Current vulnerability: Overview

Current VulnerabilityCurrent VulnerabilityFrom concepts to From concepts to integrationintegrationTom DowningGlobal Climate Adaptation Partnership

This is a slightly updated version of a presentation based on my work over the past two decades, from the World Hunger Programme to Environmental Change Institute and Stockholm Environment Institute. Some of the images may be copyrighted although all of the concepts are published.

Outline…Outline…Concepts and principlesIndicatorsVisualizationAggregate indexesLinking to adaptationPractical exercises

From Concepts to From Concepts to principlesprinciples

Quick foundations.

Photo by Stuart Franklin: AIDS clinic in Zambia

How do we translateour rich understandingof vulnerability…

… into a formal analysis?

… into strategies forreducing vulnerability andadapting to climatic risks?

…into key messages fordecision makers?

IntroductionIntroductionBeginning with a few concepts (confusados?)Focus on socio-economic vulnerability and

climatic hazardsChoosing and using indicatorsPlotting profilesAggregating indicators into indicesLinking to…

◦ Future vulnerability◦ Adaptation

Further methods

A sample of conceptsA sample of concepts Exposure to pollution Dynamic Mapped by geographic zones Biological Synergies with other environmental stresses Integrated vulnerability actual (IVA) Who is vulnerable? Perception of drought and climate change Range of coping strategies

Political ecology of vulnerability

DefinitionsDefinitionsLanguage

◦German: vulnerabiliteit◦Italian: vulnerabilidad

…if this were a mature field we would not argue over definitions

Let’s focus on the methods today!

IndicatorsIndicators

The nuts and bolts of creating indicators, much more at www.OECD.org/dataoecd/37/42/42495745.pdf

Starting with Starting with livelihood vulnerability & exposurelivelihood vulnerability & exposure

Who and what is exposed:◦ Who are the vulnerable livelihoods?◦ What livelihood activities are exposed to climatic

hazards?◦ What ecosystem and economic services support those

activities? Which hazards are significant in the area, for those

vulnerable groups? What is the relative rating for the exposure of each

vulnerable group to each climatic hazard?

Build a sensitivity matrix

Who are vulnerable?To what?

Livelihood sensitivity matrixLivelihood sensitivity matrixCLIMATIC HAZARDS Exposure

IndexDrought Dry spells Floods Warm spells

ECOSYSTEM SERVICES

Soil water ▲ ■ ▲ ◦ 75

Water supply ▲ ○ ■ ◦ 60

Wood fuel □ ◦ ○ ◦ 35

Grazing/fodder ■ ○ ■ ◦ 55

LIVELIHOODS

Smallholders ▲ □ □ ◦ 60

Emerging farmers □ ○ ○ ◦ 40

Traders □ ◦ ■ ◦ 45

Impact Index 73 40 60 20

Select indicatorsSelect indicators Which indicators portray the situation of each

vulnerable groups’ exposure to each hazard?◦ Are indicators specific to each cell?◦ Does an indicator relate to more than one group or more

than one hazard? Vulnerability to which outcomes?

◦ Our usual concern is the impact on economic and livelihood activities—productive assets, property, wealth

◦ Would we choose different indicators to look at different outcomes, such as mortality? Wellbeing, disease and psychological stress? Or displacement and migration?

Select a few groups/hazards; list a sample of key indicators with indicative data

Indicators, transformations

and profiles

Transforming data into indicators: Transforming data into indicators: Scoring in the range from minimum to Scoring in the range from minimum to maximummaximum

Two formulas depending on whether a high value in the original data corresponds to high or low vulnerability

High vulnerability = high score (reduction in crop yields during drought)

Vi = (Xi - Xi,min)/(Xi,max-Xi,min)*100Vi = the transformed vulnerability indicator (i)Xi = the indicator before it is transformedXi,max = the maximum score of the indicator (i) before it is

transformedXi,min = the minimum score of the indicator (i) before it is

transformed

High vulnerability = low score (GDP per capita)Vi = (Xi,max – Xi) / (Xi,max-Xi,min)*100

