Madagascar

Madagascar: Comprehensive Food Security and Vulnerability Analysis (CFSVA) - Conducted in Aug-Sep 2005

Attachments

Executive Summary

The Republic of Madagascar is an island nation in the Indian Ocean, off the eastern coast of Africa. It is the fourth largest island in the world and is home to 5% of the world's plant and animal species, 80% of them are unique to Madagascar. The country has a population of about 18 million people (2005) and an area of about 587,000 square kilometres.

Madagascar is prone to natural disasters, particularly cyclones and droughts. Over the past 35 years, at least 46 natural disasters, including cyclones, droughts, epidemics, floods, famines and locust infestations have been reported, which have cumulatively affected more than 11 million people. In 2004, approximately 72% of the population were living below the poverty line of 1 USD per day (2004 EPM). Eighty five percent of the poor in Madagascar live in rural areas. Labour migration is common everywhere, but is more important in the highlands than in the lowlands.

The country is classified as a low-income food deficit nation - the 2005 UNDP Human Development Report ranked Madagascar 146th of 177 countries. Agriculture (farming, livestock rearing, fishing and forestry) is the mainstay of the economy. Rice is the most important crop, followed by cassava, sweet potato and maize.

Currently, in terms of grade repetition, dropout rates, and other indicators (cited in World Bank, 2002), primary schooling in Madagascar rates poorly both in absolute terms and in comparison to other countries in the region.

Madagascar has a poorly developed transport infrastructure which constitutes a major constraint to strong economic growth which can lead to the reduction of poverty and food insecurity. Road access is a major problem throughout the country - there are about 50,000 kilometres of roadways, of which only about 6000 km are paved (1999 estimates).

Chronic malnutrition in children, resulting in stunting, is an indication of long-term undernutrition and poor consumption. In Madagascar, 45% of children are stunted at 24 months of age. Stunting is more prevalent in rural areas (46%) than in urban areas (39%).

Background

WFP Madagascar, with support from the Vulnerability Analysis and Mapping (VAM) staff from WFP Johannesburg, Maputo, and Rome, designed and implemented a Comprehensive Food Security and Vulnerability Analysis (CFSVA) in rural Madagascar.

Planning for the survey began in May 2005 with a literature review and secondary data analysis. Survey design and sampling took place in July/August and the training of enumerators and the field-testing of questionnaires was conducted mid-August. Data collection took place in August/September 2005 using Personal Digital Assistants (PDAs) for the household survey, which allowed the enumerators to collect and enter data simultaneously into a database. The data analysis began in October 2005 with the final report submitted in March 2006. The findings will serve as an important knowledge base for establishing a countrywide food security monitoring system. To enhance synergies with other agencies that also collect data on food security, the present database has been shared with the EU-funded Rural Information and Food Security System (SIRSA - Système d'Information Rurale et de Sécurité Alimentaire).

The primary objective of the 2,200 household survey was to obtain a better understanding of food insecurity and vulnerability among rural households in a non-emergency setting at sub-regional levels throughout the country. The findings serve as pre-crisis baseline information against which to measure the effects of a future shock such as a cyclone or drought. In particular, the following questions must be answered:

- Who are the hungry poor and vulnerable?

- Where do they live?

- What are the underlying causes of food insecurity and vulnerability?

- How can food aid make a difference?

Coverage and methodology

The Country Office with the support of VAM/HQ and VAM/ODJ wanted to include a health and nutrition component in the household survey. The survey was designed to draw samples of rural households at a sub-regional level. In order to achieve this, spatial analysis and principal component and cluster analyses were used to create clusters of districts that were homogeneous in terms of selected socio-demographic characteristics, risk, elevation, length of growing period, land cover and population density. From each of the 9 rural district clusters, a two-stage probability sampling method was used to select villages and households with a sample size calculated to provide an estimate of food insecurity with 90% confidence.

In total, more than 2,200 households in more than 220 rural communities in 97 districts and 22 regions were surveyed across the country. From this sample, health and nutrition information was collected for nearly 1900 women of reproductive age (15-49 years) and nearly 1,500 children aged 0-59 months. In addition, community interviews were conducted in all sample villages.

Summary findings

Overall, the CFSVA survey has identified food insecure households experiencing a problem of food availability, access and/or utilization. In addition, the analysis also identifies vulnerable households that are at risk of becoming food insecure. Their vulnerability is dependent upon their exposure to risk factors (such as natural disasters) and their ability to manage or cope with these shocks.

