Urban climate vulnerability in Cambodia: A case study in Koh Kong province
Abstract
EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.
Full text
Sa, Kimleng Article Urban climate vulnerability in Cambodia: A case study in Koh Kong province Economies Provided in Cooperation with: MDPI – Multidisciplinary Digital Publishing Institute, Basel Suggested Citation: Sa, Kimleng (2017) : Urban climate vulnerability in Cambodia: A case study in Koh Kong province, Economies, ISSN 2227-7099, MDPI, Basel, Vol. 5, Iss. 4, pp. 1-19, https://doi.org/10.3390/economies5040041 This Version is available at: https://hdl.handle.net/10419/197043 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/
economies Article Urban Climate Vulnerability in Cambodia: A Case Study in Koh Kong Province Kimleng Sa ID Faculty of Development Studies, Royal University of Phnom Penh, Phnom Penh 12105, Cambodia; [email protected]; Tel.: +855-92-276-076 Academic Editor: Peter A. G. van Bergeijk Received: 4 March 2017; Accepted: 19 October 2017; Published: 7 November 2017 Abstract: This study investigates an urban climate vulnerability in Cambodia by constructing an index to compare three different communes, Smach Meanchey, Daun Tong, and Steong Veng, located in the Khemarak Phoumin district, Koh Kong province. It is found that Daun Tong commune is the most vulnerable location among the three communes, followed by Steong Veng. Besides, vulnerability as Expected Poverty (VEP) is used to measure the vulnerability to poverty, that is, the probability of a household income to fall below the poverty line, as it captures the impact of shocks can be conducted in the cross-sectional study. It applies two poverty thresholds: the national poverty line after taking into account the inflation rate and the international poverty line defined by the World Bank, to look into its sensitivity. By using the national poverty line, the study reveals that more than one-fourth of households are vulnerable to poverty, while the international poverty threshold shows that approximately one-third of households are in peril. With low levels of income inequality, households are not highly sensitive to poverty; however, both poverty thresholds point out that the current urban poor households are more vulnerable than non-poor families. Keywords: vulnerability; urban; index; vulnerability as expected poverty; shocks JEL Classification: O18; Q54 1. Introduction Climate change is becoming a crucial problem for city dwellers, especially for urban poor in low-income countries. Informal settlements become a common practice for urban poor because they cannot afford good-quality housing (McGranahan et al. 2007). They are strongly threatened by environmental changes since they are more sensitive to the dynamics of natural resources and lack the means to improve their adaptive capacity to natural hazards (Satterthwaite et al. 2007). Human settlements located along coastal and river floodplains are the most vulnerable areas of climate change, as their economic activities highly depend on climate-sensitive resources, especially during the dry season when water sanitation cannot supply the whole population. Urban adaptation to climate change is often linked to the role of governance; however, official municipal policies frequently increase the vulnerabilities—including social, economic, and environmental vulnerabilities—of poor residents rather than reduce them (Satterthwaite et al. 2007). Recent urbanization has induced climate-related problems that disrupt the economic activities and livelihoods of Cambodians in urban areas. A rapid population growth in the city has increased the demand for water and electricity; on the other hand, a low governance capacity to supply these needs pushes the prices up and increases the cost of living, which mainly affects poor people in the city. Households living in the slum area often face health problems. UNICEF (2010) reported a persistent inequality in access to healthcare services and health threats due to communicable diseases in a poor community in Phnom Penh. Economies 2017,5, 41; doi:10.3390/economies5040041 www.mdpi.com/journal/economies
Economies 2017,5, 41 2 of 19 In Cambodia, the loss and damage caused by climate change are increasing and interrupting sustainable development. According to Ancha Srinivasan, 1 economic loss due to climate change rose to approximately 10% of the GDP in 2015. Hazards, mainly floods, resulted in a GDP loss about 4.3% in 2011, as reported in the International Disaster Database (MoE 2013). From 1984 until 2011, Cambodia encountered flooding 20 times (Chhinh 2014, p. 168). The Royal Government of Cambodia identifies the coastal zone as the most vulnerable location of climate change. Recent urbanization due to economic growth has posed many challenges not only for people living in Phnom Penh, but also for residents along the coastal zones where economic activity has been expanding substantially. These problems include waste and sanitation, water shortage, water quality (Irvine et al. 2006), and energy security (Heng 2012). Moreover, urban settlement along the coastal zone has been exposed to climate hazards such as storm surges, cyclones, seawater intrusion, and other water stresses (Hay and Mimura 2006;McGranahan et al. 2007). The coastal zone of Cambodia is comprised of four provinces: Koh Kong, Sihanoukville, Kampot, and Kep. Koh Kong is a southwestern province in Cambodia that plays an important role as an economic corridor. First, it serves as an economic gateway that links