Discussion Paper No.

56
 APRIL 2023

TOWARDS A NEW FOOD

SECURITY INDEX FOR
URBAN HOUSEHOLD
FOOD SECURITY

CAMERON MCCORDIC,1 BRUCE FRAYNE,2 NAOMI SUNU3 AND CLARE WILLIAMSON4

SERIES EDITOR: JONATHAN CRUSH5

1
School of Environment, Enterprise and Development, University of Waterloo, Waterloo, Ontario, Canada
2
School of Environment, Enterprise and Development, University of Waterloo
3
School of Environment, Enterprise and Development, University of Waterloo
4
School of Environment, Enterprise and Development, University of Waterloo
5
Balsillie School of International Affairs, Canada; University of the Western Cape, South Africa
Abstract
The multidimensionality of food security can confound both statistical modelling and clear policy nar-
ratives. That complexity can become amplified in urban areas where food security is often a function
of both local and global factors. Rather than focusing on one dimension of food security metrics, this
investigation proposes a method for building an index of urban household food access, utilization and
stability. The performance of this index is compared across three aggregation methods using household
surveys collected from five cities around the world. The findings indicate that each aggregation method
was internally consistent, although one of the aggregation methods, relying on geometric means, was
likely to be more methodologically sound among the options. This method provides a means of capturing
the multidimensional nature of food security in a way that is amenable to statistical modelling and clear
policy narratives.

Keywords
food security, urban, measurement, household

Suggested Citation
McCordic, C., Frayne, B., Sunu, N. and Williamson, C. (2023). Towards a New Food Security Index for
Urban Household Food Security. HCP Discussion Paper No. 56, Waterloo and Cape Town.

This is the 56th discussion paper in a series published by the Hungry Cities Partner-
ship (HCP), an international research project examining food security and inclusive
growth in cities in the Global South. The multi-year collaborative project aims to
understand how cities in the Global South will manage the food security challenges
arising from rapid urbanization and the transformation of urban food systems. The
Partnership is funded by the Social Sciences and Humanities Research Council of
Canada (SSHRC).

© The authors

All HCP discussion papers and other publications are available for download from the Hungry Cities
Partnership website: http://hungrycities.net.
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

Introduction thinking connects linkages between food security,

vulnerability to food prices, poverty and develop-
ment (Cohen and Barrett, 2010; Napoli et al., 2011;
Food security is a fundamentally important attri-
UNDP 2012).
bute in determining the well-being of people. Yet
the concept of food security is inherently difficult The most cited definition of food security in recent
to define (Barrett and Lentz,2010). In times past, decades was first adopted at the 1996 World Food
food security was addressed through a food supply Summit, where food security was defined as a
lens of producing more food to meet the needs of condition that exists “when all people, at all times,
increasing populations and to overcome famine have physical and economic access to sufficient, safe
(Burchi and De Muro, 2012). This approach was and nutritious food that meets their dietary needs
founded on an initial definition of food security and food preferences for an active and healthy life”
at the World Food Conference in 1974 which (FAO, 2006: 1). According to the FAO (2008a),
focused on food supply at the global scale (FAO, food security is a multidimensional construct
2006). At that time, food security was defined as encompassing food availability, access, utiliza-
“availability at all times of adequate world supplies tion, stability, and safety. The objective of food
of basic foodstuffs to sustain a steady expansion security for all will only be achieved by addressing
of food consumption and to offset fluctuations in these dimensions simultaneously (FAO, 2008a).
productions and pricing” (FAO, 2006: 1). How- Therefore, food insecurity refers to the absence
ever, there have been major paradigm shifts in or deficiency in one or more of these dimensions.
food security thinking over the years that have led The decision to use the term food security or food
to diversified definitions. In particular, there have insecurity depends on how a given argument is
been shifts from global and national to household framed and the scale of measurement being used
and individual scales, from food first to livelihood (Jones et al., 2013). The FAO (2008b) definition of
approaches, and from objective to subjective per- food insecurity can be understood as a representa-
spectives (Maxwell et al., 1996). tion of the evolution that has taken place in food
security paradigms over the years, transitioning
Central to these shifts was Amartya Sen’s (1981)
from a predominantly supply-driven understanding
entitlement approach. Sen argued that people’s
of food security to an appreciation of the impor-
entitlement to different commodities, including
tance of entitlements for food access (Sen, 1980).
food, is based on their personal endowments or
The importance of multidimensionality in food
access to commodities through trade or exchange
security paradigms has been further underscored
(Burchi and De Muro, 2012; Sen, 1981). Rather
by growing recognition of vulnerability and risk
than supply and production, such an approach pri-
in assessing food security (Barrett, 2002; Barrett,
oritized demand and consumption where disrup-
2010; Ecker and Breisinger, 2012).
tions to entitlement such as loss of employment and
trade failures negatively impact food access (Barrett Despite the diversified definition of food secu-
and Lentz,2010). As such, there was a shift in focus rity, the influence of food supply and agricultural
from food supply at the global and national level to productivity has persisted with evidence of rural-
household and individual food access (FAO, 2006). centered interventions and support for agricultural
Sen’s contribution on entitlement and access also livelihoods (Crush et al., 2018; Tuholske et al.,
created a broader platform for identifying con- 2018). This is problematic amidst increased urban-
structs such as economic, social and political dis- ization and equally high levels of food insecurity in
ruptions that influence food security (Devereux, urban and rural areas within most low- to middle-
2001; Maxwell, 1996). For example, the food secu- income countries (UN DESA, 2019; Van Wesen-
rity crisis of 2007 which triggered political riots, beek, 2018). Furthermore, urban poverty has been
resulted from lack of access to food for masses living on the increase in urban areas where almost all food
in poverty (FAO, 2009). Currently, food security is purchased, resulting in vulnerability to food price

