Вы находитесь на странице: 1из 8

Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Global Food Security


journal homepage: www.elsevier.com/locate/gfs

Urban agriculture and food security: A critique based on an assessment


of urban land constraints
Madhav G. Badami a, Navin Ramankutty b,n
a
School of Urban Planning and McGill School of Environment, McGill University, Room 400, Macdonald-Harrington Building, 815 Sherbrooke Street West,
Montreal, QC, Canada H3A 0C2
b
Department of Geography, McGill University, 805 Sherbrooke Street West, Montreal, QC, Canada H3A 0B9

art ic l e i nf o a b s t r a c t

Article history: Urban agriculture (UA) is promoted because of its contribution to food security and poverty alleviation.
Received 2 December 2013 A considerable literature highlights these benefits, but there are also criticisms that they are overstated.
Received in revised form We review these divergent perspectives and assess the potential for UA to contribute to urban food
26 September 2014
security in different regions, based on a low threshold of urban land required to grow the daily vegetable
Accepted 3 October 2014
intake for the urban poor. We find that UA is feasible in these terms in high-income countries, but its
potential is low, except in the most optimistic scenario, in low-income countries, where it might be most
Keywords: useful. We conclude that UA can only make a limited contribution in achieving urban food security in
Urban agriculture low-income countries.
Food security
& 2014 Published by Elsevier B.V.
Urban poverty
Urban land use

1. Introduction UA, as well as for cities as a whole, by increasing direct access to a


diversity of nutritious food items; allowing farming urban house-
There continues to be considerable interest in urban agriculture holds to reduce the cash purchases of food and/or supplement
(UA), as reflected in its growing importance as a research and their incomes with food sales, thereby enabling them to spend on
policy issue on the international development agenda (Ellis and other important needs such as health and children's education;
Sumberg, 1998; Mougeot, 2011). As well, there is a considerable and increasing, and providing protection from seasonal or other
and growing literature on the subject, according to which UA is temporary shortages in, urban food availability. UA is also claimed
practiced widely. The United Nations Development Programme to contribute to urban waste recycling and minimization, efficient
(UNDP) put the number of urban dwellers engaged in UA world- water use and energy conservation, reduction in air pollution and
wide at 800 million (UNDP, 1996), which was then 30% of the soil erosion, and urban beautification; community building,
global urban population (United Nations, 2013), with 200 million empowerment and solidarity; biodiversity, climate change adap-
of these producing food for sale. The first number, and the tation and resilience, and disaster prevention and relief; and
widespread nature of UA, have been cited by various authors ecological and social urban sustainability (Dubbeling and de
and agencies, including FAO (1999), Koc et al. (1999), Lee-Smith Zeeuw, 2011; Lee-Smith, 2010; Maxwell et al., 1998; Mougeot,
(2010), Mougeot (2005, 2011), Redwood (2009), and Smit et al. 2005, 2011; Satterthwaite et al., 2010; Smit and Nasr, 1992; Smit
(2001). Further, Lee-Smith (2010) concludes, based on an exam- et al., 2001).
ination of household surveys conducted mainly in Sub-Saharan In view of its benefits and potential to contribute – signifi-
Africa, that UA has been growing at least as rapidly as the urban cantly, according to some authors – to food and nutrition security,
population. income generation, and poverty alleviation, UA has been actively
A multitude of socio-economic, environmental and resource promoted and publicized by local governments, including those of
use benefits have been claimed for UA in the literature, in both Dar Es Salaam and Kampala, and national governments, including
high- and low-income countries. These benefits include contribu- those of Kenya and Tanzania, and international institutions and
tion to food and nutrition security, livelihoods and income gen- funding agencies, including FAO, IDRC, UNDP, UN-Habitat, and the
eration, and poverty alleviation for urban households practicing UN University (FAO, 1999; Lee-Smith, 2010; Mougeot, 2011).
There have been criticisms of the claims made on behalf of
n
UA in the literature by, for example, Ellis and Sumberg (1998) and
Corresponding author. Current address: The Liu Institute for Global Issues and
the Institute for Resources, Environment, and Sustainability, 6476 NW Marine
Zezza and Tasciotti (2010). Among other things, they point out
Drive, Vancouver, BC, Canada V6T 1Z2. that, notwithstanding the considerable literature, there actually is
E-mail address: navin.ramankutty@ubc.ca (N. Ramankutty). a lack of reliable data related to UA, apart from exceptions such as

http://dx.doi.org/10.1016/j.gfs.2014.10.003
2211-9124/& 2014 Published by Elsevier B.V.

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
2 M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Maxwell et al. (1998), with participation figures varying widely, that UA provides many benefits, but rather, to investigate the
a problem that is only exacerbated by the lack of definitional and extent to which UA is capable of producing sufficient food to
methodological consistency. Indeed, while the UNDP (1996) put address the food security challenges of urban dwellers. To give UA
the number of urban dwellers engaged in UA worldwide at 800 the benefit of the doubt, however, we assess the adequacy of urban
million, Zezza and Tasciotti (2010) point out that this number was land for UA by using a low threshold, namely that of meeting the
in fact an estimate by the Urban Agriculture Network, intended to daily vegetable intake, rather than for all food requirements, and
provide a “thumbnail sketch” of the status of UA. Further, there are only for the urban poor, instead of entire urban populations (we
large variations in the participation rates claimed – according to elaborate on the reasons for this choice in the next section).
the FAO (1996), also quoting an UNDP document, only 100 million Also, we recognize that there are other important resource
were estimated to earn some income from UA, as against the 200 availability, socio-economic, political-institutional, and environ-
million figure quoted in UNDP (1996). More importantly, according mental factors that might affect food production in urban areas,
to Ellis and Sumberg (1998) and Zezza and Tasciotti (2010), much including crucially the opportunity cost of land for other uses, and
of the UA literature has been motivated by an advocacy agenda, the ability to obtain secure access to land for growing food, but we
rather than by analytical rigor. set these issues aside for this paper, given our purpose of
Based on a statistical analysis of nationally representative and investigating whether there is adequate land for UA to meet the
comparable survey data on UA participation rates, income gen- restricted objective of providing the daily vegetable intake for the
eration, calorie consumption and dietary diversity in 15 countries urban poor. If this is found not to be the case, it can be argued that
in different regions of the world, Zezza and Tasciotti (2010) UA is even more limited in terms of meeting the widely claimed
conclude that, while UA is associated with urban caloric adequacy objective of meeting the food security needs of urban populations
and dietary diversity in the majority of the countries for which this as a whole, and even without considering other factors. In any
relationship was tested, UA participation rates are not significant case, our study is, we believe, useful additionally – and regardless
outside Africa and Latin America, and that the share of total of any claims made in the literature – as a thought experiment of
income derived from UA is much lower than the participation how much urban land would be required, relative to the urban
rates. Further, UA is practiced largely by lower income groups, but land available, to cater for the daily vegetable intake of the urban
not by the very poorest – who lack access to land, which is the poor, as a means of assessing the physical potential of UA.
most critical factor for UA – except perhaps for the rearing of small
livestock (Ellis and Sumberg, 1998; Lee-Smith, 2010; Ruel et al.,
1998; Zezza and Tasciotti, 2010). Ellis and Sumberg (1998), based 3. Methodology and data
on their analysis of household surveys in Africa, suggest that UA is
a livelihood response that is motivated and enabled by a host of “Food security” is a complex issue, which is not always
socio-economic and institutional factors that might be unique to analyzed or operationalized in terms of its various dimensions
that region. In view of these criticisms, Ellis and Sumberg (1998) (Coates, 2013). Similarly, what is to be considered “urban” is
and Zezza and Tasciotti (2010) conclude that the potential for UA contentious, with different definitions used for different purposes,
to contribute to urban food and nutrition security, livelihoods and and lack of consistency among different countries and agencies
poverty alleviation is limited. (Montgomery, 2008; United Nations, 2013). Therefore, we discuss
below our methodological approach and choices, and the para-
meters of our analysis, in order to specify them as clearly as
2. Research objectives and rationale possible. Also, we discuss the data we use, and related issues.

