Sección General
Revista de economía mundial 67, 2024, 297-327
ISSN: 1576-0162
DOI: http://dx.doi.org/10.33776/rem.vi67.8211
The GeoGraphy of food inSecuriTy. a Taxonomical analySiS
La geografía de La inseguridad aLimentaria. un anáLisis taxonómico
Sergio Tezanos
Universidad de Cantabria
sergio.tezanos@unican.es
Rogelio Madrueño
Center for Advanced Security, Strategic
and Integration Studies (CASSIS),
University of Bonn
rmadruen@uni-bonn.de
Recibido: marzo 2024; aceptado: mayo 2024
abSTracT
The problem of food insecurity is worsening in a multi-crisis world
simultaneously affected by socio-economic, health, governance and
environmental problems. This article characterizes the geography of food
insecurity based on six influential theories that explain the emergence of the
most dramatic manifestation of food insecurity: famines. We build a taxonomy
of 98 developing countries using a clustering procedure and identify four
groups of countries with different vulnerabilities. The multidimensional analysis
depicts a complex map of the diversity of human vulnerabilities that trigger
hunger and famines across the world.
Keywords: Hunger, famine, food insecurity, international classification,
Sustainable Development Goals
reSumen
El problema de la inseguridad alimentaria empeora en un mundo
afectado simultáneamente por diversas crisis (socioeconómica, de salud, de
gobernanza y medioambiental). Este artículo caracteriza la geografía de la
inseguridad alimentaria basándose en seis influyentes teorías explicativas de
la manifestación más dramática de inseguridad alimentaria: las hambrunas.
Construimos una taxonomía de 98 países en desarrollo mediante un
procedimiento de análisis de conglomerados e identificamos cuatro grupos
de países con diferentes vulnerabilidades. El análisis multidimensional
muestra un mapa complejo de la diversidad de vulnerabilidades humanas que
desencadenan el hambre y las hambrunas en todo el mundo.
Palabras clave: Hambre, hambruna, inseguridad alimentaria, clasificación
internacional, Objetivos de Desarrollo Sostenible
JEL Classification/ Clasificación JEL: I12, J11, Q18, O10, Q01.
fundinG
This piece of research is part of a major research project (“The geography of hunger
in a multi-crisis world”) financed by the BBVA Foundation in Spain. The BBVA Foundation
is not responsible for the opinions, comments and content included in this article and/
or the results derived from it, which are the total and absolute responsibility of their
authors.
acknowledGmenTS
We would like to thank Valpy FitzGerald, Daniel Díaz Fuentes, Ángel Martínez
González-Tablas, Francesco Burchi, Daniele Malerba and Michael Brüntrup for their
helpful and intellectually stimulating comments in this piece of research. A preliminary
version of this article was presented and discussed in the 23rd World Economy Meeting
(Santander, May 2023) and in the EADI General Conference (Lisbon, July 2023). The
former congress presentation was awarded the “José Luis Sampedro Prize”. The views
expressed in this article, however, remain solely those of the authors. Of course, the
usual caveats apply.
Revista de economía mundial 67, 2024, 297-327
1. inTroducTion
Extreme food insecurity is one of the harshest manifestations of the socio-
economic, governance, health and environmental vulnerabilities that humanity
faces in the 21st Century. The United Nations (UN) has been trying for decades
to mobilize the support of the international community to put an end to this
global problem. In particular, the second of the Sustainable Development
Goals (SDG) is precisely to eradicate world hunger by 2030.
However, the current evolution of global food insecurity is alarming. Whereas
the prevalence of undernourishment decreased at the beginning of the 21st
Century, it began to increase in 2015. According to estimates provided by the
UN Food and Agriculture Organization (FAO, 2024a), in 2022 there were 735
million undernourished people (9.2% of the world population and 146 million
more than in 2015). Moreover, the 2023 Hunger Hotspots, elaborated by the
FAO and the World Food Programme (WFP), warned that acute food insecurity
continues to escalate, impacting 175 million people in 22 countries who
urgently require assistance (FAO and WFP, 2023). Consequently, humanity
falls back on the “zero hunger goal” and the current trend leads us to a world
with more than 800 million starving people in 2030.
With this aggravating context, the aim of this article is to characterize the
current geography of food insecurity by means of an international classification.
After this introduction, section two reviews the specialized literature on the
causes of famine, identifying four main contributions: the classical economic
theories, the Food Availability Decline approach, the entitlement approach,
and the political theories. In section three we explain our analytical strategy for
building an international classification that is theoretically-based on the main
theories of famine and is built using a cluster analysis procedure. In section four
we present the results of our taxonomical investigation. Section five concludes
by summarizing the main findings and explaining the policy implications of the
analysis.
2. liTeraTure review: The cauSeS of famineS
There are three relevant concepts that are often misunderstood and thus
we need to previously define them: hunger, famine and food insecurity.
On the one hand, the FAO (2024b) defines “hunger” as:
300 Sergio Tezanos · Rogelio Madrueño
“An uncomfortable or painful physical sensation caused by
insufficient consumption of dietary energy. It becomes chronic when
the person does not consume a sufficient amount of calories (dietary
energy) on a regular basis to lead a normal, active and healthy life. For
decades, FAO has used thePrevalence of Undernourishmentindicator
to estimate the extent of hunger in the world, thus “hunger” may also be
referred to asundernourishment.”
On the other hand, the concept of “famine” is more controversial to define.1
Rubin (2016: 11) explains that a famine is a “synergistic crisis” caused by
multiple causes that results in “[…] a discrete event identifiable by an increase
in mortality caused by mass starvation and diseases”. This definition has
three important implications: first, that a famine has multiple (and reinforcing)
causes. Second, that it is a “discrete” event (with an atypical occurrence) rather
than a “normal” situation. And thirdly, that the rapid increase of deaths that
accompanies famines is not only due to starvation but also due to the diseases
caused by malnutrition.
Finally, the FAO (2024b) defines “food insecurity” in the following terms:
“A person is food insecure when they lack regular access to enough
safe and nutritious food for normal growth and development and an
active and healthy life. This may be due to unavailability of food and/or
lack of resources to obtain food. Food insecurity can be experienced at
different levels of severity”.
Therefore, food insecurity is a broader term that encompasses both hunger
and famine: the lower levels of severity of food insecurity are characterised by
low levels of undernourishment (hunger) and the hights level of food insecurity
is reached when a famine (humanitarian crisis) is triggered.
Once these definitions have been clarified, we can now revise which are
the main theories that explain the emergence of the most acute level of food
insecurity: famines.2 For the sake of simplicity, we classify this literature into
four main theories: the classical economic theories, the FAD approach, the
entitlement approach and the political theories.
2.1. claSSical economic explanaTionS of famine
The causes of famines have been analysed since the inception of Economics
as a Social Science. There are two main classical contributions to this analysis:
the Smithian and the Malthusian approaches. Their common feature is that
they carry out market-oriented analyses in order to identify the main “supply
1 For further detail, both Devereux (1993) and Rubin (2016) exhaustively review different contending
definitions of famine.
2 Comprehensive revisions of this debate can be found in Devereux (1993), Rubin (2016) and
Tezanos (2024).
301
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
cause” of a famine: either inadequate State interventions or excessive
population growth.
The Smithian approach
Adam Smith asseverated in The Wealth of Nations that ‘[…] a famine has
never arisen from any other cause but the violence of governments attempting,
by improper means, to remedy the inconveniences of dearth’ (Smith, 1776:
526). Therefore, the main cause of famines are market interventions, such
as regulating the price of food, controlling the exports of food and forcing
suppliers to sell their stock of grain, to name a few examples. The policy
recommendation for remedying these distortions is obvious: guarantee free
functioning of the food market.
This theory was not empirically proven by Smith himself, and nowadays
it is generally accepted that ‘full protection against famines appears well
beyond the grasp of the market’ (Rubin, 2016: 28). The main problem is that
Smith neglected the importance of “market failures”, such as the existence of
imperfect information and the lack of sufficient infrastructure, which result in
the creation of food monopolies and oligopolies that distort free competition.