1
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

increases (Cohen and Barrett, 2010; Frayne et al., (2015) also assess the relationship between the
2014). Food access in urban areas is also affected by HFIAS/HFIAP and other food security scales, such
access to social infrastructure and physical infra- as the Months of Adequate Household Food Provi-
structure (Frayne and McCordic, 2015). A persis- sioning (MAHFP).
tent focus on food supply reflects a misinterpreta-
tion of the food insecurity problem and will affect This paper presents an additional tool for use in
its measurement. urban food security research and policy by building
and assessing an index of food access, utilization
Although it is important to focus on all dimensions and stability of urban households. The focus is
of food security under varying scales and situations, on providing a robust methodological framework
this makes measurement complicated (Vaitla et al., to measure and compare the status of household
2017). For instance, food insecurity can be described food security along these three dimensions. The
as transitory and temporary or chronic and perma- proposed index would support predictive analytics
nent based on how long it occurs (FAO, 2008a). At to uncover the drivers of urban household food
the same time, food insecurity has other indicators security without sacrificing its multidimensional
including hunger, malnutrition, and uncertain food nature.
access resulting in moderate to severe food insecu-
rity and poor health outcomes (FAO et al., 2019).
Burchi and De Muro (2012) further suggest that the Measuring Urban Food Security
root causes of food security should be explored and
identified within the broader topic of well-being.
One of the greatest obstacles in food security
Such complexities have increased calls for appro-
research is deciphering clear narratives around the
priate measurement of food insecurity because
drivers of food insecurity while recognizing the
of serious implications for health, development
complexity of this development challenge (Carletto
programs, nutrition evaluations, vulnerable group
et al., 2013; Pérez-Escamilla et al., 2017; Ren-
identification, and informing various government
zaho and Mellor, 2010). Research on the drivers
policies ( Jones et al., 2013; Maxwell et al., 1999).
underpinning inconsistent food access in cities has
Evidently, food security requires a consistent mea-
gathered momentum in recent years (Crush and
surement scale in tandem with urbanization trends
Frayne, 2011b; Frayne et al., 2022). Several authors
due to its varying definitions, operationalization,
have noted the crippling effects of household pov-
and multidimensional nature.
erty on food security in cities of the Global South
As household food security measurements are (Maxwell, 1999; Tawodzera and Crush, 2016;)
multidimensional, previous studies indicate that particularly during food price shocks (Cohen and
household food access is only one dimension of Garrett, 2010). Research has also indicated that
urban food insecurity (as demonstrated in the scale poor households in cities can face disrupted food
produced by Ryan and Leibbrandt, 2015). Coates security under the strain of both communicable and
et al. (2007) state that the Household Food Inse- non-communicable diseases (Crush et al., 2011;
curity Access Scale (HFIAS) and Household Food Demmler et al., 2017; Smit et al., 2016).). Incon-
Insecurity Access Prevalence (HFIAP) classifica- sistent access to infrastructure can also predispose
tion continue to be effective measures of household poor urban households to food insecurity (Frayne
food access. The HFIAS and HFIAP) has been and McCordic, 2015).
used in previous studies to measure household
Stable food access in cities depends upon a func-
food insecurity and the prevalence of African food
tioning urban food system that connects food pro-
deserts in addition to access to public resources
ducers to consumers. The supply of food can occur
(McCordic and Abrahamo, 2019; Wagner et al.,
through both formal markets and supermarkets
2019; McCordic, 2016; McCordic, 2017; Frayne
(Crush and Frayne, 2011a) or informal markets
and McCordic, 2015). Frayne and McCordic

2
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

and urban food production (Battersby and Mar- Haysom (2015) notes the need for clear urban food
shak, 2017; Frayne et al., 2014; Orsini et al., 2013; research narratives to help coordinate urban policy
Skinner, 2008). The complexity of modern urban action by multiple actors in municipal government.
food systems can create significant governance However, the complexity of urban food security
challenges, particularly in developing countries challenges can hamper the effective translation of
(Smit, 2015). The multitude of actors engaged in research into policy, often because of miscommu-
the urban food system creates a disaggregated net- nication of urban food research findings (Romero-
work that is difficult to manage through central- Lankao et al., 2017). The nature of these challenges
ized governance (Haysom, 2015). Maxwell (1999) is amplified in urban environments where food
further notes that the lack of formal safety nets can access, utilization, and stability are often subser-
offload the responsibility for food security onto the vient to global economic and climate pressures
household. Localized food systems have emerged as translated through the local dynamics of market
a prominent policy solution to bolster urban food access and household entitlements (Brown and
security as well as becoming a common theme in Funk, 2008; Cohen and Garrett, 2010; Crush et
urban food studies (Sonnino, 2016). Localization al., 2012). Without the appropriate diagnostic tools,
is also a response to social justice concerns about policymakers are left with the unenviable task of
equitable household access to food. clarifying the nuanced narratives of social research.
As a result, there is an urgent need for research
To inform urban food security policy, Haysom and tools that can effectively capture the complexity of
Tawodzera (2018) have urged a renewed focus on urban food security to support statistical modelling
building food security metrics that are applicable and public policy formation. Previous approaches
to the unique characteristics of urban food systems. to overcoming this challenge have distilled satel-
Representative survey-based methods have pro- lite imagery into relevant and timely famine early
vided a foundational platform to guide policy inter- warning systems (Enunkel et al., 2014). Scenario-
ventions (Pérez-Escamilla, 2012). Freedman and based simulations have provided helpful visualiza-
Bell (2009) further note that self-reported measures tions to support policy decisions on food security
of food accessibility can be an accurate and valid impacts (He et al., 2013). Other researchers have
basis for designing food security interventions. developed novel metrics to account of the combined
influence of multidimensional factors underlying
The multidimensional nature of food security sustainable food and nutrition security (Gustafson
impacts obstructs the development of precise social et al., 2016). Each approach attempts to capture the
research methods and analytical approaches. The dynamic and complex nature of food security by
multiscaler and collateral impacts of climate change simplifying that complexity into a metric that is
have exacerbated both national food supply and valid and reliable (Prosperi et al., 2016)
household incomes, rendering opaque the images of
food security vulnerabilities and obstructing effec- Three of the most widely used self-reported
tive mitigation measures (Wheeler and von Braun, cross-cultural food security scales were developed
2013). The interdisciplinary analysis by Foran et al. by the Food and Nutrition Technical Assistance
(2014) of food security frameworks identifies several (FANTA) programme: the Household Dietary
conceptual paradigms, often in tension with one Diversity Score (HDDS) (Swindale and Bilinsky,
another and confounding effective food security 2005), the Household Food Insecurity Access Scale
interventions. As a result, there is a pressing need to (HFIAS) (Coates et al., 2007), and the Months of
develop innovative decision-support mechanisms Adequate Food Provisioning (MAHFP) (Bilinsky
to support food security policy and research (Mock and Swindale, 2010). In this discussion paper, each
et al., 2013). of these scales is discussed and assessed as measures
of food utilization, food access, and food stability.