Given the considerable on-going interest in and claims on 3.1. Target population
behalf of UA, and the critical perspectives related to its actual
and potential role in contributing to food security for urban We make conservative assumptions about the target popula-
dwellers, the objective of our paper is to present another critique tion for our analysis of the urban land requirements for UA. Since
of UA, using a more restricted perspective than those employed by the higher income groups in urban areas have a greater ability to
Ellis and Sumberg (1998) and Zezza and Tasciotti (2010). Our purchase and access food, and are therefore less likely to be food
critique is based on an assessment of the physical capacity of UA – insecure, we focus on evaluating the urban land availability for UA
and more particularly, given the centrality of land availability for to meet the food security needs of the urban poor. We focus on
food production – the adequacy of urban land to grow the food the urban poor because they form a significant share of urban
required for this purpose. populations in low-income countries, and urban poverty is grow-
A simple thought experiment demonstrates the challenge in ing as both population and poverty increasingly urbanize. Further,
this regard. Globally, the total area of land that is “urban” is the urban poor are likely to be highly vulnerable to food insecurity
estimated to range from 0.3 to 3.5 million km2, based on eight due to rising food prices, unemployment, and falling wages and
different data sets (Potere et al., 2009). The total area of land disposable household incomes, because they lack direct access to
devoted to cultivation of crops is estimated at 15 million km2 (FAO, food, rely largely on food purchases, which account for significant
2013; Ramankutty et al., 2008), which is 4–50 times larger. Even if proportions of total incomes, and lack social support networks.
100% of urban land was devoted to food production, and given that Lastly, the urban poor are potentially likely to benefit the most
half the total world population is urban, the food production in from UA; indeed, as already noted, UA is mainly practiced by the
urban areas (assuming average yields are similar) would fall short urban poor, although not necessarily the poorest (Ravallion et al.,
of needs by 2–25 times. This is not even considering that some of 2007; Ruel et al., 1998; Tacoli et al., 2013; Zezza and Tasciotti,
our nutritional needs are met from 34 million km2 of grazing land 2010).
(FAO, 2013). While this thought experiment shows its overall
scope, the global estimate provides little useful information on 3.2. Locus of food production
the geographic distribution of this challenge.
In order to better illustrate this challenge, therefore, we Much of the literature related to UA (for example, FAO 1999;
conduct an analysis of the urban land requirements for UA in a Koc et al., 1999; Lee-Smith, 2010; Mougeot, 2005; Smit et al., 2001)
geographically disaggregated manner. Our objective is not to deny either includes food production in “peri-urban” or suburban areas

