How Much Does the Food Insecurity Experience Scale Overlap with Poor Food Consumption and Monetary Poverty? Evidence from West Africa

Abstract

The Food Insecurity Experience Scale (FIES), which combines three food-access dimensions into a single indicator, is rapidly being incorporated into national statistical systems. However, there is no prediction about how one of the incorporated dimensions—subjective experiences associated with food insecurity—overlaps with poor food consumption. Using data from West Africa, this study illustrates that in 4 out of 10 countries, there is a similar prevalence of food insecurity according to the FIES among segments of the population that are likely undernourished and segments that are likely not undernourished. And in 5 out of 10 countries, there is a relatively large prevalence of food insecurity according to the FIES in the segments of the population that are least likely to be undernourished. Combined, the results offer guidance to policymakers when choosing food-access indicators and illustrate the importance of using the FIES along with other food-access measures.

1. Introduction

The Food Insecurity Experience Scale (FIES), which combines eight yes-no questions on undernourishment, undernutrition, and subjective experiences associated with food insecurity into a single indicator, is increasingly being used by policymakers across the world.¹ The measure was adopted as an official indicator to track progress towards achieving the second Sustainable Development Goal (see, e.g., UN 2017); the measure has been included in the suite of indicators used by the United Nations and partner organizations to officially declare a famine (see, e.g., IPC 2021); and nearly 100 countries have either incorporated the FIES into their national statistical systems or are in the process of doing so as of 2019.² Furthermore, a growing amount of analytical work has used the FIES to understand how and why food access is changing, including the change in response to the COVID-19 pandemic (see, e.g., FAO 2020; United Nations 2020; World Bank 2020a; Adjognon, Bloem. and Sanhoh 2021; Amare et al. 2021; and others).

However, a growing literature on interpreting subjective well-being measures in general demonstrates that it is difficult to understand how the subjective experiences with food insecurity captured by the indicator are related to undernourishment and undernutrition. This may be because there are differences in individual-specific scales used to determine whether a person worries about food consumption or what is thought to be enough to eat, or because those individual-specific scales might have little to do with actual food consumption (see, e.g., OECD 2013; Ravallion 2013; Benjamin et al. 2021; Tandon 2023). Given this ambiguity, and uncertainty over how pivotal these subjective experiences are to the overall measure, this study empirically investigates the degree to which the FIES overlaps with likely undernourishment status in 10 West African countries.

In each setting, the analysis compares poor food access as measured by the FIES to poor food access using the Food Consumption Score (FCS), which is a widely used measure that has been demonstrated to be strongly associated with undernourishment (see, e.g., Wiesmann et al. 2009; Mathiassen 2013). And second, the study further compares the FIES to monetary poverty. Although monetary poverty and food access are distinct concepts, empirically supported assumptions in models of consumer theory predict that undernourishment is more prevalent in poorer households (see, e.g., Jensen and Miller 2010). Furthermore, monetary poverty is an important targeting criterion for many interventions that have food security goals (see, e.g., Coll-Black et al. 2011).

In Nigeria, which is the setting in which the study has access to the most data with which it can rule out alternative hypotheses for the empirical patterns, the analysis illustrates that there is a similar prevalence of food insecurity according to the FIES among segments of the population that are likely undernourished and segments that are likely not undernourished. The investigation further illustrates that there is a relatively large prevalence of food insecurity according to the FIES in the segments of the population that are least likely to be undernourished—those in the top FCS and expenditure deciles.

Importantly, significant heterogeneity is found in these empirical patterns. First, the more subjective FIES questions had similar shares of the population answering affirmatively at nearly all points of the FCS and expenditure distributions, while the rest of the FIES indicators that were less subjective had significantly lower affirmative responses at the higher ends of the FCS and expenditure distributions. And second, there were sharp regional differences in how the FIES is related to each comparison measure—both in the degree to which food insecurity according to the FIES was similar in the highest and lowest segments of the FCS and expenditure distributions, and the level of food insecurity at all points along each distribution.

However, in addition to performing this analysis in Nigeria, this article also compares the FIES to the FCS and monetary poverty in a set of harmonized nationally representative surveys conducted in every country in the West African Economic and Monetary Union (WAEMU) and Chad. These estimates illustrate that the empirical patterns in Nigeria are not specific only to that country. In four out of the nine additional countries in West Africa, the empirical patterns were similar to those of Nigeria. However, the estimates also illustrate that the empirical patterns in Nigeria are not universal. In the remaining five countries, the three indicators were more strongly aligned.