Thailand and Cambodia, as well as access to the main port of Sihanoukville. Second, this Special Economic Zone (SEZ) absorbs a large amount of labor to work in garment factories and manufacturing plants in marine production. Third, it possesses many eco-tourism sites that are the important agents to improve the livelihood of the local residents. Since local people have depended strongly on the natural environment, a slight change will generate a significant impact on their income and livelihood. A large part of Koh Kong has been threatened by climate hazards. As it is situated in the coastal area, it has been endangered by climate shocks such as storm surges, droughts, floods, and seawater intrusion, especially affecting low-income citizens who hold little capacity to cope with it. According to UNEP (2013), the rainfall along the coastal area was predicted to rise 2% to 6% by 2050. Heavy rain in the rainy season has induced the risk of storm and flash flooding in the low-lying area where agricultural crops are mostly concentrated. Climate Investment Funds (CIF, p. 48) projected that in the worst-case scenario, the annual mean temperature was expected to increase by 1 ◦ C by 2025 in Koh Kong. In addition, a study of MoE (2002) conducted in Peam Krasoab, a commune in Koh Kong, found that a 1-m rise of seawater would cause 44 square kilometers (about 0.4% of Koh Kong) to lie under water permanently. Since climate hazards strongly affect middleand low-income citizens, the study aims to examine the vulnerability of climate shocks on urban residents in Koh Kong province. The two main objectives of the study are listed as follows: 1. To identify the most vulnerable region of climate hazards through the construction of an index to compare three different communes in Koh Kong. 2. To measure the vulnerability to poverty in urban locations through the comparison of poor and non-poor households. 2. Methodology 2.1. Study Area According to the national census in 2008, Koh Kong is divided into eight districts with 33 communes. This study was conducted in the Smach Meanchey district, where the provincial town of Koh Kong is located, which later was changed to Khemarak Phoumin district. It is comprised of three communes: Smach Meanchey, Daun Tong, and Steong Veng. The population of Smach Meanchey was about 29,329 which was 21% of the total population in Koh Kong, according to national census in 2008. Base on the national census (1998) conducted by the National Institute of Statistics (NIS), an urban area 1Ancha Srinivasan is ADB’s climate change specialist of Southeast Asia Department.
Economies 2017,5, 41 3 of 19 refers to any district in which a provincial town is located. With a clarification of the term “urban” due to population growth, the new definition of “urban” was adopted in 2004 and was used in the national census in 2008 based on three categories: population density exceeding 200 per Km 2 , percentage of males employed in the agricultural sector lower than 50%, and the total population of a commune exceeding 2000 (NIS 2008). Figure 1shows the location of Smach Meanchey district. It is located in the western part of Koh Kong. Although Daun Tong is relative small compared to the other two communes, high population density concentrates in this commune. Economies2017,5,x3of18 inthenationalcensusin2008basedonthreecategories:populationdensityexceeding200perKm2, percentageofmalesemployedintheagriculturalsectorlowerthan50%,andthetotalpopulationof acommuneexceeding2000(NIS2008). Figure1showsthelocationofSmachMeancheydistrict.ItislocatedinthewesternpartofKoh Kong.AlthoughDaunTongisrelativesmallcomparedtotheothertwocommunes,highpopulation densityconcentratesinthiscommune. Figure1.Mapofthestudyarea. 2.2.SamplingMethodology Thisstudyappliedaquantitativemethodbyusingasemi‐structuredquestionnaireforthe householdinterviewduringOctober2016.Itusedthenonprobabilitysamplingmethodandfollowed thestepsbelowinchoosingtherespondent.2 First,thelocationwaschosenbasedonitsgeographicalcondition.Intermsofgeography,it selectedthevillagesineachcommuneaccordingtotheirhistoryofnaturalhazards.Itonlychosethe villageswherehazardstendedtobeextremeandfrequentlyoccurredoveralongperiod,compared totheothervillagesinthecommune.Sincethestudysimplyselectedtheplaceswherehazardsused toexist,thismayhaveresultedinabiasedbehaviorinthedatacollection;however,italsoenhanced theresulttobemoreaccurateforcomparison.Next,thenumberofsamplesineachcommunewas selectedbasedontheproportionofhouseholdsineachcommunecomparedtothetotalpopulation. 2Thestudyusedpurposingsamplingbasedonresearcherknowledgetoidentifytheparticipants,sothe samplemaynottrulyrepresentthepopulationduetothelackofrandomness.Thus,theresultsofthisstudy maynotbeabletorepresenttheurbanvulnerabilityinotherregions. Figure 1. Map of the study area. 2.2. Sampling Methodology This study applied a quantitative method by using a semi-structured questionnaire for the household interview during October 2016. It used the nonprobability sampling method and followed the steps below in choosing the respondent.2 First, the location was chosen based on its geographical condition. In terms of geography, it selected the villages in each commune according to their history of natural hazards. It only chose the 2 The study used purposing sampling based on researcher knowledge to identify the participants, so the sample may not truly represent the population due to the lack of randomness. Thus, the results of this study may not be able to represent the urban vulnerability in other regions.
