Post on 23-Jul-2018
Does Disaster Experience Undermine Risk Perception and Response: An Empirical Analysis of
Typhoon-Morakot
Ching-Cheng CHANGInstitute of Economics, Academia Sinica,
Dept. of Agricultural Economics, National Taiwan University
Wan-Jung ChouAECOM Capital
Yu-Chia HuangDept. of Agricultural Economics, National Taiwan University
APEC Typhoon Symposium–Lessons Learned from Disastrous TyphoonsInstitute of Manila, the Philippines, Nov 24-25, 2015
Introduction
• Typhoon Morakot
August 7 2009, Taiwan
AfterMorakot
Warning end
Weather Watching
in typhoon Season
Focus on weather systems in South China and Taiwan region and typhoon info in North west Pacific and South China Sea
Sea warning
Land warning
Emergency response
centerTyphoon forms
12th 02:30 12th 08:30
14th 14:30 17th 09:13
14th 15:16
10th 20:00 20 times in 8 days
Provide typhoon news
in 09:00
Provideinformation twice a day
Provideinformation 4 times a day
Analysis and discussion
Schematic diagram of stages of guidance and information provided during
typhoon season in Taiwan
• Did people take precautionary actions before the
event?
• Did this event have any influence on their
perception of risk exposure?
• How would they accordingly adjust the ways of
coping with typhoon risks in the future?
Research Questions:
Purpose
• Investigate the causal relationship between risk perceptions and precautionary actions The influence of previously adopted protection
actions on risk perceptions observed at the time of observation and
in turn the effects of risk perceptions on individuals’ intention to take precautionary actions in the future.
HypothesisPrevious
experience, trust, socio-econ factors
Risk perception
Precautionary behavior
Mediating Effect
two-stage approach
Statistical Analysis
1 • Predict perceived risk levels using ordered probit model
2 • Examine its influence on precautionary intentions using probit model
Two-Stage approach
Data1. Survey Data• Where: Household Survey of Post-Morakot Social Impact and
Recovery-Wave 1.• When: June, 2010• How: face to face interview
2. Interviewees• Age: over 20• Status: household heads, primary financial supporters or the ones
most capable of answering questions in an interview
3. Observations• Raw data contains 1658 observations• representing 1658 households
QuestionsPlease indicate which of the following measures you took (1) before the typhoon Morakot. In addition, please indicate which of the following measures (2) you intend to take in the future (before next typhoon event).
• Obtain the information about this typhoon from the TV, radio and other sources (INFORMATION)
• Take part in local disaster drills(DRILLS)
• Strengthen your house’s resistance to typhoons, e.g. sandbags, water pumping machines (HOUSE)
• Prepare food, clothes and other necessities (FOOD)
• Understand and make plans of evacuation routes and temporary shelters (EVACUATION)
• Purchase or renew personal accident insurance (INSURE_ACCIDENT)
• Purchase or renew typhoon and flood insurance for your property (INSURE_PROPERTY)
• The information revealed is twofold: 1. First question (1) exhibits whether or not households took
precautionary measures before the typhoon Morakot
2. Second question (2) shows whether or not households, at the time of interview, had the intention to take the same precautionary measures in the future
• According to the phases of the hazard life cycle:
Source: Lindell and Perry (2004)
PrecautionaryMeasures Mitigation measures
Preparedness measures
Recovery measures
Precautionary measure Description Before Morakot
% of the sampleAfter Morakot
% of the sample
Preparedness
INFORMATIONObtain the information about this typhoon from the TV, radio and other sources
91% 76%
FOOD Prepare food, clothes and other necessities 85% 76%
HOUSEStrengthen your house’s resistance to typhoons, e.g. sandbags, water pumping machines
62% 66%
Mitigation
EVACUATIONUnderstand and make plans of evacuation routes and temporary shelters
44% 69%
DRILLS Take part in local disaster drills 39% 61%
RecoveryINSURE_ACCIDENT Purchase or renew personal
accident insurance 24% 48%
INSURE_PROPERTY 11 6% 35%
Table 1: Precautionary measures
Risk Perception Description Mean Std Devi.