3
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

Methodology households; the Individual Dietary Diversity Score

(IDDS) (Kennedy et al., 2011). Dietary diversity is
a key component of effective food use (Renzaho
Household Dietary Diversity Score
and Mellor, 2010) has been used as a proxy mea-
sure of food use by other studies (Codjoe et al.,
The Household Dietary Diversity Score (HDDS) is
2016; Labadarios et al., 2011). As a result, HDDS
designed to measure nutritional diversity in house-
can provide information on effective household
hold food consumption by calculating the number
food use in social survey research. The HDDS is
of food groups consumed by any member of a
also supported by a growing body of evidence for
household in the previous 24 hours (Swindale and
its external validity. Cordeiro et al. (2012) found
Bilinsky, 2005):
a strong correlation between HDDS and energy
intake in a survey of Tanzanian adolescents. HDDS
HDDS = ∑ Food Groups Consumed in the Last 24
also demonstrated a strong correlation with the
Hours
Food Consumption Score in several surveys (Jones
The score is based on the report of the household et al., 2013). Faber et al. (2009) found a strong cor-
member in charge of food preparation and/or the relation between the HDDS and the HFIAS in a
person who can reliably describe the consumption survey of Limpopo in South Africa. However, this
patterns of the household. The scale can be adapted finding was not replicated in a study by Maxwell et
to local foods and consumption patterns using the al. (2014). As they suggest, this finding may have
following food groups as a guide (Table 1). arisen because the two scales measure different
dimensions of food security. In sum, the HDDS
If the household has consumed food in any of these provides important information on a key aspect of
food groups in the past 24 hours, a score of one is effective food utilization and nutritional diversity.
inputted for that food group. Otherwise, a zero is
inputted. The HDDS is then calculated by adding
the number of food groups consumed by the house- Household Food Insecurity Access Scale
hold in the previous 24 hours:
The Household Food Insecurity Access Scale
The HDDS score for each household ranges from (HFIAS) is a survey instrument designed to mea-
0 to 12 where the higher the score, the greater the sure the frequency and intensity of food access
dietary diversity. The same methodology can be challenges experienced by a household (Coates
used to calculate a score for individuals rather than et al., 2007). The scale comprises nine Likert

TABLE 1: Household Dietary Diversity Score Food Groups

1. Bread, rice, noodles, biscuits, or any other foods made from millet, sorghum, maize, rice, wheat, or any other locally
available grain
2. Potatoes, yams, manioc, cassava, or any other foods made from roots or tubers
3. Other vegetables
4. Fruit
5. Beef, pork, lamb, goat, rabbit, wild game, chicken, duck, other birds, liver, kidney, heart or other organ meats
6. Eggs
7. Fresh or dried fish or shellfish
8. Foods made from beans, peas, lentils, or nuts
9. Cheese, yoghurt, milk, or other milk products
10. Foods made with oil, fat, or butter
11. Sugar or honey
12. Other foods such as condiments, coffee, tea
Source: Swindale and Bilinsky (2005: 4)

4
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

questions designed to measure the physical, eco- in a study of rural Tanzania. Similarly, HFIAS was
nomic, and social dimensions of challenges related also associated with an increased probability of
to food access. The questions range from minor to undernutrition among children in surveys carried
more severe experiences of these challenges in food out in Bangladesh, Vietnam and Ethiopia (Ali et al.,
access. The Likert scale accompanying each ques- 2013). Charamba et al. (2019) also found that the
tion ranges from never in the last month to more scale was internally and externally valid in a survey
than 10 times in the last month. The questions in of households in Windhoek, Namibia. However,
the scale include the following items (Table 2). some studies have questioned the effectiveness of
the scale. Dietchler et al. (2010), for example, found
If the household experienced any of these food that the HFIAS was less accurate in its classification
access challenges, respondents are asked to rank the of food security status than the Household Hunger
frequency with which this occurred the following Scale (citing potential challenges in translating the
scale: One = Rarely (once or twice in the past four concepts of the HFIAS). As with other measures of
weeks), Two = Sometimes (three to ten times in the food security, the overriding recommendation has
past four weeks), or Three = Often (more than ten been to use multiple food security measures rather
times in the past four weeks). The scores are then than attempting to rely solely on one food security
summed to provide an overall HFIAS score of from scale and disregard other dimensions of food secu-
zero to 27, where higher scores represent a higher rity (Maxwell et al., 2014). Among the multiple
frequency of experiencing food access challenges. food security scales available for measuring dif-
ferent dimensions of food security, the HFIAS is
HFIAS = ∑ Frequency of Food Access Challenges in the still considered to be the reliable measure of house-
Past 4 Weeks hold food access. The HFIAS can also be expressed
as a categorical measure by applying a scoring algo-
The HFIAS is probably the most implemented of
rithm to the scale responses in order to categorize
the three scales reviewed here and has assembled a
households into different levels of food insecurity
strong body of evidence to support its use. Knueppel
(the Household Food Insecurity Access Prevalence
et al. (2009), for example, confirmed that the
indicator or HFIAP (Coates et al., 2007).
HFIAS scores were supported by key informants

TABLE 2: Household Food Insecurity Access Scale (HFIAS) Questions

In the past four weeks:
1. Did you worry that your household would not have enough food?
2. Were you or any household member not able to eat the kinds of foods you preferred because of a lack of
resources?
3. Did you or any household member have to eat a limited variety of foods due to a lack of resources?
4. Did you or any household member have to eat some foods that you really did not want to eat because of a lack of
resources to obtain other types of food?
5. Did you or any household member have to eat a smaller meal than you felt you needed because there was not
enough food?
6. Did you or any household member have to eat fewer meals in a day because there was not enough food?
7. Was there ever no food to eat of any kind in your household because of lack of resources to get food?
8. Did you or any household member go to sleep at night hungry because there was not enough food?
9. Did you or any household member go a whole day and night without eating anything because there was not enough
food?
Source: Coates et al. (2007: 5)