There are two primary contributions of these results. First, these results corroborate previous research demonstrating that widely used food-access metrics are capturing different dimensions of poor food access (see, e.g., Maxwell, Vaitla, and Coates 2014; Broussard and Tandon 2016). But these results extend previous results to an increasingly important food-access indicator—the FIES—and further extend the results to a large set of countries in which food security is a critical issue (see, e.g., FAO 2021).³

And second, these results provide guidance for policymakers and researchers when considering whether to incorporate the FIES in national statistical systems and analytical work and how best to use the measure. Specifically, these results corroborate previous guidance on the inclusion of subjective well-being measures in policy analysis. Researchers and policy organizations often advocate for including subjective indicators in policy analysis to capture important information that is missed by more objective and traditional measures, but they also recognize that it is important to triangulate between subjective and more objective well-being measures to precisely interpret the measures (see, e.g., Stiglitz, Sen, and Fitoussi 2009; OECD 2013; Krueger and Stone 2014). Although the FIES combines information from both subjective and more objective questions, the results here illustrate that additional information on food access and well-being may be needed to understand how the FIES overlaps with undernourishment and at which point in the welfare distribution food insecurity according to the FIES is most prevalent.

The rest of the article is structured as follows. Section 2 presents a brief background on the FIES; section 3 presents a brief background on the undernourishment proxies used in the analysis; Section 4 describes the data; section 5 describes the empirical strategy; section 6 reports the empirical results from Nigeria; section 7 reports heterogeneity in the empirical results from Nigeria; section 8 reports the empirical results from the nine additional West African countries; and section 9 concludes.

2. Background on the Food Insecurity Experience Scale

As discussed in the introduction, the FIES is an experiential food-access metric that incorporates information on anxiety over poor food access, food coping strategies, and household behaviors that are consistent with low levels of consumption and poor diet quality through eight yes-no questions. There are multiple versions of the FIES module. While the questions are nearly always identical, there is variation in which household members are covered by the questions and in the reference period used. Specifically, there is a FIES module that asks about any adult member of the household, a FIES module that asks only about the respondent, a FIES module asking about the 30 days before the survey, and a FIES module asking about the 12 months before the survey.

The eight FIES component questions are ordered in severity of the food-access problem, going from least to most severe, and with affirmative responses indicating a problem with food access. As discussed in the introduction, some of the questions ask about respondents' subjective experiences with food insecurity, such as whether individuals worried about not having enough food to eat or not eating as much as one thought they should; and some of the indicators are objective behaviors associated with undernourishment or undernutrition, such as skipping meals or going a whole day without food. However, each of the eight questions in the FIES module asks whether the behavior associated with poor food access was due to a lack of money or other resources, where people might skip meals for reasons that might not be associated with food insecurity.⁴

If certain conditions are met, the sum of the eight dummy variables that comprise the FIES module responses can be used to classify the food access of individuals or households as moderately or severely food insecure (see, e.g., FAO 2016).⁵ These individual- or household-level classifications can then be used to estimate the prevalence of food insecurity for the entire country, or regions within that country. However, country-level estimates for the prevalence of severe or moderate food insecurity according to the FIES can also be calculated by applying an Item Response Theory model (the Rasch Model), which assesses the suitability of each of the component questions and tries to make the scale more comparable across countries and contexts (see, e.g., Cafiero, Viviani, and Nord 2018a).

3. How Does the Food Insecurity Experience Scale Compare to Other Measures Associated With Undernourishment?

A policy-relevant question regarding the use of the FIES is the degree to which subjective experiences associated with food insecurity overlap with undernourishment and undernutrition, and the degree to which the subjective experiences associated with food insecurity drive the overall FIES measure. As described in the introduction, there is a large literature illustrating how subjective well-being metrics might not have a strong relationship with more objective measures. In particular, answers to subjective welfare questions depend on respondent-specific scales that: (1) may not be comparable across individuals or stable over time; (2) are potentially subject to frame-of-reference effects; and (3) suffer from measurement errors, over and above those affecting traditional welfare metrics (see, e.g., OECD 2013; Ravallion 2013; Benjamin et al. 2021). Thus, it is possible that subjective experiences associated with food insecurity might not overlap well with undernourishment and undernutrition.

