Elsevier

Food Policy

Volume 33, Issue 6, December 2008, Pages 533-540
Food Policy

Measuring food insecurity: Can an indicator based on localized coping behaviors be used to compare across contexts?

https://doi.org/10.1016/j.foodpol.2008.02.004Get rights and content

Abstract

The Coping Strategies Index (CSI) was developed as a context-specific indicator of food insecurity that counts up and weights coping behaviors at the household level. It has proven useful to operational humanitarian agencies and researchers in measuring localized food insecurity, but to date has not been useful to compare the relative severity of different crises and has therefore has not been particularly useful for geographic targeting or resource allocation. This paper analyzes data from 14 surveys in crisis-affected or chronically vulnerable countries in Sub-Saharan Africa that incorporated the context-specific CSI. The paper identifies a sub-set of individual coping behaviors common to all surveys, whose severity is regarded as broadly similar by households across these studies. Data from these studies were re-analyzed using a reduced index constructed from only these behaviors. Correlations of this new index with other known food security indicators are similar to those of the complete, context-specific CSI. This suggests the possibility that an indicator based on these common behaviors could be used to compare the types of food security crises analyzed here across different contexts – particularly in Sub-Saharan Africa – to improve geographic targeting and resource allocation, according to the severity of crises. This new, more comparative indicator can be generated with no loss to the context-specific nature of the original CSI, which has proven useful for assessment and monitoring purposes.

Introduction

Food security is an ambitious concept to define,1 much less measure. Measures of household food security are needed for many different applications in situations where households are chronically vulnerable due to deepening poverty, environmental and climatic shocks, rapid economic change, and conflict. Indicators may be used to predict crises (early warning), to understand shortfalls in access to adequate food (assessment), to allocate resources (targeting) or to track the impact of interventions (monitoring and evaluation). There have been efforts to improve the quality of analysis in some of these areas (WFP, 2005, FAO, 2006). The allocation of resources continues to be subject to many considerations, but an impartial analysis of the actual requirements of disaster-affected people is all too frequently not the basis on which resources are allocated (Darcy and Hoffman, 2003). This is in part due to conflicting or overriding donor objectives, but also to a poor evidence base on which to make impartial allocations across very different contexts.

Several efforts by humanitarian agencies have attempted to address the need for comparability. The Integrated Phase Classification tool, developed by the Food and Agriculture Organization (FAO, 2006), attempts to aggregate individual indicators into a comparative analysis of food security crises. The Needs Assessment Framework developed by the UN Office for the Coordination of Humanitarian Affairs (IASC, 2005) attempts a similar task for humanitarian crises more broadly. These kinds of analyses are needed to strengthen geographic targeting and the impartial allocation of assistance. Such analysis is important on a sub-national scale to governments, national disaster management agencies or national NGOs that respond to emergencies, and on a broader scale to donors and international agencies such as the World Food Programme or International Non-Governmental Organizations. To work well however, these comparative analyzes require indicators that both capture the nature and depth of localized crises, and permit comparability between or among crises. While better information alone is no guarantee that either adequate funding will be available or that resources will be impartially allocated, doubts about the quality of analysis was one of the major reasons for both these constraints (Darcy and Hoffman, 2003).

There are few universally valid indicators of food security that are applicable in crisis situations. Nutritional status, if properly measured, is widely accepted as comparable across different contexts. But while nutritional status can be one indicator of food security status, it may equally reflect elements of health status, care practices, water quality, and other determinants of nutrition (Young and Jaspars, 2006). Some analysts suggest that measuring actual food consumption at the household level by a 24-h recall should be the “gold standard” by which other food security indicators are measured (Hoddinott and Yohannes, 2002, Weismann et al., 2006). But while 24-h recall data accurately reflects current consumption status, it does not capture other elements of the complex notion of food security. And the methodology is far too time-consuming to be useful in the applications discussed above – early warning, assessment, targeting or monitoring – all of which are very time-sensitive.

The Coping Strategies Index (CSI) was originally developed as a rapid alternative to a 24-h consumption recall. In brief, the CSI is a relatively quick and simple indicator of household food security behavior that asks a single question: “What do you do when you do not have enough food, and do not have enough money to buy food?” (Maxwell, 1996). The answers, collected over 15 years of grounded field practice with the indicator, are a series of behaviors about how households manage or “cope” with a shortfall in food consumption. These are formulated into a simple numeric score reflecting the frequency and perceived severity of these coping behaviors. In its simplest form, monitoring changes in the CSI score indicates whether household food security status is declining or improving – the higher the score, the greater the coping, and hence the higher the level of food insecurity. The CSI measures behaviors that fall into several recognized categories: those that change dietary intake; those that increase, even by unsustainable means, the amount of food available at the household level; those that reduce the number of people to provide for; and those that ration food or manage the shortfall. The CSI has been applied particularly in monitoring change – particularly in monitoring the impact of interventions – and in assessment (C-SAFE, 2004, TANGO, 2004b).

