HLEG thematic workshop on Measuring economic, social and environmental resilience, 25-26 November 2015, Rome, Italy, More information at: http://oe.cd/StrategicForum2015

1. Assessing socio-economic
resilience to natural disasters
Stephane Hallegatte, Mook Bangalore, Adrien Vogt-Schilb
Climate Change, The World Bank
November 2015
1

2. 2
Climate-related shocks and stresses,
already a major obstacle to poverty
reduction, will worsen with climate change

3. 3
Poor people are 50%
more likely to be flooded
Poor people are 130%
more likely to be affected
by a drought
Poor people are 80%
more likely to be affected
by extreme heat
Poor people are often more exposed to
natural disasters – take the case of Nigeria

4. In most (but not all) countries, poorer urban dwellers are more
likely to live in flood zones
4
Poor people are often more exposed to
natural disasters – take the case of urban
floods

5. Poor people are always much more
vulnerable to natural hazards
5
5

6. Saved at a financial
institution
Average transfer from social
protection
Indonesia Poor 8% $0.5/day
Non-poor 21% $2/day
Poor people have access to less support to cope and adapt
6

7. Saved at a financial
institution
Average transfer from social
protection
Indonesia Poor 8% $0.5/day
Non-poor 21% $2/day
Malawi Poor 4% $0.05/day
Non-poor 11% $0.17/day
7
Poor people have access to less support to cope and
adapt

8. 8
How much is social vulnerability
increasing disaster risk?

9. Where should investments in resilience
be concentrated?
Where people are most vulnerable!Where they are most cost-effective!

10. Preparation Coping
Protection
to reduce the probability and size of
losses and increase those of benefits
Coping
to recover from losses and
make the most of benefits
Insurance
to transfer resources across people and over
time, from good to bad states of nature
Knowledge
to understand shocks, internal and
external conditions, and potential
outcomes, thus reducing uncertainty
World Development Report, 2014
10
Disaster risk management is the combination of
many policies – but how to design a strategy?

11. Disaster risk management is the combination of
many policies – but how to design a strategy?
11
Risk cannot be reduced to zero: we
need to help affected people!
Prevention is the key: we should
avoid disasters!

12. A tool to answer three questions:
12
• How much is social vulnerability increasing disaster
risk?
• Where should investments in resilience be
concentrated ?
• How to design a disaster risk management
strategy?

13. The World Bank’s resilience indicator combines
information on risk to assets
13
Hazard Exposure
Vulnerability
(asset quality,
early warnings...

14. 14
We mostly report – and discuss – asset
losses

15. Poverty
The World Bank’s resilience indicator combines
information on risk and socio-economic factors
15
Over-exposure of
poor people
Less support for
poor people
Higher
vulnerability of
poor people
Hazard Exposure
Vulnerability
(asset quality,
early warnings, and
income
diversification)
Ex-post
support
(Scale-up of social
protection, access
to finance)

16. A definition of socio-economic
resilience
• Welfare losses are larger if losses are concentrated on a few
individuals, if they affect mostly poor people, or if they are amplified
by poverty traps.
• Resilience is the wedge between asset losses and welfare losses:
• Can be seen as one of four components of the risk to welfare:
16
Socio−economic resilience =
Asset losses
Welfare losses
Risk to welfare =
Expected asset losses
Socio−economic resilience
=
(Hazard) ∙ (Exposure) ∙ (Asset vulnerability)
Socio−economic resilience

17. Global data sources
Hazard
•Flood level from
GLOFRIS global
model
•Protections using
simple rules based
on GDP per capita
Exposure
•Localization of
people and assets
based on
Landscan global
data
•Case study results
for the over-
exposure of poor
people
Asset
vulnerability
•Housing quality
based on
USGS/PAGER
global database
and simple
vulnerability
functions
•Early warning
(from HFA)
reduces losses
Impact on
income
•Diversification of
income through
transfers (from
ASPIRE and others)
•Link between assets
and income, using
average capital
productivity
•Simple assumption
for the duration of
reconstruction
Coping capacity
and social
protection
•Scale-up of social
protection, based on
credit ratings and
HFA monitor
•Financial inclusion
from FINDEX
•Access to education
and health and
employment
opportunity (WDI)
Impact on
welfare
•Marginal utility of
consumption
(η=1.5)
•Share of income
of bottom quintile
(WDI)
•Poverty traps
modeled as life-
long reduced
earning
Parameter values differ for poor and non-poor people for each step
We define “poor people” as the bottom 20 percent in each country
Sensitivity analyses suggest that results are robust17

