified as more or less 'safe', has turned out to be risky after all. E.G., government bonds  where - until the 2011 crisis  in Greece - assumed to be risk free. Unfortunately, nothing could be further from the truth...

    Nothing in life is risk free
  • Tier Ratio's
    Instead of simple 'Equity to Asset Ratios', Tier 1 & 2 ratios where developed. These Tier ratios only take a fraction of the total assets into account. This leads to 'Equity to Risk-Weighted Assets Ratios' that insinuate adequate, substantial and reassuring 10-15% Capital ratios, while - in fact - they're not! These kind of ratios are misleading and create a false sense of safety....

    Tier Ratios lead people up the garden path

  • Tail Hide and Seek
    As more and more risks are valued, regulated and urge for extra capital requirements, financial institutions will try to create extra return on risks that are formally not or only 'light weighted' measured. This way substantial risks are 'pushed' into the tail, fat risk tails are created and the sight on the real risks in the company becomes misty.

    Overregulation decreases the effect of good risk management

Illustration: Comparison 'Deutsche Bank' - 'Bank of America'
To illustrate what is happening, let's compare a giant like "Deutsche Bank" (DB) with the number one on the banking list, the "Bank of America" (BOA).


Financial RatiosDeutsche BankBank of America
(x 1 bn $)              Year:2010200920102009
Assets (A)1906150122652230
Liabilities (L)1855146320371999
Shareholder Equity (SE)4937228231
SE / A - Ratio2.6%2.4%10.1%10.4%
---------------------------------
Risk-Weighted Assets (RWA)34627314561543
Assets (A)1906150122652230
RWA / A - Ratio18%18%64%69%
---------------------------------
Regulatory Capital (RC)4938230226
Risk-Weighted Assets (RWA)34627314561543
Total Capital Ratio 14.1%13.9%15.8%14.7%
---------------------------------
Tier 1 capital4334164160
Risk-Weighted Assets (RWA)34627314561543
Tier 1 Capital Ratio12.3%12.6%11.2%10.4%




















Although both banks have more or less the same 'Tier 1' and 'Total Capital Ratio', their individual risk profile is completely different. 

In the case of DB only 18% of the assets are assumed (marked) risky, while in the case of BOA around 64% is assumed risky and taken into account for a risk weighted solvency approach.

Notice that the simple gross 'Equity to Asset' Ratio (E/A-Ratio, or in short 'EAR')  of DB is only 2.6%, while the similar ratio of BOA is around 10.1%. If DB would be hit by an 5% impact loss, it would be in deep trouble.


Reflections
Our risk models have become too sophisticated and don't cover the area of 'Unkown Risk' enough. Unintentionally rand controversially, risk regulations and models make us implicitly sweep our real risks under the carpet. In principle Risks can be categorized as:

  1. Known Risk Measured
  2. Known Risk Unmeasured
  3. Unknown Risk
  4. Hidden Risk (knowingly or unknowingly)

It's time to admit that no asset or liability is completely free of risk and there's an overall substantial probability that risk - by definition - will hit eventually from an unexpected corner. To put things in perspective: In the 19th century, banks funded their assets with around 40-50% equity.

Conclusion
Including 'Unkown Risk', a simple gross E/A-Ratio (EAR) of a magnitude of 15-25% (across the total assets) would probably be the best kind of guarantee to accomplish a more sustainable financial system in the world. The new EAR could be best defined as the sum of an actuarial underpinned percentage on basis of the underlaying calculable covered risks and a TBD overall 10% 'add up' for unknown risks:

E/A-Ratio = EAR =  EAR[ calculable risk ] + EAR[ unknown risk ]

Until we've included Unkown Risk fully in our risk models, we'll stay in deep trouble.

Aftermath: 'Avatar Ratios'
To rate a company (bank), often it's not enough to look at just the traditional financial ratios. An interesting way to additionally rate a company in a more sophisticated way, is with the help of so - by me - called 'Avatar Ratios'.


Additional to financial ratios, 'Avatar Ratios' tell you more about what the intentions, (real) important issues and the 'drive' of a company and its employees are.

An 'Avatar Ratio Analysis' gives you more or less 'the embodiment' of all what drives a company. It can be constructed by making a word analysis of a crucial document or annual report of a company. In short: You simply download the annual report (or any other company characteristic document) and analyze it with a  'Word Frequency Counter' like WriteWords.

