Why We Under Prepare for Hazards

1
Why We Under Prepare for Hazards
Robert J. MeyerThe Wharton SchoolUniversity of
Pennsylvania
2
An Eternal Problem Minimizing the Societal
Impact of Natural Disasters

• A modern dilemma advanced scientific knowledge
  of the processes that generate natural disasters
  and means to protect against them has done little
  to reduce their damaging impact.
• 2004 Tsunami (est. 224,000 dead) 2005 Hurricane
  Katrina (100bn loss, 1300 dead) 2005 Earthquake
  (Pakistan) 79,000 killed 1970 Cyclone, Bay of
  Bengal 300,000 killed 1995 Kobe Earthquake
  (Japan) 6,000 killed, 80bn loss.

3
Why were these tragedies so bad?

• In almost all cases post-event analyses suggest
  that the events need not have been as a damaging
  as they were
• Decision makers knew they were living in
  risk-prone areas, knew what steps to take to
  mitigate losses, and, often, could afford to
  undertake them.

4
Example New Orleans Close Call with Hurricane
Ivan, 2004
5
Example

• September 13, 2004 Category-5 Hurricane Ivan is
  near the West Coast of Cuba heading NW into the
  Gulf, and 3 of 6 computer models predict a direct
  hit on New Orleans in 3 days
• Likely consequence catastrophe

6
(No Transcript)
7
September 14 Mayor Orders General Evacuation,
but discovers major flaws in evacuation system
Mayor Nagin said he would "aggressively
recommend" people evacuate, but that it would be
difficult to order them to, because at least
100,000 in the city rely on public transportation
and have no way to leave. Despite the potential
need for emergency housing, no shelters had been
opened in the city as of Tuesday night. Nagin
said the city was working on setting up a shelter
of "last resort" and added that the Superdome
might be used, but a spokesman for the stadium
said earlier Tuesday that it was not equipped as
a shelter.
8
Good News

• Ivan spares New Orleans (coastal Alabamians and
  Floridians not real happy, though).
• New Orleans breathes sigh of relief

9
Quiz

• If you were Ray Nagin, what should you have
  learned from this close call?
• a) That the city was fortunate to have averted a
  catastrophe, hence immediate steps should be
  taken to remedy the evacuation problems
• b) The city is safe for another 40 years
• c) The city is inherently lucky
• d) What close call?

10
One year later
11
Two Months Later Wilma

• October 2005 Wilma becomes strongest hurricane
  ever recorded in Atlantic basin, threatens South
  Florida
• South Floridians ordered to stock up (for the 4th
  time that year)
• Q What did residents learn from their own
  earlier bout with Katrina and other storms?

12
Apparently, very little
13
(No Transcript)
14
So why?

• Ultimately, decisions to undertake mitigation are
  made by individuals for whom the best course of
  personal action is highly uncertain
• While one may be aware of aggregate risk, how
  this translates to individual circumstances is
  often ambiguous
• There is inherent uncertainty about the
  cost-effectiveness of mitigation investments,
  which compete with other expenditures
• The processes that allow us to make good
  decisions in most walks of life fail when applied
  to low-probability, high-consequence events

15
The bottom line why we under-prepare

• We have limited abilities to recall the past,
  have limited abilities to foresee the future, and
  make mitigation decisions by imitating the
  behavior of neighbors who are equally myopic

16
Biases in learning from the past

• For most human endeavors, learning by
  trial-and-error is an efficient way to develop
  survival skills
• The problem when TE processes are applied to
  learning about mitigation in low-probability,
  high-consequence, settings, it will lead us to
  the wrong behaviors more often than the right
  ones.

17
The reasons

• One rarely sees positive benefits of investments
  in mitigation (most experiences are false
  alarms)
• When hazards are encountered, the implications
  they hold for optimal mitigation will tend to be
  ambiguous

18
Two major consequences

• Rapid extinguishing of normative mitigation
  behaviors and
• The prolonged persistence of superstitious
  beliefs about mitigation

19
Example Rapid forgetting and the Rebuilding of
Pass Christian, MS after Hurricane Camille
20
Richelieu Apartments, Pass Christian,
Mississippi, August 1969
21
Same Location after Hurricane Katrina (former
Pass Christian Shopping Center
22
Example the flip side of recency learning too
much from recent disasters
23
September 2005 Houston Braces for Hurricane Rita
24
FEMA, State vow not to allow this to be another
Katrina

• Action 1.5 million Texans in Galveston/Houston
  ordered to evacuate via staged plan

25
Slight problem

• 2.8 million, not 1.5 million, try to leave.
• Takes up to 13 hours to drive 45 miles
• Problem exacerbated by broken down cars, need to
  send relief supplies to people in cars
• More die during evacuation than storm

26
How observing past outcomes can be misleading
27
The hurricane-proof Dome Home Pensacola Beach,
FL 2003
28
The Dome Home after Ivan, September 2004
29
(No Transcript)
30
The Persistence of Mitigation Myths
31