To investigate the degree to which the FIES overlaps with undernourishment, the analysis compares the FIES to two measures associated with undernourishment. A strong overlap between the populations that are categorized as food insecure by the FIES and the populations that are most likely to be undernourished would be consistent with the subjective experiences associated with food insecurity overlapping with the other dimensions in the FIES.⁶ Furthermore, the comparison between the FIES and other measures that are widely used to target food security programs would be informative to policymakers when considering whether to use the FIES and potentially how best to use the measure.

First, the article compares the FIES to the Food Consumption Score (FCS), which is a weighted sum of the number of days in the past week that a household consumed foods from each of eight separate food groups.⁷ Validation studies have found that a sufficiently low FCS is strongly associated with undernourishment in a number of contexts (see, e.g., Wiesmann et al. 2009; Mathiassen 2013).

Second, the FIES is further compared to monetary poverty and total household expenditure. As discussed in the introduction, food access and monetary poverty are distinct concepts. In standard consumer theory, preferences over food and nonfood goods dictate the share of their income that households spend on food, and there are no firm predictions about how monetary poverty and food access might be related. However, when adding a penalty to consuming below one's minimum daily energy requirement (undernourishment) and making that penalty sufficiently severe relative to the marginal utility of consuming a more diversified bundle, undernourishment can be more prevalent among the poor (see, e.g., Jensen and Miller 2010).⁸

There are three streams of growing empirical support for the existence of this type of sharp penalty for undernourishment. First, the model described above is consistent with poorer households devoting a larger share of their total expenditure to food. This pattern—Engel's Law—emerges in empirical work from across the world (see, e.g., Kaus 2013). Second, there is an expanding body of evidence demonstrating that households tend to be especially averse to undernourishment (see, e.g., D'Souza and Jolliffe 2014), and that diet quality is much more responsive to income and price shocks than calorie consumption (see, e.g., Block et al. 2004; Brinkman et al. 2009). And, third, the addition of a sharp penalty on undernourishment offers an additional explanation to reconcile why traditional consumption models do a poor job of predicting how households respond to strong increases in the price of staple grains (see, e.g., D'Souza and Jolliffe 2012; Tandon 2014).⁹

4. Data

To better understand how the FIES overlaps with the FCS and monetary poverty, the study uses 11 nationally representative household surveys from 10 West African countries. Nigeria has two nationally representative household surveys that collect the FIES: the 2018/19 Nigerian Living Standards Survey (NLSS) and the 2018/19 General Household Survey (GHS). The NLSS surveyed approximately 22,000 households, and included both the FCS module and the detailed expenditure and consumption modules from which the official national poverty rate was estimated. Given that the NLSS is the survey used to report official poverty statistics, the analysis uses the survey in all the baseline empirical specifications.

Alternatively, the GHS surveyed approximately 5000 households and also included detailed expenditure and consumption modules necessary to estimate monetary poverty. Unlike the NLSS, the GHS captured the FIES in the same households at two separate points in time—once immediately after the planting season, and once immediately after the harvest season. However, the GHS did not include the FCS module.

In addition to the household surveys in Nigeria, the analysis also uses nine additional nationally representative household surveys that were harmonized across all countries in the West African Economic and Monetary Union (WAEMU) and Chad.¹⁰ The sample sizes of the surveys vary between 5,351 (Guinea-Bissau) and 12,992 (Côte d'Ivoire), and each of the surveys includes the FIES, the FCS, and the expenditure and consumption modules from which the official national poverty rates in each country were estimated.

In all surveys, the study follows FAO guidance on the construction of the FIES, which suggests that under certain conditions it is possible to use the “raw score”—the sum of the eight dummy variables that comprise the FIES module—to classify households’ food-access status. These conditions are met in all 11 nationally representative household surveys used here and, following other analysis on Sub-Saharan Africa (see, e.g., Wambogo 2018), the analysis classifiwa those households with a raw score of 7 or 8 as severely food insecure and those with a raw score of 4, 5, or 6 as moderately food insecure.¹¹, ¹², ¹³

Furthermore, in all surveys that include the FCS module, the study calculates the measure in the typical way. Each food group is given a score from 0 to 7, depending on the number of days out of the past seven on which it was consumed. The FCS is then a weighted sum of these components. Households are classified as having poor food access if they have a poor or borderline FCS, which is defined as less than or equal to 42 (WFP 2009).¹⁴

Lastly, in all surveys, the consumption aggregate used to identify monetary poverty included information on consumption of food and non-food items, and expenditures on health, education, housing, and meals consumed outside the home.¹⁵ In each country, the consumption aggregate is deflated spatially and temporally using unit values from the food consumption module.