IAA Demo
As a demo, let's analyze the IAA's Strategic Plan




FrequencyWord
21actuarial
10strategic
10develop
8associations
6standards
6priorities
6plans
6objective
6action
5promote
5profession
5member
5international
5iaa
5association
4practice
4maintain
4key
4global
4encourage
4education
3worldwide
3statement
3relationships
3plan
3including
3identify
3establish
3common
3discussion
2world
2values
2supranational
2support
2services
2risk
2relevant
2provide
2program
2professionalism
2professional
2prioritize
2principles
2organizations
2organization
2mission
2march
2management
2links
2issues
2internationale
2help
2fields
2facilitate
2countries
2contact
2conduct
2areas
2area
2approved
2among
2actuaries
2actuarielle
1voluntary
1vision
1understanding
1transparency
1traditional
1stakeholders
1soundness
1society
1social
1skills
1sections
1scope
1scientific
1role
1review
1research
1reputation
1relationship
1regional
1recommended
1recognized
1recognition
1quality
1public
1protection
1promotion
1process
1procedures
1presidents
1offered
1objectivity
1objectives
1needs
1model
1members
1knowledge
1jurisdictions
1involvement
1integrity
1improve
1guidelines
1forums
1forum
1financial
1feasibility
1experiences
1expansion
1examine
1ensure
1enhance
1disciplinary
1developing
1developed
1designation
1decisions
1decision
1credential
1create
1cooperation
1convergence
1contributing
1continuing
1constructing
1code
1changing
1availability
1audiences
1active
1achieve
1accountability
1access

With the help of WriteWords we first create (on line) a frequency table. Next we cut out irrelevant words like 'and', 'the', etc. 

Here are the results:
(1) a  scrollable frequency table of all relevant words
(2) a 'Top 22 words' frequency table



In most cases - like this one - the result of simply putting the first 10 to 15 words in the top of the frequency table behind each other, is astonishing: It creates a kind of 'Identity Statement'. Here's the result for IAA's strategic plan, where even more than 20 words give a beautiful comprised identity statement:

IAA's Strategic Plan (Identity Statement comprised with WriteWords)
Actuarial strategic develop associations. Standards, priorities plans objective action. Promote profession member international. IAA association practice maintain key. Global encourage education worldwide.

Avatar Analysis: Comparison 'Deutsche Bank' - 'Bank of America' 
Let's now go back to our banking case and compare 'Deutsche Bank' (DB) and the 'Bank of America' (BOA) with the help of a simple Avatar Ratio Analysis.

The Avatar Ratio Analys presents the word frequency (absolute numbers) and their relative frequency (= word frequency / total number of word in document). Here is the result:


Avatar RatiosDeutsche BankBank of America
Freq.Perc.
Freq.Perc.
Governance1090.06%250.02%
Risk14580.79%8520.53%
Control2730.15%1560.10%
Total G+R+C18401.00%10330.64%
------------------
Client/Customer3590.20%2500.15%
Shareholder1690.09%1620.10%
------------------
Transparent150.01%20.00%
------------------
Employee1530.08%630.04%
Director400.02%230.01%
------------------
Profit, Income10010.54%8350.52%
------------------
Tot. nr. of words161579100%184048100%


Although I'll leave the final conclusions up to you, here are some remarkable observations:
  • Total number of words
    Both companies (DB and BOA) need an enormous amount of words to explain their environment (clients, shareholders, rating agencies, etc) the essentials about what's going on in their company in a modest calendar year.

    To read an annual report of about 170,000 words, it would take an average reader (reading speed 200 to 250 words per minute) about 10-12 hours.

    Perhaps you, as an actuary, can read faster ( test it!: speed reading test ), but even at a speed of 500 wpm it would be an enormous task (5-6 hours) to fulfill.

  • Governance, Risk & Control
    It's clear that DB puts much more energy (+60%) in communicating about themes as Governance Risk and Control than BOA. Also is clear that DB is far more transparent in its communication than BOA. This does (of course) not imply that BOA's risk and control frame is inferior to DB's. It could even be the opposite. It just shows that (and how) BOA handles and communicates differently (less open) from DB.

  • Profit, Income, Shareholders + Clients and Employees
    DB and BOA weight Profit, Income and shareholders on more or less the same level. Both rank client/customer above shareholders. DB gives 'clients/customers' as well as employees double the attention of BOA!

At last
Next time you report to your board, include an Avatar Analysis of your report in your presentation!

Related Links, Sources:
- The biggest weakness of Basel III (2010)
- The Banker top 1000 list (2011) 
- Word Frequency Counter : WriteWords
- IAA's Strategic Plan  
- Bank of America: Annual Report 2010
- Deutsche Bank: Annual report 2010
- On line speed reading test 
- Spreadsheet containing this blog's DB and BOA analysis

Nov 1, 2011

Sustainable Discount Rates

Steering pension funds on a 'one point' Coverage Ratio is like trying to proof global warming on a hot summer day...... It's useless.

Why?