Economies 2017,5, 41 4 of 19 villages where hazards tended to be extreme and frequently occurred over a long period, compared to the other villages in the commune. Since the study simply selected the places where hazards used to exist, this may have resulted in a biased behavior in the data collection; however, it also enhanced the result to be more accurate for comparison. Next, the number of samples in each commune was selected based on the proportion of households in each commune compared to the total population. In total, 120 households were interviewed in Smach Meanchey, Daun Tong, and Steong Veng communes with the sample sizes of 50, 33, and 37, respectively. Lastly, it selected households via the distance from one to another, about 200 m. It targeted the head of the household to be a respondent. For ethical and confidential considerations, consent was requested from every participant. If a participant was unwilling to join the interview, the interviewer would ask for any available member in the family who could represent the household; otherwise, the interviewer would try to access the neighboring household as a substitution. Moreover, the interview did not use any recording device so as to make household feel comfortable. All the information provided only was used for study and was destroyed after six months. 2.3. Vulnerability Index The study followed the definition of the Intergovernmental Panel on Climate Change (IPCC), which identified vulnerability as “the degree to which a system is susceptible to, or unable to cope with, adverse effects of climate change, including climate variability and extremes” ( McCarthy et al. 2001, p. 6 ). It viewed vulnerability as the function of exposure, sensitivity, and adaptive capacity that is widely recognized in the study of climate vulnerability assessment. Measuring vulnerability is a challenge, and mostly proxy variables are used to construct an index in order to compare the vulnerability degree across different regions. Srinivasan et al. (2013) proposed the need to create a capable governance to reduce urban vulnerability to water shortage. Depietri et al. (2013) assessed urban vulnerability to heat waves in urban Germany via GIS with the use of a series of social and ecological indicators such as resource, infrastructure, settlement, health, death rate, and so on. Chen et al. (2013) studied vulnerability to natural hazards in China by using the PCA approach and found that employment and poverty, education, housing quality, size of family, minority, and housing size were the significant determinants of vulnerability. Andersen and Cardona (2013) introduced the Livelihood Diversification Index (LDI) to study vulnerability and resilience capacity in Bolivia. Hahn et al. (2009) developed the Livelihood Vulnerability Index (LVI) to assess vulnerability in Mozambique. Ncube et al. (2016) constructed the Household Vulnerability Index (HVI) to study the vulnerability of climate change in South Africa. Basically, it estimated the household vulnerability according to the natural assets, physical assets, financial assets, human capital assets, and social assets derived from Sustainable Livelihood Index (Solesbury 2003). The use of indicators to construct an index in the study came from the review of the past studies that were recognized for their significant contribution in vulnerability studies. A weighing indicator was one of the challenges in constructing the index. Some studies used equal weight, as in Hahn et al. (2009), while some preferred to consult with an expert, as in Chhinh and Cheb (2013). The most common method for the weighing indicator was Principal Component Analysis (PCA), as recommended in (Gbetibouo and Ringler 2009;Piya et al. 2012;Megersa 2015). This paper used the PCA approach to assign weight. A drawback of using PCA was the correlation sign produced by PCA; however, the problem was not serious, and it was recommended to ignore the issue. This was because changing the correlation sign in the component did not change the variance, so it did not affect the weighing of the variable. Although the sign was not the issue in the mathematical explanation, the study preferred to consider the second or third principal components as well when the eigenvalue was similar to the first principal component and produce better correct signs of the indicators, conforming to the past literature.
Economies 2017,5, 41 5 of 19 The first step in calculating the index is to normalize the value of variables by using the following formula: Normalized Value =Observed Value −Minimum Value Maximum Value −Minimum Value Normalization was used to create uniform data sets, to make them comparable by scale measurement. In the study, the process transformed the values to range from 0 to 1, where the higher the value, the stronger the effect. After the values were normalized, the weighing process was conducted. The weighing variable aimed to identify the relative importance of each variable among others in explaining a certain phenomenon. A review of the literature revealed three methods of weighing the indicators: expert judgment, equal weight, and econometric approaches such as principal components or factor analysis (Gbetibouo et al. 2010). The use of expert judgment was constrained due to the knowledge in different fields, and it was difficult to reach consensus among the experts. Some studies used equal weight; however, by doing so it underestimates some important variables and overestimates some unimportant variables. To avoid these problems, PCA was used in this study to produce the weight of the variables. The component score coefficient from the first PCA was used as a weight since it explained the highest variability; even so, the second and third component scores were also considered if their eigenvalues and the variabilities explained were comparable to the first component. By doing so, it was helpful to correct the wrong sign of the correlation produced by the first principal component. Next, variables were added together to form the index based on their own categories such as exposure, sensitivity, and adaptive capacity by using the formula below: Iij = n ∑ i=1 bi(xij −xmin i xmax i−xmin i) where Iij was an index value of the variable iin jcategory, bi was the weight received from the principal component—generally the first component (PCA1), xij was the value of the variable iin jcategory, xmin iwas the minimum value of variable i, and xmax iwas the maximum value of variable i. Lastly, the climate vulnerability index was calculated by the following formula:3 VI =3 qEI ×SI ×(1−ACI) where VI: vulnerability index; EI: exposure index; SI: sensitivity index; and ACI: adaptive capacity index. 2.4. Vulnerability as Expected Poverty Vulnerability is a concept that links to poverty, as it may put those who are not currently poor in poverty in the future. This idea has brought insight into the study of the wellbeing loss due to shocks, ex-post assessment. In this context, vulnerability can be explained as a situation in which shocks may cause the household income to fall below a certain threshold. The study of vulnerability in the context of poverty explains the level of household capacity to sustain a certain shock and the dynamic of household livelihood conditions. Just because a household is currently poor does not mean that they are vulnerable and vice versa. The lack of means to smooth expenditure over time has instigated the notion of studying vulnerability as expected poverty (VEP). In this study, both the urban national poverty line and international poverty line were used. The reason for using two difference poverty thresholds is to understand their sensitivities. As the 3The study used geometric mean instead of arithmetic mean.
