Multiple Criteria Philosophy and Value-at-Risk • David L. Olson • Desheng Wu

advertisement
Multiple Criteria Philosophy and
Value-at-Risk
• David L. Olson
– University of Nebraska
• Desheng Wu
– University of Toronto; University of Reykjavik
MCDM2011
Focus
• The philosophy part
– PARETO OPTIMALITY
• The enterprise risk management part
– VAR
– Treatment of investment risk
– Problems
• Models and assumptions
• If you have enough criteria, practically all
choices will be Pareto Optimal
MCDM2011
Economic Philosophy of Risk
• Thűnen [1826]
– Profit is in part payment for assuming risk
• Hawley [1907]
– Risk-taking essential for an entrepreneur
• Knight [1921]
– Uncertainty non-quantitative
– Risk: measurable uncertainty (subjective)
– Profit is due to assuming risk (objective)
MCDM2011
Contemporary Economics
• Harry Markowitz [1952]
– RISK IS VARIANCE
– Efficient frontier – tradeoff of risk, return
– Correlations – diversify
• William Sharpe [1970]
– Capital asset pricing model
• Evaluate investments in terms of risk & return relative to the market
as a whole
• The riskier a stock, the greater profit potential
• Thus RISK IS OPPORTUNITY
• Eugene Fama [1965]
– Efficient market theory
• market price incorporates perfect information
• Random walks in price around equilibrium value
MCDM2011
Empirical
• BUBBLES
– Dutch tulip mania – early 17th Century
– South Sea Company – 1711-1720
– Mississippi Company – 1719-1720
• Isaac Newton got burned: “I can calculate the motion
of heavenly bodies but not the madness of people.”
MCDM2011
Long Term Capital Management
• Black-Scholes – model pricing derivatives
• LTCM formed to take advantage
– Heavy cost to participate
– Did fabulously well
• 1998 invested in Russian banks
– Russian banks collapsed
– LTCM bailed out by US Fed
• LTCM too big to allow to collapse
MCDM2011
Real Estate
• Considered safest investment around
– 1981 deregulation
• In some places (California) consistent high rates of
price inflation
– Banks eager to invest in mortgages – created tranches of
mortgage portfolios
• 2008 – interest rates fell
– Soon many risky mortgages cost more than houses worth
– SUBPRIME MORTGAGE COLLAPSE
– Risk avoidance system so interconnected that most banks
at risk
MCDM2011
“All the Devils Are Here”
Nocera & McLean, 2010
• Circa 2005 – Financial industry urge to
optimize
– J.P. Morgan, other banks hired mathematicians,
physicists, rocket scientists, to create complex risk
models & products
• Credit default swap – derivatives based on
Value at Risk models
– One measure of market risk from one day to the
next – MAX EXPOSURE at given probability
MCDM2011
Credit Default Swap
Nocera & McLean, 2010
• 1994 J.P. Morgan
– Exxon Valdez oil spill
– Exxon faced possible $5 billion fine
• Drew on $4.8 billion line of credit from J.P. Morgan
• Morgan couldn’t alienate Exxon
– But loan would tied up lots of money
• Morgan got European Bank for Reconstruction &
Development to swap default risk for the loan for a fee
MCDM2011
Circa 2005
Nocera & McLean, 2010
• Banks want more profit
– Create products to sell to investors
• Mortgage granting agencies want fees
– Don’t worry about risk – sell to Wall Street
• Wall Street packages different mortgages into
CDOs (collateralized debt obligations)
• Prior to 2007 – CDOs consisted of corporate debt
• 2007 – shifted to mortgage debt
– Blending mortgages of different grades, locations, intended to
diversity
– View that high return required high risk
– Needed AAA rating to attractMCDM2011
investors
Ratings
Nocera & McLean, 2010
• Prior to 1970s, ratings agencies gained
revenue from subscribers
– Subscription optional
• 1970s – switched to charging issuers directly
– Investors wouldn’t buy unrated bonds
– Issuers required to get ratings
– CONFLICT OF INTEREST
• SEC decreed Moody’s, S&P, Fitch were
qualified to rate bonds
MCDM2011
Ratings Failures
Nocera & McLean, 2010
• 1929 -78% of AA or AAA municipal bonds
defaulted
• 1970s Penn Central RR
• Near default of New York City
• Bankruptcy of Orange County
• Asian, Russian meltdowns
• 1990s – Long-Term Capital Management
MCDM2011
Mortgage Abuses
Nocera & McLean, 2010
• Loan officers often convinced applicants to lie
• Part-time housekeeper earning ≈$1,300/mo
– fronted for sister, got loan
– unable to find steady work so returned to Poland
• Dairy milker earning ≈$1,000/mo purported to be
foreman earning $10,500/mo
– Didn’t speak English
– Bought house for son
– Told by lender that he was lending his credit to his son
• Janitor earning $3,900/mo
– Claimed to be account executive (for nonexistent firm)
– Closed loan on $600,000 house
MCDM2011
– Never made $30,000 down payment
Originator claimed
Correlated Investments
• EMT assumes independence across
investments
– DIVERSIFY – invest in countercyclical products
– LMX spiral blamed on assuming independence of
risk probabilities
– LTCM blamed on misunderstanding of investment
independence