The basic summary statistics for all surveys are presented in the section S3 in the Supplementary Online Appendix. Estimates for Nigeria are from the 2018/19 NLSS, and the estimates for the rest of West Africa are the average of the nine nationally representative estimates for each of the other West African countries. The summary statistics illustrate that both poverty and poor food access are widespread in all settings analyzed here. At the national poverty line, 40.1 percent of Nigerians live in poverty, and the average poverty rate at the national poverty line across the rest of the West African countries is 41.2 percent. Furthermore, severe food insecurity according to the FIES is 26 and 21 percent respectively in Nigeria and the rest of the West African countries analyzed here; and the share of the population with a poor or borderline FCS is 14.9 and 21 percent respectively in Nigeria and the rest of the West African countries analyzed.

5. Empirical Strategy

As discussed above, the study empirically investigates the prevalence of food insecurity according to the FIES in segments of the population that are very likely to be undernourished. Specifically, the analysis estimates how food insecurity according to the FIES varies across all FCS and expenditure deciles in the following linear probability model:

where Poor Access FIES is an indicator if the household i is categorized as severely food insecure by the FIES; and Decile^j_i denotes either the FCS or the expenditure decile of household i, with higher deciles corresponding to higher FCSs and higher expenditure. The estimate of β₀ represents the share of the population in the first decile (the lowest FCSs and expenditures) that is categorized as severely food insecure according to the FIES; and each β_j represents how much larger the share is in decile j than in the poorest decile.

The exact points in the FCS and expenditure distributions at which undernourishment begins to abate are difficult to precisely define, and the study does not necessarily expect food insecurity according to the FIES to decline for each decile (see, e.g., seventh versus eight FCS decile). But rather, as discussed in the background section, there are points in the FCS and expenditure distribution at which previous work has found a strong association with undernourishment. And if food insecurity according to the FIES is more prevalent among undernourished households, it would be expected that food insecurity according to the FIES would be higher at those points of the FCS and expenditure distribution than in significantly better-off points of the distribution.

Figure 1 reports the average FCS and average expenditure by decile. The figure illustrates that the bottom two deciles of the FCS distribution both have an average FCS that would qualify as poor or borderline food consumption, which is a target for emergency food assistance; and that the FCS rapidly increases for higher deciles, where the average FCS of the fifth decile is approximately 60 percent above the average in the second decile and far above any level used to target emergency food assistance. Similarly, the figure illustrates that the bottom four expenditure deciles have average expenditure below the national poverty line, and that expenditure also is increasing rapidly for higher deciles. Combined, the bottom end of each of the FCS and expenditure distributions are likely to have a significant share of households that are undernourished, and the prevalence of food insecurity according to the FIES in these deciles can be compared to rates found in higher deciles where undernourishment is less likely to be widespread.

6. Baseline Empirical Results

The estimates of specification (1) are reported in fig. 2 and illustrate that severe food insecurity according to the FIES is not only prevalent among households with a poor or borderline FCS or among the poorest households in Nigeria. The top panel reports estimates regressing the severely food insecure FIES indicator on the FCS decile indicators; and the bottom panel reports estimates regressing the indicator on the expenditure decile indicators. The figure illustrates several important patterns.¹⁶

First, the prevalence of severe food insecurity according to the FIES is very similar among FCS deciles that likely have a large share of undernourished households and FCS deciles where undernourishment is likely to be much less of a problem. It is not possible to reject the hypothesis that the coefficients on the second through the sixth decile are all equal at standard significance levels (p-value of 0.162). This is despite the fact, as demonstrated in fig. 1, that the average FCS in the second decile is at the threshold to define poor or borderline food consumption; and that the average in the sixth decile is 76 percent higher than that in the second decile and is far above any threshold typically used to define poor food access.