Malthusian approach
The British reverend Thomas Malthus argued that famines are the
consequence of food shortages due to excessive population growth. In his
Essay on the Principles of Population, Malthus (1806) explains that the world
population tends to grow ‘in geometric progression’, while the ‘means of
subsistence’ (referring mainly to food) grow ‘in arithmetic progression’, thus
irremissibly generating periodic famines, wars and epidemics that dramatically
reduce the population in order to adapt it to the level of food supply. Malthus’
policy recommendation was similar to Smith’s: a laissez faire approach, letting
the famine restore the population equilibrium.
However, this conception of hunger as a market problem associated with the
scarcity of the food supply has been contested. It seems that “contemporary”
famines have not significantly constrained the population growth, as Devereux
(2000) proved with an historical analysis of various famines occurring during the
20th Century, thus refuting Malthus. All in all, the weakness of the Malthusian
approach lies in its ceteris paribus assumption, especially in relation with
technology, which has gradually improved over time, hence raising aggregate
agricultural productivity.
2.2. food availabiliTy decline approach
This second approach also focuses on the supply side of the problem.
According to the Food Availability Decline (FAD) approach (as named by Sen
in 1981), famines are determined by a temporal scarcity of food in particular
areas. This scarcity is usually caused by natural disasters, such as floods and
droughts.
302 Sergio Tezanos · Rogelio Madrueño
In contrast with the pessimistic prediction of Malthus, the FAD approach
offers an optimistic solution to famines, which consist of exponentially increasing
the capacity to produce food, relying heavily on the technological progress
of the agricultural sector (the so-called “green revolution”). Nevertheless,
prospects for the future according to this approach are not so optimistic, given
the aggravation of the climate change.
The FAD approach has relevant theoretical limitations that are related to
its “implicit” assumptions. On the one hand, it assumes that famine-affected
countries are totally closed economies, thus neglecting the existence of
international food markets that can alleviate the episodes of food scarcity. As
a consequence, this approach is unable to explain why some countries severely
affected by droughts do not suffer famines (for example, Spain), while others do
(such as the Sahel countries). On the other hand, the FAD approach assumes
that everyone within a country is equally affected by a famine. Therefore, it is
incapable of explaining why some social groups have better access to food than
others. As Devereux (1993: 183) clearly said, ‘drought causes crop failure, but
vulnerability to drought causes famine’. That is to say, the FAD causality link
between a disruptive event (such as a drought) and famine seems theoretically
incongruent as it neglects the fundamental aspect of human vulnerabilities,
which is the actual transmission belt with famines.
2.3. enTiTlemenTS approach
The Indian economist Amartya Sen received the Nobel Prize in Economics
in 1998 for his contributions in this field. In his influential book Poverty and
famines, Sen (1981) viewed famines as “economic disasters” and not just as
food crisis. He analysed four famines: Bengal (India, 1943), Ethiopia (1972-
1974), the Sahel region (1970s) and Bangladesh (1974). His main finding was
that these crises occurred without a “significant” reduction in food availability,
from which he deduced and generalized that famines are not essentially a
supply problem, but rather a demand problem associated with poverty and
people’s lack of “entitlements” to access food markets.
Drèze and Sen (1989: 23) defined the “entitlement” of a person as ‘[…] the
set of alternative commodity bundles that can be acquired through the use of
the various legal channels of acquirement open to that person’. In particular,
the entitlement of a person has two components: their “initial endowment” and
their “entitlement mapping” (which consist of the set of alternative commodity
bundles that can be obtained given the initial endowments). Furthermore,
entitlement relations in a market economy are based on five different types
of ownership: production, trade, labour, transfer and inheritance. Households
combine these types of ownership in order to access food either producing it
or buying it. Although the well-functioning of the market is crucial to facilitate
the access to food, households that lack the appropriate entitlements can
starve even when there is food available at the local markets. Famine is
thus determined by a double failure: an “entitlement failure” that affects a
303
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
large proportion of the population, and the “State failure” to protect those
entitlements.
Therefore, according to Sen the main cause of famine is not the lack of
food but the disruption in the availability to access food. This approach allows
a more precise study of famine than the previous theories, as it distinguishes
across different socio-economic groups and identifies the victims of famine.
As Sen (1981: 162) stated, ‘the entitlement approach provides a general
framework for analysing famines rather than one particular hypothesis about
their causation’ (as is the case in the classical economic and FAD approaches).3
2.4. poliTical explanaTionS of famine
There are two political explanations of famine: on the one hand, the “political
system approach” maintains that the lack of democracy causes famines. And,
on the other hand, the “political accountability approach” considers that
governments are responsible for famines.
Political system approach
Sen (1999 and 2009) also formulated the influential hypothesis that
famines do not happen in countries with democratic regimes and free press
because in democracy political leaders must be receptive to their citizens’
demands. This proposition emphasizes the importance of the instrumental role
of democracy and political freedoms for the prevention of major economic,
political and natural disasters.
Despite the influence of this hypothesis, Sen has never empirically verified
it and, consequently, several studies have analysed the relation between
democracy and famines. There are two possible interpretations of this relation:
On the one hand, a “deterministic interpretation” which conceives
democracy as the definitive solution for famines (i.e. democratic systems
always prevent famines). Some studies have carried out qualitative case-
analysis which reject this interpretation by identifying counter-examples of
famines that took place in democratic regimes, such as the famines in Ireland
(1845-1849), Bangladesh (1974), Sudan (1986-1988), Malawi (2002), Niger
(2005) and Madagascar (2021) (see, for example, de Waal, 1997; and Rubin,
2010).
On the other hand, there is a “probabilistic interpretation” which considers
that democracy lowers the intensity and magnitude of famines. For example,
Rubin (2009 and 2010), Plümper and Neumayer (2009), Burchi (2011) and
Rossignoli and Balestri (2018) run quantitative regression analyses but did not
offer conclusive results, as some studies supported Sen’s thesis while others
rejected it (once other relevant factors are considered).
3 Constructive revisions and critiques of Sen’s entitlement approach can be found in de Waal (1990)
and Gasper (1993), among others.
304 Sergio Tezanos · Rogelio Madrueño
Political accountability approach
The political accountability approach assumes that famines are politically
determined (they are the consequence of political decisions) and hence the
analysis of famines should focus on identifying those political actors that —
directly or indirectly— promote the emergence of famines.
Several empirical studies have applied inductive methodologies for capturing
the political variables that cause famines. The first contributions (which were
published immediately after the Cold War) recognised the role of conflict in
famine causation, which was not adequately captured by the entitlement
approach. As a result, the theory of “complex political emergencies” was
elaborated, and numerous studies demonstrated the presence of certain social
groups that gain advantages from famines. For example, the seminal study by
Macrae and Zwi (1992) analysed six African famines (Angola, Ethiopia, Liberia,
Mozambique, Somalia and Sudan) that took place in 1991 and 1992, and
were caused by the use of food as a ‘weapon of war by omission, commission
and provision’ (Macrae and Zwi, 1992: 299).
More recently, several studies have further verified the importance of the
political triggers of famines. For example, Tyner and Rice (2016) argued that
the famine that took place during the Cambodian genocide (1975-1979) was
a “calculated policy”. Gooch (2017) and Kasahara and Li (2020) analysed the
Great Chinese Famine (1959-1961), which coincided with the launch of the
Great Leap Forward (the communist agricultural and industrial modernisation
plan), and concluded that the famine was motivated by the negligence of the
government. Furthermore, with an aggregated and long-term perspective, de
Waal (2018) analysed the “structural causes” and the “proximate triggers” of
famines over an extended period of 140 years (between 1870 and 2010),
and concluded that almost all famines have multiple causes but that the most
relevant ones were related to political decisions and military tactics.
