Download Chapter
Asian Development Bank

Chapter 14.5 Drought, food security and famine

Monitoring and warning systems

Photo: Asian Development Bank

There are many kinds of warning system for droughts and food shortages. They often combine hazard/meteorological monitoring, assessments of food production levels after the harvest season and other indicators of household stress, such as sales of livelihood assets. Warning systems have increasingly shifted their emphasis from the simple availability of food to considering which groups do not have access to food: this takes them logically into vulnerability and livelihoods analysis.

National and regional systems are best managed by governments and international organisations. They are also geared to large-scale disasters requiring international aid. However, local early warning systems also have an important role to play in monitoring impending food crises. They tend to draw on a wider range of indicators of food and livelihood insecurity than larger systems, rely far more on qualitative data and involve higher levels of community participation. They are better able to take account of local variations in food security and more sensitive to local coping strategies and vulnerability. They can recommend appropriate interventions to local decision-makers, who will have a better understanding of conditions on the ground and a greater sense of urgency in responding to problems. They are easier to manage than large-scale, centralised systems, but they also tend to suffer from a lack of skilled personnel and are more open to manipulation by local interests.

Drought monitoring systems look at two main indicators: rainfall and vegetation. The former is monitored by rain gauges, the latter mostly through remote sensing by satellites. In both cases, large amounts of high-quality data can be generated. These are supplemented by other meteorological data (such as rainfall forecasts) and hydrological data (such as monitoring of groundwater supplies and the level of water tables). Although rain gauges are relatively simple technologically, and collecting data often depends on local staff or volunteers, rainfall monitoring needs to take place on a large scale if it is to be of value in assessing overall needs and priorities. Management of such systems is generally taken on by government meteorological services and feeds into the well-established and effective national and international meteorological information systems. There is also growing use of telemetric rain gauges that transmit data automatically to distant monitoring stations, using radio signals or other electronic means of communication. The cost of procuring and analysing satellite data on vegetation cover is so high that this task too is generally left to international and government services.

Monitoring of rainfall and vegetation does not tell us how much food people have or need. Remote sensing does not distinguish between different kinds of vegetation, so it does not show how well crops are growing. Since different food crops vary in their levels of tolerance to drought, rainfall monitoring is of limited value as an indicator of the availability of food. It is for this reason that food security or famine information and warning systems have become an increasingly important tool for disaster managers.

Early-warning systems take many forms in terms of their institutional set-up and location, the resources available to run them and the information that they collect and process. However, all are designed to stimulate action by informing decision-makers about food security conditions and people’s needs. An efficient, effective early warning system for drought-related famine should have the following characteristics:+M. Buchanan-Smith and S. Davies, Famine Early Warning and Response – The Missing Link (London: IT Publications, 1995).

  • It should be capable of warning of large-scale famine, sensitive to changes in food security status before famine threatens and able to detect localised pockets of acute food stress.
  • It should generate a response that provides assistance at an early stage, before families and communities are reduced to destitution.
  • It should stimulate interventions that protect livelihoods before lives are threatened. This implies providing a wider range of relief than food aid, as well as a more developmental approach.

Figure 14.3 An early warning system as information system

ODI 2015

An early warning system as information system

Diagram from M. Buchanan-Smith and S. Davies, Famine Early Warning and Response – The Missing Link (London: IT Publications, 1995), p.14.

The effectiveness of early warning systems varies. There are four main reasons for this:

  1. The nature of the system and the information provided – the range of indicators used, the accuracy of the data, the timeliness of warnings.
  2. The institutional context within which the system is located, and its institutional links to decision-makers.
  3. The broader political environment. Decisions about when and how to intervene are political and therefore influenced by many other factors.
  4. Logistical obstacles to launching a timely and adequate response.

Research in a number of food crises has shown that reasons 2 and 3 are the most important in explaining if and how early warning information is used, together with variations in performance between different warning systems. Early warning systems do sound the alarm about impending food crises, but response systems often fail to act early enough.

14.5.1 Data and indicators

Food security information and famine early warning systems incorporate a wide range of indicators of the availability of food and ability to procure it. This includes data on the market price of food and other essential goods, family and community behaviour (the adoption of particular coping strategies) and employment opportunities, as well as more conventional data on rainfall and levels of groundwater, crop production (surveys before and after the harvest), nutritional status and food supplies. However, many warning systems rely too much on indicators of food production and supply, and too little on indicators of access to food, particularly purchasing power. Socio-economic indicators, which can be harder to collect and draw more on qualitative data, are less influential in overall decision-making. Coping strategies, which are difficult to monitor and interpret accurately, are also difficult to incorporate systematically into early warning systems. Officials are more likely to be impressed by conventional quantitative monitoring than community-based systems and local knowledge.

Multi-indicator systems are sensitive to the complexity of famine processes and therefore more likely to detect worsening food security early enough for interventions to protect livelihood assets and prevent starvation. However, information does not speak for itself: data have to be interpreted, and different types of data are not easily compared. For example, how does one weigh up the relative significance of data on grain prices in local markets compared to levels of rainfall or farm crop production, or sales of livelihood assets? To add to the problem, most systems depend to some extent on proxy indicators of food stress (e.g. the timing and extent of adoption of particular coping strategies). Different methods are needed to collect different kinds of information, each requiring its own skills. Formal measuring systems can be used for some aspects of food security, such as crop production and food prices. Monitoring of nutritional status has its own methods, and assessments of wider household food security status require skills in interviewing and participatory appraisal. All of these skills can be learnt and transferred, but this takes time and specialist assistance will be needed as it is unlikely that any one local organisation or project team will have all of the relevant skills in-house. Rapid staff turnover often prevents skills from becoming fixed within an organisation.

