The 2011 flood event in the Mekong Delta ...

doi:10.1111/disa.12171

The 2011 flood event in the Mekong Delta: preparedness, response, damage and recovery of private households and small businesses Do Thi Chinh, Philip Bubeck, Nguyen Viet Dung, and Heidi Kreibich1

Floods frequently cause substantial economic and human losses, particularly in developing countries. For the development of sound flood risk management schemes that reduce flood consequences, detailed insights into the different components of the flood risk management cycle, such as preparedness, response, flood impact analyses and recovery, are needed. However, such detailed insights are often lacking: commonly, only (aggregated) data on direct flood damage are available. Other damage categories such as losses owing to the disruption of production processes are usually not considered, resulting in incomplete risk assessments and possibly inappropriate recommendations for risk management. In this paper, data from 858 face-to-face interviews among flood-prone households and small businesses in Can Tho city in the Vietnamese Mekong Delta are presented to gain better insights into the damage caused by the 2011 flood event and its management by households and businesses. Keywords: Can Tho, floods, flood damage, flood loss, flood risk management, Mekong Delta, Vietnam

Introduction Vietnam is severely affected by hydro-meteorological hazards such as floods, storms, droughts and heavy rainfall as well as by geophysical hazards (Oanh et al., 2011). Among the various hazards, storms and in particular floods are the most important in respect to damage to people and assets. For instance, in the period between 1989 and 2008, floods accounted for 49 per cent of the total damage resulting from natural hazards (World Bank, 2010). In the same period, 11 flood events caused economic losses exceeding USD 100 million each and are thus considered as major loss events (World Bank, 2010). In the future, flood losses are expected to further increase due to changes in hazard and vulnerability related to global change. On the one hand, population growth, rapid socio-economic development and urbanisation increase the exposure and susceptibility to floods in Vietnam (Oanh et al., 2011). For instance, the population in the Mekong Delta is projected to grow from 17.3 million in 2011 to 18.8 million in 2020 (General Statistics Office of Vietnam, 2013; Socialist Republic of Vietnam, 2013). On the other hand, flood hazards in Vietnam are expected to increase due to the effects of climate change: for example, MONRE (2009) developed climate change and sea level rise scenarios for Vietnam, which indicate a rise of about * Paper corrected on 3 March 2016 (after initial online publication on 8 January 2016) to amend the author name order and to insert colour versions of the figures. © 2016 The Author(s). Disasters © Overseas Development Institute, 2016 Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

Do Thi Chinh, Philip Bubeck, Nguyen Viet Dung, and Heidi Kreibich

75 centimetres by 2100 compared to the period 1980–1999. Delgado et al. (2010) revealed an increasing likelihood of extreme floods despite a decrease in frequency of average floods, which has been due to an increasing variability during the past 70 years for the flood regime of the Mekong. According to the study of Dasgupta et al. (2009) on the projected effects of sea level rise on developing countries, Vietnam ranks first in terms of affected population, affected urban areas and affected gross domestic product, and second in terms of impacted land area. To mitigate current and future flood impacts effectively, an integrated flood risk management is required, which needs to be based on a sound analysis and assessment of the flood hazards, potential losses and the effectiveness of different mitigation measures (Ganoulis, 2003; Hall et al., 2003; Merz et al., 2010; Sayers et al., 2002). The risk management cycle offers a valuable framework for such an integrated approach (for example, DKKV, 2003; FEMA, 2010; PLANAT, 2004). It shows the consecutive phases—that is, preparedness, emergency response, recovery and risk reduction—that a society undergoes after it has been hit by a disaster. If a society had been affected by a flood before it might have undertaken structural and non-structural measures to be prepared for the next flood. When a flood occurs, emergency measures will be undertaken to limit adverse effects. This response includes early warning, rescuing victims and emergency measures to mitigate damage (PLANAT, 2004). The effectiveness of the response depends a lot on the preparedness of the society. The resulting damage is influenced by the flood characteristics but also by the resistance given, which might have been improved by precautionary and emergency measures. In the recovery phase the affected community will repair damage and try to regain a similar standard as before the disaster happened. Reconstruction after the flood should already contain the foundations for improved risk reduction. If society is willing to learn and invest in risk reduction, precautionary measures may be implemented without large additional effort when extensive reconstruction needs to be undertaken anyway. Despite the high exposure of Vietnam to hydro-meteorological hazards, damage and loss data as well as additional information related to the risk management cycle (preparedness, response and recovery) are lacking or only available on (highly) aggregated levels. Moreover, damage assessments usually focus on direct damage and do not consider other cost types, such as losses due to the disruption of business processes, possibly resulting in incomplete and biased assessments and decisions concerning natural hazards (Meyer et al., 2013). In the literature, losses due to the disruption of business processes are sometimes referred to as direct damages, as they occur due to the immediate impact of the hazard (see, for example, Kok et al., 2004). They are often also referred to as primary indirect damages because the losses do not result from physical damage to property but from the interruption of economic processes (for example, Smith and Ward, 1998). In this article losses due to the disruption of business processes are treated as a separate cost category, as suggested by Meyer et al. (2013, p. 1353). The necessity to consider damage categories other than direct physical damage—in order to avoid an underestimation of the true costs of natural disasters in Vietnam—has recently also been emphasised by the World Bank (2010).