All in all, the advantage of the political accountability approach is that it
contributes both to revealing the main causes of famine, and to identifying
which political actors carry the main responsibility.
3. research STraTeGy
3.1. analyTical modeL
We have conceptualised famine as the final stage of a process of extreme
human vulnerabilities that reveals a humanitarian crisis caused by a number
of interrelated “triggers”. Therefore, the first step in building an international
taxonomy of global hunger and famines is to decide which are the most relevant
classificatory variables. This decision is based on our previous literature review.
We assume that developing countries can be classified by the synergistic
interaction among the following six complex explanatory variables (“triggers”)
of famine:
305
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
i. Existence of obstacles to the free functioning of the food market (Smithian
approach).
ii. Excessive population growth in relation to each country’s capacity to
produce food (Malthusian approach).
iii. Emergence of disruptive events (droughts, floods, blockades, etc.) that
sharply reduce the amount of food available to the population (FAD
approach).
iv. Failure of entitlements to access food (entitlement approach).
v. Absence of democracy and free press (political system approach).
vi. Lack of accountability of the government in the prevention of famines
(political accountability approach).
Figure 1 depicts these various relations and constitutes the framework for
building our international classification.
3.2. SelecTion of variableS and period of analySiS
The second task for building our taxonomy is to decide which are the most
appropriate proxies to measure the six afore-mentioned theories of famine.
Table 1 summarises this information. As is always the case in Social Sciences,
the selected proxies are far from perfect, as they oversimplify the complexity
of each of the analysed theories. Nevertheless, we take this selection very
seriously in order to guarantee meaningful results and a wide geographical
coverage.
fiGure 1. inTerrelaTionS explaininG hunGer and famine
Source: authors.
306 Sergio Tezanos · Rogelio Madrueño
Table 1. dimenSionS, variableS, periodS and SourceS
# Famine approach Main cause Proxy Variable code Source Period
1 Smithian approach Barriers to the free functioning
of the food market
Index of Economic
Freedom Economic_freedom Kim (2023) 2023
2 Malthusian approach Excessive population growth
relative to food production Gross per capita food
production index Food_production FAO (2024a) 2020
3 FAD approach Food shortages due to disrup-
tions
4 Entitlement approach Entitlement failure to access
food
Multidimensional Poverty
Index Poverty UNDP and OPHI
(2023)
2010-2021 (last
available year)
5 Political system approach Absence of democracy and
free press Polity score Democracy Center for Systemic
Peace (2023) 2018
6Political accountability
approach
Government’s lack of account-
ability Voice and Accountability Accountability Kaufmann et al. (2023) 2021
War and violence
Internally displaced
persons by conflict and
violence (%)
War World Bank (2023a) 2021
Classificatory variable Hunger Prevalence of undernour-
ishment (%) Hunger FAO (2024a) 2020
Source: authors.
307
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
• Smithian approach: we use the Index of Economic Freedom, which is
co-published by The Wall Street Journal and the Heritage Foundation.
The index covers 184 countries and measures 12 economic freedoms,
grouped into the following four broad categories (Kim, 2023):
i. Rule of Law: property rights, government integrity and judicial effectiveness.
ii. Government size: government spending, tax burden and fiscal health.
iii. Regulatory efficiency: business freedom, labour freedom and monetary
freedom.
iv. Open markets: trade freedom, investment freedom and financial freedom.
Each of these 12 economic freedoms is graded on a scale of 0 to 100. We
use each country’s overall score, which is the equally weighted average of the
12 variables.
• Malthusian and FAD approaches: we proxy both theories with the
same variable in order to measure the final implication of the existence
of a relative scarcity of food, which is the decline in the amount of food
relative to each country’s population4. In particular, we use the gross per
capita food production index elaborated by FAO (2024a). The index
shows the agricultural production for each year in comparison with
the base period 2014-2016; it is based on the sum of price-weighted
quantities of different agricultural commodities after deductions of
quantities used as seed and feed.
• Entitlement approach: we use the Multidimensional Poverty Index
(MPI), elaborated by the UN Development Programme and the Oxford
Poverty and Human Development Initiative (UNDP and OPHI, 2022).
The MPI measures 10 deprivations at the household level in health,
education and standard of living. It uses micro data from household
surveys to assign a deprivation score to each person. The three
dimensions are equally weighted and the maximum deprivation score
is 100 percent. In particular, we use the MPI value, which measures the
proportion of the population that is multidimensionally poor adjusted
by the intensity of the deprivations5.
• Political system approach: we use the indicators elaborated by the
Center for Systemic Peace (Marshall and Elzinga-Marshall, 2017). The
“Polity” examines the quality of democratic and autocratic regimes,
rather than discreet and mutually exclusive forms of governance. The
Polity score captures the regime authority spectrum on a 21-point
4 The alternative of using two different proxies for measuring the Malthusian and the FAD approaches
results in a high bi-variate correlation that is problematic for cluster analysis, as we will explain in
section 3.3.
5 The MPI value is the product of the incidence of multidimensional poverty and the intensity of
poverty.
308 Sergio Tezanos · Rogelio Madrueño
scale ranging from -10 (fully institutionalized autocracies) to +10 (fully
institutionalized democracies).
• Political accountability approach: as this approach is especially broad
in pointing out possible triggers for famine, we try two different proxies.
On the one hand, we use an overall measure of the governments’ levels
of accountability. The voice and accountability indicator, elaborated
by Kaufmann et al. (2010) as part of their Worldwide Governance
Indicators, measures the level of citizens participation in selecting
their government, and the level of freedoms of expression, freedom
of association and free media. It is constructed by averaging and
rescaling the data to run from 0 to 1. On the other hand, we try to
measure the single most determinant trigger of modern famines, this
being the existence of wars and violent conflicts. We proxy the intensity
of war with the percentage of internally displaced persons by conflict
and violence. Internally displaced persons have been forced to leave
their homes, but they have not crossed an international border, due
to armed conflicts, situations of generalized violence and violations
of human rights. Displaced people are alienated from their previous
sources of income (which implies the loss of their assets and, hence, the
loss of their entitlements to food) and quite often become dependent
on aid. As Devereux (1993: 156) pointed out, ‘the problems are more
acute when those affected are farmers. Displaced from their land, they
are producing nothing for themselves nor for the market’.
Apart from these proxies, we add a classificatory variable that will help
us to interpret the cluster results: the prevalence of undernourishment (as a
percentage of total population), which is the indicator 2.1.1. of the SDG and
constitutes the “departing point” of a country’s vulnerability process that may
lead to famine.
Regarding the population of study, we focus on “developing countries”,
where problems of hunger and famine are more intense. In particular, we aim
at the 136 low and middle income countries according to the World Bank’s
income classification (World Bank, 2023b).
In relation to the period of analysis, we try to build the most contemporary
taxonomy, so we use the last available year for each of the proxies, as it is
specified in the last column of Table 1.
Finally, we try to keep a reasonable sample size in relation to the number
of clustering variables. Formann (1984) proposed a simple rule: the number
of countries must be equal or larger than 2k, where k represents the number
of clustering variables.6 With our sample of 98 countries, we limit the cluster
analysis to a maximum of six variables (table 2).
6 Nevertheless, according to Mooi and Sarstedt (2011), very few studies comply with Formann’s
criterion, as it is very restrictive.
309
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
3.3. cluSTer procedure
The third step in this research strategy is to select an appropriate
methodology for building an international taxonomy. We utilize SPSS to run
a hierarchical cluster analysis using the Ward agglomeration method. As the
variables have different scales, we standardise them with the range -1 to 1
method, and we compute the squared Euclidean distances between countries.7
The sample includes 98 of the 136 developing countries (accounting for more
than 95% of the aggregate developing countries’ population).8
A relevant preliminary analysis it to verify if there is substantial collinearity
across the six variables used to proxy different theoretical causes of famine.9
The bivariate correlation matrix (table 3) shows that two variables (Democracy
and Accountability) have a statistically significant and relatively high correlation
coefficient (close to 0.8), indicating possible collinearity. As we have considered
two proxies for the political accountability approach, we choose the variable
related to armed conflicts (War) instead of the more general variable of voice
and accountability, as the former is not highly correlated with any other
classificatory variable (thus avoiding the unwanted information redundancy).10
The subsequent step involves determining the desired number of country
groups for our classification. We take this decision using two statistical tools:
the dendrogram and the variance ratio criterion (see appendixes 1 and 2).
