Climate Change Uncertainty and Risk: from Probabilistic Forecasts to Economics of Climate AdaptationDavid N. Bresch, IED ETHReto Knutti, IAC ETHAssistants: Lea Beusch, Thomas Röösli, Marius Zumwald

David N. Bresch, Reto Knutti, ETH Zürich

Schedule25.02.19 (1) Logistics, Introduction to probability, uncertainty

and risk management (RK, DB)04.03.19 (2) Predictability of weather and climate (RK)

Exercise 1 (toy model)11.03.19 (3) Detection/attribution (RK) 18.03.19 (4) Model evaluation & calibration (RK)

Exercise 2 (toy model)25.03.19 (5) Probabilistic risk assessment model and some insurance basics01.04.19 (6) Climate change and impacts, use of scenarios (RK, DB)

Exercise 3 (toy model), preparation of presentation08.04.19 (7) Presentations of toy model work, discussion (DB, RK)15.04.19 (8) 2°C target and adaptation in UNFCCC (RK, DB)

Exercise 4 (introduction to CLIMADA)22.04.19 Ostermontag

Schedule29.04.19 (9) Basics of economic evaluation and economic decision making (DB)

Exercise 5 (impacts)06.05.19 (10) The cost of adaptation - in developing and developed regions (DB)13.05.19 (11) Shaping climate-resilient development – valuation of a basket of

adaptation options (DB)Exercise 6 (adaptation measures, preparation of presentation)

20.05.19 (12) Climate services and adaptation in SwitzerlandExercise 7 (Swiss climate scenarios CH2018 for adaptation)

27.05.19 (13) Presentations of climada exercise and final discussion (DB, RK)

Recap: Risk

Hazard:

Exposure

Vulnerability

4

1 … a positive or negative deviation from what is expected [ISO 31000] Illustration: IPCC AR5

The “effect of uncertainty on objectives”1

Risk is the combination of the probability [or likelihood] of a consequence and its magnitude:risk = probability x severity

or, to be more specific:risk = hazard x exposure x vulnerability

= (probability x intensity) x exposure x vulnerability

severity

Risk1 Management

5

Risk identification: Shared mental model, the prerequisite for awareness§ perception is based on a shared mental model

à wider sharing builds awarenessRisk analysis: Quantification, the basis for decision-making§ Risk model: the quantitative expression of a shared mental model

à allows to assess risk mitigation optionsRisk mitigation: Prioritization based on metrics, options are to § Avoid§ reduce§ prevent§ transfer : Insurance puts a rice tag on risks à incentive for prevention§ or retain the risk 1 risk = probability x severity

Risk1 Management Cycle

6

shared mental model

quantitative model

options:§ avoid§ reduce§ prevent§ transfer§ retain

1 risk = probability x severity

Dialog: Risk, Uncertainty and Decision Making

IPCC and UKCIP

1. Problem definition, Goal

7. Implementation

6. Decision (?)

NO

NO

Criteria met?

Problemdefined

correctly?

2. Decision criteria

4. Identify options

8. Monitoring

5. Options appraisal

3. Risk analysis

Note: Loss amounts indexed to 2009 Source: Swiss Re, sigma No 2/2010

Natural catastrophe damages 1970-2016, in USD billion

Natural Nat Cat damages on the riseand: Massive gap between economic and insured damage

Note: Amounts indexed to 2012 Source: Swiss Re sigma catastrophe database

Note on trend drivers

The upward trend in natural catastrophe damage is driven by:§ Higher insurance penetration§ Growing property values § Coastal value concentration § Higher vulnerabilities§ Climate change

Trend decomposition going forward ?à Economics of Climate Adaptation

Ocean Drive, FL, 1926

Ocean Drive, FL, 2000

Note on trend drivers

Roger A. Pielke et al, Normalized Hurricane Damage in the United States: 1900–2005, NATURAL HAZARDS REVIEW, FEBRUARY 2008 / 29

Note on trend drivers

Roger A. Pielke et al, Normalized Hurricane Damage in the United States: 1900–2005, NATURAL HAZARDS REVIEW, FEBRUARY 2008 / 29

Fire vs. Nat Cat à need for accumulation control

every year1 of 5’000 houses

0.2 ‰ of house value

Fire Risk

every 5’000 yearsall houses

0.2 ‰ of house value

Nat Cat Risk

Experience or burning cost approachTake historic data (often only few years), calculate expected valuePro: easy to calculate and easy to communicateCon: only looking backward, (very) incomplete sampling of phase space