Second, the prevalence of severe food insecurity according to the FIES is very similar across nearly all expenditure deciles. Only the coefficient on the highest expenditure decile is statistically different from 0; and it is not possible to reject the hypothesis that the coefficients on deciles one through nine are jointly equal to 0 at standard significance levels (p-value of 0.508). The prevalence of severe food insecurity is so similar across the expenditure distribution that it is not possible to reject the hypothesis that severe food insecurity is equal in the poor and nonpoor populations at standard significance levels.¹⁷, ¹⁸

And third, both the estimates using the FCS deciles and using the expenditure deciles illustrate that there is a large share of the population that is severely food insecure according to the FIES in the deciles that are least likely to be undernourished. Specifically, 18.1 and 18.3 percent of the population are severely food insecure according to the FIES respectively in the highest FCS decile and the highest expenditure decile, which is approximately 70 percent of the national rate reported in the summary statistics.

These results illustrate that a significant share of the population is severely food insecure according to the FIES, despite having an FCS that is far above the threshold for poor or borderline consumption or expenditure that is far above the threshold for monetary poverty. Importantly, it is possible to rule out the possibility that the particular survey that is used for this analysis—the 2018/19 NLSS—might have had difficulty in properly implementing the FIES module and that the empirical patterns are due to this, as opposed to there being a high prevalence of severe food insecurity in segments of the population that are less likely to be undernourished.

Figure 4 illustrates a very similar empirical pattern using both rounds of a completely different survey conducted by the National Bureau of Statistics—the 2018/19 General Household Survey. Figure 4a reports estimates of the baseline specification for the postplanting round; and fig. 4b reports estimates for the postharvest round. In each set of estimates, the patterns are similar to the baseline estimates reported in fig. 2 using the NLSS. In both rounds, roughly equal shares of the population were severely food insecure according to the FIES in the 2nd through the 10th expenditure deciles;¹⁹ and the top expenditure decile has a severe food insecurity rate according to the FIES of 23.9 and 14.4 percent in the postplanting and postharvest periods, respectively.

Furthermore, it is possible to illustrate that measurement errors in food-access measures are likely not driving the empirical patterns. In all food-access measures, one would expect there to be some share of the population to be misclassified and these misclassifications would result in some instances where the FIES would overlap poorly with other food-access measures. However, the high rates of severe food insecurity according to the FIES in the top FCS and expenditure deciles (approximately 18 percent) likely cannot be explained by plausible assumptions about measurement error;²⁰ and explicitly accounting for measurement error in hypothesis testing would likely make it more difficult to reject the hypothesis that estimates of food insecurity in each decile are equal, which would potentially reinforce the lack of difference between deciles.²¹, ²²

And lastly, despite the fact that the analysis finds similar empirical patterns when using two separate comparison measures, it is possible that both the FCS and monetary poverty are incorrectly identifying the segment of the population that is most likely to be undernourished. However, the study demonstrates that this explanation is not likely.

First, the analysis corroborates that the share of expenditure devoted to food, itself a measure of poor food access, is decreasing for higher levels of expenditure (Engel's law). Specifically, the study re-estimates specification (1), but uses the share of expenditure devoted to food as a continuous dependent variable. The estimates are reported in section S7 of the Supplementary Online Appendix. As expected, the share of expenditure that is devoted to food is sharply decreasing for higher-expenditure deciles. The shares in the bottom two deciles are statistically indistinguishable, but the share becomes significantly smaller for each higher-expenditure decile. By the top deciles, the share of expenditure devoted to food is around 22 percentage points lower than in the bottom expenditure decile.

Second, it is found that the regions that have the highest prevalence of monetary poverty and the highest prevalence of poor or borderline food consumption using the FCS are corroborated by other food-access surveys. Figure 3 illustrates that both measures identify the north of Nigeria as being the poorest and having the highest share of the population that has poor or borderline food consumption. Importantly, a wide variety of other sources identify the north as having the worst food access (see, e.g., FRAYM 2020; IPC 2021b). By contrast, fig. 3 also illustrates that the FIES identifies the south of the country as having the highest prevalence of severe food insecurity.²³, ²⁴, ²⁵

7. Heterogeneity in the Baseline Empirical Patterns

The article further investigates two sources of heterogeneity in the baseline empirical patterns. First, the analysis investigates the possibility that more and less subjective FIES component questions might be more prevalent at different points of the FCS and expenditure distribution. Specifically, specification (1) is re-estimated, but using as the dependent variable the number of affirmative responses a household had in the two more subjective questions (worrying about not having enough food and eating less than they thought they should), and the number of affirmative responses a household had in the other six questions of the FIES module. The results are reported in fig. 5.

