Minsky, Modellen en Mensen

Transcription

1 Minsky, Modellen en Mensen Theo Kocken CEO, Cardano

2 economics has not truly come to grips with the main difficulty, which is the inordinate practical importance of a few events Benoit Mandelbrot

3 How risk-return runs out of balance Peers P&L growth Firm Objectives Return /P&L P&L growth Constraints Risk Models Risk Assessment Perception of risk Data

4 Peer group risk is key Chuck Prince (Citi) explained it in one sentence: When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you ve got to get up and dance. We re still dancing, Remark was made.. July 2007!!!!!

5 supported by the Lender of Last Resort

6 Hyman Minski s moment will always come back Peer group pressure => leverage pressure 1) Hedge borrower 2) Leveraged Borrower 3) Ponzi Borrower Supported by behavioral flaws (overconfidence, notion of control, herding behavior etc)

7 Anti-cyclical regulations? destined to fade in prosperous times

8 Did Risk Management fail? Models? Data? Perception of Risk Numbers? And actions taken

9 Robert Lucas and poor defence on Macro models Second half 07: Macro models were accurate, says Robert Lucas (Economist, 8 aug 09): "Forecast of what could be expected conditional on a crisis not occurring." Funny, because housing decline already on its way... "Mishkin's forecast was a reasonable estimate of what would have followed if the housing decline had continued to be the only factor involved in the economic downturn".

10 Some financial risk models surely FAILED. The Rating Agencies.we paid them to see their models fail

11 Risk modelling revisited? Taleb: Ludic fallacy : Modelling as if it is a game [Ludus=game] Keynes: PROBABILITIES, but in reality we face UNCERTAINTY Modelling under the streetlight

12 More complex modelling? Fat tails? Regime shifts?? Copulas???? Multi-variate copulas + regime shifting?????? Don t try to capture event risk in long term stochastic models Stochastic models for risk return insight Event Risk analysis/stress Tests to estimate risk in extreme events => More out of the box economics than modelling

13 Keynes: Genius of The Great Depression Changed his belief in probabilities to a notion of uncertainty: We simply do not know 13

14 Models: Physics versus Economics Emanuel Derman: Physics: Playing with God Economy: Playing with the Creatures of God Andrew Lo: Physics: 3 laws explain 99% behaviour Economy: 99 laws explain 3% behaviour

15 How about the data?

16 Some data fallacies Filtering & adjustments inclined to reduce risks: Survivorship bias => markets that lost dominance excluded (Goetzman/Jorion) Easy data-fallacy => difficult periods, markets in past centuries excluded etc Short periods Recent market behaviour informative WHY??? Booming periods => lower volatility & higher leverage

17 Positive filtering behavioral? Affect-bias: If people like a situation, they perceived higher returns and lower risks (Alhakami & Slovics) => Booming markets! Markets are most risky when people have the most confidence in them

18 Misperception of risk How people look at risk output differently

19 Misperception of risk Some different reactions 1% -2% probability per annum is once in a life time Why bother about a meteor hitting us? Others. Probability of realization Χ Consequences for various stakeholders 35% probability of non-sustainable 3 consecutive years of shortfall the next 20 years is too big a risk We can t bear the consequences in a mature fund => What actions now to avoid future dramatic situations?

20 Several cognitive biases Probability of realization Cumulative versus single period Seatbelts Cumulative numbers lead to 4 times more action than single trip statistics (Slovics) Illusion of control Consequences People don t expect to bear the consequences (March & Shapira) The more complex a problem (economy!!) the more confident people are the control the situation

21 What can we do? Regulations: Avoid (too) big institutions (or make them extreme safe) Focus on event risk in system Modelling: Event risk analysis first. Robust solutions.. Data: Embrace (instead of ignore) special periods Risk perception: Higher risk- empathy by education

22 Education 1: Financial history before modelling

23 Education 2: Act on what we know about our cognitive biases Herding Overconfidence Endowment bias Holy Grail - Killer Bunny.mpg Affect bias the inordinate practical importance of a few events

24 Education 2: Act on what we know about our cognitive biases Probability of realization Χ Consequences for various stakeholders the inordinate practical importance of a few events