Transcript Behavior Finance: The Missing Element in Risk Management

Presentation to:
Risk Minds 2010
Behavioral Basis of the Market Crisis
May, 2010
J. Rizzi, CapGen Financial
([email protected])
(The ideas expressed herein are those of the author and not CapGen Financial)
2
Gregory Zuckerman
despite sophisticated models mapping past behavior
they lacked understanding of human behavior - they
need to go beyond statistics of past and
include psychology of people effect as
at their foundation markets are people and the
people effect increases uncertainty
reminding us numbers alone are never enough
3
The Setting
4
US Dollar LIBOR-OIS SPREADS
(Source: Malcom Knight, Rebuilding the Global Architecture of Financial Regulation, March, 2009)
(How many standard deviations is this?…)
(…Who cares?)
5
What Happened?
(Do you want to believe what you see..)
INVESTORS: extrapolated and believed housing prices would not fall
Government
Monetary Policy
Monitoring
CRA
Regulatory capture
Incentives: unintended consequences
Risk intermediation reduces risk oversight (originate to distribute model)
Too big too fail (TBTF) socialize losses
Principal/Agent: management captured by employees
Underwriters: Too big to manage resulted in governance breakdown
Rating Agencies: Regulatory enforced oligopoly
Diversification substitution
Knowledge
Systemic for Idiosyncratic Risk
Pseudo objectivity: math without history equals disaster
(…Or what I am telling you?)
6
Structured Finance
(Structured finance as a compensation scheme….)
Issues: Structure vs. Underlying
Substitute systematic for diversifiable risk: default risk on adverse states
Ignored joint payoff distributions
Incorrect Gaussian Copula methodology: correlation
Lacked sufficient historical data on new underlying asset class: they
extrapolated on a limited sample based on good times
Mispriced: put on real estate index earn 3X more
Moral Hazard: perfect moral hazard product for issuers, underwriters, rating
agencies and investors
Covering up strategic decline with Tail Risk
Up to 60% of large bank revenues in 2006 came from structured finance
(…disguised as a business)
7
The Problem
(Economic Capital is a lighthouse….)
• Guided by selective memories and information
• Fail to consider what we believe to be false
• Influenced by the actions of others
• Confuse preferences with prediction
• Engage in self serving attribution
• Disregard non-conforming views
(… for the soon to be shipwrecked)
8
Some Behavioral Effects in Risk Management
(Risk Management is the fig leaf …)
• Hindsight and Confirmation: I knew it all along and ignore nonconforming
evidence
• Anchoring: Unduly influenced by first impressions
• Sunk Costs: Doubling down
• Overconfidence: Infallibility of judgment. Gives raise to illusions of control
• Optimism: It will work out
• Availability: More weight given to events easily recalled
• Threshold: Once frequency drops below threshold it is ignored
• Pattern Seeking: Fooled by randomness. Gamblers fallacy
(… behind which risk taking takes place)
9
Risk and Uncertainty
(Those whom the gods wish to destroy….)
Issue: Can you reduce future to quantifiable risks calculated from existing data?
Battling Beliefs – history as data and future as output
Ergodic – future is statistical shadow of past
Nonergodic – path dependent – history matters
Uncertainty vs. Risk – the future is uncertain not just risky
Risk – calculate odds of game
Uncertainty – game changes
Result – subjective beliefs of uncertain future
Cannot calculate probability of rare events based on past
Exposure vs. experiences
Consequence: illusion of control based on flawed risk
models
(…they first teach math)
10
A MAP on the Limits of Statistics
(Source: N. Taleb)
(We observe the data….)
Considerations: Distributions and payoffs
Physical
Sciences
1
Normal (risk)
2
Distribution
Social Sciences
3
4
Fat tails/unknown
(uncertainty)
Simple
Complex
Payoffs
Quadrant 4: Normal techniques fail. Alternatives to consider:
Redundancy not optimization
Avoid predication: focus on discipline and resiliency
Time horizon is longer
Moral Hazard: bonuses tied to hidden risks
Metrics: standard metrics no longer work
Volatility absence is not equal to risk absence
Risk numbers are dangerous: framing
(…not the process)
11
Humans and Markets
(In physics you play against God….)
Markets and Hurricanes: they are different (J. Meriwether)
Hurricanes are not more likely because more hurricane insurance is
written. This is not true for financial markets.
An increase in financial insurance increases likelihood of disaster.
Those who know you sold the insurance (will trade against you) can
make it happen.
In a crisis all that matters is who holds what and at what price.
Markets are more complex than casinos. The numbers on the Roulette wheel
never change. Markets make no guarantee that yesterday’s odds will be the
same tomorrow.
(… in markets you play against God’s creatures)
12
Decisions at Risk
(It is not what we don’t know that gets us in trouble…)
Uncertainty
Beyond the data experiences
Experiences
Exposures
Black Swans
Rare Events
Large Impact
Explainable
Bias
Amplifiers
Over confidence
Incentives
Illusion of control
Bureaucracy
Hindsight bias
Opaqueness
Anchoring
(…it is what we know that ain’t so)
13
Risk Management
14
The Setting
(Performance – is it luck…)
Dimensions
Frequency
Exposure
Experience
Severity
Focus: High impact low probability events (HILPEs)
HILPEs difficult to understand and frequently ignored
History proves HILPEs do happen and can threaten survival of the unprepared
Issues
Statistical: insufficient data
Behavioral: infrequency clouds perception
Risk estimates anchored
Disaster myopia
Social: reduced from regulations collapse once behavior changes
Goodhart’s Law
Risk Adaptation
(… or skill)
15
Risk Management
(Not just that risk management fails…)
Toolbox
Avoidance
Ignore
Mitigate
Transfer
Equity
Self insure
(… but it can produce unintended consequences that amplify damages)
16
Complex Financial Institutions
(It is the system…)
Complex
Simple
High Risk Systems: prone to
endogenous normal (system)
Tight
accidents. Manmade catastrophes
Complex nonlinear interaction:
inevitable but unpredictability
uncertain
Loose
Branching paths
Feedback loops
Jumps
Tight coupling: network effects
Governance: prevent management
from imposing risks on organization
for their own benefit
Alternative
Policy Implications
costs
(A ) Tolerate and improve
(B ) Restructure
(C ) Abandon
Disaster
recovery
Strategic and
management
Processing
errors
Model risk
C
B
A
Catastrophe loss potential
(… not the event)
17
Problem and Solution
(James Grant: In financial markets…)
Problem: HIPLE
Rare Decisions
Delayed Feedback
Limited Understanding
Solution:
Firm Level: Prisoners Dilemma
Regulatory Level: Capture
(…all progress is cyclical not cumulative)
18
Conclusion
19
Thinking About Risks: the Shift
(Organizati0ns are a social…)
Classical
Independent
Stationary
Rational
Gaussian
Frictionless
Consistent beliefs
Linear Risk Reward
Complete Information
Individuals
Risk
Objective Function
Equilibrium
Shocks
Efficient
New
Memories
Unstable
Bias
Fat tails
Arbitrage limits
Inconsistency
Nonlinear
Asymmetric Information
Institutions
Uncertainty
Principal-Agent Conflicts
Creative Destruction
Endogenous
Adaptive
(…not a physical phenomena)
20
Conclusion
(Ignore behavioral finance…)
• Risk is managed by people not mathematical models
• Accept randomness
• Discipline not predictions
• Expect the unexpected
• Avoid catastrophe risk
• Focus on what you know and insure against extremes
(… at your peril)