First of all the complete pension fund balance sheet is based on market value.

As there is no substantial market for pension liabilities, this implies that pension liabilities have to be valued on basis of some kind of arbitrary (artificial) method.

In the U.S. this has led to the (irresponsible) high discount rate of 8% for state pension funds based on the 'expected' long term return without a kind of correction (subtraction) for 'risk'.

In de Dutch market, pension funds have to rate their liabilities on basis of a maturity dependable risk free interest rate, the ‘Nominal interest Rate Term Structure’ (RTS), as ordered by DNB (the Dutch Regulator).

Here's the outcome of this risk free interest rate (RTS) over the past 10 years, including the 10-y average RTS...


Ever since DNB ordered this 'artificial discounting method', pension fund board members didn't get a good night sleep. As the RTS juggles on a daily basis, every morning pension members wake up with the latest 'RTS news surprise of the day'.

You can play the RTS juggle (worm) here:


As coverage ratios are based on the RTS, they shuttle hither and thither as well and executing a long term pension fund strategy becomes more or less like riding the famous (market) bull in a rodeo show.

On a Dutch IPE congress, Angelien Kemna - chief investment officer of the €270bn asset manager APG - warned that the current swap-curve discount criterion forces pension funds to take unwise "significant long-term measures".

Kemna favors an average yield curve or a more straightened version of the current one for discounting liabilities.

The new Dutch Pensions Agreement foresees that pension funds can choose their own discount rate, as pensions are no longer guaranteed!

Indeed, it's time to stop this complex discount circus. But it's also time to stop 'one point estimate' Coverage Ratio steering.

A new look
Let's take a look at a characteristic discount rate dependence of a traditional pension fund like ABP.



Valuing ABP at an (derived average) RTS of 2.69%  (September 2011), ABP's discounted assets fail to meet the discounted liabilities, leading to a coverage ratio of around 90%.

However this kind of risk free valuing is - for sure - too conservative, as ABP's aims at an underpinned strategic expected return of 6,1% on the long term and has a convincing track record of  5 and 10-year moving average returns:

Returns (%) Pension Fund ABP 2993-2010
Year199319941995199619971998199920002001200220032004200520062007200820092010
Yearly Return16.5-1.016.411.811.912.910.03.2-0.7-7.211.011.512.89.53.8-20.220.213.5

5Y MA Return
10.910.212.69.97.33.43.03.35.27.29.72.74.24.4

10Y MA Return
7.16.67.87.57.36.52.93.84.8

Or in Graphics:

As long as a pension fund (like ABP) continues to perform (on 5 or 10-years moving average) rates that outperform the (derived average) risk free discount rate, it's seems ridiculous to force such a pension fund to discount at a 'risk free rate', as this obliges the fund to change his strategic asset mix to a less risky mix and an suboptimal return.
In turn, these suboptimal returns will lead to an asset shortage. With a vicious cycle of decreasing risk as a fatal result in the end.

Sustainable Discount Rates
In an excellent discussion paper (2006) Jürg Tobler-Oswald proves that the optimal discount rate lies between the risk free rate (RFR) and the investment strategy’s expected return (ER) depending on how good the hedge against the fund’s cash  flow  provided by its investments  is:

Discount Rate1 = RFR + FCash Flow(RFR-ER)

Another - more simple and practible - discount rate could be defined as the average between the free discount rate and the X-year (e.g. X=5, or 10) Moving Average Return of the last X-Years (MAR(X)).

Discount Rate2 = [ RFR + MAR(X) ] /2

As long as MAR(10), MAR(5) and ER stay larger than the interest rate that matches a coverage ratio of 100%, discounting by means of one of the new sustainable discount methods seems sound and safe......

Whats left is that the average (geometric) risk premiums during the last 10 years have turned out negative:

Historical Equity Risk Premiums (ERP)
ERP: Stocks minus T.BillsERP: Stocks minus T.Bonds
PeriodArithmetic Geometric Arithmetic Geometric
1928-20107.62%5.67%6.03%4.31%
1960-20105.83%4.44%4.13%3.09%
2000-20101.37%-0.79%-2.26%-4.11%

This implies (moreover) that it is important that the discounting rate of a pension fund should be based on a sustainable sound weighted mix of:
(1) proven historical performance
(2) a 'save' risk free rate
(3) realistic future return assumptions


Related Links/ Sources
- Kemna IPE article (2011)
- An investment based valuation approach for pension fund cash flows (2006)
- Ignoring the risk in risk premium in State Pensions(2011)
- DB: What went wrong? (2011)
- Actuary.org: Pension Fund Valuation and Market Values (2000)
- Aswath Damodaran: Equity-risk-premiums-2011-edition
- Dutch: ABP coverage ratio