The highest percentage of food insecure households can be found in Cluster 9 (South), followed by Clusters 2 (Western inland area), 6 (South-East / North East littoral area) and 8 (South East). The proportion of vulnerable, but not food insecure households is highest in Clusters 1 (Western littoral area), 2 (Western inland area), and 8 (South-East), followed by Clusters 4, 6, and 7. Combining food insecure and vulnerable households, Cluster 9 (South) is in the most precarious situation, followed by Clusters 2 (Western inland area) and 8 (South-East).

In determining the food security status of the rural population, exposure to natural shocks must be considered. Almost all sample villages have been affected by natural shocks, such as cyclones, floods and drought with the worst off villages being found in Clusters 2, 6, 8, and 9. More than 70% of households in Clusters 2 and 6 and nearly all in Cluster 9 experienced at least one shock. The most often reported shock was cyclone in Clusters 1- 5, floods in Clusters 6 and 8, drought in Clusters 7 and 9 and, additionally, a high level of crop diseases in Cluster 7. The most important economic shock was the increase in rice prices that had an especially strong impact on Clusters 3, 4, 6, and 7.

Strategies used to cope with all kinds of shocks consist primarily in reducing the quantity of food and the number of meals consumed per day. In the case of natural shocks, these two major coping mechanisms are supplemented by an over-reliance on forest products.

The latter strategy is especially pronounced in Clusters 1, 2, and 9. The sale of cattle formed part of the coping strategy in Cluster 9 where many of the rural households rely on livestock rearing as a livelihood activity. Taking up temporary wage labour was available as a coping strategy only to Cluster 1.

More than 80% of the households in Clusters 1,2,6,8, and 9 had bought food on credit in the previous six months and more than four-fifths of these households were still in debt at the time of the survey. The highest level of current debt was found in Cluster 1.

Household food insecurity and individual nutritional outcomes are mainly the result of exposure to recurrent disasters, poor infrastructural development, and a low level of diversity in livelihood strategies. Thus the results of the CFSVA can also be of significance for future strategies of poverty reduction, since they also point to the need for increased engagement in the areas of Education, Health, Infrastructure, and Rural Development. No doubt, there are other challenges as well such as the rising expenses to cover basic needs and the concurrent decline in purchase power or the growing level of perceived public insecurity.

Education

The lowest levels of literacy of household heads were found in Clusters 7 and 8 (around 55%) and Cluster 9 (34%) and the highest in Cluster 4 (90%) households. Literacy among spouses ranges from as low as 22% in Cluster 9 up to 77% in Clusters 4 and 5. Women's educational level was also rather low in Cluster 2 where nearly 60% of the women had never attended school at all.

Primary school enrolment was lowest in sample villages in Cluster 6 (46%) and Cluster 9 (26%), while absenteeism was most common in Clusters 1 and 2 (in 40% of the sampled households with school-aged children). School enrolment and the quality of education are also constrained by the absenteeism of teachers and a school infrastructure as well as the limited accessibility of villages. Around half of the villages in the sample have a primary school, but there is none in nearly two-thirds of the sampled villages in Cluster 9 and in half of the villages in Clusters 6 and 2. The next primary school is more than one hour's walking distance from 20% of the villages in Cluster 2 and from 40% of the villages in Cluster 9. Overcrowded classes are most common in Cluster 9 and 1 samples.

Half of the villages reported that school fees restrained school attendance. The main reason cited for leaving school was in most cases the parents' inability to buy the basic school items (Clusters 3, 6, and 7) and the families' need of the labour of the children (Clusters 2, 3, and 8). In half of all sampled villages the lack of school infrastructure, the absence of teachers, and limited levels at school account, to a large extent, for early school leaving. In Clusters 5 and 6, parents' lack of interest and thus the need for sensitisation were also mentioned.

Health and nutrition

Despite the fact that treatable health problems such as malaria, diarrhoea, respiratory infections and TB are very common, few villages have a local health centre. There is a dispensary (CSBII) in only 15% of the villages in Clusters 1 and 4. One-third of the villages in Clusters 3, 4, 5, and 6 rely on a centre that can be reached in less than one hour. The costs of even basic health care are often a limiting factor.