Example:Example:High value in the original data ~ high High value in the original data ~ high vulnerabilityvulnerabilityDecrease in crop yield during a drought

(average production is 4 t/ha in a normal year)◦ Range in the region among all small farmers

Minimum reduction of 1 t/ha Maximum reduction of 3 t/ha

◦ For a selected farmer in the region, the decrease is 2.5 t/ha

◦ Relative vulnerability for that farmer is: Vi = (2.5 – 1) / (3 – 1) * 100 = 75

◦ For a farmer that is better adapted to drought (lower vulnerability), for example with a decrease of on 1.5 t/ha Vi = (1.5 – 1) / (3 – 1) * 100 = 25

Example:Example:High value in the original data ~ low High value in the original data ~ low vulnerabilityvulnerabilityLevel of farm subsidies from government

(perhaps in food aid or cash for work) during a drought (average subsidy is $1000 in recent droughts)◦ Range in the region among all small farmers

Minimum subsidy is $500 per year Maximum subsidy is $4000 per year

◦ For a selected farmer in the region, the subsidy was $750

◦ Relative vulnerability for that farmer is: Vi = (4000 – 750) / (4000 - 500) * 100 = 93

◦ A farmer that receives greater subsidies (perhaps due to political connections) is better adapted to drought (lower vulnerability), for example with a subsidy of $2500: Vi = (4000 - 2500) / (4000 - 500) * 100 = 43

Alternative approaches to Alternative approaches to transformationtransformationExpress each indicator as a percentage

◦ Percentage reduction in production during a drought

Use standard scores:◦ Vi = Xi – Xi,mean / Xi,standard deviation◦ Note: results in positive and negative scores

Setting a threshold for the maximum

Result is a database (VI) with no units of relative vulnerability

Transform the selected indicators

visualisationvisualisationMulti-attribute profiles make sense.

Plotting profilesPlotting profilesThe structure of vulnerability is….

◦ different for different groups and hazards◦ changes over time◦ is more important than the relative index

Examples for different types of vulnerable groups and exposures

See difference between economic exposure and vulnerability to mortality

Plot the selected indicators

List profile or radar plotList profile or radar plotDrought-food security

0 2 4 6 8 10

Climate

Water & Land

Health

Economy

Socio-institutional

Demography

Drought-food security

Drought-food security

Climate

Water & Land

Health

Economy

Socio-institutional

Demography

Comparing vulnerable groups and Comparing vulnerable groups and profilesprofiles

Drought-food security

Climate

Water & Land

Health

Economy

Socio-institutional

Demography

Population-land degradation

Climate

Water & Land

Health

Economy

Socio-institutional

Demography

Globalisation-economic

Climate

Water & Land

Health

Economy

Socio-institutional

Demography

Urban-healthClimate

Water & Land

Health

Economy

Socio-institutional

Demography

Vulnerability Profile, Delanta Dawunt, Ethiopia

-0.1

0.1

0.3

0.5

0.7

0.9

1.1HH Size

Male laborers

Total Income

Total Expenditure

Crops sales price in bad year

Food Aid

Grazing land

Crop land

Mid Altitude

Road Access

Livestock holdings

Types of dairy

Low income crop (V High)

Middle income crop (High)

Crop/dairy (Mod)

Isolated, middle income crop (Mod)

High income dairy (Mod)

Five vulnerable groups in EthiopiaFive vulnerable groups in Ethiopia

Aggregate indexesAggregate indexes

A special topic of some concern as people try to allocate adaptation funds: see Hans-Martin Fussel’s review for the World Bank (2009) for a highly critical analysis.