Patterns of behavioral responses in relation to a food shortage have been documented previously by several researchers (Davies, 1996). Watts (1983) presented a sequence of options based on their reversibility and commitment of domestic resources. Modest dietary adjustments (such as eating less-preferred foods or reducing portion size), for example, are highly reversible strategies that do not jeopardize household assets. More extreme behaviors, such as sales of productive assets to purchase food, hold more long-term consequences for the household. As a food security situation worsens, households are more likely to employ strategies that are less reversible, and therefore represent a more severe form of coping and greater food insecurity (Corbett, 1988, Devereux, 1993).

Until recently, household level measures of food access have relied on proxy indicators such as food consumption (caloric intake), household income, productive assets, food storage and even under-five nutritional status, each of which are presumed to be either determinants or consequences of a particular household’s level of food security (Webb et al., 2006). Included in this category of indicators is the measure of dietary diversity as an indicator of the quality of food consumption and also as a proxy for quantity. In recent studies, dietary diversity and food frequency have proven to be among the most common and valid indicators of nutrient adequacy and/or energy intake (Weismann et al., 2006, Dewey et al., 2005, Hoddinott and Yohannes, 2002, Ruel, 2002). A second kind measure assesses various elements of food insecurity, through a series of questions about self-reported behaviors and attitudes (Webb et al., 2006, Coates et al., 2006a, Coates et al., 2006b). One such indicator is the Household Food Insecurity Access Scale (HFIAS) which identifies three key domains of household access to food: (1) perceptions of insufficient quantity of food; (2) perceptions of inadequate quality of food; and (3) anxiety or uncertainty about whether the food budget or supply is adequate to meet basic requirements (Swindale and Bilinsky, 2006).

The Coping Strategies Index is similar in many respects to other measures of food security but distinct in that it queries household behaviors directly, and factors in the severity of different behaviors. Given that no one “gold standard” indicator has emerged, particularly for use in humanitarian emergencies, different measures of food security are needed for triangulation or complementary analysis.

Attempts at developing and refining indicators of food access have revealed a number of critical considerations. First, food security is a “managed process” with predictable patterns – people can foresee a food access problem before it arises and thus begin to alter behavior long before an actual crisis hits a household (Christiaensen and Boisvert, 2000). Second, with respect to coping strategies, it must be noted that some strategies do not necessarily reflect the same severity of food insecurity, nor are they equally acceptable to vulnerable households in different cultures (Coates et al., 2006a, Coates et al., 2006b). To develop more broadly applicable measures of food security, adequate attention must be given to developing methods of translating or adapting measures from one context to another (Swindale and Bilinsky, 2006, Coates et al., 2006a, Coates et al., 2006b, Webb et al., 2006, Maxwell et al., 1999). And third, although some progress has been made, the search for more broadly applicable measures of food security continues.

Because the CSI was constructed on the basis of asking questions about locally applicable coping behaviors, it has proven useful as a context-specific measure of food insecurity, with application in situations where there is a premium on rapid data collection and analysis. Maxwell et al. (1999) tested the CSI against various benchmarks of food security, and found significant correlation with other indicators of food security including dietary intake (kilocalories per adult equivalent per day), per capita expenditure and the proportion of expenditure on food (food budget shares), and various anthropometric measures in one specific case. Since 2000, it has been widely applied in emergencies by national governments, the World Food Program, donors, and various non-governmental organizations engaged in emergency food security response, and the results have shown strong and significant correlation with many other benchmarks of food security in various settings (TANGO, 2004a, TANGO, 2004b, CARE/WFP, 2003; Eritrean Refugee and Refugee Commission, 2003; C-SAFE, 2004, Food for the Hungry, 2006). Because the CSI has proven useful as a context-specific indicator however, there has always been a caveat that findings were probably only locally applicable as well, and it has been criticized for being relatively unhelpful in comparative analysis (Kennedy, 2002).