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3

as part of UA, or discusses and counts urban and peri-urban 3.5. Choice of countries to analyze
agriculture as if they were similar activities. An FAO report
(Aragrande and Argenti, 2001) lists Bangkok and Madrid as cities We chose countries with large urban populations and a range
in which 60% of the metropolitan area is devoted to agriculture. of urban poverty rates that are representative of different geo-
If one defines UA so loosely (and capaciously), its potential for graphic regions (see Table 1). Availability of reliable and recent
contributing to food production is likely to be greatly over- data, in particular related to urban and national poverty, also
estimated. Ellis and Sumberg (1998) critique this lack of differ- determined our choice of countries. The bulk of our selected
entiation because the food outputs that are physically and eco- countries are in Sub-Saharan Africa, South Asia, and Latin America,
nomically feasible are likely to be different in urban as opposed to the regions characterized by the highest levels of urban poverty
peri-urban areas. Therefore, while definitions of UA in the litera- and malnutrition (Baker, 2008; FAO et al., 2012). We also included
ture might include food production in peri-urban areas, we focus countries in East Asia, and a few high-income countries in Europe
our analysis exclusively on food production in “built-up” urban and North America, to assess the variation in urban land require-
areas. We do so primarily because a realistic assessment of the ments for the purposes of contributing to food security for the
physical ability of UA to contribute to food production needs to urban poor in contexts with a range of urban populations, land
account for the constrained availability of land suitable for this areas, and poverty levels.
purpose in built-up urban areas. Note that in contrast to our
approach, Zezza and Tasciotti (2010), in their assessment of the 3.6. Urban land area
extent of UA, use data on agricultural activity conducted by urban
residents, regardless of the location of that activity. Lastly, we Estimating the extent of urban land is fraught with definitional
focus on food production in built-up urban areas regardless of issues, particularly related to what is considered to constitute an
whether the urban poor grow the food themselves; the urban poor “urban area”. There are many ways of estimating urban land area,
may either be the producers themselves, or be purchasers or based on urban population density thresholds to define urban
beneficiaries of this food. Of course, we recognize that mere food areas; census or jurisdictional definitions; economic criteria; and
production and availability will not guarantee food access and the presence of human-made structures (Montgomery, 2008).
security for the urban poor; we set aside this issue for this paper. While there is information on the extent of many cities of the
world, such information is incomplete. For our analysis of the
urban land requirements for UA, we use the new remote-sensing
3.3. Food items being grown
based MODIS 500m resolution data set (Schneider et al., 2010),
which, according to a recent global assessment of eight different
Next, we turn to the food items to include in our analysis.
global data sets, had the highest accuracy in terms of estimation of
In general, but particularly for the poor, cereals account for the
the extent of urban area (Potere et al., 2009). A more detailed
bulk of daily caloric intake. Despite this, we do not consider cereals
discussion of the MODIS 500 m data set and its advantages over
in our analysis because, first of all, the total harvested area for
other measures for this purpose is presented in the supplement.
cereals globally is currently 7 million km2 (FAO, 2013), which is
double the high estimate of the global urban area. In contrast, the
3.7. Urban population
total harvested area of “primary vegetables” (a set of 27 crops from
the FAOSTAT database; FAO, 2013) is 0.6 million km2, which is
We draw our urban population data at the country level from
comparable to the total global urban area. Besides, cereals, which
the United Nations ‘World Urbanization Prospects’ database
are suited for large-scale production in open spaces, are challen-
(United Nations, 2013). These estimates of course have limitations,
ging to grow in land constrained urban areas (consider paddy rice,
in that they critically depend on the definitions and estimation
as an extreme case). Vegetables, on the other hand, are one of the
approaches applied, which vary within and between countries, as
primary products of UA in practice, since they are highly valued
well as over time. Among the definitions that critically influence –
because of their nutritional content and contribution to dietary
and render extremely challenging – the estimation of urban
diversity, and are perishable (Ellis and Sumberg, 1998; Zezza and
populations is that related to “urban area”, as noted earlier (for
Tasciotti, 2010). Lastly, even though small livestock are feasible to
an in-depth discussion of the challenges, see 〈http://esa.un.org/
raise in built-up urban areas, we do not include them in our
unup/Documentation/Definition-Problems_1.htm〉; Montgomery,
analysis. Note in this regard that, although there are reports of the
2008). Notwithstanding these challenges, the urban population
importance of rearing small ruminants such as pigs and poultry in
data from the ‘World Urbanization Prospects 2011’ database is
low-income country cities (for example, World Bank/FAO, 2008),
perhaps the most reliable, up to date, inter-nationally comparable,
Zezza and Tasciotti (2010) found, based on their analysis of
and widely used and cited; besides, the sources of information on
nationally representative and comparable survey data on UA, that
which the data are based for various countries are well documen-
participation in livestock activities is much lower than in crop
ted 〈http://esa.un.org/unup/CD-ROM/Data-Sources.htm〉.
activities.
3.8. Urban poverty
3.4. Amount of food needed
The measurement of urban poverty is also fraught with con-
Finally, we need to decide the target quantity of vegetables ceptual and definitional issues, inconsistent approaches between
required to meet the dietary needs of our target population. A joint countries and over time, the lack of reliable data, and much debate
WHO/FAO expert consultation recommended a daily minimum and controversy. An important criticism of urban poverty esti-
intake of 400 g of fruits and vegetables per day (excluding starchy mates, which rely on income-based poverty measures, is that they
tubers) for the prevention of chronic diseases (WHO/FAO, 2003). underestimate and misrepresent urban poverty – see, for example,
The FAO ‘food supply’ database indicates a global average of 361 g Alkire (2013), Baker (2008), Dreze (2012), Ravallion (2010, 2013),
of vegetables per capita per day available for consumption (FAO, and Satterthwaite (2003) for these criticisms and responses. In
2013). On the basis of both of these figures, we choose a value of view of the criticisms of urban poverty estimates, and more
300 g of vegetables per capita per day as a conservative minimum particularly the variations in these estimates, we conducted our
target for our analysis of the urban land requirements for UA. analysis using three different measures (nationally reported rates,

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
4 M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Table 1
Urban agriculture potential in selected countries – key input parameters.

(1) Country (2) Region (3) United (4) Gross (5) Urban (6) % Urban (7) % Urban (8) % Urban (9) Built-up (10) National
Nations Human national income population poverty rate, poverty rate poverty rate area – vegetable yield,
Development per capita, (million) nationally (relative to $1.25/ (relative to $2/day MODIS reported by FAO
Classification (constant 2005 reported day PPP poverty PPP poverty line) 500 m (tonne/km2/yr)
PPP$) line) (km2)

Bangladesh Asia Low 1529 42.70 21.3 29.3 51.7 928.61 806.01
Canada NA Very high 35,166 27.70 11.9 – – 8611.46 2596.67
Chile LA Very high 13,329 15.40 15 1.5 2.8(   ) 3878.20 2471.80
China Asia Medium 7476 681.51 2.3(   ) 0.6(   ) 3.5 80,923.96 2313.65
(þ þ)
Colombia LA High 8315 35.34 30.3 7.3 14.0 3935.77 1839.93
DR Congo Africa Low 280(   ) 23.22 61.5( þ þ) 75.6( þ þ ) 82.1( þ þ) 2285.69 638.45
Ethiopia Africa Low 971 14.40 25.7 20.3 43.6 736.49 524.81
France Europe Very high 30,462 54.17 7.9 – – 12,163.85 2308.50
Ghana Africa Medium 1584 12.96 10.8 10.8 19.6 1219.50 496.90(–)
India Asia Medium 3468 388.29 20.9 28.9 57.6 31,240.56 1397.46
Indonesia Asia Medium 3716 122.89 9.2 11.9 31.9 11,172.13 962.51
Kenya Africa Low 1492 9.97(  ) 33.7 31.9 49.3 451.64 1254.67
( )
Mexico LA High 13,245 89.67 45.3 0.6(   ) 4.0 7266.84 1771.66
Nigeria Africa Low 2069 80.64 34.1 50.4 62.6 4631.26 611.93
Pakistan Asia Low 2550 63.97 13.1 12.3 35.4 5850.95 1262.75
Poland Europe Very high 17,451 23.31 11.0 – – 7934.64 3134.91
Tanzania Africa Low 1328 105.44 15.5 37.3 48.3 794.45 607.33
UK Europe Very high 33,296 12.35 10.0 – – 8600.94 2215.32
USA NA Very high 43,017( þ þ ) 49.70 17.4 – – 120,680.46 3221.25( þ þ )
( þ þ)

