Transcript Dr Patrick McSharry

Financial risk forecasting:
decision-making in uncertain
environments
Dr Patrick McSharry
Smith School of Enterprise and the Environment
Contact:
[email protected]
16th May 2013
Financial Sector Governance for Natural Resource Sustainability Conference
University of Oxford
International collaboration
World Forum of
Catastrophe
Programmes
www.willisresearchnetwork.com
HMG
IBM
World Bank
UNISDR
UN ECLAC
www.globalquake.org
www.Safewind.eu
Overview
• Science, big data & quantitative modelling
• Probabilistic forecasting
• Opportunities: innovative insurance products
– Parametric insurance
– Index-based insurance
– Catastrophe insurance programmes
• Technical challenges to risk management:
– Insufficient data
– Inappropriate assumptions
– Emerging trends (regulation, climate change)
Big data
•
•
•
•
•
Large quantities of data
Structured and unstructured data
Temporal and spatial
Data on individuals, households, organisations
Transactional data (searches, clicks, tweets,
purchases, comments …)
• Data about what people actually did and what
they are doing in real-time
Source: Mark Graham, OII
Challenges/Opportunities
• Computation
– Technical challenges to sift through large quantities of information
• Interpretation
– Sentiment analysis, new quantitative/qualitative collaboration
• Transparency
– Close the gap between data/models and policy
– Open access/open source approaches
• Prediction
– Evidence-based policy requires standards, evaluation protocols
and appropriate benchmarks
• Scenarios
– Probabilistic decision support for increasing competitiveness,
sustainability, reducing risk and enhancing quality of life
Occam’s razor
William of Occam studied theology at the University of Oxford
from 1309 to 1321, but never completed his master's degree
• Occam’s razor (principle of parsimony): seek the
simplest model that explains the data
• How complicated should a model be?
• How can we evaluate and compare models?
• What are the appropriate benchmarks?
• Examples:
– Parsimonious risk indices (catastrophes, climate)
– Diagnostics for decision-making
Temperature indices
• Heating and cooling degree days:
• HDDt = max(18-Tt,0) CDDt = max(Tt-18,0)
• Effective temperature combines temperature and
humidity
Simpson & McSharry (2011). An economic assessment of the impacts of climate change on tourism in Barbados. UN-ECLAC
Probabilistic forecasts
• Development of a range of
quantitative econometric time
series models (parametric and
non-parametric)
• Probabilistic forecasts have been
produced for energy demand and
production and economic growth
• Development of techniques for
evaluating probabilistic forecasts
– Value: we can assess a forecast
based on its value (economic) for
improving decision-making
– Quality (statistical assessment):
reliability, sharpness, resolution and
skill
A. Lau and P. E. McSharry (2010). Approaches for multi-step
density forecasts with application to aggregated wind power.
Annals of Applied Statistics 4(3): 1311-1341.
J. W. Taylor, P. E. McSharry and R Buizza (2009). Wind
power density forecasting using ensemble predictions and
time series models. IEEE Trans Power Conversion 24(3):
775-782.
S. Arora, M. Little & P.E. McSharry (2013), Nonlinear and
nonparametric modeling approaches for probabilistic
forecasting of the US gross national product. Studies in
Nonlinear Dynamics & Econometrics (in press).
P. Pinson, P. McSharry, H. Madsen (2010). Reliability
diagrams for density forecasts of continuous
variables: accounting for serial correlation. Quarterly Journal
of the Royal Meteorological Society 136(646), pp. 77-90.
