Increasing number and intensity of hurricanes cause significant impact on local communities every year. The ability to measure the resilience level of community is increasingly seen as a key step in disaster risk reduction. The identification of indicators that are meaningful for evaluating resilience remains a challenge. The main purpose of this work is to promote the understanding of the multidimensional nature of disaster resilience and provide a set of indicators for measuring the resilience at the city level in Fujian province of China. A set of indicators covering built environment, electricity and economic resilience was identified by cointegration test. Then, the well-chosen indicators are combined into a composite index to reflect the spatial-temporal resiliency level of coastal communities across geographical boundaries and the changing process over time. The proposed quantitative model can be used by planners and policy makers to achieve the performance goal of typhoon resilience of effected community and develop strategies for long-term sustainability of regions.
Research Article
Quantitative Approach for Evaluating the Community Resilience along Eastern Coastal City in Fujian Province
https://doi.org/10.21203/rs.3.rs-2563306/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Resilience
Composite Indicators
Hurricane Impact
Communities in coastal areas are facing increasing risk caused by hurricanes. Over the past several decades, an increasing number of tropical storms fall into hurricane category, and 1/3 of them reached category 4 and 5 (NOAA 2018. Hurricane prone areas have also experienced significant population and wealth growth (NOAA 2018), all of which contribute to increased vulnerability. Each year during hurricane season, hurricanes cause catastrophic failures in the distribution network and threaten coastal communities at exponentially higher rates. It brought huge direct and indirect damage for the affected area. In China, the population density of the eastern coastal area affected by the typhoon is twice that of the central area. Recent disasters in China such as Typoon Meranti and Typoon Rammasun provide examples of broad and lasting economic, environmental and social impacts. The impact of these disasters went beyond the geographical boundaries and scale, and had a negative impact on the government, enterprises, transportation, economic sectors and the people.
Hurricanes cause extensive damage to local economies, in some cases, recovery may take years. The hurricane disrupts normal business processes, energy distribution and consumption, and infrastructure services, and leads to the allocation and re-distribution of resources, which may be temporary or permanent (Chow and Elkind 2005; Comfort and Haase 2006; Kaiser et al. 2009).
Though large amount of funding is invested in reducing disaster risk, losses from hurricanes are still increasing year by year. Obviously, the hurricane impact on the affected area could not be fully mitigated through the current disaster reduction strategies. A more comprehensive hurricane risk mitigation method related to improve community resilience can be achieved by implementing strategies (in terms of risk reduction, risk transfer, and better preparedness and response effectiveness) is very necessary.
Resilience as the most recognized concept has recently been used to mitigate the risks of various natural disasters (Cimellaro et al., 2010; Burton et al., 2016; Masoomi and van de Lindt, 2018). In the past two decades, research has focused on developing and understanding the concept of resilience by introducing definitions and determining metrics, attributes, and dimensions (Haimes, 2009; Bhamra et al., 2011; Aven, 2011; Woods, 2015; Hosseini et al., 2016; Sharma et al., 2017). In particular, the National Academy of Sciences provided the following definition: “Community resilience is the ability to prepare and plan for, absorb, recover from, and more successfully adapt to actual or potential adverse events” (The National Academies 2012). This situation has risen great interest in understanding how to mitigate disaster impact or manage the risk. The ability of the communities could be enhanced to reduce the impact and speed up the recovery process caused by hurricanes with little or no external assistance. The goal of building resilient communities has gained significant attention from government, stakeholders, and researchers. Resilient community has better performance in withstanding and recover from the destructive disasters.
It is generally believed that resilient community is less vulnerable to disasters compared with less resilient communities. It's important to understand how to determine resilience and how to evaluate it (Cutter et al. 2008b). Many researches focus on defining and examining the factors related to disaster resilience, measuring resilience of affected area is increasingly seen as a critical step in disaster risk reduction. It is not possible to identify the effective strategies and policies to enhance the resilience of communities, monitor the changing process and compare the resilience pattern among various areas (National Academies 2012) without the assessment tools. Measuring resilience is difficult, because currently there are few clear indicators and procedures in the existing literature that suggest how to quantify resilience, how to determine whether communities become more resilient or whether the resilience level of the community is back to the pre-disaster level (Cutter, Burton, and Emrich 2010).
Community is regarded as a comprehensive system, which is composed of various interdependent systems (such as buildings, society, economy and people, etc.). The built environment of the community also constitutes the key system within the community and quantifying the response and performance of the hurricane to the community is a fundamental step towards quantifying and enhancing of community resilience. Therefore, a model that combines traditional economic measures with various engineering measures of the built environment can provide insight into a region's ability to withstand and recover from future hurricanes.
Federal Emergency Management Agency (FEMA 2018) suggested that local policy makers set up special agencies to promote resilient cities, and formulate emergency strategy according to the resilient pattern of cities, in order to ensure the full preparation before the disaster and the reasonable allocation of resources after the disaster, so as to minimize the losses caused by disasters (hurricane, earthquake and flooding, ect.) and accelerate the recovery of community.
Especially, our premise is that a community that maintains a stable relationship between these performance factors may become more resilient and sustainable over time. The purpose of this study is to establish a quantitative time series model to assess the resilience of the regional economic to hurricanes. The model could be used as a tool for evaluating the resilience of community, across geographical boundaries and over time. The results will support decision-making in disaster recovery and mitigation, and bring long-term systemic benefits to the entire community.