Data sources: Columns 3 and 4 – UN Human Development Report 2011 (United Nations, 2011); the figures in Column 4 are for 2011; Column 5 – figures for 2011, from World
Urbanization Prospects (United Nations, 2013); Column 6 – data for 2010 from the OECD Income Distribution and Poverty database (OECD, 2013) for Canada, France, Poland,
UK and USA; Data for the years 2005–2012 from the World Bank World Development Indicators database (World Bank, 2014a), for all other countries except China, for which
data for 2009 was extracted from the PovcalNet online poverty analysis tool (World Bank, 2014b) – see the Data and Analysis section for an explanation; Columns 7 and 8 –
calculated, except for China and India, as explained in the Data and Analysis section; for China and India, data for 2009 was drawn from PovcalNet; Column 9 – satellite data
from the University of Wisconsin (Schneider et al., 2010); Column 10 – national average vegetable yield data drawn from the FAOSTAT database (FAO 2013). Notes: The
maximum and minimum values in each column are indicated by þ þ and   . Abbreviations: FAO – Food and Agricultural Organization; LA – Latin America; and NA – North
America.

based on national poverty lines, and those relative to the $1.25/day and aeroponics in high-rise buildings (Despommier, 2009). At the
PPP poverty line, a common standard used internationally, and to same time, there are factors that can militate against high urban
the $2/day PPP poverty line, the median for all “developing” yields, such as urban water stress (McDonald et al., 2014), and
countries), drawn from or calculated based on data from OECD ground level ozone (Avnery et al., 2011a, 2011b; Holland et al.,
(2013) and World Bank (2014a, 2014b), to evaluate the sensitivity 2002). Also, various concerns have been raised regarding the
of our results to these different measures. In the supplement, we ability of “vertical farming” to produce the benefits claimed for it
discuss these measures in detail, as well as our rationale for (The Economist, 2010). In view of these issues, we take into
selecting them in the context of the on-going debate related to consideration the possibility of higher (as well as lower) urban
poverty estimation. vegetable yields relative to the national average yields reported by
FAO, to evaluate the sensitivity of the results of our analysis to
3.9. Vegetable yields variations in this parameter. We discuss the above issues in greater
detail in the supplement.
The FAOSTAT database (FAO, 2013) provides national-level data
on yields of ‘primary vegetables’ (a set of 27 crops). While yields 3.10. Scenario analysis
are sometimes obtained through field measurements, they are
most often calculated as the ratio of total production to total We conduct our analysis of the urban land requirements for
harvested area. Assuming that the national-level yield data apply UA for various selected countries in different regions, using data
to urban areas, we use this data to assess the physical ability, in compiled from various sources on the parameters we have
terms of the amount of urban land that would be required, to discussed (see Table 1).
produce a basic level of daily intake of vegetables for the urban The metric we compute is the percentage of the total urban
poor, in different countries. area needed to grow 300 g of vegetables per capita per day to feed
However, the possibility of urban vegetable yields being higher the urban poor in each of our selected countries, using the
than the national average values reported in the FAOSTAT database following expression:
must be considered. Indeed, there are several cases reported in the
100  ðUrbanPop  UrbPovRate  V egConsÞ=ðVegYield  UrbAreaÞ
literature, related to a range of vegetable crops in various coun-
tries, of vegetable yields that are higher (as well as lower) in urban where UrbanPop¼ total urban population (number of people);
relative to rural areas (see for example, Altieri et al., 1999; Central UrbPovRate¼the urban poverty rate (% of total urban population),
Agricultural Census Commission, 2003; Duchemin et al., 2009; as discussed above; VegCons ¼annualized target per-capita vege-
Grewal and Grewal, 2012; Potutan et al., 2000; Yi-Zhong and table consumption of 300 g/capita/day¼ 109.5 kg/capita/year;
Zhangen, 2000; Yoveva et al., 2000). Also, it is claimed that higher VegYield ¼vegetable yield (kg/km2/year), based on the FAOSTAT
yields, among other benefits, could potentially be achieved database, and as discussed below; and UrbArea¼total urban area
through “vertical farming”, employing drip irrigation, hydroponics (km2) from the MODIS 500 m data set.

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 5

Table 2
Best case, medium case and worst case scenarios – assumptions.

Best case scenario Medium case scenario Worst case scenario

% Urban Lowest rate, among nationally reported urban For Canada, France, Poland, UK and USA, the Highest rate, among nationally reported urban
poverty poverty rate, urban poverty rate relative to $1.25/ national urban poverty rate poverty rate, urban poverty rate relative to $1.25/
rate day PPP poverty line, and urban poverty rate For the other countries, the median rate, among day PPP poverty line, and urban poverty rate
relative to $2/day PPP poverty line (Columns 6,7 nationally reported urban poverty rate, urban relative to $2/day PPP poverty line (Columns 6,7
and 8 in Table 1), for countries other than Canada, poverty rate relative to $1.25/day PPP poverty and 8 in Table 1), for countries other than Canada,
France, Poland, UK and USA line, and urban poverty rate relative to $2/day France, Poland, UK and USA
PPP poverty line (Columns 6,7 and 8 in Table 1)
Built-up Built-up area – MODIS 500 m, multiplied by 1.5 Built-up area – MODIS 500 m Built-up area – MODIS 500 m
area
Vegetable National vegetable yield reported in the FAOSTAT National vegetable yield reported in the FAOSTAT National vegetable yield reported in the FAOSTAT
yield database, multiplied by 4.76, the ratio of the 75th database database, multiplied by 0.40, the ratio of the 25th
percentile to the median values of the urban yields percentile to the median values of the urban
relative to rural yields, for a range of vegetables in yields relative to rural yields, for a range of
various countries, based on several sources vegetables in various countries, based on several
sources