MCDM2011
PRACTICAL ALTERNATIVES
• Warren Buffet
• George Soros
MCDM2011
Warren Buffett
• Conservative investment view
– There is an underlying worth (value) to each firm
– Stock market prices vary from that worth
– BUY UNDERPRICED FIRMS
– HOLD
• At least until your confidence is shaken
– ONLY INVEST IN THINGS YOU UNDERSTAND
• NOT INCOMPATIBLE WITH EMT
MCDM2011
George Soros
• Humans fallable
• Bubbles examples reflexivity
– Human decisions affect data they analyze for future
decisions
– Human nature to join the band-wagon
– Causes bubble
– Some shock brings down prices
• JUMP ON INITIAL BUBBLE-FORMING
INVESTMENT OPPORTUNITIES
– Help the bubble along
– WHEN NEAR BURSTING, BAIL OUT
MCDM2011
12 Investment Opportunities
daily data – 6/14/2000 to 7/6/2009
Change each day from prior
Mean, Standard Deviation, Avoid Chinese, Avoid US (except Berkshire)
•
•
•
•
•
•
•
World Index
• Hong Kong index
USA1
• Treasury Yield Bond
USA2
• DJSI World Index
Chinese index
• Royce Focus Fund
Eurostoxx
• Berkshire Hathaway
Japanese index
• Equal
20 Nondominated portfolios
MCDM2011
Idea
• Identify Pareto optimal set
– 2 criteria
• Maximize mean (return)
• Minimize standard deviation (risk)
– 3 criteria
• Avoid Chinese (China, HongKong)
– 4 criteria
• Avoid US (USA1, USA2, Treasury, DowJ, Royce Focus)
MCDM2011
Data – 2 Criteria
Min Var
World
USA1
USA2
China
Europe
Japan
HongKong
Treasury
DowJ
Royce
Berkshire
Fidelity
0.023
0
0
0.011
0
0.016
0
0.031
0.002
0
0.031
0.887
CC@0.6
CC@0.8
CC@0.9
CC@0.95
Max
Return
0
0.005
0.011
0.014
0
0
0
0
0
0
0
0
0
0
0
0.022
0.014
0.013
0.012
0
0
0
0
0
0
0.005
0.013
0.014
0.014
0
0.006
0
0
0
0
0.025
0.030
0.030
0.030
0
0.010
0.018
0.012
0.010
0
0.014
0.001
0
0
1
0.042
0.034
0.033
0.033
0
0.876
0.885
0.886
0.886
0
MCDM2011
Data Additional Criteria
1 to 4 criteria
World
USA1
USA2
China
Europe
Japan
HongKong
Treasury
DowJ
Royce
Berkshire
Fidelity
Add 5th (max China) Add 6th (min US)
Nondominated
Dominated
Dominated
Nondominated
Dominated
Weak nondom
Nondominated
Nondominated
Nondominated
Nondominated
Nondominated
Nondominated
Nondominated
MCDM2011
Weak nondom
POINT
• Investments will be portfolios
– Mixtures of investments
• The data still demonstrates the point
– IF YOU INCLUDE ENOUGH CRITERIA, HARD TO FIND
DOMINATED SOLUTIONS
– There must be a reason the market cleared
• Keeney MAUT models
– Typically 80 criteria
• Government choices
– Whatever is first choice, hearings will stifle
MCDM2011
Better Models
Cooper [2008]
• Efficient market hypothesis
– Inaccurate description of real markets
– disregards bubbles
• FAT TAILS
• Hyman Minsky [2008]
– Financial instability hypothesis
• Markets can generate waves of credit expansion, asset inflation,
reverse
• Positive feedback leads to wild swings
• Need central banking control
• Mandelbrot & Hudson [2004]
– Fractal models
• Better description of real market swings
MCDM2011
Models are Flawed
• Soros got rich taking advantage of flaws in
other peoples’ models
• Buffett is a contrarian investor
– In that he buys what he views as underpriced in
underlying long-run value (assets>price);
• holds until convinced otherwise
– Avoids buying what he doesn’t understand (IT)
MCDM2011
Nassim Taleb
• Black Swans
– Human fallability in cognitive understanding
– Investors considered successful in bubble-forming
period are headed for disaster
• BLOW-Ups
• There is no profit in joining the band-wagon
– Seek investments where everyone else is wrong
• Seek High-payoff on these long shots
– Lottery-investment approach
• Except the odds in your favor
MCDM2011
Fat Tails
• Investors tend to assume normal distribution
– Real investment data bell shaped
– Normal distribution well-developed, widely understood
• TALEB [2007]
– BLACK SWANS
– Humans tend to assume if they haven’t seen it, it’s impossible
• BUT REAL INVESTMENT DATA OFF AT EXTREMES
– Rare events have higher probability of occurring than normal
distribution would imply
•
•
•
•
Power-Log distribution
Student-t
Logistic
Normal
MCDM2011
Human Cognitive Psychology
• Kahneman & Tversky [many – c. 1980]
– Human decision making fraught with biases
• Often lead to irrational choices
• FRAMING – biased by recent observations
– Risk-averse if winning
– Risk-seeking if losing
• RARE EVENTS – we overestimate probability of rare
events
– We fear the next asteroid
– Airline security processing
MCDM2011
Animal Spirits
• Akerlof & Shiller [2009]
– Standard economic theory makes too many
assumptions
• Decision makers consider all available options
• Evaluate outcomes of each option
– Advantages, probabilities
• Optimize expected results
– Akerlof & Shiller propose
• Consideration of objectives in addition to profit
• Altruism - fairness
MCDM2011
APPROACHES TO THE PROBLEM
• MAKE THE MODELS BETTER
– The economic theoretical way
– But human systems too complex to completely
capture
– Black-Scholes a good example
• PRACTICAL ALTERNATIVES
– Buffett
– Soros
MCDM2011
Download