The 2011 flood event in the Mekong Delta

For Vietnam, damage data are available from the global EM-DAT database, which provides information on the human impact and economic damage per event (EM-DAT, 2012). However, due to the high level of aggregation (for example, one economic damage number per event), this dataset provides only limited insights for the development of specific risk management strategies at the national, regional or local level. More detailed data on the consequences of flooding are available from the Central Committee for Flood and Storm Control (CCFSC) of the Vietnamese government, which provides quantitative information on damage to housing, agriculture, transportation and victims. While this rather detailed database is certainly a valuable information resource, insights into the different stages of the risk management cycle other than flood damage—such as flood preparedness, early warning and emergency measures as well as flood recovery—are not available, to the best of our knowledge. Such detailed information related to the different aspects of the risk management cycle is needed, however, for the development of sound and effective flood risk management strategies (Bouwer et al., 2007; Viavattene et al., 2012; Vis et al., 2003). For instance, it can provide insights into whether the early warning system is well functioning or needs to be improved. Moreover, cost types other than direct damage, such as losses owing to the disruption of business processes, are not accounted for, even though these can make up a substantial part of overall losses (Hallegatte, 2008). To account also for these cost categories is important, because optimal investments in risk management correspond to the point on the cost–risk curve at which the marginal benefit of an increment of risk reduction is equal to the cost of buying that increment (Hall et al., 2012). Thus, if we fail to better understand the total costs of floods, including losses due to the disruption of business processes and costs of risk mitigation, societies will be increasingly under-protected. Similar to previous quantitative studies in Europe (for example, Kreibich et al., 2007, 2009; Thieken et al., 2007), this article presents data from face-to-face interviews with 480 flood-prone households and 378 small businesses in Can Tho city to gain detailed insights into flood preparedness, early warning and emergency measures, flood damage and recovery, which can be used for the improvement of risk mitigation strategies. Careful analyses of a flood, society’s response and possibilities for prevention and preparation enable the identification of strengths and weaknesses of the risk management strategy and as such enable its enhancement (Kienholz et al., 2004). There are various concepts and definitions for risk or vulnerability (Renn, 2008; Thywissen, 2006). This study follows the natural sciences oriented concept, which defines risk as the probability of suffering loss and defines vulnerability as the degree of loss to a given element at risk resulting from the occurrence of a natural phenomenon of a given magnitude. Vulnerability is composed of two elements: exposure and susceptibility (Merz and Thieken, 2004). Can Tho, which is situated at the heart of the Mekong Delta, was chosen as a case study area due to its economic importance for the region and its high exposure to floods. Moreover, the low-lying delta is considered to be especially vulnerable to the effects of climate change and the projected increase in flood hazards (Dasgupta et al., 2009; Delgado et al., 2010; MONRE, 2009).