Previous research related to disaster resilience often proposed multi-domain including social, economic, institutional, infrastructural, ecological and community elements (Cutter et al. 2008; Gunderson 2009; NRC 2010). Indicators in each capital are identified to explain the impact and performance of affected area related to wind hazards (Ewing et al. 2005, 2006a, b, 2007, 2009; Ewing and Kruse 2002). More specifically, a large amount of variables include the atmospheric characteristics (i.e., wind speed, central pressure, total rainfall), physical infrastructure, socio-economic composition, insurance situation and government funding, etc. For example, Liu et al. (2006) developed an index including 37 economic and social indicators to evaluate the impact of disasters in the parish of Orleans, the metropolitan area of New Orleans, Louisiana, and Mississippi. Zhang et al. (2008) found that the operating conditions of enterprises after the hurricane are consist of multiple interrelated components, not only physical damage, but also employee injury, infrastructure damage and local economic condition are also evaluated. Due to the complex and transient economic environment in which enterprises operate, it may be quite challenging to properly separate the impact of hurricanes from the normal business cycle (Dahlhammer and Tierney, 1996; Webb et al., 2000 need more reference). In the framework proposed by Bruneau et al. (2003), Tierney and Bruneau (2007), the interaction and dynamic performance of engineering and economic indicators are investigated. The importance of the built environment in disasters can also be explained by examining human efforts, salary level and some social indicators (Bartuska 2007). The damaging degree of physical infrastructure plays a crucial role in capturing the interactions with activities in disaster cycle of communities (NIST 2008).
Stevenson et al. (2010) discussed the use of building permits to monitor spatial and temporal patterns of disaster recovery, with a special focus on three cities along the Gulf Coast of Mississippi. The changes in the stock of buildings, building innovation, infrastructure and built environment could be well described by the patter of building permits (Cui et al., 2015).
Although a great deal of knowledge is available on this subject, many indicators are qualitative rather than quantitative and often interact with each other to varying degrees. Therefore, the choice of which variables to evaluate resilience is still under active research.
Fekete et al. (2012), Frazier et al. (2014) and other scholars believe that all areas of the system are an important part of resilience research, and the comprehensive consideration of resilience indicators can be used to evaluate and analyze the resilience of different areas in the city. Although a large amount of high-quality information can be used to evaluate resilience, the number of identified indicators by researchers are huge. It is necessary to use effective methods to choose the resilience indicators and quantify them as resilience index. In recent years, the research focuses on constructing the resilience index by the proposed resilience indicators and combining it with GIS to evaluate and compare the resilience level of city.
The community resilience baseline is established by Cutter et al. (2010, 2014) through the data unit scale adjustment and data standardization, and the resilience index of the eastern coastal cities in the United States is obtained. Taking the census block as the spatial unit, Kontokosta et al. (2018) calculated the scores of infrastructures, community connectivity, physical infrastructure, economy and environment of New York City by 24 indicators, and the distribution map of resilience index under hurricane scenario is drawn, which can be used to identify the less resilient communities.
Serre et al. (2018) used the method of linear summation to get the resilience index in the fields of economy, society and technology. The indicators of resilience index include population age, highest education level, key infrastructure, etc. Only a small number of research focus on developing resilience index and drawing the distribution map on the city scale by quantifying the existing resilience index and combining with GIS, most of the distribution map is static.
However, most of these studies are on the level of qualitative and quantitative analysis, less empirical research methods based on mathematical statistics and mathematical models are used to study the temporal and spatial characteristics of resilience in various fields. The combination of qualitative and quantitative analysis is applied in many researches.
Moreover, there are a large number of existing resilience indicators, which brings some difficulties for the promotion and real-time update of indicators. Therefore, it is necessary to screen resilience indicators systematically and generate dynamic distribution maps to reflect the temporal and spatial variation characteristics of resilience in multi-field combination.
The impact of hurricane in Xiamen
Tropical Storm Cimaron landed in Zhangzhou, Fujian on July 13, 2013 and gradually moved towards Xiamen. Tropical Storm Cimaron brought heavy rainstorms to Xiamen that lasted more than ten hours, causing serious water accumulation and serious traffic jam in the urban area of Xiamen. Hurricane Meranti made landfall in Xiamen, Fujian, China on September 15, 2016, and brought significant damage to Xiamen. Around 650,000 trees were fell down, 17,907 houses sustained extensive damage, and the affected area covered 105,000 acres of crops, and the direct economic loss reached 10.2 billion RMB (The data comes from Xiamen Municipal People's Government: http://www.xm.gov.cn/). The strong wind left extensive power outages, recording the largest power failure in the history of Xiamen. The water supply was also cut off and the infrastructure system experienced severe damage.
The Impact Of Hurricane In Fuzhou
From 2010 to 2020, Fuzhou experienced a total of 4 hurricanes. They are Hurricane Soulik, Hurricane Nesat, Tropical Storm Haitang and Hurricane Maria. Hurricane Soulik brought strong winds of magnitude 9 to 12 to Fuzhou, heavy rainfall occurred over northeastern part of Fuzhou, which caused the closure of highways, interruption of power lines, and damage to crops. Hurricane Nesat and Hurricane Haitang landed in Fuzhou one day apart. They both caused a local heavy rainstorm in Fuzhou. Affected by this, many accidents occurred on the Fuzhou highway, and traffic was severely blocked. The continuous rainstorm also led to power outages in many areas of Fuzhou. Hurricane Maria as the most famous hurricane affecting Fuzhou area made landfall in Fuzhou on July 11, 2018. This is one of the most devastating hurricanes in Fuzhou in the past ten years. 726,800 people were affected due to the landfall of Hurricane Maria. According to the statistics, 254 houses collapse, 22.85 thousand hectares of crops were damaged, 811 communications and 735 power supply were interrupted. In addition, a lot of flights were cancelled, ships and high-speed railways were suspended. The total direct economic loss was estimated around 22.6 billion RMB (The data comes from Baidu Wikipedia: https://baike.baidu.com/). Table 1 below is a summary of the major hurricanes and tropical storms affecting the study area. And the geographical location of the two study areas is shown in Fig. 1.
