a Faculty of Engineering Technology, Water Engineering and Management, University of Twente, P.O. Box 217, 7500 AE, The Netherlands b Center for Energy and Environmental Studies (IVEM), University of Groningen, Nijenborg 4, 9747 AG Groningen, The Netherlands

Introduction

At present, the world faces enormous challenges over food

security (Millenium Ecosystems Assesment, 2005), which threaten

the availability and quality of natural resources such as arable

lands, freshwater and natural areas (FAO, 2003; Hoekstra &

Chapagain, 2008; WWF, 2007). The potential impacts of climate

change are likely to worsen this situation (Fischer, Van Velthuizen,

Shah, & Nachtergaele, 2002). Globally, food consumption gives rise

to the greatest use of land (FAO, 2003; Penning de Vries et al., 1995)

and freshwater (Falkenmark, 1989; FAO, 2003; Hoekstra &

Chapagain, 2008; Rockstrom, 1999; Rosegrant & Ringler, 1998)

and is an important cause of greenhouse gas emissions (Carlsson-

Kanyama, Engström, & Kok, 2005; Kramer, 2000). The current

growth in the world population requires the production of more

food. As well as population growth, most areas of the world

have shown economic development that resulted in increased

purchasing power, causing not only a demand for more food

(Latham, 2000) but also for different food. Studies on human

nutrition have shown that worldwide a nutrition transition is

taking place, in which people shift towards more affluent food

consumption patterns (FAO, 2003; Grigg, 1995; Popkin, 2002).

Globalization of nutrition includes shifts from local markets

towards global trade in commodities and processes in which

people and ideas spread throughout the world (Lang, 2002) and

thereby change consumption. Since the beginning of the eigh-

teenth century, the nutrition transition that accompanied eco-

nomic development has caused large shifts in food consumption

patterns in Europe and the United States (Fogel & Helmchen, 2002).

When economic development occurs in developing countries as

well, as is the case in China today (IMF, 2010), nutritional changes

put additional pressure on limited natural resources.

The use of natural resources for food is the combined effect of a

specific consumption pattern and production system. Several

scientists describe the complex links between sustainable

consumption and the limited availability of natural resources

(e.g. Hertwich, 2005). Duchin (2005) provides an overview of

studies on energy and land required for food and suggests that a

shift from affluent consumption patterns towards a Mediterra-

nean-type pattern, characteristic of Greece in the 1960s, has

favorable impacts on the environment.

Appetite 55 (2010) 597–608

A R T I C L E I N F O

Article history:

Received 11 March 2010

Received in revised form 1 September 2010

Accepted 14 September 2010

Keywords:

Dietary change

Economic development

Natural resource use

Nutrition transition

Food consumption patterns

A B S T R A C T

This study analyzes relationships between food supply, consumption and income, taking supply, meat

and dairy, and consumption composition (in macronutrients) as indicators, with annual per capita GDP

as indicator for income. It compares food consumption patterns for 57 countries (2001) and gives time

trends for western and southern Europe. Cross-sectional and time series relationships show similar

patterns of change. For low income countries, GDP increase is accompanied by changes towards food

consumption patterns with large gaps between supply and actual consumption. Total supply differs by a

factor of two between low and high income countries. People in low income countries derive nutritional

energy mainly from carbohydrates; the contribution of fats is small, that of protein the same as for high

income countries and that of meat and dairy negligible. People in high income countries derive

nutritional energy mainly from carbohydrates and fat, with substantial contribution of meat and dairy.

Whenever and wherever economic growth occurs, food consumption shows similar change in direction.

The European nutrition transition happened gradually, enabling agriculture and trade to keep pace with

demand growth. Continuation of present economic trends might cause significant pressure on natural

resources, because changes in food demand occur much faster than in the past, especially in Asia.

Abbreviations: A%, average supply of nutritional energy from animal sources (%); E%,

energy percentage; FAO, Food and Agriculture Organization of the United Nations;

GDP, gross domestic product; GE, grain equivalents; G-K dollars, 1990 International

Geary-Khamis dollars; PPP, purchasing power parity; WHO, World Health

Organization.

* Corresponding author.

E-mail address: [email protected] (P.W. Gerbens-Leenes).