Economies 2017,5, 41 6 of 19 Cambodian national poverty line was calculated based on the prices in 2009, it will be adjusted according to inflation to determine the prices in 2015. It was used as a baseline, while the international poverty line was used to observe the rate of change from the baseline. This study used the concept of Vulnerability as Expected Poverty (VEP) developed by Shubham Chaudhuri to study the vulnerability in the context of poverty. He identified vulnerability as the probability that the household income or consumption may fall below the poverty line due to shocks such as climate hazards or financial crisis, and so on, as expressed by this formula: ln Ch=Xhβ+eh(1) ln Ch represents daily income in log form; 4Xh is the bundle of household characteristics and climate exposure index; β is a vector of parameters; and eh represents the disturbance that captures idiosyncratic factors (shocks). Another assumption was that the variance eh , which captures idiosyncratic factors and reflects the inter-temporal variance of income, is provided by Equation (2). Since eh could not be observed, the solution was to examine from the sample. σ2 e,h=Xhθ(2) Chaudhuri et al. (2002) expressed that vulnerability to poverty was not a linear function, since it depended not only on the mean of income level but also inter-temporal variance stemming from income; thus, it involved not only the prediction of future income but also the expected disturbance as well. Since it was assumed that the disturbance term possesses heteroscedasticity in nature, it will not produce efficient estimators. To resolve this issue, Amemiya (1977) carried out three-step Feasible Generalized Least Square (FGLS) to transform variables by assigning weight to produce efficient estimators: β and θ . To reduce the methodological problems, the data was checked before processing. The dependent variable was checked for normality and was transformed into a log form to avoid outliers. For explanatory variables, they were checked for multicollinearity by using a correlation matrix. If the correlation is higher than 70%, one variable is dropped out of the equation. As expected from the assumption of ramdom error, the residual eh is not normally distributed that created the non-constant variance, so weight was used to transform the data as a remedial measure. Starting with Equation (1), an OLS regression was run to find the estimated residual eh, then the square of estimated residual was used to regress on Xhthrough the OLS procedure again. ˆ e2 OLS,h=Xhθ+µh(3) Then, the predicted value was used from the regression as weight, and Xhˆ θ was transformed from Equation (3) into: ˆ e2 OLS,h Xhˆ θOLS =Xh Xhˆ θOLS θ+µh Xhˆ θOLS =Xhˆ θFGLS +ui(4) Xhˆ θFGLS was the consistent estimated variance in Equation (2), shown above. Through the FGLS procedure, this variance σ2 e,hwas unbiased and can be written as a standard deviation as follows: ˆ σe,h=qXhˆ θFGLS (5) At this stage, the issue that some estimated values are not always positive may occur, so another procedure may need to be used, such as a logistic specification that would force the predicted value to 4 As the study focused on the urban area, it was more appropriate to assume that households saved money, so income was used rather than consumption for its nature of heteroscedasticity.
Economies 2017,5, 41 7 of 19 always be positive. 5 However, in this study, we will ignore this issue and simply drop those data from the estimation. Subsequently, this standard deviation was employed to transform Equation (1) into: lnCh qXhˆ θFGLS =Xh qXhˆ θFGLS β+eh qXhˆ θFGLS (6) The OLS estimation from Equation (6) will produce a consistent and efficient β . To be sure, the residual test, the Shapiro-Wilk test, was used to check the normality. Finally, by using β and θ , the expected log of income and the variance of log income of each household were predicted as: ˆ E(lnCh|Xh)=Xhˆ β(7) ˆ V(lnCh|Xh)=ˆ σ2 e,h=Xhˆ θ(8) Assuming income as log-normal distributed, the above equation can estimate the probability that a household with characteristic Xh will be poor (household vulnerability level). Let Φ refer to the cumulative density of the standard normal; the estimated probability equation is as follows: ˆ V=ˆ Pr(lnCh<lnz|Xh)=Φ lnz −Xhˆ β qXhˆ θ (9) where lnz is the natural log of minimum income; at this level a household will be considered vulnerable (at or below the poverty line). Xhˆ β was the expected mean of household income, while Xhˆ θ was the predicted variance. 3. Results 3.1. Household Characteristics Table 1summarizes the basic characteristics of the households. On average, the head of household was aged around 44 years old with five years of education. Normally, a household had five members with two people in the family earning income, and there was at least one person in a family who had finished secondary school on average. Average monthly household income and consumption were about 464 dollars/month and 388 dollars/month, respectively. Daily income and consumption per person were around 3.3 dollars/day and 2.75 dollars/day on average. There was a small difference between the income and consumption levels that reflected the lower rate of saving. Table 1. Household characteristics. Household Characteristics Total Daun Tong Steong Veng Smach Meanchey Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Age of household head 43.93 13.40 43.80 13.50 45.60 13.90 42.70 13.10 Education of head of household 4.91 4.00 4.30 3.60 4.60 4.10 5.50 4.20 Household sizes 5.04 2.14 5.70 2.10 5.00 2.40 4.70 1.90 Household members who finished grade 9 1.07 1.19 1.50 1.30 0.70 1.20 1.00 1.10 Household members who earn revenue 2.13 1.06 2.40 1.10 2.10 1.10 1.90 1.00 5See: Elbers, Lanjouw, and Lanjouw (Elbers et al. 2001).