The results illustrate that the baseline empirical patterns are more evident in the more subjective questions than in the less subjective ones. Specifically, fig. 5a illustrates that there is a strong similarity between the number of affirmative answers to the subjective questions across the FCS distribution, both among deciles where there is a significant share of undernourished households and among deciles in which this is less likely to be the case. It is not possible to reject the hypothesis that the coefficients on the third through the eighth decile are equal at standard significance levels (p-value of 0.475). This is in stark contrast to the estimates using the responses to the less subjective questions in fig. 5b, where there is a significantly lower share of affirmative responses for higher FCS deciles.²⁶

Furthermore, the highest FCS decile has a higher share of affirmative responses to the more subjective questions than to the less subjective questions. Specifically, the highest FCS decile answers 53 percent of the more subjective questions affirmatively on average (1.07 affirmative responses out of 2 questions), while the average for the less subjective questions is 36 percent (2.18 affirmative responses out of 6 questions).²⁷

And second, the study also investigates the possibility that the baseline empirical patterns might vary based on household characteristics. Specifically, the analysis re-estimates specification (1) separately for eight separate subpopulations, based on education of household adults, the sector in which households were located, access to basic services, the availability of resources at the community level, and the region in which households were located.

For the vast majority of the characteristics, there was little variation in the empirical patterns. Similar to the baseline patterns in fig. 2, there were broadly similar shares of the population that were severely food insecure according to the FIES in all subgroups across FCS deciles that likely did and likely did not contain a significant share of undernourished households (see, e.g., second through the sixth decile); and there was a significant share of households in the highest FCS decile that were severely food insecure for all subgroups.²⁸

However, there were substantial differences in the results based on the region in which households lived. The analysis re-estimates specification (1) for the 37 percent of the population that live in the North East and North West zones of the country, which are regions in which poverty is widespread and there is an ongoing conflict (see, e.g., Lain and Vishwanath 2022); and the analysis re-estimates specification (1) for the rest of the country. The results illustrate that the source of the two empirical patterns—the similar prevalence across much of the FCS distribution and the high share of severe food insecurity in the top deciles in the national figures—are potentially being driven by different regions of the country.²⁹

The estimates for the two northern zones are reported in fig. 6a. The prevalence of severe food insecurity is nearly identical in the 2nd through the 10th FCS decile, with the confidence intervals for the coefficients on all of those deciles overlapping. Furthermore, the hypothesis test of all the coefficients on the second through the 10th decile being jointly equal has a p-value of 0.100. And this pattern is in stark contrast to the estimates from the rest of Nigeria in fig. 6b, where the share that is severely food insecure is sharply declining for higher FCS deciles. One can reject the hypothesis that all the coefficients between the second and the sixth decile are jointly equal as well as the hypothesis that all the coefficients between the second and the tenth are jointly equal at the 1 percent significance level (p-values of 0.001 and 0.000 respectively).

However, there is also another striking difference between the regions aside from the similarity in the food insecurity rate according to the FIES across the expenditure deciles. The share of the population that is severely food insecure in the bottom FCS decile in the rest of Nigeria is over double that of the northern zones; and despite the rapidly declining share that is food-insecure in the rest of Nigeria for higher FCS deciles, the share of the population that is severely food insecure in the top FCS decile is significantly higher in the rest of Nigeria than in the northern zones (21.8 percent versus 9.4 percent).³⁰

8. The Relationship Between the FIES and the FCS and Monetary Poverty in Other West African Countries

The study further re-estimates specification (1) using both the FCS and the expenditure deciles for the rest of the countries in the WAEMU and Chad. The results are reported in tables 1 and 2. Combined, the results illustrate that the empirical patterns in Nigeria are not unique only to that country. Of the nine additional countries reported in tables 1 and 2, three have empirical patterns that are similar to that of Nigeria—namely that there is little difference across much of the FCS or expenditure distribution in the share of the population that is severely food insecure according to the FIES, and that there is a large share of the population in the top deciles that is severely food insecure according to the FIES. Furthermore, a fourth country exhibits the latter pattern.