Damage

Frequency

Underwriter experience

Scenario

Scenario analysis is used in general …§ as a risk management tool to assess the potential impact of an event or

development to anticipate and understand risks§ as a tool to spot new business opportunities and to discover strategic

options§ as foresight in contexts of accelerated change, greater complexity and

interdependency§ for evaluation of highly uncertain events that could have a major impact § to steer mitigation strategies, implementation and monitoring by reviewing

and tracking different possible developments

Definition: A scenario is a snapshot that describes a possible and plausible future. Scenario analysis is a systematic approach to anticipate a broad range of plausible future outcomes

Forecast§ Focuses on certainties,

disguises uncertainties§ Conceals risks§ Results in a single-point

projections§ Sensitivity analysis§ Quantitative > qualitative

Scenario § Focuses on uncertainties,

legitimizes recognition of uncertainties

§ Clarifies risk§ Results in adaptive

understanding § Diversity of interpretations§ Qualitative > quantitative

The presentCurrentrealities(mental maps)

The futureAlternativefuture images

The path

Forecast Scenarios

Scenario approach

Damage

Frequency

Underwriter experience

Construct some possible consistent states, requires understanding of driversPro: spot wise sampling of phase spaceCon: still incomplete sampling of phase space

Scenario Considerations

?

Probabilistic modeling

Damage

Frequency

Underwriter experience

Construct all possible states, requires ‘system’ understandingPro: full sampling of phase spaceCon: huge effort Note: still scenarios, NOT forecasts

Scenario Considerations

Probabilistic modeling

Natural Catastrophe Risk Assessment Model

CLIMADAprobabilistic event-based simulation(open source)

vulnerability

exposure

weather à hazard

outputs:risk analysis+ mapping, + (early) warning ...

example: building damage

animated:https://vimeo.com/223292292

low middle high

Based on COSMO (1 km), MeteoSwiss. grey-red: hazard, green-blue: exposure, orange-red: risk (damage)Animated: https://vimeo.com/223292292

CLIMADA

CLIMADA: High-resolution (1x1km) impact modelLothar, 26 Dec 1999

CLIMADA

CLIMADA: High-resolution (1x1km) impact modelIrma, 30 Aug 2017

grey to red: hazard intensity

CLIMADA

CLIMADA: High-resolution (1x1km) impact modelIrma, 30 Aug 2017

grey to red: hazard intensity

Tropical cyclones in the North Atlantic

historic~100 years

probabilistic~10‘000 years

Tropical cyclones in the North Atlantic

historic~100 yearshistoric~100 years

probabilistic~10‘000 years

Tropical cyclones in the Indian ocean

historic, ~ 25 years probabilistic, ~ 5‘000 years

Understanding the hazard

Tropical cyclone intensity – the wind field (1/3)

http://en.wikipedia.org/wiki/Hurricanes

Tropical cyclone intensity – the wind field (2/3)

We use the Holland wind field model

§ The 1-min sustained wind at

gradient wind level (boundary

layer height & no surface

effects) is modelled using the

Holland 2008 approach. It

models the first-order vortex of a

tropical cyclone.

§ The translational speed (also

called celerity) is added

geometrically.

§ Holland, G. J., 1980: An analytic model of the wind and pressure

profiles in hurricanes. Monthly Weather Review, 108, 1212-1218.

§ Vickery, P.J. and D. Wadhera, 2008: Statistical models of Holland

pressure profile parameter and radius to maximum winds of

hurricanes from flight-level pressure and H*wind data. J. Appl.

Meteor. Clim.

Tropical cyclone intensity – the wind field (3/3)

§ A topography module corrects the Holland model for large scale topography, but still assumes a gradient level wind, i.e. at boundary layer height.

§ The approach of Vickery et al. 2009 corrects the wind for the boundary layer and surface properties.

§ Vickery, P.J. et al., 2009: A Hurricane Boundary Layer and Wind Field Model for Use in Engineering Applications. J. Appl. Meteor. Clim.

§ and an update of the Holland approach:Holland, G. J., 2008: A Revised Hurricane Pressure–Wind Model, Monthly Weather Review, 136, 3432-3445.

Landfall correction (not part of Exercise)

Note on validation

§ historical catalogue, 34 storms (upper panels)

§ probabilistic event set, 3391 storms (lower panels)

§ red line shows an extreme value distribution (Gumbel)

Both the historical dataset and especially the probabilistic set need to be carefully validated, as an example, we show the validation of intensity distributions:

Damage calculation

asset value damage PAA MDD , …. = ….. ???