Half of the sampled villages in Clusters 3, 4, and 9 had access to a nutritional centre to be reached in less than 3 hours. In Cluster 1 and 2, people in most of the villages with no local nutrition centre had either never heard of one or knew of the existence of a nutrition centre that took 6 hours to reach. .

Women in Clusters 5 and 6 reported an average of three pregnancies and live births as compared to four in the other clusters. Stillbirths were most frequent in Clusters 5 and 9 (31% of the sample women), while the lowest was found in Cluster 4 (17%). The highest percentage of children described as being 'very small' or 'smaller than normal' at birth was found in Clusters 1 and 9 (35%), followed by Cluster 7 (33%). The likelihood of being low birth weight was lowest in the Cluster 5 sample (13%). The rate of reported child deaths was highest in Cluster 8 (44%) and Cluster 2 (40%) and lowest among women in Clusters 5 (21%) and 4 (23%).

Women in Clusters 4 and 5 had the highest mean body-mass index (BMI) of the sample while those in Clusters 8 and 2 had the lowest. Nearly half the women in the Cluster 7 sample were found to be underweight (< 45 kgs), followed by nearly 40% in Cluster 8. Nearly 20% of the women in the Cluster 7 sample were stunted (< 145 cms). This rate was 15% in the Cluster 6 sample. Overall, it appears that women in Cluster 7 are the worst off in terms of nutritional outcomes while those in Cluster 9 show the lowest prevalence of underweight and stunting (none) in adult women.

By district cluster, the highest two-week-period prevalence of diarrhoea in women was found in Cluster 1 (21%), followed by Cluster 7 (16%), while the lowest was found in Cluster 4 (7%) and Cluster 3 (8%). However, the prevalence of fever was 31% in Clusters 7 and 9, followed by Cluster 1 (30%) and Cluster 2 (28%) and lowest in Cluster 3 (13%). More than 10% of the women in Cluster 1 and Cluster 7 reported suffering from both illnesses in the 2 weeks prior to the survey.

The prevalence of acute child malnutrition was highest in Cluster 8 (10.8%), followed by Cluster 5 (10.6%) and Cluster 2 (9.9%). Wasting was lowest in Cluster 1 (4.3%) and Cluster 9 (5.8%). The highest prevalence of underweight in pre-school children was found in Cluster 7 where more than 45% of the children in the sample had low weight for their age. More than 40% of the children in Clusters 4 and 8 were also underweight. The prevalence of underweight children was lowest in Clusters 1, 2, 5, and 9. The greatest problems in chronic child malnutrition lie along a ridge to the east of the centre of the island. In Cluster 4, nearly two-thirds of the sampled children were stunted, followed by half the children in Cluster 3 and 6 samples. The prevalence of chronic malnutrition was lowest in Clusters 1 and 2 where only around 30% of the children were malnourished.

The prevalence of severe underweight was highest in Cluster 8 (12%), followed by Cluster 4 (11%) and lowest in Clusters 5 and 9 (5%). Around 20% of the children in the sample were severely stunted with the highest prevalence found in Cluster 4 (32%), followed by Clusters 8 (26%) and 7 (25%).

The highest disease prevalence in young children is found mostly in the western part of the country. More than 30% of the sample children in Clusters 1, 2, and 8 had experienced non-specific fever in the 2 weeks prior to the survey. The prevalence of acute respiratory infection (ARI) was highest in Clusters 8 (24%) and 1 (20%) and less than 10% in Clusters 3 and 4. The 2-week period prevalence of diarrhoea was also highest in the children in Clusters 1 and 2 (21%), and 7 (20%).

Infrastructure

Most villages in the sample have no direct access to public transportation. Less than onequarter of the villages is connected to the bus network in Clusters 3, 6, 7, and 8, and only 5% of the villages in the Cluster 9 sample. It takes sometimes as much as 6 hours to reach a bus line from most villages in Clusters 1, 2, 6, 7, and 9, where no bus passes through the village. While bus roads are generally served throughout the year, accessibility is difficult for to 5 months in Clusters 7, 2, and 1 and for as many as 7 months on average from the sampled villages in Cluster 4.

Less than 30% of the villages in the sample host a market. The walking distance to the nearest market from villages without a marketplace of their own varies between 1 to 3 hours for more than half of the villages in all Clusters. The worst disruptions in market supply have been experienced by the sampled villages in Clusters 8 and 9. Only about 10% of the sample households in Cluster 4 and 9 reported going to a market 4-7 times a week.