5
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

Months of Adequate Household Food score needs to be decided (Munda and Nardo,
Provisioning 2005a). Although this is usually a decision made
on theoretical grounds, the index can either give
The Months of Adequate Household Food Pro- equal weight to each scale’s contribution or give
visioning (MAHFP) provides a measure of the disproportionately weight to each scale’s contribu-
stability with which households have maintained tion based on these grounds. Second, given that
adequate food provisioning over the previous year HDDS, HFIAS, and MAHFP are measured on
(Bilinsky and Swindale, 2010). The scale is admin- different scales of varying magnitude, the scales
istered using the following questions: need to be normalized to ensure that they are com-
parable (Nardo et al., 2005). Third, the means by
Were there months, in the past 12 months, in which the scores are aggregated (averaged) can sig-
which you did not have enough food to meet your nificantly impact the stability of the overall index.
family’s needs? These predominantly revolve around the theo-
retical implications of compensability (or the extent
If yes, which were the months in the past 12 to which performance on each scale can be traded
months during which you did not have enough off) (Munda and Nardo, 2005b). Some means of
food to meet your family’s needs?” (Bilinsky and aggregation (like arithmetic mean or Bordo ranking
Swindale, 2010:4). procedures) are perfectly compensable, in that poor
performance on one scale can be traded off for
If a given month is identified by the respondent, 1
improved performance on another scale. Alterna-
is inputted for that month. Otherwise, 0 is inputted
tively, Condorcet classification procedures ensure
for any months not identified by the respondent.
that performance on each scale cannot be traded off
The scale is calculated by subtracting the sum of
for performance on another scale.
the inputted numbers from 12 (thus, higher scores
on the scale are associated with greater household To support clear policy narratives and statistical
food stability). modeling, an index of urban food access, utiliza-
tion, and stability will thus need to address the
MAHFP = 12 – ∑ Months of Inadequate Food Provi-
issues of compensability and weighting. Such an
sioning in the Last Year
index needs to provide (a) a means of normaliza-
tion that is not relative; (b) a weighting scheme that
Unlike the HDDS and the HFIAS, there have been
ensures equal priority to all included measures, and
fewer studies assessing the validity or reliability of
(c) a means of aggregation that is consistent with the
this measure in spite of its widespread use in studies
theory underlying food security measurement. The
of urban food security (Battersby, 2011; Frayne and
remainder of this paper compares three methods
McCordic, 2015; Frayne et al., 2010). One study
for building a single index of urban household food
has identified common predictors of the MAHFP
access, utilization, and stability using the HDDS,
and other food security scales (Harris-Fry et al.,
HFIAS, and MAHFP measures. The aim is to pro-
2015). In sum, the MAHFP is s an empirically sup-
vide an internally consistent and externally valid
ported measure of food stability, but without the
means of scoring household food security across
same degree of external validation as the other mea-
these three dimensions.
sures reviewed above.
To create a single index from the HDDS, HFIAS
Each of the reviewed food security scales provides
and MAHFP scales, all three need to be normalized
a measure of different dimensions of food security.
so that they can be expressed on the same numeric
To combine the measures into one overarching
scale. Given the focus on building an index to
index, there are several considerations to take into
support comparisons of household food security
account. First, the relative weighting of each food
across geographical regions and time, relative nor-
security scale’s contribution to the overall index
malization techniques (such as standardization and

6
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

ranking) were not viable approaches. Instead, we sensitive to outliers, which can skew the overall
implemented min-max normalization to transform mean. Furthermore, the arithmetic mean assumes
each food security score on a scale of zero to one. that the scales included in the index can be aggre-
All transformed scales are denoted with the super- gated linearly. As with other aggregation methods,
script ′ notation (e.g., HDDS′). The HDDS and the arithmetic mean also assumes that each of the
MAHFP scores were transformed using the fol- scales included in the index are independent of one
lowing equations: another. The FSAM was calculated using the fol-
lowing equation:
HDDS – 0
HDDS′ =
12 – 0 Step 1. Sum the normalized HDDS, HFIAS, and
HDDS scores and divide by 3:
MAHFP – 0
MAHFP′ =
12 – 0 HFIAS′ + HDDS′ + MAHFP′
FSAM =
3
While this approach easily converts the magnitude
of each scale, the Min-Max transformation does The Food Security Geometric Mean (FSGM)
not account for the reversed direction of the HFIAS index provides an alternative approach to aggre-
(where, unlike HDDS and MAHFP, higher scores gating the food security scales in an unweighted
denote more severe food insecurity). In this case, fashion. In this case, the mean is calculated as the
the Min-Max Normalization Equation was modi- nth root of a product of n scales. The geometric
fied in order to reverse the direction of the HFIAS mean has a few advantages over the arithmetic
scale in addition to its magnitude: mean. First, the geometric mean is less sensitive to
outliers (displaying imperfect compensability) than
HFIAS – 27 the arithmetic mean. Second, the geometric mean
HFIAS′ =
0 – 27 is a unitless measure and can aggregate scales with
varied degrees of magnitude. Finally, geometric
means are the preferred method of aggregation for
FSGM Index Aggregation
ratios. That said, there is one important caveat to
the use of geometric means when aggregating mul-
We used three methods for aggregating the nor-
tiple scales. The inclusion of a score of 0 for any
malized food security scales: arithmetic means,
of the scales included in the index will result in a
geometric means, and harmonic means. Each of
geometric mean of 0 (regardless of the scores for
these methods of aggregation provides its own
the other scales). In order to overcome this chal-
properties and benefits. As a result, testing the dif-
lenge, the scale for each of the food security scales
ferences between each of the means provides an
was shifted by one before calculating the geometric
opportunity to assess the utility of each approach.
mean of the food security scales (providing FSGM’).
The approaches are all predicated on the notion that
Once the geometric mean was calculated, one was
each food security scale should equally contribute
subtracted from the geometric mean to provide the
toward the overall food security index.
FSGM:
The Food Security Arithmetic Mean (FSAM)
Step 1. Add 1 to each normalized HFIAS, MAHFP,
index provides the unweighted arithmetic mean
and HDDS score, multiply the sums together and
of the three normalized food security scores. This
find the cubed root of the product:
approach allows for trade-offs in performance across
these indices (perfect compensability). In other 3
FSGM′ = √ (HFIAS′ + 1) * (HDDS′ + 1) * (MAHFP′ + 1)
words, poor household performance on one scale
can be traded off for better household performance
on another scale. This aggregation approach is also

7
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

Step 2. Subtract 1 from the resulting cubed root: scales to consistently measure one domain (as
would be indicated by tests of internal consistency).
FSGM = FSGM′ - 1 In order to preserve the linear aggregation of the
scales, it is important to determine whether there
The Food Security Harmonic Mean (FSHM) pro- is a positive linear relationship between the scales
vides an alternative approach to the scale aggrega- that would support the aggregation of the scales.
tion challenge. Harmonic means express the quo- A negative linear relationship between any of the
tient of n divided by n reciprocal scales. Expressed scales would indicate that increasing scores on one
differently, the harmonic mean is also the reciprocal food security scale was associated with decreasing
of the arithmetic mean of the reciprocals for each scores on another food security scale in the index
scale. Harmonic means are helpful when averaging (creating internal inconsistency in the index). As a
across different rates. Like the geometric mean, result, one or more of the sub-scales may not be
however, including a score of 0 for any of the positively correlated with the overall index score.
underlying scales creates challenges for the calcu- In order to assess whether this is the case, we cal-
lation. In this case, the harmonic mean would be culated descriptive statistics of the index scores for
undefined if any of the underlying scales included 0 the different aggregation methods. Pearson’s R cor-
as a score. As a result, similar to the FSGM calcula- relation analysis was used to determine the linear
tion, one was added to each of the underlying food strength and direction of the correlation between
security scales. Once the FSHM was calculated, each of the underlying scales and the overall index
one was subtracted from the overall mean to bring score using each aggregation method.
the FSHM back to its original normalized scale.