Both methods recommend a four-cluster solution which results in limited
7 See Tezanos and Sumner (2013) for a detailed methodological explanation on the use of cluster
analysis for building an international development taxonomy.
8 There are two types of missing countries: those with limited statistical information (Cuba, Eritrea,
Kosovo, Lebanon, Libya, North Korea, Uzbekistan and Palestine), and insular nations with populations
below one million.
9 When highly correlated variables are used, they tend to be overrepresented in the cluster analysis.
That is why Mooi and Sarstedt (2011) and Everitt et al. (2011) do not recommend to use pairs of
variables with correlations above 0.9.
10 We further assume that the Polity score is a useful proxy for both democracy and accountability,
as democracy is a political system designed for making governments accountable for their citizens.
Table 2. deScripTive STaTiSTicS
NMinimum Maximum Mean Std. Deviation
Economic_freedom 110 29.5 71.8 55.29 7.78
Food_production 113 73.32 159.06 100.97 13.04
Poverty 113 0.0004 0.6013 0.13 0.15
Democracy 102 -9 10 3.37 5.60
Accountability 113 -1.92 1.16 -0.41 0.78
War 109 0 31.24 1.56 4.30
Hunger 104 2.4 53.1 15.67 13.15
Valid N (listwise) 98
Source: authors.
310 Sergio Tezanos · Rogelio Madrueño
Table 3. correlaTion maTrix
Economic_
freedom
Food_
production Poverty Democracy Accountability War Hunger
Economic_
freedom
Pearson
Correla-
tion
1 0.125 -0.302** 0.356** 0.536** -0.051 -0.429**
Sig.
(2-tailed) 0.192 0.001 0 0 0.601 0
N110 110 110 102 110 106 103
Food_
production
Pearson
Correla-
tion
0.125 1 -0.012 0.182 -0.008 0.082 -0.196*
Sig. (2-tai-
led) 0.192 0.904 0.068 0.936 0.396 0.046
N110 113 113 102 113 109 104
Poverty
Pearson
Correla-
tion
-0.302** -0.012 1 -0.123 -0.347** 0.277** 0.672**
Sig. (2-tai-
led) 0.001 0.904 0.217 0 0.004 0
N110 113 113 102 113 109 104
Democracy
Pearson
Correla-
tion
0.356** 0.182 -0.123 1 0.789** -0.261** -0.18
Sig. (2-tai-
led) 0 0.068 0.217 0 0.008 0.074
N102 102 102 102 102 101 99
Accountability
Pearson
Correla-
tion
0.536** -0.008 -0.347** 0.789** 1 -0.381** -0.390**
Sig. (2-tai-
led) 0 0.936 0 0 0 0
N110 113 113 102 113 109 104
War
Pearson
Correla-
tion
-0.051 0.082 0.277** -0.261** -0.381** 1 0.445**
Sig. (2-tai-
led) 0.601 0.396 0.004 0.008 0 0
N106 109 109 101 109 109 102
Hunger
Pearson
Correla-
tion
-0.429** -0.196* 0.672** -0.18 -0.390** 0.445** 1
Sig. (2-tai-
led) 0 0.046 0 0.074 0 0
N103 104 104 99 104 102 104
* Significant correlation at the 0.05 level (2-tailed).
** Significant correlation at the 0.01 level (2-tailed).
Source: authors.
311
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
dissimilarities within each group. In particular, the most dissimilar country in
the sample is Syria, being the last country to be grouped.
A final methodological step involves investigating which variables have
a greater influence on distinguishing between country groups. The one-way
ANOVA analysis verifies if there are significant differences across clusters. In
our case, the six variables are statistically significant (table 4) and the most
relevant variable for discerning groups (i.e. the one with the largest F statistic)
is Democracy, followed by Poverty, Hunger and War. By contrast, the least
influential variables are Economic freedom and Food production per capita.
This last result coincides with our previous literature review, as both classical
economic theories (Smith’s and Malthus’) and the FAD approach have been
empirically refuted as reasonable explanations for the emergence of famines.
4. diScuSSion of reSulTS
The cluster analysis classifies 98 developing countries into four groups, with
each group comprising countries from diverse geographical regions (map 1).
In order to characterize the four groups of countries we compute the
cluster “centroids” (which are the average values of the variables for the
countries included in each cluster). We focus our description of the country
groups on those variables that have low variability within each cluster in order
Table 4. one-way anova analySiS
Sum of Squares df Mean Square F Sig.
Economic_freedom Between Groups 854.42 3 284.81 6.2420 0.001
Within Groups 4,289.28 94 45.63
Total 5,143.70 97
Food_production Between Groups 3,022.73 3 1007.58 6.8160 0.000
Within Groups 13,895.14 94 147.82
Total 16,917.86 97
Poverty Between Groups 1.54 3 0.51 52.7030 0.000
Within Groups 0.91 94 0.01
Total 2.45 97
Democracy Between Groups 2,304.32 3 768.11 116.0780 0.000
Within Groups 622.02 94 6.62
Total 2,926.34 97
War Between Groups 757.42 3 252.47 19.5940 0.000
Within Groups 1,211.22 94 12.89
Total 1,968.64 97
Hunger Between Groups 9,834.62 3 3278.21 41.6140 0.000
Within Groups 7,404.92 94 78.78
Total 17,239.54 97
Source: authors.
312 Sergio Tezanos · Rogelio Madrueño
to avoid spurious interpretations (as centroids with high dispersion are not
representative of that cluster) (Table 5).11
• Cluster 1 (C1): poor and conflicted countries facing severe risk of
famine
14 countries, 11 of which are located in Sub-Saharan Africa, two in the
Middle-East region (Syria and Yemen) and the remaining one in Central Asia
(Afghanistan). They have the highest poverty and hunger rates and most of
them are severely damaged by conflicts. Among them, seven are autocratic
regimes and the remaining seven are weak democracies. Although they have a
high prevalence of undernourishment, this outcome is not clearly related with
the levels of per capita food production; a result that provides evidence for
refuting the “food scarcity” theories (the Malthusian and the FAD approaches).12
11 See appendix 3 for detailed information on cluster memberships and the values of each variable
for each developing country.
12 The relation between food scarcity and hunger is incongruent. There are countries with both
a high prevalence of undernourishment and low levels of food production (the clearest example is
Somalia, the C1 country with both the lowest per capita food production and the largest proportion
of hungry people). Whereas other countries have high rates of undernourishment but relatively high
levels of per capita food production (such as Burundi and Mozambique).
map 1. world diSTribuTion of cluSTerS
Source: authors.
313
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
Syria requires further explanation. As we said before, it is the most
dissimilar country in the whole data set. The cluster procedure locates Syria
in C1 because of the similarities with the other 13 countries in this group
in all variables except on the MPI. The problem is that there is no updated
information on the MPI, so we needed to include the last available year (2009),
before the Syrian war started in 2011. However, according to a recent report
by the UN Office for the Coordination of Humanitarian Affairs (OCHA, 2022),
extreme poverty has risen drastically since the beginning of the conflict, which
presumably means that the MPI value is also much higher than it was in 2009,
clearly locating Syria close to the average of this group (implying that this
country is not actually an outlier).