|
Storm Name |
Landfall Date |
Landfall Location |
Intensity |
Economic Loss (billion) |
Fatalities |
|---|---|---|---|---|---|
|
Tropical Storm Cimaron |
July 18, 2013 |
Zhangzhou |
Category 1 |
$15.52 |
1 |
|
Typhoon Meranti |
September15,2016 |
Xiamen |
Category 5 |
$10.2 |
28 |
|
Typhoon Soulik |
July 13, 2013 |
Fuzhou |
Category 4 |
$2.49 |
0 |
|
Typhoon Nesat |
July 30, 2017 |
Fuzhou |
Category 2 |
$7.6 |
0 |
|
Tropical Storm Haitang |
July 31, 2017 |
Fuzhou |
Category 2 |
0 |
|
|
Typhoon Maria |
July 11, 2018 |
Fuzhou |
Category 5 |
$22.6 |
0 |
Data Description For Xiamen And Fuzhou City
The annual electricity consumption data in Xiamen can be obtained from the prospective database (https://d.qianzhan.com/) ranging from 2002 to 2019. Xiamen Municipal Bureau of Statistics has provided the cumulative value of monthly real estate development investment and total retail sales of consumer goods (http://tjj.xm.gov.cn/). Data starts from January 2009 to September 2020. The annual electricity consumption data in Fuzhou, the cumulative value of monthly real estate development investment and total retail sales of consumer goods are all from the Fuzhou Bureau of Statistics (http://tjj.fuzhou.gov.cn/) and the sample period is from January 2009 through September 2020. Quarterly electricity consumption data are converted into monthly data by low to high frequency conversion method, and then Catmull-Rom Spline methodology is applied to fill in missing January data from 2010 to 2019 on real estate development investment and total retail sales of consumer goods. Before proceeding with data analysis, as usual, we begin by seasonally adjusting the data series using the Census X-12 methodology. Subsequently, the consumer price index is used to transfer real estate development investment and total retail sales of consumer goods into actual price. The natural log is taken of each variable and used in the analysis that follow.
Table 2 and Table 3 provide descriptive statistics of the three variables used in Xiamen and Fuzhou in this study respectively. E denotes the natural log of electricity consumption; R denotes the natural log of real estate development investment and T denotes the natural log of total retail sales of consumer goods. Δ denotes first-difference operator, and the adjusted sample period is January 2010–September 2020. Figures 2 and Figs. 3 present the three variables in levels in Xiamen and Fuzhou, respectively.
Table 2 Descriptive Statistics in Xiamen
For the three variables in Xiamen (electricity consumption, real estate development investment and total retail sales of consumer) (in levels), standard deviation ranges from 0.224 to 0.490, greatest variation is real estate development investment and the most stable is electricity consumption. Based on the analysis of first difference, the three variables all have seen positive growth over the sample period.
Table 3 Descriptive Statistics in Fuzhou
Among the three variables in Fuzhou, the standard deviation of three variables present poor stability comparing with the variables in Xiamen, varying from 0.438 to 0.758. The most stable is total retail sales of consumer goods, electricity consumption has the greatest variation. The three variables in Fuzhou show upward tendency over the sample period.
There are several points worth noting. As can be seen from the figures, the electricity consumption in Fuzhou began to rose sharply due to the rapid economic development of Fuzhou in 2016. Among them, a significant decline is involved in the Fuzhou's Real Estate Development Investment after the landfall of hurricane Haitang. Fuzhou's Real Estate Development Investmen return to its pre-disaster level after one year. It can be observed that an obvious drop is shown in Xiamen's Real Estate Development Investment following the landfall of Typhoon Meranti and the Corona Virus Disease. Real Estate Development Investment in Xiamen returned to pre-disaster level after nine months of Typhoon Meranti.
Though obvious changings are not observed during most of hurricane period, it doesn’t imply that there is no hurricane-related impact on the region. The three variables may capture broader measures of the dynamic system. A closer examination is needed to reveal the fundamental relationship among the three indicators. Thus, a formal investigation is conduct in this study to the time series properties of these variables, and an effective tool is developed to evaluate the resilience level of community.
Due to the significant differences in architecture, infrastructure, economy situation among coastal cities, the degree of damage and duration of impact following the landfall of hurricanes are different. At present, the evaluation of urban resilience is mainly based on the selection of resilience evaluation indicators. A large number of indicators (age, real estate value, income etc.) have been identified by the academic community (Stevenson et al., 2010; Olsen et al., 2013; etc.); Tatum et al., 2012), but due to the differences in economic, social and public management systems among countries, the application of existing models and indicators is very limited.
Hurricane is usually considered as external shocking or interruption that pushes a system from operating in equilibrium to a situation of disequilibrium. The existence of common factors shows that these measures are cointegrated or linked for a long time. If the system does not return to a long-term stable equilibrium, then the system is not cointegrated. The government needs to take measures to help the system reach the normal operation of the balanced state. However, in the case of cointegration, the system does return to equilibrium through self- adjustment process and shows the ability of resilience.