18. A global picture of socio-
economic resilience18

19. Welfare risks from river floods in 90
countries
19

20. The socio-economic resilience of 90
countries to river floods
20

21. Country-level resilience indicator
and a scorecard to identify
priority policies • Resilience is correlated with GDP
per capita (correlation = 0.6).
• $1000 increase in GDP per
capita increases resilience by 0.8
points.
• But GDP per capita only explains
a fraction of the variation: other
factors count, such as inequality,
social protection, etc.
• All countries can act to increase
resilience, regardless of their
income level21

22. Resilience “anomaly”
Resilient countries have:
• Low inequality
• High share of social protection and transfers
• High ability to scale-up social protection
(good financial management and
preparedness)
• High financial inclusion
• Universal access to health care and
education
• Low unemployment
22

23. Resilience is largely independent of risk and
exposure
23

24. Risk management scorecards to
identify policy priorities
An application to 90 countries
24

25. Illustration on the case of
Malawi
25

26. 26
High flood risks in Malawi
Welfare risk = expected annual
welfare losses = 2.1% of GDP

27. 27
The four drivers of welfare risk,
including resilience
And the 14 drivers of resilience
2 drivers of asset vulnerability

28. These sub-indicators are
weighted using our simple
model based on welfare theory
The weighting is country-
specific, depending on the
context28

29. 29
They are ranked based on the
efficacy to reduce welfare risk
This ranking is country specific

30. Socio-economic resilience in Malawi
is 51%
It means that each $1 of flood losses
is equivalent to a $2 loss in welfare
This is due to losses that are
concentrated on poor people30

31. 31
Poverty reduction and social
protection have high efficacy
Increasing the income share of the
bottom 20% by 1.5 points or the
social protection toward them by 6
points reduces risk by 10%

32. 32
Poverty reduction and social
protection have high efficacy
Malawi is lagging along these
dimensions, so progress is likely
possible

33. Risk much smaller in Sweden, mostly thanks to better protection
But also lower asset vulnerability and higher resilience in Sweden
Let’s focus on resilience
Comparing Malawi and Sweden
33
Malawi Sweden

34. Malawi Sweden
34

35. Malawi Sweden
35

36. Malawi Sweden
36

37. Malawi Sweden
37

38. Malawi Sweden
38

39. Malawi Sweden
39

40. (Very) preliminary application to
the Philippines, at the provincial
level
!! This is based on preliminary data, focus only on river floods !!
40

41. Model and data sources
Hazard
•Flood level from
GLOFRIS global
model (DOST has
better estimates)
•Protections using
simple rules based
on GDP per capita
Exposure
•Localization of
people and assets
based on
Landscan global
data
•Case study results
for the over-
exposure of poor
people
Asset
vulnerability
•Housing quality
based on roof and
wall structures
from family
surveys
•Early warning
proxied with
access to radio
and cell phone
ownership
Impact on
income
•Diversification of
income (from family
surveys)
•Link between assets
and income, using
average capital
productivity
•Simple assumption
for the duration of
reconstruction
Coping capacity
and social
protection
•Scale-up of social
protection
•Financial inclusion
(Central Bank)
•Access to education
and health and
employment
opportunity
Impact on well-
being
•Marginal utility of
consumption
(η=1.5)
•Poverty incidence
•Poverty traps
modeled as life-
long reduced
earning
Parameter values differ across provinces and –within each
province – for poor and non-poor people41

42. 42
Risk to well-being combines risk to assets and
socio-economic capacity
Risk to assets (% of GDP) Socio-economic capacity Risk to well-being (% of GDP)

43. 43 Application to the 2005 floods in Mumbai

44. Dynamics of exposure and vulnerability
Exposure Vulnerability
Household
income
(Rs./month)
Share of
population in
survey (%)
Share of
population
exposed (%)
<5000 24% 41%
5001–7500 28% 34%
7501–10000 23% 19%
10001–15000 12% 5%
15001–20000 6% 1%
>20000 6% 1%
n=21,691 n=930
44

45. Ex-post measure of resilience and how to improve
Baseline
The policy
effect on
asset and
welfare losses
Asset losses Rs. 35 billion
Net present value of consumption losses Rs. 39 billion (+13%)
Welfare losses Rs. 60 billion (+71%)
Social economic resilience 57%
45
-5.0
0 0 0
+0.9
-5.0
-2.7
-3.7
-3.0
-2.6
Reduce exposure by
5%
Increase post-
disaster support
(from 10 to 25% of
the losses)
Cut reconstruction
duration (by a third)
Increase income
diversification (from
10% to 25%)
Increase income
share of bottom 20
percent (by 10%)
Relativechange(%)
Asset losses Welfare losses

46. This is still under development, so
comments are welcome and useful
shallegatte@worldbank.org
avogtschilb@worldbank.org
mbangalore@worldbank.org
46