We are effectively evaluating the physical ability of UA, in terms available urban area. Midway between these extremes are Ghana,
of the adequacy of urban land, to grow the food required for India and Kenya, which will need to devote more than 20% (but
meeting the food security needs of the urban poor, based on low under 100%), and Indonesia and Pakistan, which will need
thresholds that are favorable to UA. These thresholds consist of a between 10% and 20% of their urban area. Note that even 10–
basic level of daily intake of vegetables (rather than the full daily 20% of the urban area in low-income country cities is a very high
caloric intake of cereals, vegetables and other food items), for the proportion to devote to growing vegetables, given the many other
urban poor (rather than for urban populations as a whole). The vitally important functions that are required of urban land,
only assumption we make that sets a high bar for UA is that the particularly given the high population densities.
urban area is restricted to the built-up area, but we believe this The expression for the proportion of urban area required for
assumption is justified, for the reasons discussed previously. growing the basic daily vegetable intake level of 300 g/capita/day
Our analytical approach is essentially similar to that of for the urban poor can be rewritten as
Martellozzo et al. (2014), in which one of us was a co-author,
except that we consider the urban poor specifically, while they 100  PopDensUrbPoor  VegCons=VegYield
consider the entire urban population. Also, we evaluate the urban
land requirements for UA under different scenarios, which they where PopDensUrbPoor¼population density of the urban poor,
do not. which is equal to (Urban population  Urban poverty rate)/Urban
Specifically in this regard, we test the sensitivity of our area, and VegCons and VegYield are as described earlier. Since the
evaluation of the ability of UA, in terms of the adequacy of urban per-capita vegetable consumption is a constant (the assumed
land to produce a basic level of daily intake of vegetables for the target of 300 g/capita/day), the proportion of urban area required
urban poor in our selected countries, to different estimates of for growing this basic daily vegetable intake for the urban poor is
urban poverty rates, as well as to alternative assumptions regard- directly proportional to the population density of the urban poor,
ing urban area, and vegetable yields, by means of three scenarios. and inversely proportional to the vegetable yield. Whereas the
While the different estimates and assumptions related to population density of the urban poor varies by a factor of around
urban poverty rates, urban area and vegetable yields that we 122 (from around 110 people/km2 in Chile to 13,460 people/km2
used in each of our scenarios are explained in detail in the for Bangladesh) in the Medium case scenario, for example, the
supplement, these estimates and assumptions are summarized in vegetable yields in this scenario only vary by about six-fold (from
Table 2 below. 497 t/km2 in Ghana to 3221 t/km2 in the USA). Therefore, the
proportion of urban area required for growing the basic daily
vegetable intake for the urban poor is driven far more by the
4. Results and discussion population density of the urban poor (i.e., higher the urban poor
population densities, the greater the urban area needed), than by
The proportion of urban area required for growing the basic vegetable yields. This is evident in Fig. 2, in which we plot the
daily vegetable intake level of 300 g/capita/day for the urban poor proportion of urban area required against the population density
is found to vary by more than two orders of magnitude across the of the urban poor in the Medium case scenario for all the countries
different countries in our sample, in each of our scenarios – see in our sample. This relationship is similar in the other scenarios,
Table 3, in which we present the key output parameters that we and is therefore not shown.
calculated, and Fig. 1, in which the countries are arranged in At the same time, of course, vegetable yields are important.
increasing order of proportion of total urban area required for this The urban land area requirements are so very large in the case of
purpose, in the Medium case scenario. At one extreme, in the Bangladesh, Democratic Republic of Congo, Ethiopia, Nigeria, and
Medium case scenario, the five high income countries, besides Tanzania, not only because of high densities of urban poor
Chile and China, would need to devote less than 3% of their urban populations, but also due to their low vegetable yields compared
area to this end, despite the high threshold for the poverty rate for to other nations. Note also that in Fig. 2, while Kenya has a higher
the high income countries, based on headcounts below 50% of the urban poor density relative to Tanzania and Ethiopia, and an urban
median income, after taxes and transfers (OECD, 2013). At the poor density not much lower than that of DR Congo (and even
other extreme are five countries (Bangladesh, Democratic Republic Nigeria), its urban area needs for UA are significantly lower than
of Congo, Ethiopia, Nigeria, and Tanzania), which would need for these other countries, because its vegetable yield is much
more than, and in some cases considerably over, 100% of their higher.

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
6 M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Table 3
Urban agriculture potential in selected countries – key output parameters.

(1) Country (2) Region Total vegetable consumption target of urban Urban poor density (1000 people/ % Urban area – MODIS required to produce
poor (1000 t/yr) km2) vegetables for urban poor

Best case Medium case Worst case Best case Medium case Worst case Best case Medium case Worst case

Bangladesh Asia 995.9 1368.9 2418.6 6.53 13.46 23.79 18.6 182.9 807.8
Canada NA – 361.0 – – 0.38 – – 1.6 –
Chile LA 24.6 47.4 252.9 0.04 0.11 0.60 0.04 0.5 6.6
China Asia 447.8 1716.4 2611.9 0.03 0.19 0.29 0.03 0.9 3.5
Colombia LA 281.9 543.2 1172.4 0.44 1.26 2.72 0.5 7.5 40.5
DR Congo Africa 1563.8 1923.4 2087.9 4.17 7.69 8.34 15.0 131.8 357.7
Ethiopia Africa 319.9 405.3 687.7 2.64 5.03 8.53 11.6 104.9 444.8
France Europe – 468.6 – – 0.35 – – 1.7 –
Ghana Africa 153.3 153.8 278.6 0.77 1.15 2.09 3.5 25.4 114.9
India Asia 8886.1 12,287.5 24,490.0 1.73 3.59 7.16 2.9 28.1 140.2
Indonesia Asia 1237.9 1604.4 4288.2 0.67 1.31 3.51 1.6 14.9 99.7
Kenya Africa 348.0 368.0 538.8 4.69 7.44 10.89 8.6 64.9 237.7
Mexico LA 60.9 391.7 4447.9 0.05 0.49 5.59 0.1 3.0 86.4
Nigeria Africa 3011.1 4451.1 5531.2 3.96 8.78 10.91 14.9 157.1 487.9
Pakistan Asia 864.1 917.6 2477.1 0.90 1.43 3.87 1.6 12.4 83.8
Poland Europe – 280.7 – – 0.32 – – 1.1 –
Tanzania Africa 209.6 504.8 653.4 1.61 5.80 7.51 6.1 104.6 338.6
UK Europe – 544.2 – – 0.58 – – 2.9 –
USA NA – 4915.3 – – 0.37 – – 1.3 –

Notes: Total vegetable consumption target calculated based on urban population, % urban poverty rate, and per-capita vegetable consumption target of urban poor; Urban
poor density calculated based on urban population, % urban poverty rate, and MODIS 500 m built-up area; % Urban area – MODIS required to produce vegetables for urban
poor calculated based on total vegetable consumption target, vegetable yield, and MODIS 500 m built-up area. For explanations of the assumptions used in calculating these
parameters in the Best case, Medium case, and Worst case scenarios, see the Data and Analysis section, and Table 2.