The remainder of this paper is organised as follows: the following section provides an overview of flood impacts in Vietnam and the available damage data. The next section provides information on the study area, the methods applied and the sample characteristics of the surveyed households and businesses. Results on flood preparedness, emergency measures, flood impacts and recovery are provided and discussed in the subsequent section, which is followed by the conclusion.

Flood damage and risk management in Vietnam Due to its geographical and climatic boundary conditions, floods are common phenomena all over Vietnam and occur especially in the central coast region, the Mekong Delta and the Red River Delta (Socialist Republic of Vietnam, 2004; Tran and Shaw, 2007). Floods in Vietnam are classified into four types (Socialist Republic of Vietnam, 2004): fluvial floods, flash floods, pluvial floods and storm surges. Depending on the climate (monsoon) regime and the geographic zones, floods in Vietnam occur at different times of the year. For the northern part of Vietnam, the flood season is mainly from April to October. In the central coastal part and the highlands, floods predominantly occur between July and November. In the southern part of Vietnam, flood season occurs in the period between July and the end of November (Long and Dung, 2009). Moreover, Vietnam is prone to coastal flooding due to its 3,260 kilometre-long coastline and low-lying delta areas (Bubeck et al., 2012b; Tran and Shaw, 2007). Flood damage data for Vietnam are collected by the Central Committee for Flood and Storm Control (CCFSC), which is controlled by the Ministry of Agriculture and Rural Development, by the General Statistics Office and by the Ministry of Planning and Investment (Oanh et al., 2011). Damage data are collected from the level of local communities up to the national level by means of a template that lists more than 150 indicators (Oanh et al., 2011). According to the CCFSC, data on flood damage due to fluvial floods, flash floods, pluvial floods and storm surges have been collected for the whole country since 1990. Damage data contained in the CCFSC database predominantly relate to fluvial floods. However, uncertainty is introduced by multi-hazard events, for example, in the CCFSC statistics coinciding fluvial and pluvial floods are categorised as a fluvial flood event; coinciding storm surges and fluvial floods are categorised as storm surges. According to the CCFSC database, Vietnam suffered flood damage of about USD 162 million on average per year in the period from 1990 to 2009.2 Most of the damaged or destroyed items listed in the database are provided in units, such as number of houses, kilometres of railroads, hectares of rice fields, hospitals, road bridges or power stations. Total damage in monetary terms is also provided (Booij, 2004). Some of the damage data are listed for a specific event; others are reported for a period of several months or on a yearly basis. While these rather detailed insights into flood damage in Vietnam are certainly a valuable source of information, they provide little insight into how to improve flood risk management in order to reduce flood losses.


Figure 1. Direct flood damage of all sectors (USD) at 2011 price levels and number of damaged houses in the Vietnamese Mekong Delta, 1991–2011

Source: authors.

An overview of total flood damage of all sectors (housing, agriculture, transportation) and the number of damaged houses recorded for the Mekong Delta is provided in Figure 1. In addition to the data contained in the CCFSC database3 Figure 1 comprises information from reports of the Mekong River Commission (Mekong River Commission, 2010) and the Centre for Hydro-Meteorological Services for Vietnam (Tinh, 2011). As can be seen from Figure 1, floods frequently cause economic losses in the Mekong Delta. However, it also shows that before the destructive flood in 2011 there were about eight years with only minor flood events that hardly caused any damage. The highest loss record has been observed during the historic flood event in 2000, with about USD 500 million. This extreme event triggered various activities for flood risk reduction. The ‘living with flood policy’ was put into effect, aiming at an integrated risk management. Structural and non-structural measures were newly implemented or improved: for example, residential clusters were planned, dykes were built and improved, early warning systems were enhanced and preparedness programmes were implemented (Nien, 2008). Generally, the flood risk management in Vietnam follows a top-down approach (Ehlert, 2010). The central level, which includes the government, is responsible for observing the implementation of the annual flood and storm preparedness solutions and plans, and for issuing legal documents and regulations related to this matter. The CCFSC assists the government and is responsible for the supervision of the ministries, agencies and provinces in preparation and implementation of annual plans and solutions for flood and storm control for their territories (ADB, 2007; Ngan and Long, 2009).