Step 1. Add 1 to each normalized HFIAS, MAHFP,

and HDDS score, sum the inverse of these totals, Application
and calculate 3 divided by the resulting sum:
To illustrate the utility of the New Food Security
3
Index, this section of the paper draws on data from
FSHM′ = 1 +
1 +
1
(HFIAS′ + 1) (HDDS′ + 1) (MAHFP′ + 1) household food security surveys conducted by the
Hungry Cities Partnership in five cities: Maputo,
Step 2. Subtract 1 from the resulting quotient: Mozambique; Kingston, Jamaica; Nairobi, Kenya;
Mexico City, Mexico; and Nanjing, China. These
FSHM = FSHM′ - 1 surveys all administered the same household survey
instrument and administered the survey in the same
Each of these scales represent unweighted approaches manner to adult household respondents in each ran-
to aggregating the food security scales. It should be domly selected household. However, the language
noted that, given the properties of each approach, in which the survey and scales were administered
there is a ranked order to a given household’s did vary. The survey scales were administered in
FSAM, FSGM and FSHM scores. More specifi- English (in Kingston and Nairobi), Portuguese (in
cally, the following order (FSHM≤FSGM≤FSAM) Maputo), Spanish (in Mexico City), and Mandarin
will be consistent for all households. (in Nanjing).

Index Internal Consistency Each survey sampled households from across the
city using a combination of random systematic
The indices constructed represent aggregated sampling with sample sizes distributed across city
measures of urban household food access, utiliza- subdistricts using approximate proportionate allo-
tion, and stability. Given the multidimensional cation (based on the most recently available census
nature of these scales, it is not necessary for the data for the city). The sample sizes varied between
the cities but for this analysis 500 households were

8
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

randomly selected from each city data set to pro- there were city-specific differences in both the
vide an equal contribution of each to the effects magnitude and spread of the scores for the different
observed. Among the final 2,500 household sam- indices (Figure 2). The households in Nanjing
ples, the response rates for the food security scales recorded the highest scores on the FSAM, FSGM
varied from 96% to 99%. and FSHM as well as the smallest spread. The
households in Maputo indicated the lowest scores
The three indices (the FSAM, FSGM and FSHM) across the three indices as well as the greatest spread.
demonstrated a similar distribution. However,

FIGURE 1: Histograms of FSAM, FSGM and FSHM

FIGURE 2: Distribution of Mean FSAM, FSGM and FSHM Scores Across Cities

9
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

The descriptive statistics behind Figure 3 are The Pearson’s R correlations of the indices and
presented in Table 3. The differences observed sub-scales revealed significant and high positive
in the variation of these three indices across the linear correlations across the three indices (Table
cities may be the result of a combination of factors: 4). In addition, each of the underlying food secu-
for example, the predominant clustering of high rity scales also demonstrated significant and positive
scores in a given city (as observed in Nanjing) correlations with the three indices. However, while
or the reduced range of scores (as observed in the relative strength of the relationship between
Nanjing and Nairobi). The association between these sub-scales and the three indices did not vary
the clustering of scores within and between the substantially across indices, the relative strength of
three index scores in each city gives an indication these relationships did vary by city. For example,
of the spread inherent in the underlying food Maputo, Nairobi, Mexico City and Kingston all
security scales in these cities. indicated very strong linear correlations between
the food security scales and the three indices. In
TABLE 3: Descriptive Statistics of the FSAM, Nanjing, however, the MAHFP and HFIAS indi-
FSGM and FSHM Scores Across Cities
cated weak to moderate positive relationships with
City Statistics FSAM FSGM FSHM
the overall food security scale. This observation
Mean 0.66 0.64 0.61
likely resulted from the fact that few households
Median 0.7 0.67 0.64
Maputo received less than a perfect food security score on
Std. Dev. 0.18 0.18 0.18
(n=484) these scales in the Nanjing household sample.
Min 0.06 0.05 0.05
Max 0.97 0.97 0.97 TABLE 4: Correlation of HFIAS, MAHFP, and
Mean 0.69 0.66 0.64 HDDS Scores with FSAM, FSGM, and FSHM
Median 0.72 0.68 0.65 Across Cities
Kingston Maputo
Std. Dev. 0.15 0.16 0.16 FSAM FSGM FSHM
(n=474) (n=484)
Min 0.03 0.03 0.03
FSAM 1 .997** .987**
Max 1 1 1
FSGM .997** 1 .996**
Mean 0.88 0.87 0.86
FSHM .987** .996** 1
Median 0.89 0.88 0.88
Nanjing HDDS’ .632** .683** .733**
Std. Dev. 0.07 0.08 0.09
(n=488) HFIAS’ .875** .857** .829**
Min 0.59 0.55 0.51
MAHFP’ .815** .790** .760**
Max 1 1 1
Kingston
Mean 0.72 0.71 0.69 FSAM FSGM FSHM
(n=474)
Median 0.74 0.73 0.71
Nairobi FSAM 1 .995** .981**
Std. Dev. 0.15 0.15 0.15
(n=493) FSGM .995** 1 .995**
Min 0.22 0.2 0.18
FSHM .981** .995** 1
Max 1 1 1
HDDS’ .622** .684** .740**
Mean 0.77 0.76 0.74
HFIAS’ .848** .818** .776**
Mexico Median 0.78 0.77 0.75
MAHFP’ .685** .645** .602**
City Std. Dev. 0.12 0.13 0.14
Nanjing
(n=494) Min 0.06 0.05 0.05 FSAM FSGM FSHM
(n=488)
Max 1 1 1
FSAM 1 .998** .992**
Mean 0.75 0.73 0.71
FSGM .998** 1 .998**
Median 0.77 0.75 0.72
Total FSHM .992** .998** 1
Std. Dev. 0.16 0.16 0.17
(n=2433) HDDS’ .954** .966** .974**
Min 0.03 0.03 0.03
HFIAS’ .457** .415** .377**
Max 1 1 1
MAHFP’ .273** .244** .219**