Fig. 1. The percentage of urban area required for growing the basic daily vegetable
Fig. 2. Percentage urban area required against the population density of the urban
intake level of 300 g/capita/day for the urban poor in different countries of the
poor in the Medium case scenario for all the countries in our sample. The countries
world, across 3 different scenarios with varying assumptions about urban areas,
in the oval near the origin of the scatter plot are, in order of increasing urban poor
urban vegetable yields, and urban poverty rates. For explanations of the assump-
density, Chile, China, Poland, France, USA, Canada, Mexico and UK. For an
tions used, see Section 3 and Table 2. Note the logarithmic scale on the y-axis. Also
explanation of the assumptions used in calculating the x- and y-axis parameters,
note that the best and worst-case scenarios are not estimated for Poland, USA,
see Section 3 and Table 2.
Canada, France, and the UK.

assumptions in the Best case relative to the Worst case scenario.


As for the proportion of urban area required to grow the basic First, the ratio of the highest urban poverty rates (assumed in the
daily vegetable intake for the urban poor, across the three Worst case scenario) to the urban poverty rates in the Medium
scenarios, in the countries other than Canada, France, Poland, UK case was on average higher (by 1.35 times) than the ratio of these
and USA, first of all note that there is a very large variation in this rates in the Medium case relative to the lowest urban poverty rates
regard for Chile, China and Mexico, but particularly Mexico, in the Best case scenario. Secondly, the built-up area in the Best
because of the large differences, in particular between the highest case scenario was assumed to be 1.5 times that in the Medium and
and lowest urban poverty rates in the Worst and Best cases Worst case scenarios. But most importantly, whereas the vegetable
respectively, and most strikingly for Mexico (see Table 1). Sec- yields in the Worst case scenario were assumed to be 0.4 times
ondly, while the urban area required is of course considerably (that is, 2.5 times lower than) those in the Medium case, the yields
higher in the Worst case scenario, and considerably lower in the in the Best case were assumed to be 4.76 times those in the
Best case scenario, relative to the Medium case scenario, the Medium case (see Table 2).
figures in the Worst case are uniformly closer to those in the UA, in terms of the low threshold of growing the basic daily
Medium case than are the Best case figures. This is because of our vegetable intake for the urban poor, appears feasible in (the most

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 7

optimistic) Best case scenario, even in countries in which it would security, where it is needed the most. It appears that, purely on the
not be feasible, as already discussed, in the Medium case scenario, basis of physical land availability in urban areas, and not even
let alone the Worst case scenario. But even in the Best case accounting for other barriers, including importantly the opportu-
scenario, 10–20% of the urban area would be required in Bangla- nity cost of land for other uses, UA can only make a limited
desh, DR Congo, Ethiopia and Nigeria – see Table 3 and Fig. 1. contribution to urban food security, let alone food security gen-
Further, the Best case scenario is, in our considered view, by far the erally, in low-income countries.
most unlikely. Also, we believe that the Medium case scenario, or In these countries, UA of course contributes to improving food
even a scenario that is slightly less favorable to UA is, all things availability and access, and to food security and nutritional status
considered, close to being the most likely scenario. This is because, of the urban households that engage in it (and others as well, to
while it is possible that urban area might be underestimated in the the extent that UA products are sold). But given our findings, our
Medium case scenario, so is poverty, based on the criticisms of the considered view, echoing Ellis and Sumberg (1998) is that, while
$1.25/day PPP rate discussed earlier. Note in this regard that the UA activities should by no means be discouraged, there is no
$2/day PPP urban poverty rates, which many have argued most convincing rationale for governments to actively promote UA. This
closely represents reality in low-income country cities, were as is particularly so in light of the range of urban priorities vying for
much as 2.7 times higher than $1.25/day PPP urban poverty rates limited administrative resources, land and water availability con-
in our analysis. As for vegetable yields, it is hard to believe that a straints, and given that UA is largely a supplement in terms of food
figure that is 4.76 times higher (as in the Best case scenario), or availability and access for the urban poor.
indeed anything very much higher than the average national The question then is how might food security be achieved for
vegetable yield as indicated in the FAOSTAT database can in the large numbers of the poor in the densely populated cities in
practice be achieved in urban areas in low-income countries on low-income countries? Food production and availability are of
a sustained basis, and on a mass scale, given all of the issues course an essential part of the answer to that question, but as we
related to water stress and ground level ozone discussed earlier, noted earlier, mere food production and availability will not
and not even considering other barriers in this context. But even if guarantee food security for the urban poor. This challenge will
we assume, for the sake of argument, that yields and urban areas require a multi-pronged response, including policies to promote
are as high as in the Best case scenario, and use the most realistic employment and income generation for them, which in turn will
$2/day PPP urban poverty rate as in the Worst case scenario, the increase their purchasing power and food affordability, and food
urban area requirements for growing the basic daily vegetable delivery at affordable prices through well run public distribution
intake for the urban poor would be 13 and 19% in Kenya and systems.
Tanzania, 20–30% in DR Congo, Ethiopia and Nigeria, and as high
as 45% in Bangladesh.
Acknowledgments