§ damage is the damage ‘from ground up’, from the first dollar, so to speak

§ asset value is the total value of the asset

§ MDD is the Mean Damage Degree (the damage for a given intensity at an affected asset) - how strongly an asset is damaged. Range 0..1 (from none to total destruction)

§ PAA is the Percentage of Assets Affected (the percentage of assets affected for a given hazard intensity) - how many assets are affected. Range 0..1 (from none affected to all affected)

The damage is calculated for each single asset at each location for each scenario or event, i.e:

Damage calculation (2/2)So far, the hazard intensity did not show up in the calculation, did we miss something? Well, the damage is a function of the hazard intensity, hence:

MDD = f(hazard intensity) PAA = f(hazard intensity)

where hazard intensity is the hazard's intensity at each asset for each event. Since the damage also depends on the asset type, we have in fact:

MDD = f(hazard intensity, asset type) PAA = f(hazard intensity, asset type)

Mea

n D

amag

e R

atio

Notes on damage function

Hazard Intensity

Uncertainty of the hazard intensity

+Uncertainty of the

damageresults in

Convoluted Distribution

Insurance - Basics

Insurability

§ mutuality: numerous exposed parties must join together to form a risk community, to share and diversify the risk à large number

§ fortuitous or randomness: time of occurrence must be unpredictable, occurrence itself must be independent of the will of the insured

§ assessability: damage probability and severity must be quantifiable§ similarly exposed business à large number§ plus: economic viability: private insurers must be able to obtain a risk-

adequate premium

Insurance is the mutual cover of a fortuitous, assessable need of a large number of similarly exposed business

Alfred Manes, 1877-1963

average result 0.15, stddev 0.60

Effect of insurance (1/4)

Time series of annual result

Effect of insurance (2/4)

-1.00

-0.50

0.00

0.50

1.00

2010

2015

2020

2025

2030

price for prevention

Time series of annual result (after prevention)

average result 0.19 (+25%), stddev 0.56 (-8%), prevention price 0.01à effect of prevention: stabilize result, reduce volatility

Effect of insurance (3/4)

Time series of annual result (after prevention and insurance)

average result 0.17 (+12%), stddev 0.43 (-29%), prev+ins price: 0.13à effect of insurance: reduce (extreme) volatility

price for preventionand insurance

result stdev price*raw 0.15 0.60

+ prevention 0.19 (+25%) 0.56 (-8%) 0.01à cost-effective adaptation (net gain of 0.04 at cost of 0.01 )

+ insurance 0.17 (+12%) 0.43 (-29%) 0.01+0.12à substantial reduction of volatility, result increase even after deduction

of prevention cost and insurance premium à affordable!

insurance alone 0.12 (-17%) 0.45 (-25%) 0.153à prevention (strongly) incentivizes insurance

*price is already taken into account in result

Effect of insurance (4/4)

Insurance conditions

from ground-up damagefgu (total area) to damage after conditions, conditions get applied in the following sequence:

1) coinsurance2) deductible3) cover (we show the overspill)4) share (see also 1-share)

hence we calculate as follows:damage after conditions=min[

max(damagefgu*{1-coinsurance}-deductible,0),cover] *shareproportional

non-

prop

ortio

nal

proportional: damageafter=damagebefore*sharenon-proportional: damageafter=min(max(damagebefore-deductible,0),cover)

cove

r

Note on insurance conditions – the realityIn reality, risk transfer happens on several (repeated) hierarchical levels, such as:

Portfolio of assets

Policy 1 Policy 2 Policy n...

Site 1 Site 2 Site m...

Coverage 1 Coverage 2 Coverage p... Dam

age

accu

mul

atio

n

e.g. Building

e.g. ZIP code 12345

Note on insurance conditions – the real reality

Forms of insurance§ Risk transfer can be agreed upon based on different triggers:

§ indemnity1, also called incurred or occurred damage

§ parametric, also called index

§ modelled (well, a form of parametric)

§ and with different partners, such as:

§ policyholders – from macro (e.g. large corporates in Texas) to micro (e.g. smallholder farmers in Ethiopia à Example)

§ insurers (reinsurers insure them)

§ other reinsurers, called retrocession

§ capital market, called insurance-linked security (ILS) or often also Cat Bond (à Example)

§ public sector (PPP à Example) 1specific or market-share…

Microinsurance case study - Ethiopia

http://www.swissre.com/rethinking/crm/The_R4_Rural_Resilience_Initiative.html

Cat Bond case study

http://www.swissre.com/rethinking/crm/experts_on_multicat_mexico.html

Efficient monopolies

Source: Efficient Monopolies. The Limits of Competition in the European Property Insurance Market, Thomas von Ungern-Sternberg, Oxford University press, 2004.ISBN 0-19-926881-9

Efficient monopolies

Efficient Monopolies. The Limits of Competition in the European Property Insurance Market, Thomas von Ungern-Sternberg, Oxford University press, 2004.

40% more expensive for the same cover