PROB_DISASTER
In a scale from ‘1’ (very unlikely) to ‘4’ (very likely), how likely do you think that a typhoon disaster would occur in the region of your residency?
3.29 0.87
IMPACT_SAFETY
In a scale from ‘1’ (very mildly) to ‘4’ (very seriously), to what extent do you think a typhoon disaster would threaten the safety of your life?
3.58 0.60
IMPACT_PROPERTY
In a scale from ‘1’ (very mildly) to ‘4’ (very seriously), to what extent do you think a typhoon disaster would result in the loss of your personal property?
3.50 0.67
Table 2: Risk perceptions
• 1st question measures one’s subjective judgment of the chances of a typhoon-induced disaster in the future.
• 2nd and 3rd question stated one’s perceivedlevels of potential impact on his/her personal safety and of the loss of his/her personal property, as a result of a typhoon disaster.
Table 3: Summary Statistics of the Sample
Variable Mean Variable MeanAlready adopted precautionary measures Geographic factors
The number of preparedness measures taken 2.38 Nantou 3%The number of mitigation measures taken 0.83 Chiayi 14%The number of recovery measures taken 0.30 Tainan 5%
Experience_disaster 0.63 Kaohsiung 48%Level of trust Pintong 17%
In central government 3.10 Taitong 11%In community 3.66 Tainan City 1%
Socio-demographic factorsHealth impact in Morakot
Female 40% The number of death 0.15Age 52.37 The number of injury 0.11Income
Below 12k 21%12k and above, below 36k 38%36k and above, below 60k 27%60k and above, below 108k 11%108k and above 3%
Education (illiterate) 11%Aboriginal 39%
Results1.Risk perceptions
• Confirm the association between precautionary behaviour taken before Morakot and households risk perceptions after typhoon Morakot.
• Whether or not former actions could reduce risk perceptions in a later stage is inconclusive.
• A negative correlation between trust in the central government and perceived impact of property loss is observed
Table 4 : The estimation results for three different types of risk perception
Ordered probit PROB_DISASTER IMPACT_SAFETY IMPACT_PROPERTYCoef. Std. Err. Coef. Std. Err. Coef. Std. Err.
The number of preparedness measures taken
-0.166*** 0.046 -0.022 0.048 0.078 0.046
The number of mitigation measures taken 0.175*** 0.049 0.238*** 0.056 0.194*** 0.053
The number of recovery measures taken -0.140** 0.065 -0.113 0.072 -0.028 0.070
Experience_disasterbefore Morakot -0.037 0.075 0.043 0.082 0.125 0.079
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level
Level of Trust PROB_DISASTER IMPACT_SAFETY IMPACT_PROPERTY
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
In central government 0.209*** 0.041 -0.010 0.046 -0.168*** 0.046
In community -0.154*** 0.044 -0.053 0.049 -0.039 0.048
Socio-demographic PROB_DISASTER IMPACT_SAFETY IMPACT_PROPERTY
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Female 0.256*** 0.074 0.146 0.079 0.071 0.077
Age 0.005 0.003 -0.009*** 0.003 -0.004 0.003
Income 0.098*** 0.028 0.051 0.030 0.140*** 0.031
Education (illiterate) -0.341*** 0.124 0.049 0.134 0.032 0.130
Aboriginal 0.140 0.084 -0.183 0.095 -0.329*** 0.090
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level
Geographic factors(base = Tainan City) PROB_DISASTER IMPACT_SAFETY IMPACT_PROPERTY
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Nantou 1.126*** 0.385 -0.206 0.402 -0.035 0.386Chiayi 0.562 0.332 0.370 0.363 0.702** 0.347Tainan -0.001 0.346 -0.384 0.375 -0.053 0.359Kaohsiung 0.741** 0.323 0.436 0.352 0.606 0.335Pintong 0.719** 0.335 0.129 0.366 0.116 0.347Taitong 0.615 0.342 0.420 0.376 0.760** 0.360
Health impact in Morakot PROB_DISASTER IMPACT_SAFETY IMPACT_PROPERTY
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Death 0.248*** 0.095
Injured 0.414*** 0.126
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level
• Focused on the determinants of households’ intentions to take 7 different types of precautionary measures, respectively.