Stationary tests as the preliminary step for cointegration test are conduct in this study to examine the stationary situation of each variables. The null hypothesis is generally defined as the presence of a unit root and the alternative hypothesis is either stationarity, trend stationarity or explosive root depending on the test used. A time series is said to be stationary if its statistical properties do not change with time. Augment Dickey Fuller (ADF) test and Phillips and Perron (PP) test are common methods used to examine whether the time series is stationary. In our study, the level of variables is test firstly. If the variable is not stationary in level, then the first difference of variable will be tested to determine the stationary status. A non-stationary series could be stationary in first difference, which is called integrated of order 1.
ADF test includes fitting regression model with ordinary least squares (OLS), but sequence correlation will bring a problem. To illustrate this point, the regression of extended Dickey-Fuller test includes the lag of the first difference of X. Based on the following equation (Dickey and fuller 1981), the ADF test is performed based on the following equation (Dickey and Fuller 1981):
ΔXt = κ + δt +(ρ − 1)Xt−1 +∑N ψiΔXt−i + υt (1) i=1 (1)
where Xt is the individual time series under investigation, in this study, they are electricity consumption, real estate development investment and total retail sales of consumer goods. Δ is the first- difference of the variables, t is a linear time trend, υt is a covariance stationary random error, and the number of lags on the augmenting term N is determined by Akaike's information criterion to ensure sequence uncorrelated residuals. If (ρ − 1) < 1, the testing result is statistically significant at certain level (5% or 1%), then the null hypothesis that X is a non-stationary sequence is rejected.
An alternative unit root test is developed by Phillips and Perron (1988), allowing for weak dependence and heterogeneity in error terms. The proposed method is robust to broad range sequence correlation and time-dependent heteroscedasticity. PP tests for unit root is generally expressed as:
X t = ξ0 + ξ1(t − T2) + λXt−1 + γt (2)
where (t-T/2) is the time trend, T is the sample size, and γt is the error term.
In contrast to the alternative hypothesis that Xt is stationary around the determined trend (Ha: λ < 1), null hypothesis of the unit root Ho: λ = 1 is tested. MacKinnon (1991) critical values is applied to determine the statistical significance of ADF test and PP test.
Table 4 Unit Root Tests of Three Variables in Xiamen
Notes: E denotes the log of electricity consumption, R denotes the log of real estate development investment and S denotes the log of total retail sales of consumer goods. Δ denotes first-difference operator. A*denotes statistically significant at the one percent level or less based on MacKinnon (1991) critical values.
Table 5 Unit Root Tests of Three Variables in Fuzhou
The results of the ADF and PP unit root tests in Xiamen and Fuzhou are presented in Table 4 and Table 5 respectively and suggest that each variable is nonstationary in levels. The impact on one of the series alone is permanent because the series will not return to its long-term average over time. However, the first difference of each time series is stationary. As mentioned above, it is possible that a linear combination of nonstationary I (1) variables in levels may be stationary. In this study, these variables are called cointegration variables, although each variable may not return to a specific value alone for a long-term trend, there is a common factor that leads the group (consisting of two or more variables) to have a common tendency to converge in the long term. Given the results of the ADF and PP tests, a group test called cointegration test is proceeded to examine the cointegrating properties of the variables.
Cointegration Test
The cointegration test analyzes a non-stationary time series process, the variance and mean of the process change with time. In other words, cointegration is a test method used to examine the possible long-term co-movement among time series variables. The most popular cointegration tests are Engle-Granger and Johansen Test. If the linear combination of two or more variables has a lower order of integration, they will be cointegrated. For example, if a linear combination of I (0) can be used to model a set of I (1) variables, there is cointegration. The order of integration I (1) reveal that a set of differences can transform a non-stationary variable into stationarity. If two or more variables are cointegrated together, a common factor is existed behind the variables, and the system could return to equilibrium without external adjustment, which could prove the ability of resilience. If the system could not return to long-run equilibrium, government action or assistance are needed to help the affected area return to normal operation state.
The maximum likelihood method of Johansen (1988) and Johansen and Juselius (1990) is applied in this study to determine the number of cointegrating vectors. The testing process is based on the P-order vector autoregression (VAR) model of order p:
$$\varDelta {X}_{i}=\mu +{{\Gamma }}_{1}{\Delta }{X}_{t-1}+\dots +{{\Gamma }}_{k}{\Delta }{X}_{t-k-1}-{\Pi }{X}_{t-k}+{\epsilon }_{t}$$
3
Where
$${{\Gamma }}_{i}=-I+{{\Pi }}_{1}+\dots +{{\Pi }}_{i}$$
4
And
$${\Pi }=I-{{\Pi }}_{1}-\dots -{{\Pi }}_{k}$$
5
Where Xt is the vector of the non-stationary I(1) variable under study, and εt is the independent Gaussian variable(zero mean and Ω variance) in k dimension. The vector Γi contains short-term parameters that capture the imbalance characteristics of the data. The Π (k × k) matrix includes the information related to long-term relationships among the three variables under the examination.
The cointegration test determines the rank of the matrix Π. In order to determine the rank, two tests were carried out, namely the trace test and the maximum eigenvalue test. For both of the test statistics, the initial Johansen test is a test of the null hypothesis that there is no cointegration. The forms of alternative hypotheses are various.