The National Hydro-Meteorological Center and the regional Hydro-Meteorological Stations are responsible for flood forecasting and flood warning in cooperation with the CCFSC, the national broadcasting radio (Voice of Vietnam) and the national television (Vietnam Television) (ADB, 2007). According to different flood characteristics, risk management strategies differ between the regions (CCFSC database, see endnote 1). In Northern Vietnam, risk management is particularly focussed on strengthening the dyke system, on diverging flood courses and on improving safety standards of flood mitigation works. In Central Vietnam, the so-called ‘pro-active prevention, mitigation and adaptation’ strategy is applied, which includes the construction of upstream reservoirs and dyke systems. In the Mekong Delta, as mentioned above, the ‘living with flood’ strategy is applied, which uses structural and non-structural measures to minimise the damage caused by floods as well as to take advantage of floods for sustainable development (Shaw, 2006). For instance, there are two different systems of dykes in the Mekong Delta—fullprotection dyke systems and semi-protection dyke systems. The semi-protection dykes are built to protect summer and autumn rice crops at the beginning of the flood season but also to exploit the sediment and other benefits of the flood during mid-flood season (Huu, 2011). The full-protection dyke systems are mainly implemented in areas with shallow flood water levels to protect people and agriculture in residential clusters from flooding. The residential clusters were built after the flood event in the Mekong Delta in 2000, to move people from flood prone to safe locations (Birkmann, 2011; Dun, 2010; Huu, 2011). In Can Tho city, 24 residential clusters were established from 2002 to 2010, moving about 5,390 households from flood prone areas to new locations (Can Tho CFSC, 2004–2011). Relocation helps people avoid flooding, but it is also associated with challenges and social risks including delayed infrastructure services such as electricity and water, new jobs with possibly lower incomes, and new schools and neighbours (Birkmann, 2014a). Phase 2 of this project continued until 2013 to establish a further eight residential clusters. In Can Tho city, a new 4,760-metre-long dyke was built and the old dyke system was improved. Dyke construction was accompanied by the development of a zoning concept, based on criteria such as topography, soil quality, existing irrigation and drainage networks, and inundation depth. Since 2005 the World Bank has supported flood risk management in Can Tho city. Additionally, flood warning systems and community education and training programmes are conducted to improve awareness and skills of Vietnamese communities to cope with floods. A crop diversification strategy was implemented to reduce potential losses caused by floods (Birkmann, 2011). However, such non-structural strategies remain underdeveloped (Reynaud et al., 2013), and they have limited influence on urban planning. Generally, flood risk management and adaptation are not sufficiently integrated in general procedures of urban planning in Vietnam, and in Can Tho city in particular (Birkmann et al., 2014b). In Can Tho city some climate change action plans are in place and measures include, for example, establishing flood depth markers in


residential areas, establishing forecasting and warning systems for meteorological and hydraulic risks, and training and flood prevention rehearsals to improve flood coping capacity (Can Tho City People’s Committee, 2010a). However, adaptation strategies are not included in action plans of Can Tho city yet as most of the adaptation measures are still debated or planned (Garschagen, 2013). Flood relief and recovery programmes were implemented by both political and social organisations such as the Vietnam Fatherland Front, the trade union or the youth and women association, which support affected areas (Socialist Republic of Vietnam, 2007). Support funds are also be raised by international organisations such as the Red Cross, and by overseas Vietnamese or local private companies. Most governmental support is focussed on the recovery of the public infrastructure, and support for affected households is very limited (Garschagen, 2013). Currently there is no insurance industry in Vietnam operating in the field of disaster risk. The responsibility for disaster financing mechanisms is not clearly defined between the governmental and private sectors. However, it is specified that individuals are responsible for prevention and self-protection against natural disasters. Some individuals are also financially supported by national and local authorities to undertake damage mitigation measures, but there are no clear rules about who will be subsidised (Lempertt et al., 2003). Garschagen (2013) describes major mismatches between state and non-state action for disaster risk reduction owing to low integration between both domains.