10
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

Nairobi Geometric means may provide a more reasonable

FSAM FSGM FSHM
(n=493) conceptual grounds for the index, given that the
FSAM 1 .998** .992** approach supports imperfect compensability.
FSGM .998** 1 .998**
FSHM .992** .998** 1 The proposed index should not be interpreted as
HDDS’ .705** .738** .769** a comprehensive measurement of household food
HFIAS’ .871** .859** .842** access, utilization and stability, but rather as an
MAHFP’ .740** .715** .688** indication of these food security characteristics. For
Mexico City example, the definition of food utilization covers a
FSAM FSGM FSHM
(n=494) broader set of features than that measured by the
FSAM 1 .996** .983** HDDS and the scale by which food stability is
FSGM .996** 1 .996** measured can vary from the 12-month recall period
FSHM .983** .996** 1 covered by the MAHFP. As a result, this index pro-
HDDS’ .620** .677** .727** vides one proxy for the measurement of these food
HFIAS’ .770** .729** .681** security concepts. With the development of more
MAHFP’ .673** .631** .586** refined food security scales, it may be possible to
*p<.05; **p<.01 develop more sophisticated indices of food security
to better account for the characteristics not covered
by the current proposed index.
Conclusion This paper provides a means of calculating an index
of urban household food access, utilization and
This paper proposed a new index of urban house- stability that is methodologically sound, reliable
hold food access, stability, and utilization by com- and externally valid. Given the data normalization
bining three common food insecurity metrics (the technique implemented, this index also provides a
HFIAS, HDDS and MAHFP) via min-max nor- means of comparing performance on these pillars of
malization and geometric means for index aggrega- food security across geographies and points in time.
tion. Geometric means are less sensitive to outliers Furthermore, the index provides a means of simpli-
(demonstrating partial non-compensability) and fying multiple dimensions of urban household food
the preferred method of aggregation for ratios and security into one metric that is amenable to both
skewed data (as is the case in normalized data and statistical modelling and clear policy narratives.
food security data, which can be highly skewed). Given the complexity of both the characteristics
The issue of compensability is also an impor- and the drivers of urban food security, the utility of
tant theoretical consideration among these index the index methodology proposed here could sup-
aggregation techniques. Perfect compensability (as port further research into sustainable food insecu-
allowed by an arithmetic mean) would allow house- rity mitigation measures under the twin pressures
holds to trade off their performance across their of climate change and rapid urbanization. Future
food access, utilization, and stability. These pillars research could investigate the utility of this index
of food security are likely meant to be interpreted as a means of measuring household food security
as fundamental necessities in order for food secu- status in cities not included in this paper.
rity to exist (rather than compensable entities). As
a result, from a theoretical perspective, it is unlikely
that arithmetic means would work as a means of
measuring across urban household food access, uti-
lization and stability. While perfect non-compen-
sability is not likely to be helpful (given that avail-
able methods, like Condorcet methods, would only
support household ranking rather than scoring).

11
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

References Household Food Insecurity Access Scale (HFIAS) for

Measurement of Household Food Access: Indicator Guide
1. Ali, D., Saha, K., Nguyen, P., Diressie, M., Ruel, (Version 3). Washington, DC: FANTA.
M., Menon, P. and Rawat, R. (2013). “Household 15. Codjoe, S., Okutu, D. and Abu, M. (2016). “Urban
Food Insecurity is Associated with Higher Child Household Characteristics and Dietary Diversity: An
Undernutrition in Bangladesh, Ethiopia, and Vietnam, Analysis of Food Security in Accra, Ghana.” Food &
But the Effect is Not Mediated by Child Dietary Nutrition Bulletin 37: 202-218.
Diversity.” Journal of Nutrition 143: 2015-2021. 16. Cohen, M. and Garrett, J. (2010). “The Food Price
2. Barrett, C. (2002). “Food Security and Food Assistance Crisis and Urban Food (In)security.” Environment and
Programs.” Handbook of Agricultural Economics 2B: 2103- Urbanization 22: 467-482.
2190. 17. Cordeiro, L., Wilde, P., Semu, H. and Levinson,
3. Barrett, C. (2010). “Measuring Food Insecurity.” F. (2012). “Household Food Security is Inversely
Science 327: 825-828. Associated with Undernutrition Among Adolescents in
4. Barrett, C. and Lentz, E. (2010). “Food Insecurity.” Kilosa, Tanzania.” Journal of Nutrition 142: 1741-1747.
In Oxford Research Encyclopedia of International Studies. At: 18. Crush, J. and Frayne, B. (2011a). “Supermarket
https://oxfordre.com/internationalstudies/view/10.1093/ Expansion and the Informal Food Economy in
acrefore/9780190846626.001.0001/acrefore- Southern African Cities: Implications for Urban Food
9780190846626-e-438 Security.” Journal of Southern African Studies 37: 781-807.
5. Battersby, J, (2011). “Urban Food Insecurity in Cape 19. Crush, J. and Frayne, B. (2011b) “Urban Food
Town, South Africa: An Alternative Approach to Food Insecurity and the New International Food Security
Access.” Development Southern Africa 28(4): 545-561. Agenda.” Development Southern Africa 28: 527-544.
6. Battersby, J. (2012). “Beyond the Food Desert: Finding 20. Crush, J., Frayne, B., and McCordic, C. (2018).
Ways to Speak About Urban Food Security in South “Divorcing Food and Agriculture: Towards an Agenda
Africa.” Geografiska Annaler, Series B 94: 141-159. for Urban Food Security Research.” In B. Frayne,
7. Battersby, J. and Marshak, M. (2017). Mapping the J. Crush and C. McCordic (Eds.), Food and Nutrition
Invisible: The Informal Food Economy of Cape Town, South Security in Southern African Cities. New York: Routledge,
Africa. Urban Food Security Series No. 24, AFSUN, pp. 1-8.
Cape Town. 21. Crush, J., Frayne, B. and Pendleton, W. (2012). “The
8. Bilinsky, P. and Swindale, A. (2010). Months of Adequate Crisis of Food Insecurity in African Cities.” Journal of
Household Food Provisioning (MAHFP) for Measurement Hunger & Environmental Nutrition 7: 271-292.
of Household Food Access: Indicator Guide (Version 4). 22. Crush, J., Drimie, S., Frayne, B. and Caesar, M.
Washington, D.C.: FANTA. (2011). “The HIV and Urban Food Security Nexus in
9. Brown, M. and Funk, C. (2008). “Food Security Africa.” Food Security 3: 347-362.
Under Climate Change.” Science 319: 580-581. 23. Deitchler, M., Ballard, T., Swindale, A. and Coates,
10. Burchi, F. and De Muro, P. (2012). “A Human J. (2010). Validation of a Measure of Household Hunger for
Development Approach and Capability Approach Cross-Cultural Use. Washington D.C.: FANTA.
to Food Security: Conceptual Framework and 24. Demmler, K., Klasen, S., Nzuma, J. and Qaim,
Informational Basis.” Working Paper, Regional Bureau M. (2017). “Supermarket Purchase Contributes to
for Africa, United Nations Development Programme. Nutrition-Related Non-Communicable Diseases in
11. Carletto, C., Zezza, A. and Banerjee, R. (2013). Urban Kenya.” PloS One 12: e0185148.
“Towards Better Measurement of Household Food 25. Devereux, S. (2001). “Sen’s Entitlement Approach:
Security: Harmonizing Indicators and the Role of Critiques and Counter-Critiques.” Oxford Development
Household Surveys.” Global Food Security 2: 30-40. Studies 29: 245-263.
12. Charamba, V., Nickanor, N. and Kazembe, L. (2019). 26. Ecker, O. and Breisinger, C. (2012). “The Food
“Validation of the HCP Survey Tool for Measuring Security System: A New Conceptual Framework.”
Urban Food Insecurity: An Item Response Theory Discussion Paper No. 01166, IFPRI, Washington D.C.
Approach.” HCP Discussion Paper No. 37, Hungry 27. Enenkel, M., See, L., Bonifacio, R., Boken, V. et
Cities Partnership. al. (2015). “Drought and Food Security: Improving
13. Coates, J. (2013). “Build it Back Better: Deconstructing Decision-Support Via New Technologies and
Food Security for Improved Measurement and Action.” Innovative Collaboration.” Global Food Security 4: 51-55.
Global Food Security 2: 188-194. 28. Faber M, Schwabe C, and Drimie S (2009). Dietary
14. Coates, J., Swindale, A .and Bilinsky, P. (2007). diversity in relation to other household food security