Aggregating to indicesAggregating to indicesCommon practice

◦Add up each indicator◦V* = ∑ (Vi)/ N(i)

Vi is the individual indicator N(i) is the number of indicators

◦Hoovering analogy: V corresponds to the weight of dirt collected in the bag of a vacuum cleaner (but difficult to verify)

Aggregation

Aggregation: ExampleAggregation: ExampleCommon practiceFive indicators with scores in the

range of 0 to 100:◦10 + 20 + 60 + 90 + 20◦Sum = 200◦Average = 80

The unweighted score (V*) is the average across all five indicators = 80

Aggregation: WeightingAggregation: WeightingEach layer in the VI data base can

be weighted before adding up◦V* = ∑ (Vi * Wi / ∑Wi)

If no weights are applied◦all indicators are equally important◦A high score‘cancels out’ a low score

(weak sustainability: may not correspond to the implicit conceptual model)

Weighting: ExampleWeighting: Example Weighting each indicator Five indicators with scores in the range of 0 to 100:

◦ 10 + 20 + 60 + 90 + 20 Weights applied to each indicator

◦ 40; 15; 15; 15; 15◦ Assumes first indicator is most important, all others are

equally important◦ Sum of weights = 100 (this is convenient but not

necessary) Weighted index:

◦ 10*40/100 + 20*15/100 + 60*15/100 + 90*15/100 + 20*15/100

◦ = 4 + 3 + 9 + 13.5 + 3 = 32.5 The weighted score (V*) is the sum of the weighted

indicators (the cross product sum in Excel)

Approaches to weightingApproaches to weightingExpert decidesStakeholders decide

◦ Reach a consensus in consultation◦ Choices revealed by asking for preferences

between pairs to establish a rank order to the indicators (Saaty algorithm)

Vulnerable groups decideFitting to outcome

◦ Multivariate statistics: predict an outcome such as economic losses during a drought

Explore different perceptions◦ Let stakeholders and experts try out different

weights and see if the resulting maps are similar. If not, why not?

Aggregation in the sensitivity Aggregation in the sensitivity matrixmatrixAcross the rows = exposure to

different climatic hazards◦Could be weighted by relative

frequency (or magnitude) of each hazard

Down the columns = socio-economic sensitivity to the impacts of climatic hazards◦Could be weighted by the prevalence

of each livelihood or importance of each sector to the regional economy

Other aggregation modelsOther aggregation modelsFlags and thresholds

◦ Count the number of indicators that exceed a threshold for concern

◦ Assumes that vulnerability is related to the number of critical stresses rather than a balance between each indicator

Dependencies and cascades◦ Some indicators are modelled on others, for

example changes in non-farm income trigger different values the sensitivity of maize to drought

◦ Assumes that there is a causal relationship between indicators

Try out different aggregation techniques

……commentarycommentary The ‘standard model’ of a basket of indicators is not

adequate. Whichever approach is chosen, it should meet specific

criteria:◦ Represents a conceptual model of vulnerability processes

that coheres with theoretical and observed evidence◦ Complies with best practice in the mathematics of

aggregating different indicators (issues of scale, numerical representation, ranking, uncertainty)

◦ Is validated (or verified) as consistent with the evidence and robust

◦ Is well documented and transparent, providing a consistent trace from assumptions to results

Linking to adaptationLinking to adaptation

The SEI working paper (Downing et al. 2005) has more material on dynamic vulnerability, and more up to date!

Linking to future Linking to future vulnerability vulnerability

How sensitive are the indicators, plots and aggregate indices?◦ What would the plot look like for an historical event, such as the 1982

El Nino?

◦ How big would you expect changes to be in the future?

What narrative explains the relative numbers?How does the changing structure of vulnerability

influence the choice and interpretations of indicators?

Discuss how scenarios of future vulnerability might change the plots: we will look at scenarios in more detail on Wednesday

Future vulnerability and adaptation

Verification and validationVerification and validation

How robust are the plots and aggregate indices to different assumptions?◦ Uncertainty in the original data◦ Different transformation and weighting techniques

How important are different perceptions and values?

Linking to adaptationLinking to adaptationHow does this synthesis of current

vulnerability link to planning adaptation?◦Targets vulnerable groups and hazards

as a group below a threshold (note limitations of aggregate rankings!)