• Cluster 2 (C2): democratic regimes with chronic problems of hunger
These are 26 countries from different continents that have, on average,
the second highest prevalence of undernourishment. Some of these countries
Table 5. cluSTer cenTroidS
Economic_freedom Food_production Poverty Democracy War Hunger
C1
Mean 51.06 104.52 0.4023 0.14 8.51 35.31
N 14 14 14 14 14 14
Std. Deviation 5.21 12.03 0.1543 5.25 8.66 10.55
Minimum 39.40 86.38 0.0210 -9 0 18
Maximum 59.00 126.85 0.6013 6 31.24 53.1
C2
Mean 53.72 96.37 0.2077 6.12 0.39 22.44
N 26 26 26 26 26 26
Std. Deviation 6.84 9.10 0.1022 1.66 0.67 11.65
Minimum 33.10 73.32 0.0079 3 0 3.1
Maximum 64.80 116.32 0.3840 9 2.40 48.5
C3
Mean 53.23 97.45 0.1216 -4.36 0.53 12.72
N 22 22 22 22 22 22
Std. Deviation 6.97 7.46 0.1121 2.17 1.61 9.76
Minimum 32 89.11 0.0008 -9 0 2.4
Maximum 64.40 120.24 0.3270 -1 6.95 35.8
C4
Mean 58.83 108.80 0.0273 7.61 0.74 6.43
N 36 36 36 36 36 36
Std. Deviation 7.06 15.83 0.0476 1.66 2.21 4.02
Minimum 43 82.56 0.0004 2 0 2.4
Maximum 71.80 159.06 0.2629 10 10.16 18.6
Total
Mean 55.11 102.34 0.1499 3.46 1.71 16.21
N 98 98 98 98 98 98
Std. Deviation 7.28 13.21 0.1589 5.49 4.51 13.33
Minimum 32 73.32 0.0004 -9 0 2.4
Maximum 71.80 159.06 0.6013 10 31.24 53.1
Source: authors.
314 Sergio Tezanos · Rogelio Madrueño
produce very low levels of food in per capita terms (in particular, Haiti has
the lowest production of the whole sample), although there is an important
variability in this indicator. Similarly, poverty rates vary significantly across
countries; whereas Benin, Guinea-Bissau, Madagascar and Mali have very high
MPI values, Ecuador and Honduras have much lower values. A remarkable
feature is that C2 countries have relatively high standards of democracy and,
in general terms, are not affected by war and conflicts.13
• Cluster 3 (C3): autocratic regimes with chronic problems of hunger
22 countries with autocratic regimes and a relatively high prevalence of
undernourishment (only two countries within this cluster, China and Kazakhstan,
have rates below 2.5%). They share the important feature of a relatively low
incidence of conflicts. The notable exceptions are Sudan and Cameroon, which
have the largest proportion of internally displaced population in this cluster
(they are the only countries with more than 1% of their population internally
displaced).
It is worth mentioning that the information on Sudan is prior to the recent
break out of the civil war. In its current violent situation, Sudan is closer to C1
countries and it is already facing a severe risk of famine.
• Cluster 4 (C4): democratic regimes with moderate poverty and hunger
This is the largest cluster, with 36 countries scattered across all the
developing regions. They are all democratic regimes (Algeria being the country
with the weakest democracy). Poverty rates are comparatively low (in fact,
much lower than the rest of the clusters, with the sole exception of Senegal)
and the prevalence of undernourishment is also lower than in the other three
groups, although India, Iraq and Nicaragua have considerably higher rates than
the rest of the C4 countries (above 15%).
We explore the dissimilarities across clusters by means of a “web graph”.
In order to facilitate the interpretation, figure 2 rescales the magnitudes of the
six variables to a range of 0 to 100. The graph shows that C1 has the highest
indicators of poverty, hunger and war. The main characteristics in C2 are the
lowest incidence of wars and per capita food production. C3 does not have
either a minimum value, or a maximum one. And C4 has the highest scores in
terms of democracy, food production and economic freedom (although this
last variable is highly dispersed within the cluster), as well as the lowest ratios
of poverty and hunger.
Figure 3 summarises the cluster results organizing them into four quadrants
in terms of the four classificatory variables which, according to the ANOVA
analysis (see previous section), have the greatest discriminating power (these
13 A notable exception is Honduras, with almost 2.5% of its population internally displaced due to
conflicts.
315
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
are democracy, poverty, hunger and war). The cluster results show positive
associations between poverty and hunger on the one hand, and between
democracy and absence of conflicts, on the other hand. A remarkable feature
is that each cluster of countries has specific characteristics and vulnerabilities,
and, therefore, the multidimensional world map of hunger and famines cannot
be easily represented in a “linear” way.
5. concluSionS and implicaTionS
Despite the efforts made by the UN to mobilize international support
towards the 2030 Agenda for Sustainable Development, humanity falls back
on the SDG-2 of “zero hunger” as the current trend leads us to a world with
more undernourished people in 2030 than we had in 2015. This terrible trend
is a consequence of a multi-crisis world, simultaneously affected by socio-
economic, health, governance and environmental problems.
Hunger and famine are different phases of a multivariate and synergistic
process of aggravation of human vulnerabilities and deprivations. On the one
hand, hunger means undernourishment, and this problem becomes chronic in
some societies. On the other hand, famine is a humanitarian crisis characterized
by extreme levels of mass starvation that result in a sharp increase in mortality
and morbidity. Hunger and famine are thus different concepts but intimately
fiGure 2. compariSon of cluSTerS’ averaGeS
Source: authors.
Note: the centroids have been rescaled to a range of 0 to 100.
316 Sergio Tezanos · Rogelio Madrueño
connected in the continuum of human food insecurity, as chronic and severe
hunger can be the prelude for the emergence of famine.
In order to shed light on the complexity of the current geography of food
insecurity, we build an international classification of developing countries in
relation to the main determinants of famine. The following five results are
derived from this research effort:
1. There is a vast literature that analyses the determinants of famines which
can be classified into four main theories. Each of these theories identifies
“triggers” of famine, some of which are related to the supply side of the
food market, others to the demand side, while more recent explanations
highlight the importance of several political factors. No single theory offers
a comprehensive and universal explanation of the causes of famines,
applicable to every food security crisis (irrespective of where and when it
took place). On the contrary, some of these theories offer complementary
explanations which, brought together, help us to understand the complexity
of the process of human vulnerabilities that leads to famine. In this sense,
a “systemic approach” for understanding famines seems necesary for
fiGure 3. SpaTial repreSenTaTion of cluSTerS’ characTeriSTicS
Source: authors.
317
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
overcoming the single-factor explanations and for better understanding the
synergetic formation and evolution of these crises.
2. In order to build an international taxonomy, we assume that the complex
process of vulnerabilities that generate food insecurity stems from the
synergetic interaction of six possible causal explanations of famine
derived from the specialized literature: i) the existence of obstacles to the
free functioning of the food market (Smithian approach); ii) the excessive
population growth in relation to each country’s capacity to produce
food (Malthusian approach); iii) the emergence of disruptive events (like
droughts and floods) that sharply reduce the amount of food available
to the population (FAD approach); iv) the failure of the entitlements to
access food (entitlement approach); v) the absence of democracy and free
press (political system approach); and vi) the lack of accountability of the
government (political accountability approach).
3. We run a hierarchical cluster analysis in order to classify 98 countries
(accounting for more than 95% of the population in the developing world)
into four country groups with distinguishable features.
4. Although the six variables used in the classification are statistically
significant, only four of them are relevant in discriminating groups
(Democracy, Poverty, Hunger and War). By contrast, the variables related to
the supply side of the food market (Smith’s, Malthus’ and FAD approaches)
do not significantly contribute to explaining the cross-country differences
in terms of vulnerabilities to hunger and famine.
5. This multidimensional world map of food insecurity cannot be represented
in a linear way and, hence, our taxonomy depicts a complex map of the
variety of human vulnerabilities that trigger hunger across the world.