The maximum eigenvalue test starts with the test of rank (Π) = 0 and the alternative hypothesis is that rank (Π) = 1. If the rank of the matrix is zero, the maximum eigenvalue is zero, there is no co-integration, and the test ends. If the maximum eigenvalue λ1 is not zero, the rank of the matrix is at least one, in other words, more cointegration vectors may exist. Then test whether the second largest eigenvalue λ2 is zero. If the eigenvalue is zero, the tests are completed, and there is only one cointegration vector. If the second largest eigenvalue λ2 = 0, there are more than two variables, and more cointegration vectors may exist. The testing process continues until the null hypothesis that the eigenvalue is equal to zero cannot be rejected. In summary, the maximum eigenvalue test first tests whether the largest eigenvalue is zero relative to the alternative that the next largest eigenvalue is zero.
The test statistic is as follows:
$$LR\left({r}_{0}, {r}_{0}+1\right)=-Tln(1-{\lambda }_{r0+1})$$
6
Where LR (r0, r0 + 1) is the likelihood ratio test statistic, used to test the null hypothesis of rank (Π) = r0 and the alternative hypothesis of rank (Π) = r0 + 1. For example, the likelihood ratio test statistic LR (0,1) = -T ln(1-λ1) is used to test the hypothesis of rank (Π) = 0 relative to rank (Π) = 1.
Trace Test
The trace test is to test whether the rank of the matrix Π equals to r0. The null hypothesis is that rank(Π) = r0. The alternative hypothesis is that r0 < rank(Π) ≤ n, where n is the maximum number of possible cointegrating vectors(reference). If the null hypothesis is rejected, the next null hypothesis is rank (Π) = r0 + 1, and the alternative hypothesis is r0 + 1 < rank (Π) ≤ n. For the subsequent tests, it repeats the testing process above, the testing value plus one each time, until it reaches the number n.
The likelihood ratio test statistic is
$$LR\left({r}_{0},n\right)=-T\sum _{i={r}_{0}+1}^{n}\text{l}\text{n}(1-{\lambda }_{i})$$
7
Where \(LR({r}_{0},n)\) is the likelihood ratio statistic. For example, the likelihood ratio test statistic \(LR\left(0,n\right)=-T \sum _{i=1}^{n}\text{l}\text{n}(1-{\lambda }_{i})\) is applied to test the null hypothesis versus the alternative that rank(Π) ≤ n.
Johansen and Juselius (1990) and Osterwald-Lenum (1992) provide critical values for these two tests. A critical value is used to determine whether to reject the null hypothesis. If the absolute value of the test statistic is greater than the critical value, you can reject the null hypothesis. Table 6 - Table 9 report results for the cointegration tests. The alternative hypothesis of having a long run cointegration relationship among three time series is accepted. The results indicate that there is exactly one cointegrating vector among the three analyzed variables in city Fuzhou and Xiamen.
|
Hypothesized |
Eigenvalue |
Trace |
Critical Value at 5% |
Prob.** |
|---|---|---|---|---|
|
None* |
0.165008 |
35.66217 |
29.79707 |
0.0094** |
|
At most 1 |
0.084002 |
12.94023 |
15.49471 |
0.1170 |
|
At most 2 |
0.014848 |
1.884886 |
3.841466 |
0.1698 |
Trace test indicates 1 cointegrating at the 0.05 level
|
Hypothesized |
Eigenvalue |
Max-Eigen Statistic |
Critical Value at 5% |
Prob.** |
|---|---|---|---|---|
|
None* |
0.165008 |
22.72194 |
21.13162 |
0.0296** |
|
At most 1 |
0.084002 |
11.05534 |
14.26460 |
0.1515 |
|
At most 2 |
0.014848 |
1.884886 |
3.841466 |
0.1698 |
Max-Eigenvalue test indicates 1 cointegrating at the 0.05 level
|
Hypothesized |
Eigenvalue |
Trace |
Critical Value at 5% |
Prob.** |
|---|---|---|---|---|
|
None* |
0.165077 |
34.72380 |
29.79707 |
0.0125** |
|
At most 1 |
0.070924 |
11.99147 |
15.49471 |
0.1573 |
|
At most 2 |
0.021374 |
2.722376 |
3.841465 |
0.0989 |
Trace test indicates 1 cointegrating at the 0.05 level
|
Hypothesized |
Eigenvalue |
Max-Eigen Statistic |
Critical Value at 5% |
Prob.** |
|---|---|---|---|---|
|
None* |
0.165077 |
22.73232 |
21.13162 |
0.0295** |
|
At most 1 |
0.070924 |
9.269095 |
14.26460 |
0.2644 |
|
At most2 |
0.021374 |
2.722376 |
3.841466 |
0.0989 |
Max-Eigenvalue test indicates 1 cointegrating at the 0.05 level
Construct Index
In this section, the composite indicator construction method recorded in the Conference Board (2001) will be used to construct the index to evaluate the resilience of affected city based on the selected variables in the last section. Its components include real estate development investment, electricity consumption and retail sales of consumer goods at the city level. The composite index takes into account the built environment, the ability of customer purchasing and energy distribution, thus representing the disaster resilience capacity of the given community. The process of calculating composite index involves transforming the indicators to similar ranges, aggregating the time series indicators in a composite form and recalculating the composite index in the base year. The aggregation procedure can be described in detail in the following 5 steps:
In the first step, monthly changes are computed for each component over the entire sample period. The subscript of t and t-1 represent the current month and last month, respectively, and m represent specific components of the index. If the component is not in the form of percentage change, the symmetry alternative could be used to substitute the traditional percentage change formula:
$${r}_{i,t}= \frac{200\times \left({X}_{i,t}-{X}_{i,t-1}\right)}{\left({X}_{i,t}+{X}_{i, t-1}\right)}$$
8
where, \({r}_{i,t}\) is month-to-month changes, \({X}_{i,t}\) is time series data for component \(i\)at month \(t\), in our case, \({X}_{i,t}\) is electricity consumption, real estate development investment and total retail sales of consumer goods.