• A tornado is approaching your house. The best
  way to prevent the house from suffering damage
  is
• Close all the doors and windows to create a tight
  seal
• Open a few windows to relieve pressure when the
  funnel passes near or over
• Neither of these actions will have any effect on
  reducing damage

32

• Opinions (95 Pennsylvanians)
• Close all the doors and windows (15)
• Open a few windows (55)
• Neither of these actions will have any effect on
  reducing damage (30)

33
Hurricanes in the Lab
34
The Hurricane Simulation

• Respondents were endowed with a residence of
  known value, and were paid at the end of the
  simulation the difference between this endowment
  and the cost of mitigation and storm repairs.
  Mitigation measures do not improve the value of
  the home--they only reduce storm losses.
• At the start respondents are told their expected
  length of tenure in the home and its location
• Respondents could gather information about
  hurricanes, mitigation, and make mitigation
  purchases by clicking control buttons in the
  simulation

35
(No Transcript)
36
(No Transcript)
37
The Explanation

• In the absence of an unambiguous correct course
  of action, mitigation decisions were driven by
  short-run negative feedback
• There was no evidence of learning either from
  observing the misfortunes of others or close-call
  encountersthe damage had to be real
• In time lag effects vanished, but investments
  remained well below optimum.

38
Biases in seeing into the future

• As bad as we are at learning from the past, we
  seem to be worse at accurately anticipating the
  future consequences of current behaviors

39
Key biases

• Projection bias we have a hard time envisioning
  future hedonic states that are different from the
  one we are in
• Optimism Bias we are prone to imagine the are
  prone to the best rather than worst-case
  scenarios, causing errors in protective planning

40
Examples New Orleans post 2004 Hurricane
planning, failure to evacuate in the face of
Hurricane Katrina
41
Optimistic Planning and the 1935 Labor Day
Hurricane
42
September 2,1935 (Labor Day)

• 675 WWI vets are in make-shift camps in the Fla
  Keys, working to build a highway to Key
  West
• 7 AM Weather Bureau warns there is a CHANCE that
  a hurricane MIGHT affect the area that night or
  early Tuesdaybut it looks to be heading to Cuba

43
The decision

• The only way to evacuate the Vets is by a train
  from Miami
• No train had been scheduled because of the
  holiday a special one would have to be ordered.
• the FERA supervisor in Jacksonville must decide
  whether and when to order an evacuation

44
The Decision

• The calculation it usually takes 2.5 hours to
  ready a train and reach the camps
• Hence, no need for an immediate evacuation if
  the threat looks real come noon/early afternoon,
  send the train (better be safe than sorry).

45
What happened

• 130 PM Weather service revises forecastgales
  to begin soon, hurricane conditions late that
  night
• 2PM Evacuation Train ordered
• Problem Cars are in Miami, Engine in Homestead
• Engine is Pointed in the Wrong Direction
• Train does not leave Homestead until 5PM

46
5 PM
47
7PM
48
8PM no further progress
49
10 PM Landfall Long Key 200 mph 26.35
50
Morning 452 Dead 279 VFW Camp Workers
51
Biases in leaning from Others

• Given the tremendous uncertainty that surrounds
  mitigation decisions, many homeowners tend to
  make decisions by imitating the decisions of
  others or following social norms
• The problem, of course, is that such a heuristic
  works only if the norms are rational

52
Example the Wharton Earthquake Simulations
53
Procedure

• Participants played a series of real-time games
  in which they lived with other players in a
  hypothetical country prone to earthquakes.
• They could make investments in permanent
  improvements that reduced damage from quakes
• They were paid based on the initial value of
  their home plus earnings minus earthquake damage
  and mitigation investments

54
The Screen Layout
55
The Manipulations

• For half of all communities mitigation was
  ineffective (optimal investment0), for half it
  was highly effective (optimal100)
• Ss played 3 blocks of 10-minute games
• After 1 warm up game, 1 player in each community
  was secretly informed of the true effectiveness.
  Other players knew that the community had an
  informed player, but his/her identity was not
  revealed

56
(No Transcript)
57
(No Transcript)
58
Informed player is told that mitigation is highly
effective
59
Informed player is told that mitigation is highly
effective
60
(No Transcript)
61
So why dont we prepare?

• As human decision makers we have evolved to be
  quite skilled at learning quickly from frequent,
  unambiguous, feedback, and planning for the short
  term
• Problem effective mitigation decisions requires
  skills that are just the opposite to that for
  example, a willingness to persistently invest in
  costly actions that do not have an observable
  positive payoff

62
Solutions the obvious

• Legislation policies need to be put into place
  that protect policy makers and residents from
  themselves e.g. through building codes,
  long-term commitments to funding, required
  hazard-response plans
• Education residents need to be taught not just
  about hazard risks, but also trained to be better
  long-term decision makers

63
Solutions, the less obvious

• Problem forming effective legislation and
  education programs requires us to know much more
  than we currently do about human decision making
  in low-probability, high-consequence settings.
  While we know much about the physical science of
  hazards, we know much less about the associated
  psychological science. Bridging this gap should
  be a major goal of research funding in the
  natural hazards area in the years to come.