Apart from these research results, it is also worth reflecting on five policy
implications that are derived from our taxonomical analysis:
1. International classifications on global food insecurity serve a purpose for
identifying groups of countries with similar vulnerabilities and, in this sense,
are useful for guiding international development policies by highlighting a
set of geographical priorities. But our classification should not be confused
with an “early warning system” to prevent famines, as the later requires
detailed information at subnational levels and is based on current and
prospective conditions —not on past conditions, as it is the case in our
classification. For the purpose of alerting on famine risk, the Integrated
Food Security Phase Classification (IPC), elaborated by the FAO and other
14 organizations, is currently the most advanced available mechanism.
2. The greatest challenges for meeting the SDG-2 are located in the 40 countries
grouped in clusters 1 and 2. These countries have chronic problems of
hunger and are affected by severe human vulnerabilities. Therefore, the
international community needs to strengthen the cooperation efforts (both
318 Sergio Tezanos · Rogelio Madrueño
Nort-South and South-South initiatives) in these most vulnerable countries,
not only focusing on alleviating food security crises, but specially on solving
the multiple causes that generate these crises (such as building peace and
strengthening governments’ accountability).
3. In this international context, C4 countries have the potentiality to act as
donors from the Global South, as they have the best scores in terms of
democracy, food production, poverty and hunger. Therefore, they can share
with other developing countries their own experiences on fighting against
food insecurity, thus enriching the South-South cooperation system.
4. Despite the acute situation of clusters 1 and 2, “only” 42% of the world’s
starving people live in these countries, basically because India (located in
cluster 4) still accounts for 31,7% of the global undernourished population
(almost 230 million people). This figure sharply contrasts with the fact that
India has experienced a considerable reduction in the reception of Official
Development Assistance since it started being classified as a middle income
country. Beyond this simple income classification, it is obvious that meeting
the SDG-2 requires India to solve the vulnerabilities that generate hunger
with the collaboration of the international community.
5. Although we need more regional and international cooperation to improve
global food security, we are currently moving away from this ambition due
to the aggravation of the geopolitical rivalries in an increasingly multipolar
world. In particular, it is worth mentioning that solving the global food crisis
requires reintegrating both Ukraine and Russia into the world food markets,
which makes it even more urgent to put an end to the war.
We think that the taxonomic procedure that we propose in this article has
four main advantages:
Firstly, it is “innovative” because the multivariate statistical technique of
cluster analysis has not been previously used (to the best of our knowledge) to
analyse the geography of food insecurity.
Secondly, it is “objective” in the sense that we build the classification using
a meticulous statistical procedure. The only arbitrary decision that we have
taken is the selection criteria of the analysed countries. We included the so-
called “developing countries”, as these are the countries, in the 21st Century,
with a higher incidence of hunger and famine.
Thirdly, it is “multidimensional” and “synergetic”, as our taxonomy is based
on six different approaches that offer interrelated explanations of hunger and
famines.
And fourthly, our statistical procedure allows a “fine discrimination” of
reasonably homogenous groups of countries that share vulnerabilities to food
insecurity.
Nevertheless, our analysis also has four limitations that must be considered:
Firstly, there is the difficulty of measuring six complex theoretical
explanations of famine with a reduced set of “simple” proxies. We have carefully
319
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
selected and justified the indicators but, as always the case in Social Sciences,
the selected proxies are far from perfect and they over simplify the complexity
of each of the theories.
Secondly, cluster analysis has some margin of error, as happens in any
other multivariate statistical technique, and, in particular, it relies heavily on
the researchers’ selection of the similarity measure and the agglomeration
method. We have guided these decisions both on statistical theory and on
the coherence and interpretability of the results, and we have offered all the
necessary information to understand the statistical procedure.
Thirdly, international classifications have the intrinsic weakness of
“generalization”, which implies that this type of analysis should be understood
as complementary to other, more detailed, qualitative case-studies.
And fourthly, our analysis does not seek to test the validity of the six
causative approaches to famine. Instead, we depart from these theoretical
explanations of famine to build a theoretically-based taxonomy. Our future
line of research is thus moving to a causality analysis.
Ultimately, the main motivation for studying extreme food insecurity is to
contribute to its ending. With this modest piece of research we try to raise
awareness on this (aggravating) global problem and to improve our knowledge
on the formidable and necessary challenge of advancing towards the “zero
hunger” goal.
referenceS
Burchi, F. (2011) Democracy, institutions and famines in developing and
emerging countries. Canadian Journal of Development Studies, 32(1): 17-
31. https://doi.org/10.1080/02255189.2011.576136
Center for Systemic Peace (2023) Integrated Network for Societal Conflict
Research (INSCR) Data Page, available at https://www.systemicpeace.org/
inscrdata.html
De Waal, A. (1990) A Re-assessment of Entitlement Theory in the Light of
the Recent Famines in Africa. Development and Change, 21 (3): 469-490,
https://doi.org/10.1111/j.1467-7660.1990.tb00384.x
De Waal, A. (1997) Famine crimes. London: Villiers Publications.
De Waal, A. (2018) The end of famine? Prospects for the elimination of mass
starvation by political action. Political Geography, 62: 184-195. https://
doi.org/10.1016/j.polgeo.2017.09.004
Devereux, S. (1993) Theories of Famine. New York: Harvester Wheatsheaf.
Devereux, S. (2000) Famine in the 20th century. IDS working paper 105.
Retrieved from https://www.ids.ac.uk/files/dmfile/wp105.pdf.
Drèze, J., and Sen, A. (1989) Hunger and public action. Oxford: Clarendon
Press.
Everitt, B.S., Landau, S., Leese, M., Stahl, D. (2011) Cluster analysis. Chichester:
John Wiley and Sons.
320 Sergio Tezanos · Rogelio Madrueño
Food and Agriculture Organization of the United Nations (FAO) (2024a)
FAOSTATS, available at https://www.fao.org/faostat/es/#data
Food and Agriculture Organization of the United Nations (FAO) (2024b) Hunger
and food insecurity, available at https://www.fao.org/hunger/en/
Food and Agriculture Organization of the United Nations (FAO) and World
Food Programme (WFP) (2023) Hunger Hotspots. FAO-WFP early warnings
on acute food insecurity: November 2023 to April 2024 Outlook. Rome.
https://doi.org/10.4060/cc8419en
Formann, A. K. (1984), Die Latent-Class-Analyse: Einfuhrung in die Theorie und
Anwendung, Weinheim, Beltz.
Gasper, D. (1993) Entitlements Analysis: Relating Concepts and
Contexts; Development and Change. 24 (4): 679-718, https://doi.
org/10.1111/j.1467-7660.1993.tb00501.x
Gooch, E. (2017) Estimating the Long-Term Impact of the Great Chinese Famine
(1959–61) on Modern China. World Development, 89: 140-151. https://
doi.org/10.1016/j.worlddev.2016.08.006
Kasahara, H. and Li, B. (2020) Grain exports and the causes of China’s Great
Famine, 1959–1961: County-level evidence. Journal of Development
Economics, 146, 102513, https://doi.org/10.1016/j.jdeveco.2020.102513
Kaufmann, D., Kraay, A. and Mastruzzi, M. (2010) “The Worldwide Governance
Indicators: Methodology and Analytical Issues”, World Bank Policy
Research Working Papers, No. 5430, available at https://ssrn.com/
abstract=1682130
Kaufmann, D., Kraay, A. and Mastruzzi, M. (2023) The Worldwide Governance
Indicators. Interactive Data Access, available at https://info.worldbank.org/
governance/wgi/Home/Reports
Kim, A. B. (2023) 2023 Index of Economic Freedom, Washington DC: The
Heritage Foundation, available at https://www.heritage.org/index/pdf/2023/
book/2023_IndexOfEconomicFreedom_FINAL.pdf
Macrae, J. and Zwi, A. B. (1992) Food as an instrument of war in contemporary
African famines: A review of the evidence. Disasters, 16 (4): 299-321.
https://doi.org/10.1111/j.1467-7717.1992.tb00412.x
Malthus, T. R. (1806) An essay on the principle of population: or, A view of
its past and present effects on human happiness. With an inquiry into our
prospects respecting the future removal or mitigation of the evils which it
occasions. 3rd edition. London: J. Johnson.