In the second step, the monthly changes are adjusted to balance the volatility of each component. Standard deviations of the changes in each component are computed, inverted, and their sum are also obtained. The adjusted monthly change in each component is multiplied by the corresponding standardization factors.
$${w}_{i,year}= \frac{{\frac{1}{std}}_{i,year}}{{\sum }_{1}^{n}\left({\frac{1}{std}}_{i,year}\right)}$$
9
Where \(n\) is the number of components, and standardization factors are updated annually and consistent with the annual benchmark revision of the Conference Board.
The third step is comprised of the adjusted monthly contributions of each component are added up to obtain the growth rate of the index. The sum of adjusted contributions can be obtained in this step, that is, the monthly growth rate of the index.
$${c}_{i,t}={w}_{i,year}\times {r}_{i,t}$$
10
where \({c}_{i,t}\) is monthly contribution for each component (The Conference Board 2001).
In the fourth step, the sum of the growth rate of the composite index is adjusted so that its trend is similar to coincident index. This is achieved by adding an adjustment factor to the monthly growth rate of the index. For example, the trend adjustment factor of the leading index is calculated by subtracting its average monthly growth rate from the average monthly growth rate of the coincident index. The symmetric percentage change formula is used to calculate the index level in the fifth step. The index is calculated from the initial value of 100 in the first month of the sample period. In the final step, the index will be re-adjusted to the average of 100 in 2010, the historical index is multiplied by 100, and then divided by the 12-month average in 2010. \({s}_{t}\) is the sum of the monthly contributions across the component, and \(n\) is the number of components, the number is three in our study.
$${s}_{t}=\sum _{i=1}^{n}{c}_{i,t}$$
11
These annually updated standardized factors should be consistent with the annual benchmark of the Conference Board. The third step is to assign the standardized coefficient wi and month-to-month changes ri;t to each component. The monthly contribution of each component is calculated as follows: where ci;t is monthly contributions of each component (The Conference Board 2001).
In the next step, the initial index is calculated recursively based on the symmetrical percentage change formula, starting from 100 in the first month of the sample.
$${I}_{j}={I}_{j-1}\times \frac{(200+{s}_{j})}{(200-{s}_{j})} (j=\text{2,3},\dots ,m)$$
12
where \(m\) is the total number of month in the sample period. The results of Typhoon Resilience Index in Xiamen and Fuzhou are shown in Figures below.
Typhoon Resilience Index based on non-parametric approach(introduced above)in Xiamen and Fuzhou are presented in Fig. 4 and Fig. 5. The dot indicates the months that Hurricanes made landfall and Corona Virus Disease. Decomposition of macroeconomic time series into trend and cycle components is essential for many macroeconomic concepts. Several tools are developed to extract cycles from raw data. Hodrick precott (HP) filter has become the main tool for decision and policy makers to estimate business cycle (Kaiser 1999). This method was used by Hodrick and Prescott to analyze postwar U.S. business cycles in 1980. It is a common smoothing method in the field of macro-economy, which can get the smooth estimation of the long-term trend component of the series by removing the short-term fluctuations (Hodrick and Prescott 1980).
The HP filter decomposes a time series into two components: a long-term trend component and stationary cycle (Hodrick and Prescott 1980; Kydland and Prescott 1990, and Prescott 1986). It is a two-sided linear filter that needs to specify a parameter known as λ, and calculate the smoothed series of Y by minimizing the variance of Y around S. This parameter λ tunes the smoothness of the trend and depends on the periodicity of the data and the main period of the cycle that is of interest to the analyst. The econometric program Eviews uses a default value of 14400 to analyze monthly data. That is, the HP filter chooses s to minimize:
$$(\sum _{t=1}^{T}{({y}_{t}-{s}_{t})}^{2}+\lambda \sum _{t=2}^{T-1}\left(\right({s}_{t+1}-{s}_{t})-{({s}_{t}-{s}_{t-1}))}^{2})$$
13
The parameter λ controls the smoothness of the resulting trend. The bigger the λ it is, the smoother the resulting trend. As \(\lambda ={\infty }\), then \({s}_{t}\)is calculated by fitting \({y}_{t}\)to the linear trend model by Ordinary Least Square (OLS). Figure 6 and Fig. 7 present the Typhoon Resilience Index after applying Hodrick-Prescott filter in Xiamen and Fuzhou, respectively. The black line shows the Typhoon Resilience Index from February 2009 to September 2020. The light grey line shows the trend of Typhoon Resiliency Index, the exact number could be obtained in the right vertical axis.
In Fig. 6, the index values are 133 and 159 for Typhoon Cimaron and Typhoon Meranti, respectively. Policymakers and businesses owners monitor economic trends to understand how the economy performs. They need relatively accurate and timely information on the state of the economy for strategic planning (Sarker, Peng, etc. 2020).The dark grey line indicates the difference(left vertical axis) between the Typhoon Resiliency Index and the trend. It is clearly presented that when Hurricane Cimaron (9.7) made landfall, the difference between Hurricane Resiliency Index and trend is lower than that of Hurricane Meranti (13.1). In addition, it is clear that when Hurricane Maria landed, the difference between the index and its respective trend (66) is much higher than that when Hurricane Soulik (22) landed. This information indicates that the resilience level in Fuzhou to Maria is better than that to Soulik under the same conditions.