5. Conclusions and policy implications This paper benefited from parallel work on a global assessment
of urban agriculture conducted by the following colleagues:
Our analysis shows that UA is feasible – in terms of urban land Dr. Federico Martellozzo, Jean-Sébastien Landry, Dr. Dany Plouffe,
availability to grow the basic daily vegetable intake for the urban Verena Seufert, and Dr. Pedram Rowhani (and N. Ramankutty as a
poor – in the high-income countries. There is, however, little need co-author). The major difference between their analysis and ours
for UA from the point of view of urban food security in these is that we consider the urban poor specifically, and evaluate the
countries, given the availability of abundant, inexpensive food, and urban land availability for UA under different scenarios. We also
their generally high affordability. Of course, this is not to deny undertake a thorough review of the UA literature. We would like
the many benefits of UA in high-income countries, including the to thank these colleagues for the ideas that spilled over into our
provision of nutritious food to the most needy, the productive use paper. N. Ramankutty would like to acknowledge funding from the
of land that might otherwise be wasted, the greening of cities, and Natural Science and Engineering Research Council (NSERC) of
the social benefits of community building. Canada Discovery Grant RGPIN/341935-2012 and the Gordon and
On the other hand – and despite claims made in the literature Betty Moore Foundation.
regarding its ability to contribute to urban food security – UA has
low potential, or would be infeasible, in terms of land availability
in urban areas to achieve even the low threshold of growing the Appendix A. Supplementary material
daily vegetable intake for the urban poor, except in the most
optimistic (but also least likely) scenario that we considered, in Supplementary data associated with this article can be found in
precisely those contexts where it might most usefully contribute the online version at http://dx.doi.org/10.1016/j.gfs.2014.10.003.
to urban food security; namely, in cities in low-income countries,
which are characterized by very large populations, the bulk of References
whom are poor, and high urban densities. Indeed, as our analysis
shows, the low-income countries, where the vast majority of Alkire, S., 2013. Why the poorest of the poor need MPI 2.0. In: Samman, E. (Ed.),
the world's urban poor live, and malnutrition levels are high, Eradicating Global Poverty: A Noble Goal, But How Do we Measure it? Overseas
would require significant proportions of – and in some cases, Development Institute, London, pp. 2–4.
Altieri, M.A., et al., 1999. The greening of the “barrios”: urban agriculture for food
nearly all or more than – their urban areas for this purpose, except security in Cuba. Agric. Hum. Values 16, 131–140.
in the most optimistic scenario. But even in this least likely Aragrande, M., and O. Argenti, 2001. Studying food supply and distribution systems
scenario, in which assumptions regarding urban poverty rates, to cities in developing countries and countries in transition, "Food into Cities"
Collection, DT/36-01E, 148 pp, Food and Agriculture Organization of the United
total urban area and vegetable yields were the most favorable to
Nations, Rome.
UA, as much as 10–20% of the urban area would be required in Avnery, S., et al., 2011a. Global crop yield reductions due to surface ozone exposure:
some low-income countries. 1. Year 2000 crop production losses and economic damage. Atmos. Environ. 45,
Taken as a whole, our analysis raises serious doubts as to the 2284–2296.
Avnery, S., et al., 2011b. Global crop yield reductions due to surface ozone exposure:
extent to which food production in urban areas is possible at a 2. Year 2030 potential crop production losses and economic damage under two
scale sufficiently large to allow UA to contribute to urban food scenarios of O3 pollution. Atmos. Environ. 45, 2297–2309.