Study area, methods and sample characteristics Study area The Mekong Delta begins in the vicinity of the Cambodian capital Phnom Pen, where the Bassac River bifurcates from the Mekong River to form its largest distributary channel. Both the Mekong and the Bassac then further split into nine channels which are called the ‘Nine Dragons’ (Mekong River Commission, 2010). The Mekong Delta in Vietnam comprises an area of about 39,000 kilometres squared (Hung et al., 2012) and is of major economic importance. It is also referred to as the ‘rice bowl’ of Vietnam, producing approximately 50 per cent of Vietnam’s staple food crops, about 90 per cent of Vietnam’s rice export and almost 60 per cent of the country’s export of fishery products (Be et al., 2007). With about 1.2 million inhabitants in 2011 (General Statistics Office of Vietnam, 2013), including about 400,000 inhabitants living in the urban core of the municipality, Can Tho is the largest city in the Mekong Delta and considered its economic, educational and cultural centre. Can Tho city became administratively independent at the beginning of 2004 when the former Can Tho province split into two new administrative units: Can Tho city and Hau Giang province. The city is located on the south bank of the Bassac River, which is the most westerly distributary of the Mekong (Nguyen, 2010).


Floods in the Mekong Delta occur on a recurring basis during the flood season from July to November, and regular inundations of large areas are a prerequisite for the livelihoods of about 17 million people in the delta (Hung et al., 2012). They distribute fertile sediments, enable fish to migrate, flush out toxic materials, serve as pest control as they decimate rat and insect populations, and replenish groundwater reservoirs (Be et al., 2007). At the same time, large-scale flood events above usual water levels pose a serious hazard that have repeatedly caused severe economic damage and loss of life in past decades (Mekong River Commission, 2005; Oanh et al., 2011), such as during the major flood in 2000 that also had a severe impact on Can Tho city (Huong and Pathirana, 2011). Different flood influencing factors have been identified for the Vietnamese Mekong Delta, including the discharge of the Mekong upstream of Kratie (Cambodia), the way the flood wave is buffered by the Tonlé Sap lake system, tidal influences of the East Sea and the Gulf of Thailand, and local precipitation in the Delta (see Figure 2) (Hung et al., 2012). Moreover, floods are nowadays influenced by a vast system of dykes and channels that were built in the Vietnamese part of the delta for flood control (Dung et al., 2013; Hoi, 2005). Whether a flood is associated with predominantly negative or beneficial impacts depends on the speed of onset, the timing of the flood and the peak water levels (Ngan and Long, 2009). In line with the projected rise in sea level and the estimated increase in precipitation during the wet season, which is expected to be especially pronounced in the south of Vietnam (MONRE, 2009), flood hazards could increase considerably in the Mekong Delta. An increased likelihood of extreme discharge and consequently flood events along the Mekong in past decades has been reported by Delgado et al. (2010). The most recent flood event in the Mekong Delta occurred from August to December 2011 and heavily impacted Can Tho city, causing substantial damage to various economic sectors. A map displaying the flood extent and maximum inundation depths of the 2011 flood event is provided in Figure 2. At the main inflow point of the Mekong to the delta at Kratie in Cambodia (see Figure 2), a maximum water level of 22.9 metres was gauged on 24 September. This maximum water level was even higher than that of the historic flood event in 2000 (Mekong River Commission, 2011), while water levels gauged further downstream at Tan Chau and Chau Doc in Vietnam did not reach the record levels of the flood in 2000. One reason for the much lower flood volume in 2011 is that the natural flood retention via the Tonlé Sap lake was reducing water levels considerably more than in 2000. As inundation in Can Tho city is not only caused by river water levels but also tidal influences (Hung et al., 2012), many areas were flooded twice per day for about five to eight hours over a period of three to five months during the flood season in 2011. In areas near the river course, houses were flooded continuously for about one month. Overall about 27,800 houses were inundated in Can Tho city, and USD 11.3 million of damage was caused to buildings, infrastructure, agriculture and aquaculture (Can Tho CFSC, 2011).