Economies 2017,5, 41 8 of 19 Table 1. Cont. Household Characteristics Total Daun Tong Steong Veng Smach Meanchey Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Mean Standard Deviation Monthly household income (dollars) 463.93 244.47 518.39 183.78 443.83 263.71 442.85 263.12 Monthly household consumption (dollars) 387.55 176.53 453.03 153.10 356.96 171.27 367.02 186.53 Daily income per person (dollars) 3.28 1.68 3.37 1.64 3.12 1.38 3.32 1.92 Daily consumption per person (dollars) 2.75 1.15 2.90 1.22 2.60 1.16 2.71 1.09 3.2. Gini Coefficient and Inequality Monthly incomes of 120 households were collected to construct the Lorenz curve to measure income inequality in the Smach Meanchey district. Using the data from this survey, the Gini coefficient or Gini index was calculated. This index ranges from 0 to 1, where a higher value represents greater inequality. According to the data, the Gini index was around 0.28 (see Figure 2), so it can be considered as an indication of low inequality among households in the Smach Meanchey district. With low-income inequality, it would be useful to the study of the degree of the sensitivity of vulnerability to poverty as the assumption of income indifference across households. However, the lowest 10% controlled only about 3% of the total income, while the top 10% occupied almost one-fourth of the entire income. Economies2017,5,x8of18 Figure2.IncomeinequalityintheSmachMeancheydistrict. 3.3.ClimateVulnerability 3.3.1.ExposureIndex TherateofexposurewasrelativelystrongerinSteongVengandDaunTongcommunes comparedtoSmachMeanchey,withindicesof0.456,0.450,and0.289,respectively(Table2).PCA revealedtherelativeimportanceofstormsandtyphoonsinexplainingtheexposuredegreeon householdlivelihoodinthethreecommunescomparedtofloodinganddrought,aslargecomponent scoreswerecontributedbythesevariables. Table2.Summaryofexposureindex. DescriptiveComponentScore Coefficient Daun Tong Steong Veng Smach Meanchey Index Thenumberoffloodsaffectinghouseholdsthat occurredduringthelast5years0.0390.0020.0040.004 Impactoffloodingonlivelihood0.1070.0290.0400.041 Periodofinsufficientcleanwaterusageperyear duringthelast5years0.1030.0270.0340.038 Impactofdroughtonlivelihood0.0430.0320.0290.027 Frequencyoftyphoonsaffectinghousehold livelihoodperyearduringthelast5years0.2860.0620.0630.035 Impactoftyphoonsonlivelihood0.2890.1680.1330.099 Frequencyofstormsaffectinghouseholdlivelihood peryearduringthelast5years0.3120.0360.0530.016 Impactofstormsonlivelihood0.3290.0940.0990.028 Total 0.4500.4560.289 Eigenvalue2.526 %ofvarianceexplained31.575 3.3.2.SensitivityIndex Threevariablesgreatlycontributedtoexplainingthesensitivityofclimatehazards:resource‐ dependency,accessibilitytocleanwaterduringdrought,andhealthcareservices.Forresource‐ dependency,itwasbecausehouseholdsdependedstronglyonthefisheryasasourceoftheir revenue.Withstormsandtyphoons,theywereunabletooperatetheirbusinesses,whichcauseda rapiddeclineintheirincome.Additionally,withoutsufficientwatersupplyduringthedrought period,somefamiliesboughtwateratarelativelyhigherpricewhilepoorfamiliesmanagedtouse waterfromwells.However,privatewellscannotbedugdeeperduetosaltwater,andthewaterwas notcleanenoughtouse,andoftenlinkedtohealthissues.Thesensitivitydegreewasfoundtoeb 0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 %ofIncome %ofPopulation EqualityLine LorenzCurve Figure 2. Income inequality in the Smach Meanchey district. 3.3. Climate Vulnerability 3.3.1. Exposure Index The rate of exposure was relatively stronger in Steong Veng and Daun Tong communes compared to Smach Meanchey, with indices of 0.456, 0.450, and 0.289, respectively (Table 2). PCA revealed the relative importance of storms and typhoons in explaining the exposure degree on household livelihood in the three communes compared to flooding and drought, as large component scores were contributed by these variables.