However, it has been noted in surveys since 2000 that some of the individual behaviors that the CSI measures recur across different contexts, suggesting that a version of the CSI could be constructed that would permit a broader comparative analysis. With data from 14 different surveys in crisis-affected or chronically vulnerable countries in Sub-Saharan Africa, this paper addresses two questions:

  • 1.

    From these studies, which were all based on locally constructed sets of coping behaviors, can a sub-set of coping behaviors be identified that are applicable across all of the studies?

  • 2.

    If so, does a CSI indicator based solely on this subset of behaviors deliver similar results when compared to the various benchmarks examined in the original 1999 study as described above? In other words, would this “reduced” new version of the CSI measure food insecurity as well as the original version?

If so, the implication would be that a CSI indicator could be developed that would facilitate comparisons across contexts and locations, and thus contribute to geographic targeting and resource allocation, in addition to the already-accepted applications of the CSI.

Section snippets

Methods

Data from 14 studies that incorporated the CSI were compared to determine which individual coping behaviors were common to all; and for those that were common, whether they were similar in terms of frequency of occurrence and perceived severity. To investigate this, correlations were tested between the CSI and other food security or livelihoods indicators in the same data set using the “full” or locally specific CSI. Additional correlations were run using the “reduced” or common set of

Results

Results of reviewing individual surveys revealed that there was indeed a core set of behaviors that were common across all the studies and that seemed to follow a relatively similar order of frequency and severity. Table 2 summarizes the eleven most commonly mentioned individual behaviors (out of about 25 mentioned in the various studies), and ranks them in order of frequency of individual behaviors reported. As Table 2 makes clear, across all the studies re-analyzed, very nearly the same set

Conclusions and recommendations for usage of methodology

Several conclusions can be reached from this analysis. First, there is a core set of behaviors that are relied on across a variety of contexts, in more or less the same order of frequency, and each is perceived as having relatively the same severity. This enabled a test of how well a “reduced” CSI would perform compared to the original. Second, in general terms, the “reduced” CSI reflects food insecurity nearly as well as the “full” or context-specific CSI. This suggests that cross-contextual

Acknowledgements

The authors gratefully acknowledge data sets from the World Food Programme, the International Food Policy Research Institute, CARE International, and the C-SAFE consortium; the research assistance of Tom Spangler; and funding from CARE and from the Bill and Melinda Gates Foundation. The authors are also grateful for feedback from Patrick Webb, Jennifer Coates, Gregory Collins and three blind reviewers.

References (28)

  • J. Corbett

    Famine and household coping strategies

    World Development

    (1988)
  • D. Maxwell et al.

    Alternative food security indicators: revisiting the frequency and severity of ‘coping strategies

    Food Policy

    (1999)
  • D. Maxwell

    Measuring food insecurity: the frequency and severity of coping strategies

    Food Policy

    (1996)
  • CARE and World Food Programme, 2003. The Coping Strategies Index: A Tool for Rapidly Monitoring Food Security in...
  • L. Christiaensen et al.

    On Measuring Household Food Vulnerability: Case Evidence from Northern Mali

    (2000)
  • J. Coates et al.

    Commonalities in the experience of household food insecurity across cultures: what are measures missing?

    Supplement to the Journal of Nutrition

    (2006)
  • Coates, J., Swindale, A., Bilinsky, P., 2006. Household Food Insecurity Access Scale (HFIAS) for Measurement of...
  • Consortium for Southern African Food Security Emergency (C-SAFE), 2004. Malawi Final Survey: Report of Findings....
  • Darcy, J., Hoffman, C., 2003. According to need? Needs assessment and decision-making in the humanitarian sector....
  • S. Davies

    Adaptable Livelihoods. Coping with Food Insecurity in the Malian Sahel

    (1996)
  • S. Devereux

    Goats before ploughs: dilemmas of household response sequencing during food shortages

    IDS Bulletin

    (1993)
  • Dewey, K., Cohen, R., Arimond, M., Ruel, M., 2005. Developing and Validating Simple Indicators of Complementary Food...
  • Eritrean Refugee and Relief Commission (ERREC), CARE International and World Food Program, 2003. Eritrea Rural...
  • Food and Agriculture Organization, 2006. The Integrated Phase and Humanitarian Phase Classification Technical Manual....
  • Cited by (101)

    • Dealing with vulnerabilities in tribal food security—a study on Jhargram district of West Bengal, India

      2022, Indigenous People and Nature: Insights for Social, Ecological, and Technological Sustainability
    View all citing articles on Scopus
    View full text