Figure 2. Simulated flood extent and maximum inundation depth of the 2011 flood event along the Mekong River and in the Mekong Delta

Source: generated by the authors from the simulation results of Dung et al. (2012) and data from the Southern Institute of Water Resources Research of Vietnam and the Mekong River Commission.

Methods and sample characteristics Detailed insights into the flood preparedness, flood warnings and emergency measures, direct flood damage and losses due to the disruption of production processes and flood recovery (see following section) were obtained by means of a questionnaire that was implemented among 480 flood-prone households and 378 businesses in four different districts in Can Tho: • • • •

Ninh Kieu; Binh Thuy; Cai Rang; and Omon (Figure 3).

The questionnaire comprised about 70 questions for households and 88 questions for businesses and covered the following topics: flood hazard characteristics of the 2011 flood event, flood preparedness, flood warning and emergency measures, flood


damage to households and business contents, building structures and losses due to business interruption, risk perceptions, and the business and socio-economic characteristics of the respondents. The following questions were asked: • Two questions about the significance of flooding and flood experience. • Seventeen questions about the flood in 2011 such as severity, type, flood warning, emergency measures and their effectiveness, time of flooding, flood duration, water depth, flow velocity, contamination and erosion. • Twelve questions about flood damage such as structural damage to building, content items damaged, monetary damage to building and contents, labour and costs needed for repairing, cleaning efforts, damage to pets and agricultural products and compensation. • Eleven specific questions for small businesses about damage to equipment, goods and products, disruption of business, costs of cleaning and reconstruction and sale decrease. • Seven questions about flood mitigation behaviour such as mitigation measures undertaken at the building level (information measures, adaptation measures, structural measures) and factors possibly influencing these activities such as risk perceptions or socio-economic characteristics. • Three questions about governmental flood risk management. • Four questions about the housing situation including the year of construction, building value, ownership and the duration of residence. • Four questions about the contents such as pets, vehicles, appliances and stored chemicals. • Nine questions about the interviewee and household characteristics such as name, gender, age, education, number of household members, children, old people and income. • Seven specific questions for small businesses about their profile: opening year, size of business, average monthly sale, sources of funding, investment, value of products and equipment. In addition to the answers collected through the interviews, the interviewers also collected and validated information themselves, for instance regarding house characteristics such as floor space, construction materials or floor elevation. The value of the houses was estimated on basis of this information and the ‘Regulation of prices for houses, works, and structural compensation in Can Tho city, Decision 12/2010-QĐUBND’ (Can Tho City People’s Committee, 2010b), which provides prices per square metre categorised for different construction materials. Additionally, the interviewee was asked about the value of the house, which was used to cross check the estimated value. Prior to the main survey, the questionnaire was pre-tested in November 2011 with 40 respondents from the different districts in Can Tho. Based on the experience from the pre-test, parts of the questionnaire were improved. For instance, in the pre-test, direct flood damage and repair costs were elicited as a single cost item. However,


the pre-test revealed that many respondents had not yet, or only partly, repaired the damaged assets at the time of the interviews. Therefore, the two cost items were elicited separately from each other in the main survey. The interviewee was asked about all single parts of the building and pieces of contents that were damaged by the flood as well as about an estimate of the building and contents damage. Common marked prices were associated with all damaged items and summed up for the estimate of building and contents damage. These estimates were additionally cross checked with information provided by households whose members had already repaired or replaced similar damaged items. Following this procedure, we tried to minimise uncertainty in damage estimates, especially in the cases where total cost of repair or replacement were not clear at the time of the interview. In the results section mean and median values of damage data are provided due to high skewness of damage data, which usually exhibit large outliers. As outliers have a large influence on mean values, we additionally also provide median values. The main survey was then administered in January and February 2012 using faceto-face interviews, which were carried out by the first author who was supported by eight experienced Vietnamese interviewers, both male and female. The first author had trained the interviewers and supervised them closely during the survey: she accompanied the interviewers every day, introduced them to the interviewees and randomly cross checked the collected information after the interviews. Interviews Figure 3. Study area and spatial location of the sampled households and small businesses