• each measure functions uniquely towards reducing the impact of typhoon events on households.
• Households may exhibit various degrees of willingness to take each of these measures and the knowledge behind,
• conveys useful implications for hazard risk management strategies at the household level.
2.Precautionary intentions
INFORMATION FOOD HOUSEProbit model estimation Coef.
(Std. Err.)dy/dx
(Std. Err.)Coef.
(Std. Err.)dy/dx
(Std. Err.)Coef.
(Std. Err.)dy/dx
(Std. Err.)
PREVIOUSLY ADOPTED THE SAME MEASURE
-0.177(0.186)
-0.050(0.052)
-0.055(0.145)
-0.015(0.041)
0.633***(0.096)
0.212***(0.030)
RISK PERCEPTIONProb_Disaster (very high) 1.864***
(0.593)0.522***
(0.165) 2.182***
(0.605)0.612***
(0.168) 2.660***
(0.532)0.889***
(0.172) Impact_Safety (very
seriously)2.000***
(0.706)0.560***
(0.196) 1.383**(0.690)
0.388**(0.193)
1.434**(0.624)
0.480**(0.207)
Impact_Property (very seriously)
-3.473***(0.652)
-0.973***(0.177)
-3.427***(0.652)
-0.962***(0.177)
-3.228***(0.585)
-1.079***(0.189)
EXPERIENCE_DISASTER -0.399***(0.095)
-0.112***(0.026)
-0.325***(0.094)
-0.091***(0.026)
-0.296***(0.086)
-0.099***(0.028)
LEVEL OF TRUSTCentral government+ -0.311***
(0.079)-0.087***
(0.022)-0.342***
(0.082)-0.096***
(0.023) -0.401***
(0.073)-0.134***
(0.023) Local community+ 0.169***
(0.060)0.047***
(0.017)0.190***
(0.061)0.053***
(0.017) 0.284***
(0.056)0.095***
(0.018)
Table 6: Predictors of the intention to take preparedness measures
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level ‘+’ For the purpose of simplicity, these variables were treated as quasi continuous variables. However, these variables are categorical in nature. One must note that a general trend of their association with households’ intention to take precautionary actions can be confidently identified in this approach but at the same time, one must interpret the corresponding marginal effects with caution.
θφθ
INFORMATION FOOD HOUSE
Probit model estimation Coef.(Std. Err.)
dy/dx(Std. Err.)
Coef.(Std. Err.)
dy/dx(Std. Err.)
Coef.(Std. Err.)
dy/dx(Std. Err.)
SOCIO-DEMOGRAPHIC FACTORS
Female -0.139(0.102)
-0.039(0.028)
-0.121(0.102)
-0.034(0.029)
-0.124(0.093)
-0.041(0.031)
Age -0.011**(0.004)
-0.003**(0.001)
-0.015***(0.004)
-0.004***(0.001)
-0.012***(0.004)
-0.004***(0.001)
Income+ 0.031(0.043)
0.009(0.012)
0.026(0.042)
0.007(0.012)
-0.005(0.039)
-0.002(0.013)
Education (illiterate) 0.101(0.173)
0.028(0.048)
0.112(0.171)
0.031(0.048)
0.205(0.160)
0.069(0.053)
Aboriginal -0.937***(0.116)
-0.262***(0.030)
-0.998***(0.119)
-0.280***(0.031)
-0.643***(0.113)
-0.215***(0.036)
CONSTANT 2.253***(0.446)
2.568***(0.434)
1.030***(0.397)
Number of obs. 1135 1135 1135 Log likelihood -563.46 -564.75 -666.89Pseudo R2 0.099 0.101 0.082‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level ‘+’ For the purpose of simplicity, these variables were treated as quasi continuous variables. However, these variables are categorical in nature. One must note that a general trend of their association with households’ intention to take precautionary actions can be confidently identified in this approach but at the same time, one must interpret the corresponding marginal effects with caution.