Marshall, M. G. and Elzinga-Marshall, G. (2017) Global Report 2017: Conflict,
Governance and State Fragility, Center for Systemic Peace, Vienna, Virginia
(USA)
Mooi, E. and Sarstedt, M. (2011) A concise guide to market research. Berlin:
Springer-Verlag.
Plümper, T. and Neumayer, E. (2009) Famine mortality, rational political
inactivity, and international food aid. World Development, 37(1), 50–61.
https://doi.org/10.1016/j.worlddev.2008.05.005
321
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
Rossignoli, D. and Balestri, S. (2018) Food security and democracy: do inclusive
institutions matter? Canadian Journal of Development Studies, 39(2): 215-
233. https://doi.org/10.1080/02255189.2017.1382335
Rubin, O. (2009) The merits of democracy in famine protection—Fact or
Fallacy? European Journal of Development Research, 21(5): 699–717.
https://doi.org/10.1057/ejdr.2009.37
Rubin, O. (2010) Democracy and famine. Abington: Routledge.
Rubin, O. (2016) Contemporary Famine Analysis. Cham: Springer Briefs in
Political Science.
Sen, A. (1981) Poverty and famines: an essay on entitlement and deprivation.
Oxford: Oxford University Press.
Sen, A. (1999) Development as freedom. New York: Knopf.
Sen, A. (2009) The idea of justice. Harvard Massachusetts: Belknap Press.
Smith, A. (1776 / 2000) An inquiry into the nature and causes of the wealth of
nations, Princeton: Princeton Review.
Tezanos, S. and Sumner, A. (2013) Revisiting the Meaning of Development:
A Multidimensional Taxonomy of Developing Countries. Journal of
Development Studies, 49 (12): 1728-1745. https://doi.org/10.1080/002
20388.2013.822071
Tezanos, S. (2024) Why Do Famines Still Occur in the 21st Century? A Review
on the Causes of Extreme Food Insecurity. Under review.
Tyner, J.A. and Rice, S. (2016) “To live and let die: Food, famine, and
administrative violence in Democratic Kampuchea, 1975–1979”. Political
Geography, 52: 47-56. https://doi.org/10.1016/j.polgeo.2015.11.004
United Nations Development Programme (UNDP) and Oxford Poverty
and Human Development Initiative (OPHI) (2022) 2022 Global
Multidimensional Poverty Index (MPI): Unpacking deprivation bundles to
reduce multidimensional poverty. New York.
United Nations Office for the Coordination of Humanitarian Affairs (OCHA)
(2022) Syrian Arab Republic: 2023 Humanitarian Needs Overview,
available at https://reliefweb.int/attachments/5a13538d-a71c-4688-
88c7-4f7ce8f4b4e0/hno_2023-rev-1.12.pdf
World Bank (2023a) World Development Indicators, available at https://
databank.worldbank.org/source/world-development-indicators
World Bank (2023b) World Bank Country and Lending Groups, available at
https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-
world-bank-country-and-lending-groups
322 Sergio Tezanos · Rogelio Madrueño
appendix 1. dendroGram of counTrieS
Source: authors.
323
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
appendix 2. variance raTio criTerion (vrc)
# clusters VRCkwk
2 179.92 ..
3 209.37 4.22
4 243.05 -31.09
5 245.63 -30.51
6 217.70 ..
Source: authors.
324 Sergio Tezanos · Rogelio Madrueño
appendix 3. Cluster membership of developing countries
For reference Variables considered for the classification
Country Cluster
membership Population
GNI per capita,
Atlas method
(current US$)
Economic_
freedom
Food_
production Poverty Democracy Accountability War Hunger
Afghanistan 1 40,099,462 390 48 104.22 0.2717 -1 -1.5709 10.7582 29.8
Burkina Faso 1 22,100,683 830 58.3 93.64 0.5234 6 -0.1093 7.1491 18
Burundi 1 12,551,213 220 39.4 122.65 0.4089 -1 -1.4052 0.1514 40
Central African Republic 1 5,457,154 480 45.7 98.04 0.4613 6 -1.1956 12.6806 52.2
Chad 1 17,179,740 640 49.8 103.71 0.5170 -2 -1.4201 2.2818 32.7
Congo (Democratic
Rep.) 1 95,894,118 550 47.6 94.01 0.3312 -3 -1.1982 5.5676 39.8
Ethiopia 1 120,283,026 940 49.6 104.57 0.3666 1 -1.0691 2.9838 24.9
Guinea 1 13,531,906 1,020 54.2 118.34 0.3732 4 -0.9876 0.0000 30
Mozambique 1 32,077,072 480 51.3 113.18 0.4170 5 -0.6125 2.2914 32.7
Niger 1 25,252,722 590 54.9 106.87 0.6013 5 -0.3871 0.8870 19.8
Somalia 1 17,065,581 430 49 86.38 0.5140 5 -1.7492 17.3917 53.1
South Sudan 1 10,748,272 1040 53 98.01 0.5802 -7 -1.7265 12.7369 40
Syria 1 21,324,367 760 59 126.85 0.021 -9 -1.9167 31.2413 40
Yemen 1 32,981,641 840 55 92.77 0.2452 -7 -1.6845 13.0042 41.4
Benin 2 12,996,895 1,350 61 100.91 0.3677 7 -0.2375 0.0208 7.4
Botswana 2 2,588,423 6,430 64.8 92.44 0.0726 8 0.4576 0.0000 21.9
Côte d’Ivoire 2 27,478,249 2,420 61.6 104.3 0.2359 4 -0.4711 1.0991 4.4
Ecuador 2 17,797,737 5,960 54.3 88.98 0.0079 5 0.1061 0.0000 15.4
Gabon 2 2,341,179 6,440 55.8 90.24 0.0697 3 -0.8942 0.0000 17.2
Gambia 2 2,639,916 740 58 87.43 0.1980 4 -0.1029 0.0000 21.6
Guatemala 2 17,109,746 4,940 63.2 96.74 0.1335 8 -0.4619 1.4202 16
Guinea-Bissau 2 2,060,721 760 46 92.77 0.3407 6 -0.2363 0.0000 31.7
Haiti 2 11,447,569 1,430 50 73.32 0.1996 5 -0.9521 0.1485 47.2
Honduras 2 10,278,345 2,490 59.5 95.34 0.0512 7 -0.5871 2.4031 15.3
/ ... /
325
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
For reference Variables considered for the classification
Country Cluster
membership Population
GNI per capita,
Atlas method
(current US$)
Economic_
freedom
Food_
production Poverty Democracy Accountability War Hunger
Kenya 2 53,005,614 2,080 52.6 96.92 0.1708 9 -0.3663 0.3585 26.9
Lesotho 2 2,281,454 1,210 48.1 94.71 0.0844 8 -0.0198 0.0000 34.7
Liberia 2 5,193,416 630 47.9 89.09 0.2593 7 -0.0211 0.0000 38.3