This paper focus on identifying the factors that could be applied to evaluate the resilience level of affected communities, and the identified indicators are constructed into index to present the resilience level of affected area across geographical boundaries and over time. The potential application of the index is mainly focused on two aspects. First, it provides a new method for assessing disaster resilience, and it could not only be used to compare the various resilience levels among affected areas, but also to evaluate the resilience level of the same county in different time periods. Continuously updating index to address temporal and spatial dynamism and uncertainty of the system.
In addition, the model could also be used to implement both short term and long-term strategy for speeding up the recovery process and support sustainability development of community. Great regional differences exist in the light of how quickly the local economy responds to hurricane impact and government actions. Therefore, under the consideration of regional difference, a customized approach of federal assistance in post disaster recovery may be more ideal and feasible.
For the future work, more hurricane-prone regions will be added to the model to examine the cointegration relationship. The results could help us explore the fundamental way to improve the resilience level of community. And the assessment approach will also be combined with GIS to provide a dynamic tool for effective communication of risks and implement priority mitigation strategies.
Acknowledgement
This material is based upon work supported by National Natural Science Foundation of China (Grant No. 52008110), Natural Science Foundation of Fujian Province (Grant No. 2020J05195). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the sponsors.
Ethical statements
Conflict of interest:
The authors have no conflicts of interest to declare.
Consent to participate:
All authors designed the study, collected data, wrote the manuscript and revised it.
Consent to publish:
All authors agree to publish this manuscript. There is no conflict of interest.
- Aven, T. (2011). On some recent definitions and analysis frameworks for risk, vulnerability, and resilience. Risk Analysis: An International Journal, 31(4), 515-522.
- Bartuska, T. J. (2007). Understanding environment (s): built and natural. The built environment: A collaborative inquiry into design and planning, 33-43.
- Bhamra, R., Dani, S., and Burnard, K. (2011). Resilience: the concept, a literature review and future directions. International journal of production research, 49(18), 5375-5393.
- Burton, H. V., Deierlein, G., Lallemant, D., & Lin, T. (2016). Framework for incorporating probabilistic building performance in the assessment of community seismic resilience. Journal of Structural Engineering, 142(8), C4015007.
- Chow, E., and Elkind, J. (2005). Hurricane Katrina and US energy security. Survival, 47(4), 145-160.
- Cimellaro, G. P., Reinhorn, A. M., & Bruneau, M. (2010). Framework for analytical quantification of disaster resilience. Engineering structures, 32(11), 3639-3649.
- Comfort, L. K., & Haase, T. W. (2006). Communication, coherence, and collective action: The impact of Hurricane Katrina on communications infrastructure. Public Works management & policy, 10(4), 328-343.
- Cui, Y., Liang, D., & Ewing, B. T. (2015). Empirical analysis of building permits in response to hurricane landfalls. Natural Hazards Review, 16(4), 04015009.
- Cutter, S. L., Barnes, L., Berry, M., Burton, C., Evans, E., Tate, E., & Webb, J. (2008). A place-based model for understanding community resilience to natural disasters. Global environmental change, 18(4), 598-606.
- Cutter, S. L., & Finch, C. (2008). Temporal and spatial changes in social vulnerability to natural hazards. Proceedings of the national academy of sciences, 105(7), 2301-2306.
- Cutter, S. L., Burton, C. G., & Emrich, C. T. (2010). Disaster resilience indicators for benchmarking baseline conditions. Journal of homeland security and emergency management, 7(1).
- Cutter, S. L., Ash, K. D., & Emrich, C. T. (2014). The geographies of community disaster resilience. Global environmental change, 29, 65-77.
- Dahlhamer, J. M., & Tierney, K. J. (1998). Rebounding from disruptive events: business recovery following the Northridge earthquake. Sociological spectrum, 18(2), 121-141.
- Ewing, B. T., & Kruse, J. B. (2002). The impact of project impact on the Wilmington, North Carolina, labor market. Public Finance Review, 30(4), 296-309.
- Ewing, B. T., Kruse, J. B., & Thompson, M. A. (2005). Empirical examination of the Corpus Christi unemployment rate and Hurricane Bret. Natural Hazards Review, 6(4), 191-196.
- Ewing, B. T., Kruse, J. B., & Schroeder, J. L. (2006). Time series analysis of wind speed with time‐varying turbulence. Environmetrics, 17(2), 119-127.
- Ewing, B. T., Hein, S. E., & Kruse, J. B. (2006). Insurer stock price responses to Hurricane Floyd: An event study analysis using storm characteristics. Weather and forecasting, 21(3), 395-407.
- Ewing, B. T., Kruse, J. B., & Sutter, D. (2007). Hurricanes and economic research: An introduction to the hurricane Katrina symposium. Southern Economic Journal, 74(2), 315-325.
- Ewing, B. T., Kruse, J. B., & Sutter, D. (2009). An overview of Hurricane Katrina and economic loss. Journal of Business Valuation and Economic Loss Analysis, 4(2).
- FEMA, 2018c, Building the Performance You Need, A Guide to State-of-the-Art Tools for Seismic Design and Assessment, FEMA P-58-7, prepared by the Applied Technology Council for the Federal Emergency Management Agency, Washington, D.C.