Economies 2017,5, 41 15 of 19 5. Conclusions This study contributes to identify the most vulnerable location of climate change. To measure the climate vulnerability, we chose an index approach to compare climate vulnerability among three communes in Koh Kong. It was suggested that Daun Tong communes should be prioritized for intervention policies, as it experienced the highest climate vulnerability with an index of about 0.53 compared to the other two regions: Steong Veng and Smach Meanchey with indices of 0.52 and 0.43, respectively. In addition, the study also investigates vulnerability in the context of poverty by perceiving vulnerability as the probability that a household income will fall below the poverty line. It revealed that, by using the international poverty line threshold, more than one-third of households have been facing vulnerability to poverty, while this value was only about 28% when the national poverty line was used. By using two poverty thresholds, the study showed that households were not sensitive to falling into the poverty. However, both poverty thresholds indicated that a higher proportion of poor families faced vulnerability to poverty compared to non-poor households. By assuming that vulnerability arises as a result of shocks such as climate hazards and changes in household characteristics, the study found that improving education, income diversification, and physical assets such as housing could be effective methods to lower vulnerability. Lastly, borrowing should be done cautiously. This study pointed out that access to a loan may not be a wise decision as a household may be unable to pay back the loan with interest, and it induced a high vulnerability for the subsequent period. 6. Further Study This study did not categorize the types of loans, whether a household borrowed from formal or informal sectors. Some studies suggested that access to authorized loans could improve the livelihood of a household, while borrowing from the informal sector would likely induce vulnerability. The next study should categorize the types of loan for a better comparison. Moreover, not all the variables that were highlighted were incorporated into the study due to methodological constraint. To illustrate, since households used various types of energy, the study could not capture it well, and it would lead to a bias to include it. Consultation with an expert is important to find good proxies that fit within the context. Next, the use of PCA for weighing is effective; however, one issue is the correlation sign produced by PCA. For example, in the sensitivity study, the correlation signs of the road conditions and accessibility to healthcare were the opposite of what was expected. The next study is advised to conduct a rigorous study and consultation about the relationships among variables; otherwise, it would lead to a misinterpretation of the results. Lastly, Since VEP works on many assumptions, some may not be fulfilled. One of these is heteroscedasticity of income. Whether income, consumption, or another variable should be used as a proxy is a crucial problem. In this study, income was used instead of consumption because it tended to vary across households more than consumption, and reflected the social status of the rich and poor households, so it was suitable for the assumption of heteroscedasticity. However, the record of household income and expenditure cannot avoid some errors due to the accessibility of data. The next study should implement a clear and standardized procedure to record household income and expenditure that is recommended by a formal institution in order to improve the validity of the data. Acknowledgments: Author sincerely acknowledges the reviewers’ comments and the supports from Chhinh Nyda and Thath Rido for their technical comments. Author also would like to show gratitude and appreciation to the Urban Climate Change Resilience in Southeast Asia (UCRSEA) project, which is funded by the Social Sciences and Humanities Research Council of Canada and the International Development Research Centre (IDRC) for funding support to conduct this research. Conflicts of Interest: The author declares no conflict of interest.
Economies 2017,5, 41 16 of 19 Appendix A Lists of indicators in this study were summarized in the 4 tables below. In totally, 8 variables were used to study the exposure degree (Table A1). It could be categorized into four main groups: flood, drought, Typhoon, and storm. Similarly, another 8 variables were constructed to measure the sensitivity (Table A2). 15 variables were used to assess the adaptive capacity in Table A3. In terms of vulnerability as expected poverty, a list of variables was summarized in Table A4. Table A1. Indicators in exposure study; which the positive sign (+) refers to positive contribution to higher level of exposure. Exposure Indicators Descriptive Type of Measurement Expected Sign Flood The number of floods affecting households that occurred during the last 5 years Scale + Impact of flooding on livelihood Ordinal + Drought Period of insufficient clean water usage per year during the last 5 years Scale + Impact of drought on livelihood Ordinal + Typhoon Frequency of typhoons affecting household livelihood per year during the last 5 years Scale + Impact of typhoons on livelihood Ordinal + Storm Frequency of storms affecting