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i
8 M.G. Badami, N. Ramankutty / Global Food Security ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Baker, J.L., 2008. Urban Poverty: A Global View. Urban Papers UP-5. The World Ramankutty, N., Evan, A.T., Monfreda, C., Foley, J.A., 2008. Farming the planet: 1.
Bank, Washington, DC. Geographic distribution of global agricultural lands in the year 2000. Glob.
Central Agricultural Census Commission, 2003. Ethiopian Agricultural Sample Biogeochem. Cycles 22, GB1003. http://dx.doi.org/10.1029/2007GB002952.
Enumeration 2001/02, Chapter V – Area and Production of Crops and Crop Ravallion, M., 2013. Two goals for fighting poverty. In: Samman, E. (Ed.), Eradicating
Utilization, Federal Democratic Republic of Ethiopia. Global Poverty: A Noble Goal, but How do we Measure it?. Overseas Develop-
Coates, J., 2013. Build it back better: Deconstructing food security for improved ment Institute, London, pp. 2–4.
measurement and action, Global Food Security, 2(3), 188-194. Ravallion, M., 2010. World Bank's $1.25/day Poverty Measure – Countering the
Despommier, D., 2009. Growing Skyscrapers: The Rise of Vertical Farms. Scientific Latest Criticisms; Reply to Angus Deaton's Paper, “Price Indexes, Inequality and
American, pp. 80–89. the Measurement of World Poverty”. Accessible at: 〈http://econ.worldbank.org/
Dreze, J. 2012. Poverty, targeting and food security. Symposium on Ending WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/0,,con
Hunger – A Symposium on the Proposed National Food Security Bill. Seminar. tentMDK:22510787  pagePK:64165401  piPK:64165026  theSiteP
Issue #634. Accessible at: 〈http://www.india-seminar.com/2012/634/634_jean_ K:469382,00.html〉.
dreze.htm〉. Ravallion, M., Chen, S., Sangraula, P. 2007, New Evidence on the Urbanization of
Dubbeling, M., de Zeeuw, H., 2011. Urban agriculture and climate change adapta- Global Poverty. World Bank Policy Research Working Paper 4199. The World
tion: ensuring food security through adaptation. In: Otto-Zimmermann, K. Bank, Washington, DC.
(Ed.), Resilient Cities: Cities and Adaptation to Climate Change. Proceedings of Redwood, M., 2009. Agriculture in Urban Planning: Generating Livelihoods and
the Global Forum 2010, Local Sustainability 1. Springer Science þ Business Food Security. Earthscan, London, UK.
Media B.V., Amsterdam, pp. 441–449. Ruel, M.T. et al., 1998. Urban challenges to food and nutrition security: a review of
Duchemin, E., Wegmuller, F., Legault, A.-M., 2009. Urban agriculture: multi- food security, health and caregiving in the cities. FCND Discussion Paper No. 51.
dimensional tools for social development in poor neighbourhoods. Field International Food Policy Research Institute, Washington, DC.
Actions Sci. Rep. 2, 1–8. Satterthwaite, D., 2003. The millennium development goals and urban poverty
Ellis, F., Sumberg, J., 1998. Food production, urban areas and policy responses. reduction: great expectations and nonsense statistics. Environ. Urban. 15,
World Dev. 26, 213–225. 181–190.
FAO, 1996. The State of Food and Agriculture. Food and Agricultural Organization, Satterthwaite, D., McGranahan, G., Tacoli, C., 2010. Urbanization and its implications
Rome. for food and farming. Philos. Trans. R. Soc. B 365, 2809–2820.
FAO, 2013. FAOSTAT Database. Food and Agriculture Organization, Rome 〈http:// Schneider, A., Friedl, M.A., Potere, D., 2010. Mapping global urban areas using
faostat3.fao.org/home/index.html〉 (Accessible at:). MODIS 500-m data: new methods and datasets based on ‘urban ecoregions’.
FAO, 1999. Spotlight: Issues in Urban Agriculture. Food and Agricultural Organiza- Remote Sens. Environ. 114, 1733–1746.
tion, Rome 〈http://www.fao.org/ag/magazine/9901sp2.htm〉 (Accessible at:). Smit, J., Nasr, J., 1992. Urban agriculture for sustainable cities: using wastes and idle
FAO, WFP, IFAD, 2012. The State of Food Insecurity in the World 2012: Economic land and water bodies as resources. Environ. Urban. 4, 141–152.
Growth is Necessary but not Sufficient to Accelerate Reduction of Hunger and Smit, J., Nasr, J., Ratta, A., 2001. Urban Agriculture: Food, Jobs and Sustainable Cities.
Malnutrition. Food and Agricultural Organization, Rome. The Urban Agriculture Network, Inc., New York.
Grewal, S.S., Grewal, P.S., 2012. Can cities become self-reliant in food? Cities 29, Tacoli, C., Bukhari, B., Fisher, S., 2013. Urban Poverty, Food Security and Climate
1–11. Change. International Institute for Environment and Development, London, UK.
Holland, M., et al. 2002. Economic Assessment of Crop Yield Losses from Ozone The Economist, 2010. Vertical Farming: Does it Really Stack Up? The Economist
Exposure. Official report prepared under Contract EPG 1/3/170 between DEFRA Technology Quarterly. December 11, pp. 9–10.
and the Natural Environment Research Council, under The UNECE International UNDP. 1996. Urban Agriculture: Food, Jobs, and Sustainable Cities. UNDP Publica-
Cooperation Programme on Vegetation. Centre for Ecology & Hydrology, tion Series for Habitat II, vol. 1. United Nations Development Programme,
Environment Centre Wales, Bangor, Gwynedd, UK. New York.
Koc, M., et al., 1999. For Hunger-proof Cities: Sustainable Urban Food Systems. United Nations, 2011. Human Development Report 2011: Sustainability and Equity
International Development Research Centre, Ottawa. – A Better Future for All. United Nations Development Program, New York.
Lee-Smith, D., 2010. Cities feeding people: an update on urban agriculture in United Nations, 2013. World Urbanization Prospects, retrieved October 2013.
equatorial Africa. Environ. Urban. 22, 483–499. Accessible at: 〈http://esa.un.org/unup/〉.
Martellozzo, F., et al., 2014. Urban agriculture: a global analysis of the space WHO/FAO, 2003. Diet, Nutrition and the Prevention of Chronic Diseases, Report of a
constraint to meet urban vegetable demand. Environ. Res. Lett. 9, 064025. Joint WHO/FAO Expert Consultation, 149 pp, World Health Organization, Food
Maxwell, D., Levin, C., Csete, J., 1998. Does urban agriculture help prevent and Agriculture Organization of the United Nations, Geneva.
malnutrition? Evidence from Kampala. Food Policy 23, 411–424. World Bank, 2014a. World Development Indicators Database. The World Bank,
McDonald, R.I., et al., 2014. Water on an urban planet: Urbanization and the reach Washington, DC 〈http://data.worldbank.org/data-catalog/world-developmen
of urban water infrastructure. Glob. Environ. Change 27, 96–105. t-indicators〉 (Accessible at:).
Montgomery, M.R., 2008. The Urban Transformation of the Developing World. World Bank, 2014b. PovcalNet – an online poverty analysis tool. The World Bank,
Science 319, 761–764. Washington, DC 〈http://iresearch.worldbank.org/PovcalNet/index.htm〉 (Acces-
Mougeot, L.J.A., 2011. International Support to Research and Policy on Urban sible at:).
Agriculture (1996–2010): achievements and challenges. International Develop- World Bank/FAO, 2008. Urban Agriculture for Sustainable Poverty Alleviation and
ment Research Centre, Ottawa. Food Security. The World Bank, Washington, DC and Food and Agricultural
Mougeot, L.J.A., 2005. Introduction. In: Mougeot, L.J.A. (Ed.), Agropolis: the Social, Organization, Rome 〈http://www.fao.org/fileadmin/templates/FCIT/PDF/UPA_
Political, and Environmental Dimensions of Urban Agriculture. Earthscan -WBpaper-Final_October_2008.pdf〉 (Accessible at:).
and the International Development Research Centre, London and Ottawa, Yi-Zhong, C., Zhangen, Z., 2000. Shanghai: trends towards specialised and capital-
pp. 1–24. intensive urban agriculture. In: Bakker, N., et al. (Eds.), Growing Cities Growing
OECD, 2013. OECD Income Distribution and Poverty Database. Accessible at: 〈http:// Food: Urban Agriculture on the Policy Agenda – A Reader on Urban Agriculture.
stats.oecd.org/Index.aspx?DatasetCode=IDD〉. Deutsche Stiftung für internationale Entwicklung (DSE), Feldafing, Germany,
Potere, D., Schneider, A., Angel, S., Civco, D.L., 2009. Mapping urban areas on a pp. 467–475.
global scale: which of the eight maps now available is more accurate? Int. J. Yoveva, A., et al., 2000. Sofia: urban agriculture in an economy in transition. In:
Remote Sens. 30, 6531–6558. Bakker, N., et al. (Eds.), Growing Cities Growing Food: Urban Agriculture on the
Potutan, G.E., et al., 2000. Urban agriculture in Cagayan de Oro: a favorable Policy Agenda – A Reader on Urban Agriculture. Deutsche Stiftung für inter-
response of city government and NGOs. In: Bakker, N., et al. (Eds.), Growing nationale Entwicklung (DSE), Feldafing, Germany, pp. 501–518.
Cities Growing Food: Urban Agriculture on the Policy Agenda – A Reader on Zezza, A., Tasciotti, L., 2010. Urban agriculture, poverty, and food security:
Urban Agriculture. Deutsche Stiftung für internationale Entwicklung (DSE), Empirical evidence from a sample of developing countries. Food Policy 35,
Feldafing, Germany, pp. 413–428. 265–273.

Please cite this article as: Badami, M.G., Ramankutty, N., Urban agriculture and food security: A critique based on an assessment of
urban land constraints. Global Food Security (2014), http://dx.doi.org/10.1016/j.gfs.2014.10.003i