Madagascar 2 28,915,653 490 58.9 89.6 0.3840 6 -0.2701 0.0097 48.5
Malawi 2 19,889,742 620 53 116.32 0.2311 6 0.0846 0.0000 17.8
Mali 2 21,904,983 820 55.9 116.14 0.3761 5 -0.7775 1.4882 9.8
Myanmar 2 53,798,084 1,170 49.6 99.73 0.1758 8 -1.6614 1.2064 3.1
Namibia 2 2,530,151 4,650 59.2 94 0.1847 6 0.5723 0.0000 18
Nigeria 2 213,401,323 2,080 54.4 93.46 0.2544 7 -0.6366 1.5126 12.7
Pakistan 2 231,402,117 1,470 48.8 106.98 0.1982 7 -0.8405 0.0449 16.9
Papua New Guinea 2 9,949,437 2,460 54.6 92.15 0.2633 5 0.0230 0.2412 21.6
Sierra Leone 2 8,420,641 500 52 97.34 0.2929 7 -0.0636 0.0653 27.4
Tanzania 2 63,588,334 1,100 59.5 100.13 0.2842 3 -0.7116 0.0000 22.6
Timor-Leste 2 1,320,942 1,140 46.3 88.3 0.2215 8 0.4591 0.0000 26.2
Zambia 2 19,473,125 1,030 48.7 105.65 0.2317 6 -0.3670 0.0000 30.9
Zimbabwe 2 15,993,524 1,530 33.1 102.7 0.1099 4 -1.1369 0.0000 30
Angola 3 34,503,774 1,710 52.6 91.69 0.2824 -2 -0.8409 0.0000 20.8
Bangladesh 3 169,356,251 2,570 52.7 106.44 0.1041 -6 -0.7696 0.2521 11.4
Cambodia 3 16,589,023 1,580 57.1 99.59 0.1703 -4 -1.4359 0.0000 6.3
Cameroon 3 27,198,628 1,590 52.9 90.54 0.2321 -4 -1.1602 3.3421 6.7
China 3 1,412,360,000 11,880 48 100.64 0.0161 -7 -1.6366 0.0000 2.4
Comoros 3 821,625 1,580 50.4 93.51 0.1808 -3 -0.7404 0.0000 20.4
Congo 3 5,835,806 1,970 48.5 89.3 0.1117 -4 -1.2381 0.9767 31.6
Egypt 3 109,262,178 3,350 49.1 89.92 0.0197 -4 -1.5098 0.0000 5.1
Eswatini 3 1,192,271 3,650 51.4 97.91 0.0813 -9 -1.2973 0.0000 11
Jordan 3 11,148,278 4,170 60.1 89.83 0.0015 -3 -0.7992 0.0000 16.9
Kazakhstan 3 19,000,988 8,880 64.4 109.31 0.0016 -6 -1.1355 0.0000 2.4
/ ... /
326 Sergio Tezanos · Rogelio Madrueño
For reference Variables considered for the classification
Country Cluster
membership Population
GNI per capita,
Atlas method
(current US$)
Economic_
freedom
Food_
production Poverty Democracy Accountability War Hunger
Lao 3 7,425,057 2,500 49.2 96.3 0.1083 -7 -1.6817 0.0000 5.1
Mauritania 3 4,614,974 1,950 55.3 96.14 0.3270 -2 -0.7652 0.0000 10.1
Morocco 3 37,076,584 3,620 59.2 94.43 0.0267 -4 -0.6073 0.0000 5.6
Rwanda 3 13,461,888 840 57.1 98.17 0.2310 -3 -0.9554 0.0000 35.8
Sudan 3 45,657,202 650 32 101.86 0.2794 -4 -1.4682 6.9540 12.8
Tajikistan 3 9,750,064 1,150 49.7 120.24 0.0290 -3 -1.7052 0.0000 8.6
Thailand 3 71,601,103 7,090 63.2 92.77 0.0021 -3 -0.7906 0.0573 8.8
Togo 3 8,644,829 960 57.2 98.16 0.1796 -2 -0.7944 0.0000 18.8
Turkmenistan 3 6,341,855 6970 46.2 89.11 0.0008 -8 -1.9147 0.0000 3.5
Uganda 3 45,853,778 760 54.2 97.02 0.2810 -1 -0.8218 0.0037 30
Viet Nam 3 97,468,029 3,590 60.6 100.96 0.0077 -7 -1.3042 0.0000 5.7
Albania 4 2,811,666 6,110 66.6 106.14 0.0027 9 0.0912 0.0000 3.9
Algeria 4 44,177,969 3,660 45.8 101.95 0.0054 2 -1.0095 0.0000 2.70
Argentina 4 45,808,747 9,960 50.1 104.68 0.0015 9 0.6196 0.0000 3.7
Armenia 4 2,790,974 4,850 65.3 84.66 0.0007 7 0.0592 0.0301 3.5
Bolivia 4 12,079,472 3,290 43 102.11 0.0378 7 -0.1099 0.0000 13.9
Bosnia and Herzegovina 4 3,270,943 6,810 63.4 120 0.0083 7 -0.3133 2.8126 2.4
Brazil 4 214,326,223 7,740 53.3 107.59 0.0163 8 0.2782 0.0098 4.1
Colombia 4 51,516,562 6,190 65.1 96.3 0.0197 7 0.0997 10.1618 8.2
Costa Rica 4 5,153,957 12,310 65.4 93.57 0.0020 10 1.0914 0.0000 3.4
Dominican Republic 4 11,117,873 8,100 63 111 . 21 0.0088 7 0.3018 0.0000 6.7
El Salvador 4 6,314,167 4,260 59.6 101.86 0.0325 8 -0.0552 0.0000 7.7
Georgia 4 3,708,610 4,700 71.8 116.97 0.0012 7 0.0153 8.2241 7.6
Ghana 4 32,833,031 2,280 59.8 103.63 0.1112 8 0.4673 0.0000 4.1
Guyana 4 804,567 9,410 59.5 110.99 0.0066 7 0.2467 0.0000 4.9
India 4 1,407,563,842 2,150 53.9 110.49 0.0688 9 0.1127 0.0359 16.3
Indonesia 4 273,753,191 4,180 64.4 108.15 0.0140 9 0.1552 0.0267 6.5
/ ... /
327
The geography of food insecuriTy. a Taxonomical analysis
revisTa de economía mundial 67, 2024, 297-327
For reference Variables considered for the classification
Country Cluster
membership Population
GNI per capita,
Atlas method
(current US$)
Economic_
freedom
Food_
production Poverty Democracy Accountability War Hunger
Iraq 4 43,533,592 4,760 46 131.1 0.0327 6 -0.9630 2.7266 15.9
Jamaica 4 2,827,695 5,190 67.4 98.43 0.0108 9 0.6280 0.0000 6.9
Kyrgyzstan 4 6,691,800 1,180 55.8 100.71 0.0014 8 -0.6055 0.0000 5.3
Mexico 4 126,705,138 9,590 63.7 106.65 0.0281 8 -0.0742 0.2991 6.1
Moldova 4 2,615,199 5,370 61.3 82.56 0.0035 9 0.0476 0.0000 6.7
Mongolia 4 3,347,782 3,730 63.9 159.06 0.0281 10 0.3189 0.0000 3.6
Montenegro 4 619,211 9,340 57.8 102.4 0.0049 9 0.1745 0.0000 2.4
Nepal 4 30,034,989 1,220 49.7 106.51 0.0744 7 -0.0891 0.0000 5.5
Nicaragua 4 6,850,540 1,950 54.8 123.57 0.0745 6 -1.2870 0.0000 18.6
North Macedonia 4 2,065,092 6,190 65.7 101.97 0.0014 9 0.1415 0.0053 3.3
Paraguay 4 6,703,799 5,740 62.9 109.53 0.0188 9 0.0091 0.0000 8.7
Peru 4 33,715,471 6,460 66.5 109.64 0.0292 9 0.1816 0.1780 8.3
Philippines 4 113,880,328 3,550 61.1 93.76 0.0242 8 -0.1505 0.0948 5.2
Senegal 4 16,876,720 1,570 60 158.24 0.2629 7 0.1926 0.0498 7.5
Serbia 4 6,834,326 8,460 65.2 112.65 0.0004 8 -0.1236 0.0000 3.3
South Africa 4 59,392,255 6,530 56.2 103.59 0.0249 9 0.7885 0.0000 6.9
Sri Lanka 4 22,156,000 4,030 53.3 123.23 0.0112 6 -0.0696 0.0542 3.4
Suriname 4 612,985 4,410 48.1 95.09 0.0112 5 0.3769 0.0000 8.2
Tunisia 4 12,262,946 3,540 54.2 115.9 0.0029 7 0.1871 0.0000 3.1
Ukraine 4 43,792,855 4,120 54.1 102.03 0.0008 4 0.0763 1.9501 2.8
Source: authors. See Table 1 for detailed information on each variable.