12
 TOWARDS A NEW FOOD SECURITY INDEX FOR URBAN HOUSEHOLD FOOD SECURITY

indicators. International Journal of Food Safety, Nutrition Food Security and Women’s Dietary Diversity in Rural
and Public Health 2(1): 1–15. Bangladesh: A Cross-Sectional Study.” Journal of Health,
29. FAO (1996). The Rome Declaration on World Food Population and Nutrition 33:2.
Security. Rome: FAO Food and Agriculture 44. Haysom, G. (2015). “Food and the City: Urban Scale
Organization Statistics Division, Food System Governance.” Urban Forum 26: 263-281.
30. FAO (2006). “Food Security Policy Brief.” Food 45. Haysom, G. and Tawodzera, G. (2018) “’Measurement
and Agriculture Organization of the United Nations, Drives Diagnosis and Response’: Gaps in Transferring
Rome. Food Security Assessment to the Urban Scale.” Food
31. FAO (2008a). The State of Food and Agriculture 2008. Policy 74: 117-125.
Rome: Food and Agriculture Organization of the 46. He, J., Liu, Y., Yu, Y., Tang, W. et al. (2013). “A
United Nations. Counterfactual Scenario Simulation Approach for
32. FAO (2008b). An Introduction to the Basic Concepts of Food Assessing the Impact of Farmland Preservation Policies
Security. Rome: FAO Food Security Program. on Urban Sprawl and Food Security in a Major Grain-
Producing Area of China.” Applied Geography 37: 127-
33. FAO (2009). “High Food Prices: The Food Security
138.
Crisis of 2007-2008 and Recent Food Price Increases:
Facts and Lessons.” At: http://www.fao.org/fileadmin/ 47. Jones, A., Ngure, F., Pelto, G. and Young S. (2013).
user_upload/ISFP/High_food_prices.pdf “What Are We Assessing When We Measure Food
Security?” Advances in Nutrition 4: 481-505.
34. FAO, IFAD, UNICEF, WFP and WHO (2019). The
State of Food Security and Nutrition in the World 2019. 48. Kennedy, G., Ballard, T. and Dop, M. (2011).
Rome: FAO. Guidelines for Measuring Household and Individual Dietary
Diversity. Rome: FAO.
35. Foran, T., Butler, J., Williams, L., Wanjura, W. et al.
(2014). “Taking Complexity in Food Systems Seriously: 49. Knueppel, D., Demment, M, and Kaiser, L. (2009).
An Interdisciplinary Analysis.” World Development 61: ”Validation of the Household Food Insecurity Access
85-101. Scale in Rural Tanzania.” Public Health Nutrition 13:
360-367.
36. Frayne, B. and McCordic, C. (2015). “Planning
for Food Secure Cities: Measuring the Influence of 50. Labadarios, D., Mchiza, Z., Steyn, N., Gericke, G. et
Infrastructure and Income on Household Food Security al. (2011). “Food Security in South Africa: A Review
in Southern African Cities.” Geoforum 65: 1-11. of National Surveys.” Bulletin of the World Health
Organization 89: 891-899.
37. Frayne, B., Crush J., and McLachlan, M. (2014).
“Urbanization, Nutrition and Development in 51. Lappé, F., Clapp, J., Anderson, M., Broad, R. et al.
Southern African Cities.” Food Security 6:101-112. (2013). “How We Count Hunger Matters.” Ethics and
International Affairs 1: 1-9.
38. Frayne. B., McCordic, C. and Shilomboleni, H. (2014).
“Growing Out of Poverty: Does Urban Agriculture 52. Maxwell, S. (1996). Food Security: A Post-Modern
Contribute to Household Food Security in Southern Perspective.” Food Policy 21: 155-170.
African Cities?” Urban Forum 25: 177-189. 53. Maxwell, D. (1999). “The Political Economy of
39. Frayne, B., Dordi, T., McCordic, C., Sunu, N., and Urban Food Security in Sub-Saharan Africa.” World
Williamson, C. (2022). “A Bibliometric Analysis of Development 27: 1939-1953.
Urban Food Security.” Urban Transformations 4: 1-22. 54. Maxwell, D., Vaitla, B. and Coates, J. (2014). “How
40. Frayne, B, Pendleton, W., Crush, J., Acquah, B. et Do Indicators of Household Food Insecurity Measure
al. (2010). The State of Urban Food Insecurity in Southern Up? An Empirical Comparison from Ethiopia.” Food
Africa. Urban Food Security Series No. 2, AFSUN, Policy 47: 107-116.
Cape Town. 55. Maxwell, D., Ahiadeke, C., Levin, C.., Armar-
41. Freedman D. and Bell B. (2009). “Access to Healthful Klemesu, M. et al. (1999). “Alternative Food Security
Foods Among an Urban Food Insecure Population: Indicators: Revisiting the Frequency and Severity of
Perceptions Versus Reality.” Journal of Urban Health 86: Coping Strategies.” Food Policy 24: 411-429.
825-838. 56. McCordic, C. (2016). “Urban Infrastructure and
42. Gustafson, D., Gutman, A., Leet, W., Drewnowski, Household Vulnerability to Food Insecurity in Maputo,
A. et al. (2016), “Seven Food System Metrics of Mozambique.” PhD Thesis, University of Waterloo,
Sustainable Nutrition Security.” Sustainability 8: 196- Waterloo.
213. 57. McCordic, C. (2017). “Household Food Security and
43. Harris-Fry, H., Azad, K., Kuddus, A., Shaha, S. et al. Access to Medical Care in Maputo, Mozambique.”
(2015). “Socio-Economic Determinants of Household HCP Discussion Paper No. 7, Hungry Cities
Partnership, Waterloo.
13
HUNGRY CITIES PARTNERSHIP DISCUSSION PAPER NO. 56