household livelihood per year during the last 5 years Scale + Impact of storms on livelihood Ordinal + Table A2. Indicators in sensitivity study; which the positive sign (+) refers to positive contribution to higher rate of sensitivity while the negative sign ( − ) refers to negative contribution to lower sensitivity degree. Descriptive Type of Measurement Expected Sign Damage to property and livestock due to climate hazards Ordinal + Number of family member(s) injured due to floods, storms, and landslides Scale + The level of household dependency on natural resources Ordinal + Agricultural dependency for income Ordinal + Road conditions after flooding ordinal − Distance from market (minutes of traveling) Scale + Lacking clean water during drought Ordinal + Accessibility to healthcare (level of receiving health services per year) Ordinal + Table A3. Indicators in adaptive capacity study; which the positive sign (+) refers to positive contribution to higher level of adaptive capacity while the negative sign ( − ) refers to negative contribution to lower adaptive capacity. Descriptive Unit of Measurement Expected Sign Housing quality Ordinal + Tools and technology to access climate information (TV, radio, mobile phone) Ordinal + Self-protection tools such as sandbags, life-jackets, and so on Scale + Number of family members who finished grade 9 Scale + Number of family members who earn income Scale + Training or vocational course related to climate change attended by family members Scale + Assistance from government Ordinal +
Economies 2017,5, 41 17 of 19 Table A3. Cont. Descriptive Unit of Measurement Expected Sign Availability of supportive policy Ordinal + Diversification of income sources Scale + Rice reserve during a shock Ordinal + Ownership (animal, livestock) Ordinal + Amount of borrowing from formal and informal sectors (debt: monthly) Scale + Information sharing related to climate hazards with neighboring Ordinal + Amount of social support from relatives and community during and after disaster Ordinal + Dependency ratio Ordinal + Table A4. Variables in VEP study in which positive sign (+) refers to positive contribution to higher vulnerability while the negative sign (−) refers to negative contribution to lower vulnerability. For binary dummy variable, 0 indicates the absence while 1 indicates the presence of the effect, simply put it refers to (No, Yes) answer. Descriptive of Independent Variables Unit of Measurement Expected Sign Age of respondent Scale (years) − Household sizes Scale (persons) + Education of the head of household Scale (years) − Climate hazards (exposure index) Scale (Index) + Agricultural dependency Ordinal (rating) + Level of healthcare accessibility Ordinal (rating) − Housing quality Ordinal (rating) − Assistance from government during a shock Dummy (0.1) − Income diversification Dummy (0.1) − Possession of livestock asset Dummy (0.1) + Debt accessibility Dummy (0.1) − Access to information related to climate hazards Ordinal (rating) − References Amemiya, Takeshi. 1977. The maximum likelihood and the nonlinear three-stage least squares estimator in the general nonlinear simultaneous equation model. Econometrica: Journal of the Econometric Society 45: 955–68. [CrossRef] Andersen, Lykke E., and Marcelo Cardona. 2013. Building Resilence against Adverse Shocks: What Are the Determinants of Vulnerability and Resilence? Development Research Working Paper Series No. 02/2013, Institute for Advanced Development Studies, La Paz, Bolivia. Chaudhuri, Shubham, Jyotsna Jalan, and Asep Suryahadi. 2002. Assessing Household Vulnerability to Poverty from Cross-Sectional Data: A Methodology and Estimates from Indonesia. Discussion Paper, No. 0102-52, Columbia University, New York, NY, USA. Chen, Wenfang, Susan L. Cutter, Christopher T. Emrich, and Peijun Shi. 2013. Measuring social vulnerability to natural hazards in the Yangtze River Delta region, China. International Journal of Disaster Risk Science 4: 169–81. [CrossRef] Chhinh, Nyda. 2014. Climate Change Adaptation in Agriculture in Cambodia. London: Edward Elgar Publishing. Chhinh, Nyda, and Hoeurn Cheb. 2013. Climate Change Vulnerability: Household Assessment Levels in the Kampong Speu Province, Cambodia. In Climate Change Vulnerability Assessment in Kampong Speu Province, Cambodia. Phnom Penh: Phnom Penh Royal University of Phnom Penh, pp. 53–62.
Economies 2017,5, 41 18 of 19 Climate Investment Funds (CIF). 2014. Climate Change Impact Modeling and Vulnerability Assessments for Koh Kong and Mondulkiri Provinces in Cambodia. Phnom Penh: ADB. Depietri, Yaellav, Torsten Welle, and Fabrice G. Renaud. 2013. Social vulnerability assessment of the Cologne urban area (Germany) to heat waves: Links to ecosystem services. International Journal of Disaster Risk Reduction 6: 98–117. [CrossRef] Elbers, Chris, Jean O. Lanjouw, and Peter Lanjouw. 2001. Welfare in villages and towns: Micro-level estimation of poverty and inequality. Econometrica 71: 355–64. [CrossRef] Gaiha, Raghav, and Katsushi Imai. 2008. Measuring Vulnerability and Poverty Estimates for Rural India. Working Paper UNU-WIDER 2008.40, Unitded Nation University, Helsinki, Finland. Gbetibouo, Glwadys Aymone, and Claudia Ringler. 