For Guinea-Bissau, in table 1, it is not possible to reject the hypothesis that all the coefficients on the 2nd through the 10th FCS deciles are equal at conventional significance levels (p-value of 0.159);³¹ in table 2, only the coefficient on the 10th expenditure decile is statistically significant at conventional levels; and again in table 2, it is not possible to reject the hypothesis that all the coefficients between the second and the ninth expenditure deciles are jointly equal to 0 at conventional significance levels (p-value of 0.152). Similar patterns to Nigeria are found for Mali and for Senegal as well. However, for other countries, the decline across both the FCS and expenditure deciles is much more rapid, and there are significantly lower severe food insecurity rates according to the FIES for higher deciles.

However, in addition to the little variation in the share of severe food insecurity according to the FIES across the FCS and expenditure distribution, there is also a relatively high prevalence of severe food insecurity in the top deciles both overall and relative to the bottom decile. In Chad and in the three countries similar to Nigeria mentioned above—Guinea-Bissau, Mali, and Senegal—the share of the population in the top FCS decile that is severely food insecure is between 54 and 61 percent of the national rate, which is approaching the figures reported for Nigeria.

Importantly, in the case of Chad, approximately 35 percent of the population in the top FCS decile is severely food insecure according to the FIES. This is nearly three times the rate in the next-highest WAEMU country analyzed. And this is despite the fact that the internationally comparable poverty rates and inequality measures are similar between Chad and many of the other countries in the region.³² Furthermore, the average FCS in the top FCS decile in Chad is 98.4, which is significantly over double the threshold defining poor or borderline food consumption and similar to the average FCS in the top decile of other WAEMU countries.

But when evaluating the external validity of the baseline results, it is important to account for the size of the country. In the case of West Africa, the population of Nigeria is an order of magnitude larger than any of the other countries analyzed here. The population of each of the WAEMU countries and Chad vary between 0.8 and 12.3 percent of Nigeria's population; the sum of the population of those countries is still 67 million people less than the population in Nigeria;³³ and Nigeria accounts for approximately 20 percent of the entire population of Sub-Saharan Africa, which is the region which contributes the largest number of undernourished individuals in the world (see, e.g., FAO 2021). Thus, even though the FIES has a high prevalence in populations that are unlikely to be undernourished in 5 out of the 10 countries analyzed, the results are applicable to a majority of the population in the region.

9. Conclusion

The results demonstrate that there are several important contexts in which the prevalence of the FIES is high among populations that are unlikely to be undernourished. The results extend critical comparisons across food-access measures to an increasingly important food-access measure and further extend the results to a range of countries where food access is a critical issue. Furthermore, the results suggest that the measure should be used alongside other measures of food access to more precisely interpret the results.

However, there are a number of issues that this study is not able to address. In particular, the countries analyzed here are all from West Africa and have a significant share of the population that is severely food insecure. Thus, the degree to which the results might generalize to other contexts is not known.

Additionally, the analysis is unable to identify exactly how or why the FIES might overlap with undernourishment more in some contexts but not in others. However, the heterogeneity in the baseline empirical patterns presented above offer some potential mechanisms that might explain some of these differences. In particular, the stark regional differences are consistent with conflict and shocks having substantial impacts on how people experience food insecurity and report those experiences. But these and other potential reasons are left to future work.

Conflict of Interest

There was no external funding used in this research and the authors have not conflict of interest to declare.

Data Availability

The Nigeria Living Standards Survey used in this article is available to be requested from the World Bank's microdata library at the following link: "https://microdata.worldbank.org/index.php/catalog/3827"; the Nigeria General Household Survey, Panel 2018-2019, Wave 4" used in this article is available to be requested from the World Bank's microdata library at the following link: "https://microdata.worldbank.org/index.php/catalog/3557". The household surveys used from the other WAEMU countries and Chad can only be shared with the permission of the national statistics office in each individual country.

Author Biography

Jonathan Lain is an Economist with the Poverty & Equity Global Practice of the World Bank, Washington DC, USA; his email address is ([email protected]). Sharad Tandon (corresponding author) is a Senior Economist with the Poverty & Equity Global Practice of the World Bank, Washington DC, USA; his email address is ([email protected]). Tara Vishwanath is a Lead Economist with the Poverty & Equity Global Practice of the World Bank, Washington DC, USA; her email address is ([email protected]). The authors would like to thank the editor and three anonymous referees for comments on an earlier draft. The views expressed here are those of the authors and may not be attributed to the World Bank. A supplementary online appendix for this article can be found at The World Bank Economic Review website.

Footnotes

References