Relevant information may have been collected by other people and for other purposes (e.g. a Ministry of Agriculture will collect agricultural production data). It will have to be obtained from those users, which may not be easy, especially in countries with very bureaucratic administrations. Information from other sources may have been collected or aggregated on a different basis from that of the local monitoring system. For example, government officials prefer to use administrative areas as their units of analysis and their sampling methods may not take account of geographical or social differentiation within those areas, but food insecurity can be very localised and unevenly distributed. Disaggregation of data by age, sex or occupation is likely to vary between different data sets, as will the timing and frequency of data collection. Warning systems tend to overlook the value of community-based data collection, drawing mostly on local technical services: communities and even local governments are often left out of the process. Information can be gathered for its own sake, without sufficient thought being given to what field agencies need to know. Systems may be unable to process all the information they collect, especially where they use a wide range of data sources and indicators.

At the end of this process, information has to be packaged in a way that is intelligible to decision-makers and that helps to guide them towards appropriate action. This link to action must be kept firmly in mind when planning and running early warning systems. The system may have to supply information to a wide variety of users, ranging from government policy-makers to field managers. Each group may want different kinds of data, which may have to be presented in different ways.

14.5.2 Assessment tools

A number of tools can be used to assess food insecurity and warn of impending crises. Vulnerability and capacity assessments (see Chapter 3) are widely used by agencies working in drought-prone areas. The Household Economy Approach (HEA), which collects and analyses food security information, has been used extensively since it was first developed in the early 1990s (see Case Study 14.5: Monitoring household food security). A number of other approaches are available, including a food security and resilience ‘screening tool’ developed by the International Institute for Sustainable Development.+See http://www.iisd.org/cristaltool/download.aspx#cristal-food-security; IISD, Climate Resilience and Food Security: A Framework for Planning and Monitoring (Winnipeg: International Institute for Sustainable Development, 2013), http://www.iisd.org/pdf/2013/adaptation_CREFSCA.pdf.

Where appropriate, data should be exchanged between vulnerability/food security assessments and early warning systems, to avoid duplication of effort. However, there is no single measure of food security that is valid and reliable, comparable over time and space, and which captures its diverse elements. In practice, many different approaches and means of measurement are used, focusing on different aspects. Relying on just one measure of analysis in project design risks missing important dimensions of food insecurity, but agencies find it difficult to know which indicators are best suited to particular situations.

Case Study 14.5 Monitoring household food security

In the early 1990s Save the Children developed the Household Economy Approach (HEA) to analyse household food security. The approach has been taken up by many other agencies. The best known application of HEA is to emergencies, for instance in early warning systems, needs assessments and scenario planning, but it is also widely used in poverty reduction to guide development plans and assess the impacts of social protection, market and other livelihood interventions.

Central to the HEA is understanding households’ livelihoods and everyday circumstances as an essential part of predicting how they will react in a crisis. Baseline data collection provides information on how households normally obtain food and cash income, their connections with the market and social or kinship networks, their assets (land, food stocks, livestock, cash, goods, tools) and expenditure patterns. Data collection is based mainly on fieldwork using participatory techniques that involve community members, individually and in groups, though a very broad range of assessment tools can be used.

The next steps are to identify potential problems – hazard threats and changes in agricultural, economic or security conditions – that could affect access to food, and to develop scenarios showing what the impact of such changes would be. For example, the impact of reduced crop production, milk yields and income from livestock sales or wage labour can be translated into an estimate of the likely impact on food availability. Similarly, the potential role of various coping strategies can be estimated.

P. Holzmann et al., The Household Economy Approach: A Guide for Programme Planners and Policy Makers (London: Save the Children, 2008), https://www.savethechildren.org.uk/sites/default/files/images/HEA_Guide.pdf.

14.5.3 Maintaining local systems

Food security information systems are complex and difficult to manage. They can also be costly because of the considerable staff time required to collect and interpret data. This is true even with participatory data-gathering methods that involve community members, because the information still has to be drawn together from different sources, analysed and then packaged for decision-makers and field workers. As data often have to be gathered from people dispersed over a wide geographical area, transport and subsistence costs can also be high.

Systems need to be maintained continuously to give reliable data of patterns of food supply and demand over time. A secure funding stream is therefore needed. Lack of resources has damaged a number of government and NGO-run warning systems. The project-based approach that NGOs generally adopt is an insecure foundation for such work because of its fixed time spans and the difficulty of obtaining repeat funding from donor agencies.

Both national and local warning systems must be integrated into the institutions that manage them. Many systems are purpose-built and tend to stand alone. Those who set up early warning systems should plan their external links as carefully as their internal mechanisms. Local systems often feed into national ones, but unless decentralised data are available for all the areas at risk, this can distort decision-making by giving undue prominence to particular districts.

Other problems include the lack of integration between different agencies’ early warning systems, which hinders collective analysis and action. Failure to standardise data across systems is a major issue. Agencies fail to learn lessons from each other’s experiences and even from their own similar programmes elsewhere.