Contents lists available at ScienceDirect

Appetite

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

doi:10.1016/j.appet.2010.09.013

http://dx.doi.org/10.1016/j.appet.2010.09.013

mailto:[email protected]

http://www.sciencedirect.com/science/journal/01956663

http://dx.doi.org/10.1016/j.appet.2010.09.013

Agriculture is the basis for the production of food, providing

commodities such as wheat and raw milk. The number of

agricultural commodities that are important for the global supply

of food is limited to about 21 when expressed in terms of weight of

annual global production (FAO, 2010). The 15 main categories of

crop commodities, expressed as annual global production (kg), are

sugar cane, root crops, vegetables, maize, paddy rice, wheat, fruits,

potato, sugar beet, cassava, soybean, barley, pulses, oil seed rape

and sorghum; the six main animal commodities are raw milk, pork,

poultry, beef, mutton and goat’s meat (FAO, 2010). These

commodities provide the ingredients for a large number of

different food items, such as pizza or cheese. The mass of food

items mainly consists of only four different components: water

and the three macronutrients carbohydrates, fats and proteins

(FAO, 2010; Voedingscentrum, 1998a, 1998b; Whitney & Rolfes,

1999). Food items form the basis of food consumption patterns,

defined as the consumption of specific food items and their

combination in dishes and meals. These patterns show large

temporal and spatial differences, mainly caused by the availability

of commodities, cultural aspects and economic factors (Whitney &

Rolfes, 1999). The requirements for specific natural resources for

each food item are determined by the production system. In

general, food items show large differences in the requirement for

land (Gerbens-Leenes & Nonhebel, 2005), energy (Kramer, 2000)

and freshwater (Hoekstra & Chapagain, 2007), resulting in

substantial variations in requirements for natural resources

between food consumption patterns. As a rule, affluent western-

style food consumption patterns need more natural resources than

those of poor developing countries (e.g. Duchin, 2005; Gerbens-

Leenes & Nonhebel, 2002; Hoekstra & Chapagain, 2007). Food

items that are typical of affluent patterns are fats, drinks and foods

derived from animal sources, such as milk, cheese and meat. These

items have a substantial impact on natural resources, either

through heavy consumption (for example of beer), or through large

specific resource requirements per unit of food. For example, in

western countries the contribution of fats to land requirements is

about 25% and that of meat about 30% (Gerbens-Leenes &

Nonhebel, 2002), while the contribution of meat to energy and

freshwater requirements is also about 30% (Gerbens-Leenes &

Hoekstra, 2007; Kramer, 2000). It is therefore important to identify

the relationship between economic growth and more affluent food

consumption patterns.

Hundreds of detailed studies from the nutritional, social and

agricultural sciences, as well as food security studies, are

available. Nutritional and social studies express consumption in

terms of specific food items (e.g. Mennell et al., 1992; Receveur,

Boulay, & Kuhnlein, 1997; Whitney & Rolfes, 1999), agricultural

studies on global food security simplify consumption to basic

and affluent diets and show them in grain equivalents (GE) (e.g.

Penning de Vries et al., 1995), while other studies address food

security as the average per capita availability of commodities

(FAO, 2003). The agricultural and food security studies often

show time trends, emphasizing the need to increase agricultural

production. The effect of income on food consumption patterns

is recognized as one of the factors that determine food choice

(e.g. Van der Boom-Binkhorst et al., 1997; Von Braun, 1988; Von

Braun & Paulino, 1990; Ivens et al., 1992; Musaiger, 1989;

Vringer & Blok, 1995; Wandel, 1988; Whitney & Rolfes, 1999; De

Wijn & Weits, 1971). Among the poorest people, be they

individuals or nations, diets tend to be composed principally of

cheap starchy staple foods: wheat, rice, potatoes, cassava and

the like (Jobse-van Putten, 1995; Poleman & Thomas, 1995).

Existing research often focuses on health issues and changes in

time. General relationships between economic change and the

rate of change in food consumption patterns are also important

to be explored.

The nutrition transition began in developed countries 300

years ago. It coincided with great economic growth (Maddison,

2003). If developing countries follow the same route, it would

mean a major shift in the balance between global food demand

and supply, with considerable consequences for natural

resources. It is therefore important to investigate whether there

are general relationships between economic growth and food

consumption patterns. This is also important when growth is

negative. The FAO, for example, estimated that in 2007 75 million

people were pushed into undernourishment as a result of higher

food prices mainly caused by an increase in commodities used for

bio-energy, bringing the total number of hungry people in the

world to 923 million (FAO, 2008). It is possible that the current

financial crisis will diminish purchasing power and so increase the

risk of a drop in food intake.