◦Suggests different coping strategies that would be effective

◦Screening options in multi-criteria analysis

◦Monitoring success

Further methodsFurther methods

Clustering on more than one dimension◦ Bubble charts◦ Sensitivity to climate impacts vs adaptive capacity

Rule-based, dynamic models of representative vulnerable groups

VI:Layers of indicatorsProfilesAggregate indices

Narratives:Cultural values

Historical transformationsDynamic processesComplex networks

Food Insecurity: Present Food Insecurity: Present StatusStatus

Food Insecurity: Present Status

4

6

8

10

12

14

2 4 6 8 10 12 14

Food Availability

Fo

od

Acc

ess

SUSTAINABLE DEVELOPMENT

AT-RISK

Orissa

Bihar

HimachalPradesh

Gujarat

Punjab

WestBengal

TamilNadu

Rajasthan

Maharashtra

Haryana

Karmataka

UttarPradesh

Assam

AndhraPradesh Kerala

MadhyaPradesh

Food Insecurity: Food Insecurity: Links to Climate ChangeLinks to Climate Change

Food Insecurity: Worst Case?

4

6

8

10

12

14

2 4 6 8 10 12 14

Food Availability

Fo

od

Acc

ess

-Disaster morbidity-Social infrastructure losses-Consequences of availability & access+Adaptation interventions?

-Energy costs & reduced irrigation-Loss of market infrastructure in disasters-Increased transport costs

+Local sourcing for markets

-Heat stress & water shortage-Drought & storms-Salinisation & loss of coastal lands

+ CO2 enrichment

Food Insecurity: Worst Food Insecurity: Worst Case?Case?

Food Insecurity: Worst Case?

4

6

8

10

12

14

2 4 6 8 10 12 14

Food Availability

Fo

od

Acc

ess

AT-RISK

SUSTAINABLE DEVELOPMENT

Vulnerability hot spots?Vulnerability hot spots?

ADAPTIVE CAPACITY

IMPACTS L H

H Vulnerable 3 billion $2000 pc

Adaptable 1.5 billion $15,500 pc

L

Residual risk 0.5 billion $3000 pc

Safe 0.5 billion $12,000 pc

Vulnerability hot spots?Vulnerability hot spots?

IMPACTS VS ADAPTIVE CAPACITY

MOST VULNERABLE (37)DEVELOPMENT OPPORTUNITIES (26)RESIDUAL RISKS (51)SUSTAINABLE LIVELIHOODS (37)

This is an early experiment with hot spots by Downing (using the ExternE results). There are huge problems with the conceptual approach, indeterminacy in the data and problems in interpreting the results. It is an open question whether a ‘hot system’ approach can deliver more reliable indices!

Decision treeDecision tree• Stakeholders participate in

developing decision trees

• Machine-assisted techniques

• Link to scenarios of driving forces as well as decisions

Sukaina Bharwani (SEI) and Michael Fisher (Kent) developed an automated rule tree approach that is promising, especially where tacit information is critical to the decision process.

Agent based modellingAgent based modelling

Temperature

Ann. Temp

Strategy

Existing Knowledge*Altittue*Delayed

ImmediateSustainedMultiplierLabour

Irrigation

Environment

Soil type

Economy

Sterling*Dollar*Euro*

WorldPrices*Crop production

Weather

SpringSummerAutumnWinter

Ann. TempAnn. Prec

Capital

InputOutput

Production

Crop

Irrigation

Crop

Rainfall

SpringSummerAutumnWinter

Ann. PrecWorld

Agent

Agent

Agent

Java Expert SystemShell (Jess)

RePast

Political ecology of vulnerability

Who are vulnerable?To what?

Indicators, transformations

and profilesAggregation

Future vulnerability and adaptation

Practical exercisesPractical exercises

There is only one here, something we hope to expand!

Break out exerciseBreak out exercise Build a sensitivity matrix Select a few groups/hazards; list a sample of

key indicators with indicative data Transform the selected indicators Plot the selected indicators Try out different aggregation techniques Discuss how scenarios of future vulnerability

might change the plots: we will look at scenarios in more detail on Wednesday