Village access to water is dependent on the proximity of rivers or lakes in most Clusters. About half of the villages in all Clusters experience difficulties with the water supply. This percentage is much lower in Cluster 2 (one-quarter) and highest (two-thirds) in Clusters 5 and 9. Households in Clusters 1, 5, 9 and especially 8 have the least access to drinking water from an improved source.

Rural development

Most villagers' main economic activity at the community level is the marketing of crops, then the sale of cattle (Cluster 9) and cash crops. To a lesser extent, handicrafts are of importance in Clusters 2 and 4.

Crop diversity is very low in Cluster 5 (mostly rice) and low in Clusters 1, 2 (mostly rice), and Cluster 6 (mostly cassava). The main harvest does not adequately provide food for the majority of households - supply is insufficient for more than 6 months in Cluster 9, less so in Clusters 2, 6, 7, and 8.

In terms of assets diversity, sample households Clusters 7 and 8 were the worst off. In Clusters 2 and 8, only 66% of the households owned farming equipment at all. Land ownership was limited in Cluster 5 where just over 70% of sample households had access to farming land. The average size of owned and/or cultivated plots was the lowest in Clusters 6 and 5. Only 19 % of households in Cluster 1 and 25% in Cluster 6 cultivated a vegetable garden.

The total monthly expenditure on food by the sampled households was the highest in Clusters 2 and 6 and relatively high in Clusters 1, 5, 7, and 8. Households in Clusters 1, 2, 5 and 6 were the most reliant on purchases for the food they consume while the least reliant on purchases were households in the Cluster 4 and 9 samples.

There is no community granary in the majority of the villages and even if there is a community storage facility, very few people make use of them. Mutual mistrust and the lack of leadership skills were cited to account for the lack of associations within the communities.

Villagers were asked which aspects of their lives they thought were most important to develop. In Clusters 1-4 they mentioned means to improve agricultural production, including seeds, fertilizers and insecticide as their immediate needs from a short-term perspective. Needs related to drinking water were considered crucial in Clusters 1 and 7- 9. Water management issues related to irrigation and drainage were mainly mentioned in Clusters 3 and 5. Needs related to education such as building schools, literacy initiatives, and the recruitment of teachers were seen as of key importance in Clusters 4, 6, and 7 and, to a lesser extent, in Clusters 1, 2, and 8. Interestingly, the problem of road infrastructure was seen as a secondary issue in most Clusters.

Role of food aid

The results of the CFSVA point to serious food security and vulnerability problems among households in Clusters 9, 2, 6 and 8, indicating the need for a close monitoring of the food security and nutritional situation is these areas. While some data is provided by the SAP (Système d'Alerte Précoce) and the SIRSA (Système d'Information Rurale et de Sécurité Alimentair ) in the South, sentinel sites could be established to extend the coverage of food security monitoring in the country.

Food assistance though school-feeding activities should be continued in the South, given the high level of household food insecurity related to the frequent exposure to shocks and the limited options for livelihood strategies in this area. This could have an impact not only on household food security, but constitute an investment in the future of rural households through improved learning as well. Another rationale supporting school feeding activities in this region is to compensate for the increased caloric intake of children due to the long walking distances to reach school. Nonetheless, the construction of additional schools remains an imperative. In case school feeding activities were to be extended, Cluster 6, 7 and 8 in the Province of Fianarantsoa should benefit, and a special consideration should be given to the Western inland area, given the high proportion of food insecure households combined with a low enrolment rate and a high percentage of absenteeism.

Food for Work/Food for Training programmes are to be continued combined with safety net strategies in areas vulnerable to natural disasters and recurrent drought. The development of a disaster preparedness plan within the national framework, including an efficient early warning system and eventual contingency plan is crucial, giving a special consideration to districts in Clusters 2, 6 and 9.In the fight against malnutrition a nationwide monitoring of the nutritional situation is necessary to ensure a timely provision of food aid where necessary. For vulnerable groups, such as women and children less than five years of age, fortified blended food aid can continue to play a significant role in improving their health and nutrition status. In many areas, however, the nutritional situation is an outcome of the poor utilization of food. In these areas sensitization on nutrition (dietary diversity, vitamin A and iron intake), neonatal health care, and the provision of basic care for young children is necessary. Training on crop diversification is another area to be pursued.

(pdf* format - 6.62 MB)