2009. Mapping South African Farming Sector Vulnerability to Climate Change and Variability: A Subnational Assessment. IFPRI Discussion Paper, International Food Policy Research Institute (IFPRI), Washington, DC, USA; Center for Environmental Economics and Policy in Africa (CEEPA), Pretoria, South Africa. Gbetibouo, Glwadys A., Claudia Ringler, and Rashid Hassan. 2010. Vulnerability of the South African farming sector to climate change and variability: An indicator approach. Natural Resources Forum 34: 175–87. [CrossRef] Hahn, Micah B., Anne M. Riederer, and Stanley O. Foster. 2009. The Livelihood Vulnerability Index: A pragmatic approach to assessing risks from climate variability and change—A case study in Mozambique. Global Environmental Change 19: 74–88. [CrossRef] Hay, John, and Nobuo Mimura. 2006. Supporting climate change vulnerability and adaptation assessments in the Asia-Pacific region: An example of sustainability science. Sustainability Science 1: 23–35. [CrossRef] Heng, Pheakdey. 2012. Cambodia’s Energy Security Is at Risk. Available online: https://www.cambodiadaily.com (accessed on 24 October 2016). Irvine, Kim, Murphy Tom, M. Sampson, D. Va, Stephen J. Vermette, and Tao Tang. 2006. An overview of water quality issues in Cambodia. In Effective Modeling of Urban Water Systems, Monograph, 14. Guelph: Computational Hydraulics International. Kelly, Mick P., and Neil W. Adger. 2000. Theory and practice in assessing vulnerability to climate change andFacilitating adaptation. Climatic Change 47: 325–52. [CrossRef] Ligon, Ethan, and Laura Schechter. 2003. Measuring vulnerability. The Economic Journal 113: C95–C102. [CrossRef] Liu, Chunling, Dominic Golding, and Gang Gong. 2008. Farmers’ coping response to the low flows in the lower Yellow River: A case study of temporal dimensions of vulnerability. Global Environmental Change 18: 543–53. [CrossRef] Ludeña, Carlos E., and Sang Won Yoon. 2015. Local Vulnerability Indicators and Adaptation to Climate Change: A Survey. Washington: Inter-American Development Bank. McCarthy, James J., Osvaldo F. Canziani, Neil A. Leary, David J. Dokken, and Kasey S. White. 2001. Climate Change 2001: Impacts, Adaptation, and Vulnerability. Camberidge: Camberidge University Press. McGranahan, Gordon, Deborah Balk, and Bridget Anderson. 2007. The rising tide: assessing the risks of climate change and human settlements in low elevation coastal zones. Environment and Urbanization 19: 17–37. [CrossRef] Megersa, Dawit. 2015. Measuring Vulnerability to Poverty. Uppsala: SLU, Department of Economics. Mendoza, Maria Emilinda T., Bui Dung The, Heng Naret, Vicente G. Ballaran, and Jamie Kim B. Arias. 2014. Assessing vulnerability to climate change impacts in Cambodia, the Philippines and Vietnam: An analysis at the commune and household level. Journal of Environmental Science and Management 17: 78–91. Mirza, M. Monirul Qader. 2003. Climate change and extreme weather events: Can developing countries adapt? Climate Policy 3: 233–48. [CrossRef] MoE. 2002. Physical Framework Plan Koh Kong Province. Phnom Penh: Ministry of the Environment. MoE. 2013. Cambodia Climate Change Strategic Plan 2014–2023. Phnom Penh: National Climate Change Committee. Moser, Caroline O. N. 1998. The asset vulnerability framework: Reassessing urban poverty reduction strategies. World Development 26: 1–19. [CrossRef] Ncube, Mkhululi, Nomonde Madubula, Hlami Ngwenya, Nkulumo Zinyengere, Leocadia Zhou, Joseph Francis, Talentus Mthunzi, Crespo Olivier, and Tshilidzi Madzivhandila. 2016. Climate change, household vulnerability and smart agriculture: The case of two South African provinces: Original research. Jàmbá: Journal of Disaster Risk Studies 8: 1–14. [CrossRef]
Economies 2017,5, 41 19 of 19 NIS. 2008. General Population Census of Cambodia 2008. Phnom Penh: National Institute of Statistics, Ministry of Planning. Oni, Omobowale, and Sulaiman Yusuf. 2008. Determinants of Expected Poverty among Rural Households in Nigeria. African Economic Research Consortium Research Paper, Nairobi, Kenya. Opiyo, Francis E. O., Oliver V. Wasonga, and Moses M. Nyangito. 2014. Measuring household vulnerability to climate-induced stresses in pastoral rangelands of Kenya: Implications for resilience programming. Pastoralism 4: 1–15. [CrossRef] Piya, Luni, Keshav Lall Maharjan, and Niraj Prakash Joshi. 2012. Vulnerability of rural households to climate change and extremes: Analysis of Chepang households in the Mid-Hills of Nepal. Paper presented at the 2012 International Association of Agricultural Economists (IAAE) Triennial Conference, Foz do Iguacu, Brazil, August 18–24. Satterthwaite, David, Saleemul Huq, Hannah Reid, Mark Pelling, and Patricia Romero Lankao. 2007. Adapting to Climate Change in Urban Areas: The Possibilities and Constraints in Low-and Middle-Income Nations. London: IIED, vol. 1. Solesbury, William. 2003. Sustainable Livelihoods: A Case Study of the Evolution of DFID Policy. London: Overseas Development Institute. Srinivasan, Veena, Karen C. Seto, Ruth Emerson, and Steven M. Gorelick. 2013. The impact of urbanization on water vulnerability: A coupled human–environment system approach for Chennai, India. Global Environmental Change 23: 229–39. [CrossRef] UNEP. 2013. Mainstreaming of Climate Change Adaptation into the Sub-National Development Planning in Cambodia. Nairobi: United Nations Environment Programme. UNICEF. 2010. Health Service Access among Poor Communities in Phnom Penh. New York: UNICEF. © 2017 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).