Global Risks (impact vs likelihood)
Source: World Economic Forum
Global Trends
• By 2050, world population will be 9 billion with
70% living in cities
• Security of food, water and energy
• Today, 925 million people experience hunger,
with food price spikes and volatility threatening
the sustainability of global food security
• Interconnected risks: Rising food prices 
Tunisian street vendor, Mohamed Bouazizi, sets
himself on fire and dies  Ben Ali toppled from
from power  Arab Spring
Model Risk
• Numerous major losses which were not adequately
modelled
– Japan ($37bn); NZ quake ($12bn); Thai flood ($10bn)
• Vendor model changes
– Modelled versus real world & uncertainty
• Regulation associated with Solvency II and increased
focus on 1 in 200 year events
– Compliance issues relating to model uncertainty and
risk assessment
Systemic Risk
• Exposure is more interconnected than ever
– Sovereign risk, Arab Spring, Spatiotemporal extremes
• Supply chain risk (outsourcing, lean
manufacturing, just in time inventory)
• Local and regional models are inadequate for
assessing potential global losses
• Thailand flooding was an unexpected loss
– $10 bn (forecasted to reach $20 bn)
– Contingent business interruption
– Semiconductors, car manufacturing
Emerging Risk
Oil spills:
1989 Exxon: 0.75 million barrels
2010 BP: 4.9 million barrels
Exxon's share price dropped by
6%, that of BP declined by 53%
before starting to recover.
This suggests that the share
price drawdown risk due to oil
spill catastrophes has increased
by a factor of nine in the last two
decades (environmental/social
media/reputational risk).
20 Apr: Deepwater Horizon offshore oil rig explosion in the Gulf of Mexico, kills 11 workers & injures 17 others.
 01 Jun: “top kill” operation was unsuccessful.
 10 Jun: US government signals it will take legal action to force BP to stop paying a dividend to shareholders.
 14 Jun: President Obama compares the oil spill's impact to that of the 9/11 terrorist attacks.
 22 Jun: Hayward hands day-to-day control of the spill operation to Bob Dudley.
 15 Jul: BP announces that it has stopped its Gulf of Mexico leak for the first time since April.
Insufficient historical data
• Ongoing collaboration with EU partners in the
SafeWind consortium
• Assessment of risk of extreme wind speed to
measure resource
• Wind farms sites only have a few years of data
• Dynamical atmospheric models
• Reanalysis data (~ 50 years)
50-year return wind speeds
Reanalysis data gives superior return period estimates for all records of duration less
than 20 years
Anastasiades & McSharry (2012). Wind Energy.
Catastrophe Modelling
Risk = Hazard x Exposure x Vulnerability
Under certain assumptions,
extreme value theory or
simulation can be used to
perform extrapolation in order
to assess the probability of
extreme losses.
Hurricane counts
Exceedance probability
Losses over time
Recent decline in losses
may be a result of a
combination of decreased
hazard (low hurricane
activity) and improved
engineering standards
along the coast of the US.
The exact explanation
requires a separation of the
effects due to the hazard,
exposure and vulnerability.
While engineering
standards may have
reduced the exposure, it
is likely that there has been
an increase in the
exposure along the coast.
Risk = hazard x exposure x vulnerability
Loss summary
Loss versus wind speed
Catastrophe risk
• Member of the Willis Research Network working on
parsimonious models for catastrophe risk assessment
• Assessment of economic losses arising from disasters
• Demonstration of ability of atmospheric models to
generate synthetic hurricanes
• Potential for global windstorm risk assessment
Loss estimates
from calibrated
HIGEM model
(solid) versus
commercial CAT
model (dashed).
Commercial EP curves
Risk Forecasting
• Quantitative predictive modelling informed by
qualitative analysis
• Fusion of scientific knowledge and empirical
investigations based on data analysis
• Provide prospective rather than retrospective
risk analysis
• Bridge gap between state of the art modelling
and real-world decision making
Global Temperatures
Global temperatures
Stranded Assets Risk
• Hazard: regulation, pricing, technology, society
and physical climate change
• Exposure: portfolio weights
• Vulnerability: carbon intensity of business
• Financial losses: quantify using VaR and EP curves
Future Research
• Performance-based evaluation of risk models
– Comparison with simple benchmarks in order to provide
quality control for end-users
• Risk indices based on parsimonious models
– Hurricane risk index using atmospheric models
– Stranded asset risk
• Risk forecasting applications
– Risk assessment using reanalysis data
– Lighthill Risk Network (open-access CAT model)
• Email: [email protected]