58. McCordic, C. and Abrahamo, E. (2019). “Family 72. Sen, A. (1981). Poverty and Famines: An Essay on
Structure and Severe Food Insecurity in Maputo and Entitlement and Deprivation. Oxford: Clarendon Press.
Matola, Mozambique.” Sustainability 11: 267–281. 73. Skinner, C. (2008). “The Struggle for the Streets:
59. Mock, N., Morrow, N. and Papendieck, A. (2013). Processes of Exclusion and Inclusion of Street Traders
“From Complexity to Food Security Decision- in Durban, South Africa.” Development Southern
Support: Novel Methods of Assessment and Their Africa 25: 227-242.
Role in Enhancing the Timeliness and Relevance of 74. Smit, W. (2016). “Urban Governance and Urban Food
Food and Nutrition Security Information.” Global Food Systems in Africa: Examining the Linkages.” Cities 58:
Security 2: 41-49. 80-86.
60. Munda, G. and Nardo, M. (2005a). Non-Compensatory 75. Smit, W., de Lannoy, A., Dover, R., Lambert, E.
Composite Indicators for Ranking Countries: A Defensible et al. (2016). “Making Unhealthy Places: The Built
Setting. Ispra: European Commission Joint Research Environment and Non-Communicable Diseases in
Centre. Khayelitsha, Cape Town.” Health and Place 39: 196-203.
61. Munda, G. and Nardo, M. (2005b). Constructing 76. Sonnino, R.(2016). “The New Geography of Food
Consistent Composite Indicators: The Issue of Weights. Ispra: Security: Exploring the Potential of Urban Food
European Commission Joint Research Centre. Strategies.” Geographical Journal 182: 190-200.
62. Napoli, M. (2011). “Towards a Food Security 77. Swindale, A. and Bilinsky, P. (2005) Household Dietary
Multidimensional Index (FIMI).” Master in Human Diversity Score (HDDS) for Measurement of Household
Development and Food Security, Roma Tre Food Access: Indicator Guide. Washington DC: Food and
University, Rome. Nutrition Technical Assistance Project.
63. Nardo, M., Saisana, M., Saltelli, A., Tarantola, S. et 78. Tawodzera, G., Riley, L. and Crush, J. (2016). The
al. (2005) Handbook on Constructing Composite Indicators: Return of Food: Poverty and Urban Food Security in
Methodology and User Guide. Paris: OECD. Zimbabwe After the Crisis. Urban Food Security Series
64. Orsini, F., Kahane, R., Nono-Womdim, R. and No. 22, AFSUN, Cape Town.
Gianquinto, G. (2013). “Urban Agriculture in the 79. Tuholske, C., Andam, K., Blekking, J., Evans, T. and
Developing World: A Review.” Agronomy for Sustainable Caylor, K. (2020). “Comparing Measures of Urban
Development 33: 695-720. Food Security in Accra, Ghana.” Food Security 12: 417-
65. Pérez-Escamilla, R. (2012). “Can Experience-Based 431.
Household Food Security Scales Help Improve Food 80. UNDP (2012). Africa Human Development Report
Security Governance?” Global Food Security 1: 120-125. 2012: Towards a Food Secure Future. New York: United
66. Pérez-Escamilla, R., Gubert, M., Rogers, B. Nations.
and Hromi-Fiedler, A. (2017). “Food Security 81. Vaitla, B., Coates, J., Glaeser, L., Hillbruner, C. et
Measurement and Governance: Assessment of the al. (2017). “The Measurement of Household Food
Usefulness of Diverse Food Insecurity Indicators for Security: Correlation and Latent Variable Analysis
Policy Makers.” Global Food Security 14: 96-104. of Alternative Indicators in a Large Multi-Country
67. Petkovová, L., Hartman, D. and Pavelka, T. (2020). Dataset.” Food Policy 68: 193-205.
“Problems of Aggregation of Sustainable Development 82. Van Wesenbeeck, C. (2018), “Disentangling Urban
Indicators at the Regional Level.” Sustainability 12: and Rural Food Security in West Africa.” West African
7156-7176. Papers No. 15, OECD Publishing, Paris.
68. Prosperi, P., Allen, T., Cogill, B., Padilla, M. and 83. Wagner, J., Hinton, L., McCordic, C., Owuor, S. et al.
Peri, I. (2016). “Towards Metrics of Sustainable Food (2019). “Do Urban Food Deserts Exist in the Global
Systems: A Review of the Resilience and Vulnerability South? An Analysis of Nairobi and Mexico City.”
Literature.” Environment Systems and Decisions 36: 3-19. Sustainability, 11: 1963.
69. Renzaho, A. and Mellor, D. (2010). “Food Security 84. Wheeler, T. and Von Braun, J. (2013). “Climate
Measurement in Cultural Pluralism: Missing the Point Change Impacts on Global Food Security.” Science 341:
or Conceptual Misunderstanding?” Nutrition 26: 1-9. 508-513.
70. Romero-Lankao, P., McPhearson, T. and Davidson,
D. (2017). “The Food-Energy-Water Nexus and Urban
Complexity.” Nature Climate Change 7: 233-235.
71. Ryan, J. and Leibbrandt, M. (2015). “Multidimensional
Food Insecurity Measurement.” SALDRU Working
Paper No. 160, University of Cape Town.

14