Table 7: Predictors of the intention to take mitigation measures
DRILLS EVACUATIONProbit model estimation Coef.
(Std. Err.)dy/dx
(Std. Err.)Coef.
(Std. Err.)dy/dx
(Std. Err.)PREVIOUSLY ADOPTED THE SAME MEASURE
0.278***(0.103)
0.095***(0.035)
-0.335***(0.099)
-0.104***(0.030)
RISK PERCEPTIONProb_Disaster 2.086***
(0.484)0.714***
(0.161) 4.808***
(0.529)1.492***
(0.147) Impact_Safety 1.857***
(0.595)0.636***
(0.201) 1.014
(0.624)0.315
(0.193) Impact_Property -0.825
(0.577)-0.283
(0.197) -0.813
(0.589)-0.252
(0.183) EXPERIENCE_DISASTER -0.078
(0.086)-0.027
(0.030) -0.088
(0.089)-0.027
(0.028) LEVEL OF TRUST
Central government+ -0.136**(0.068)
-0.047**(0.023)
-0.389***(0.072)
-0.121***(0.021)
Local community+ 0.226***(0.056)
0.077***(0.019)
0.340***(0.058)
0.105***(0.017)
Table 7: Predictors of the intention to take mitigation measures
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level ‘+’ For the purpose of simplicity, these variables were treated as quasi continuous variables. However, these variables are categorical in nature. One must note that a general trend of their association with households’ intention to take precautionary actions can be confidently identified in this approach but at the same time, one must interpret the corresponding marginal effects with caution.
DRILLS EVACUATIONProbit model estimation Coef.
(Std. Err.)dy/dx
(Std. Err.)Coef.
(Std. Err.)dy/dx
(Std. Err.)SOCIO-DEMOGRAPHIC FACTORS
Female -0.028(0.093)
-0.009(0.032)
-0.271***(0.097)
-0.084***(0.030)
Age -0.006(0.004)
-0.002(0.001)
-0.015***(0.004)
-0.005***(0.001)
Income+ -0.085**(0.039)
-0.029**(0.013)
-0.151***(0.040)
-0.047***(0.012)
Education (illiterate) 0.107(0.156)
0.037(0.054)
0.457***(0.162)
0.142***(0.050)
Aboriginal 0.066(0.107)
0.023(0.037)
-0.498***(0.114)
-0.155***(0.034)
CONSTANT -1.338***(0.433)
-0.399(0.432)
Number of obs. 1135 1135Log likelihood -683.38 -623.55Pseudo R2 0.101 0.114
‘***’ denotes at 99% confidence level; ‘**’ for at 95% confidence level ‘+’ For the purpose of simplicity, these variables were treated as quasi continuous variables. However, these variables are categorical in nature. One must note that a general trend of their association with households’ intention to take precautionary actions can be confidently identified in this approach but at the same time, one must interpret the corresponding marginal effects with caution.
Conclusions1. Certain predictors have both direct and indirect effects on
the intentions to take certain precautionary measures,when indirect effects are mediated by risk perceptions.
2. The direct and indirect effects in some cases cancounteract each other.• For example, Aboriginal households, On one hand, have
lower perceived impact concerning property loss andhence are less likely to buy/renew property insurance.
• On the other hand, they are more likely to buy/renewproperty insurance, when the influence of perceivedimpact remains constant.
households’ attitudes and socio-demographic factorsought to be taken into account in the development ofcommunication strategies..
3. The extent to which one trusts the central government and local communities in their disaster response capacity explains his/her intention to take precautionary actions. • Higher trust in the central government is associated
with weaker intention to take preparedness and mitigation measures.
• Higher trust in local communities’ capacity is correlated to stronger intention to take preparedness, mitigation and recovery measures.
indicates their dependency on the government and hence results in weaker intention to take self-protect actions. recognise the necessity of precautionary actions at the community level.
The End &Comments Welcome