Source: generated by the authors from spatial location of the sampled households and small businesses undertaken by Do Chinh, mapping data of the Southern Institute for Water Resources Research Vietnam and satellite imagery from the Global Land Cover Facility database.4


were undertaken during the day between 8 am and 6 pm. Flood prone households that suffered damage during the 2011 flood event were identified through the first author’s own observations during the peak of the flood season in September 2011 in Can Tho, which were complemented by expert interviews with members of the Districts’ People’s Committees. Only those households and small businesses reporting damage during the 2011 flood event were included in the survey. In the case of households this included damage to content or building structure; in the case of businesses it also included losses due to the disruption of production processes. The spatial locations of the sampled households and small businesses are presented in Figure 3. A detailed picture of the sample characteristics is provided in Table 1. It shows that the sample is biased towards women, which can be explained by the fact that many men leave the house during the day for work, while women more often stay at home to do housework and run the small business. Additionally, migratory labour (often men) is a widespread phenomenon in Vietnam. According to estimates of the Asia Foundation, there are about 26 million migrant workers in Vietnam. Nearly one third of the population of Ho Chi Minh city, which is the economic centre of Vietnam adjacent to the Mekong Delta, is comprised of migrant workers in search of better employment opportunities (Taylor, 2011). The mean age of the respondents is 51 for households and 49 for shop owners. From the people interviewed in the households, 55 per cent never went to school or completed only primary school, whereas this share was only 37 per cent for the business people. In terms of income, Table 1 shows that the most frequent (mode) income category of all respondents ranges from USD 74 to 185 monthly. However, 30 per cent of the households and about 14 per cent of the businesses have an income below this range. The large majority of the respondents owned the house or the shop they live and work in, with a slightly higher share of owners for households as opposed to businesses. More than 30 per cent of the households live in wooden buildings, whereas only 16 per cent of the businesses are located in such building types. This difference in building material between residential and business buildings is also reflected by differences in building values. Additionally, private households are generally affected by slightly higher water depths and longer flood durations. More detailed information on the surveyed businesses is provided in Table 2. The 378 surveyed businesses were grouped into five different types of business branches in line with the categorisation used by the Statistical Office of Can Tho city (2012), namely: grocery shops (118 interviews), eating and drinking shops (99 interviews), home appliance shops (56 interviews), services such as hairdressers or copy shops (72 interviews) and production facilities such as handcrafts or noodle factories (33 interviews). A large majority of the businesses (82 per cent) employ only one or two staff members and can thus be considered as small businesses. Average monthly sales for all businesses are about USD 190. The highest monthly turnover was reported by the home appliance group with USD 475, while production facilities reported the lowest turnover with about USD 143. The home appliance group also has the highest value of equipment and goods and products, which can be explained by the fact that this group sells the most expensive goods.


Table 1. Sample characteristics of the interviewed households and small businesses Variable

Percentage of households (n=480 respondents)

Percentage of businesses (n=378 respondents)

Gender Male Female

32.5 67.5

35.2 64.8

Age 18–25 26–35 36–45 46–55 56–65 66–75 >75

3.7 9 18.8 31.9 24.8 8.1 3.7

3.7 11.9 24.4 26.8 22.5 8.8 1.9

Education Never went to school Primary school completed Secondary school completed High school completed Vocational University and higher

11.1 44.2 27.3 11.1 1.7 4.6

5.6 31.6 32.5 23.6 1.9 4.8

Monthly income per household (USD)