- Fekete A (2012) Spatial disaster vulnerability and risk assessments: challenges in their quality and acceptance. Nat Hazards 61:1161–1178
- Frazier, T. G., Thompson, C. M., & Dezzani, R. J. (2014). A framework for the development of the SERV model: A Spatially Explicit Resilience-Vulnerability model. Applied Geography, 51, 158-172.
- Gunderson, L. (2009). Some comparisons between ecological and human community resilience. INSCT, 14.
- Haimes, Y. Y. (2009). On the definition of resilience in systems. Risk Analysis: An International Journal, 29(4), 498-501.
- Hodrick, Robert J., and Edward C. Prescott, (1980) “Postwar U.S. Business Cycles: An Empirical Investigation,” Carnegie Mellon University discussion paper no. 451.
- Hosseini, S., Barker, K., & Ramirez-Marquez, J. E. (2016). A review of definitions and measures of system resilience. Reliability Engineering & System Safety, 145, 47-61.
- Johansen, S. (1988). Statistical analysis of cointegration vectors. Journal of economic dynamics and control, 12(2-3), 231-254.
- Johansen, S., & Juselius, K. (1990). Maximum likelihood estimation and inference on cointegration—with appucations to the demand for money. Oxford Bulletin of Economics and statistics, 52(2), 169-210.
- Kaiser, M. J., Yu, Y., & Jablonowski, C. J. (2009). Modeling lost production from destroyed platforms in the 2004–2005 Gulf of Mexico hurricane seasons. Energy, 34(9), 1156-1171.
- Kontokosta, C. E., & Malik, A. (2018). The Resilience to Emergencies and Disasters Index: Applying big data to benchmark and validate neighborhood resilience capacity. Sustainable cities and society, 36, 272-285.
- Kydland, F. E., & Prescott, E. C. (1990). Business cycles: Real facts and a monetary myth. Federal Reserve Bank of Minneapolis Quarterly Review, 14(2), 3-18.
- Liu, A., Fellowes, M., & Mabanta, M. (2006). Special edition of the Katrina index: A one year review of key indicators of recovery in post-storm New Orleans. Washington, DC: Brookings Institution.
- Masoomi, H., & van de Lindt, J. W. (2018). Restoration and functionality assessment of a community subjected to tornado hazard. Structure and Infrastructure Engineering, 14(3), 275-291.
- National Institute of Standards and Technology (2008) Visiting committee on advanced technology. http:// www.nist.gov/director/vcat/upload/report07.pdf
- National Research Council (NRC) (2010) Private-public sector collaboration to enhance community disaster resilience: a workshop report. National Academy Press, Washington
- NOAA, 2018, Costliest U.S. Tropical Cyclones Tables Updated, National Oceanic and Atmospheric Administration (NOAA), National Centers for Environmental Information (NCEI), National Hurricane Center (NHC), https://www.nhc.noaa.gov/news/Updated Costliest.pdf, accessed April 5, 2020.
- Olsen AH, Porter KA (2013) Storm surge to demand surge: exploratory study of hurricanes, labor wages and material prices. Nat Hazard Rev 14:247–257
- Osterwald-Lenum, M. (1992). A note with quantiles of the asymptotic distribution of the maximum likelihood cointegration rank test statistics. Oxford Bulletin of Economics and statistics, 54(3), 461-472.
- Prescott, E. C. (1986). Theory ahead of business-cycle measurement. In Carnegie-Rochester conference series on public policy (Vol. 25, pp. 11-44). North-Holland.
- Sarker, M. N. I., Peng, Y., Yiran, C., & Shouse, R. C. (2020). Disaster resilience through big data: Way to environmental sustainability. International Journal of Disaster Risk Reduction, 101769.
- Serre, D., & Heinzlef, C. (2018). Assessing and mapping urban resilience to floods with respect to cascading effects through critical infrastructure networks. International Journal of Disaster Risk Reduction, 30, 235-243.
- Sharma, V., Marin, D. B., Koenig, H. K., Feder, A., Iacoviello, B. M., Southwick, S. M., & Pietrzak, R. H. (2017). Religion, spirituality, and mental health of US military veterans: Results from the National Health and Resilience in Veterans Study. Journal of affective disorders, 217, 197-204.
- Stevenson, J. R., Emrich, C. T., Mitchell, J. T., & Cutter, S. L. (2010). Using building permits to monitor disaster recovery: A spatio-temporal case study of coastal Mississippi following Hurricane Katrina. Cartography and Geographic Information Science, 37(1), 57-68.
- Tatum, M. C., and F. Terrell. 2012. “Hurricane reconstruction in the United States Gulf Coast.” Int. J. Disaster Resilience Built Environ. 3 (3): 199–219. https://doi.org/10.1108/17595901211263602.
- The National Academies, (2012). Disaster Resilience: A National Imperative. The National Academies Press, Washington, DC.
- Webb, G. R., Tierney, K. J., & Dahlhamer, J. M. (2000). Businesses and disasters: Empirical patterns and unanswered questions. Natural Hazards Review, 1(2), 83-90.
- Woods, D. D. (2015). Four concepts for resilience and the implications for the future of resilience engineering. Reliability Engineering & System Safety, 141, 5-9.
- Zhang, Z., Wu, Z., Martinez, M., & Gaspard, K. (2008). Pavement structures damage caused by Hurricane Katrina flooding. Journal of geotechnical and geo-environmental engineering, 134(5), 633-643.