When food consumption patterns are expressed in terms of

food items, differences among the patterns are large and studying

them requires a great amount of data and time. This paper

presents an analysis of nutritional changes due to economic

growth, so being situated between detailed consumption pattern

analyses in terms of specific foods items, which are only valid for a

limited group of consumers, and the coarse agricultural analyses

based on simplified diets in terms of GE. The paper expresses

changes in terms of macronutrients and related nutritional

energy. The specific aims of this study are to quantify shifts in

food consumption patterns that accompany economic develop-

ment. The research questions are: (i) what are the trends in

national per capita food supply, measured in terms of nutritional

energy and macronutrients that follow economic changes? (ii)

what are the trends in individual per capita food consumption, i.e.

the food actually eaten, measured in terms of nutritional energy

and macronutrients, that accompany economic changes? (iii) in

which regions will large changes in food supply and consumption

occur in the next 10 years? To analyze the impact of economic

changes on food consumption patterns this study addresses

similarities among patterns in terms of composition, rather than

differences in terms of food items. This approach makes it possible

to identify general trends. By differentiating between national per

capita supply and individual consumption, it also shows trends for

the gap between supply and consumption. The study analyzes

cross-sectional and time series relationships, revealing general

trends. These trends provide a better understanding of the

connection between food consumption and environment and can

contribute to environmental studies that aim to indicate

transition pathways towards a more sustainable use of natural

resources.

[()TD$FIG]

National

agricultural

production

Import Export

Gap

Consumer

supply

Consumption

(food eaten)

Fig. 1. Simplified food system overview. Consumer supply is defined as per capita

national food availability and is a function of national production + import � export.

Consumption is food actually eaten. The difference between supply and consumption

is the gap caused by food chain losses.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608598

Food systems

Food systems include production and related supply of

commodities and foods as well as consumption, defined here as

the food actually eaten. Figure 1 shows a simplified overview of the

system. Agricultural production provides crop and animal com-

modities. On a national level, supply is a function of national

production plus imports minus exports. Consumer supply is defined

here as the per capita national availability of food, consumption as

actual consumed food. The difference between supply and

consumption is the gap caused by losses in the food chain.

Production

Starting in 1961, FAO food balance sheets (FAO, 2010) provide

information for almost all countries in the world on the annual

market supply of agricultural commodities. The food industry

selects and processes agricultural commodities to manufacture

food items (Catsberg & Kempen-van Dommelen, 1997). It often

divides commodities into several fractions based on composition

characteristics, such as the fat, protein or carbohydrate content

(the macronutrients). The fractions form the basic ingredients for

food items, when industry joins in and processes ingredients.

Soybeans, for example, are split into an oil and an oil cake fraction

(Kramer & Moll, 1995). Oil is a basic ingredient in margarines, oil

cake in livestock feed. In the western world, technological

developments in agriculture, transportation and food conservation

at the end of the 19th century prompted the expansion of the food

industry and food preparation shifted from households to industry

(Jobse-van Putten, 1995), so stimulating the nutrition transition.

Figure 1 shows that agriculture, inside or outside the national

borders, determines consumer supply available for consumption.

Sometimes wastesarereused,forexamplemanureor waste fromthe

food industry for livestock feed (Nonhebel, 2004). In every link of the

system losses take place. For example, the food industry processes

1.4 kg of wheat to manufacture 1 kg of flour (Kramer & Moll, 1995). A

study on household consumption in Sweden has estimated that

during meals 10% of the food remains behind on the plate and is

wasted (Karlsson,2001).Theselosses cause a gap betweenconsumer

supply and consumption, i.e. the food actually eaten.

Consumption

Consumer supply concerns the food available for consumption at

a national level. Consumers buy it in shops or sometimes produce it

in their gardens (Fernandes & Nair, 1986; Pallot & Nefedova, 2003).

The repeated arrangements of consumption, characterized by types

and quantities of food items and their combination in dishes and

meals, are termed food consumption patterns (Gerbens-Leenes &

Nonhebel, 2002). Factors such as preferences, habits, availability,

tradition, culture and income influence these patterns (Van der

Boom-Binkhorst et al., 1997; Von Braun, 1988; Von Braun & Paulino,

1990; Ivens et al., 1992; Musaiger, 1989; Vringer & Blok, 1995;

Wandel, 1988; Whitney & Rolfes, 1999; De Wijn & Weits, 1971). For

instance, when income increases, peoplespend more money on food

(Pindyck & Rubinfeld, 2005; Vringer & Blok, 1995).

Consumption, or human nutrition, concerns the food actually

eaten. For nutrition the composition of food in terms of

macronutrients, its fats, carbohydrates and proteins, is important,

because they provide energy and are essential for the functions of

the human body. Food surveys ask respondents what they have

eaten and provide detailed information on the composition of

consumed food (see also Appendix B). Humans can derive energy

from different combinations of macronutrients. This flexibility

contributes to variations in the macronutrient composition of

nutrition and to differences in food consumption patterns.

The composition of food

Four components, water, carbohydrates, fats and proteins,

dominate the composition of every commodity (FAO, 2010;

Voedingscentrum, 1998a, 1998b; Whitney & Rolfes, 1999). The

macronutrient content of commodities, such as wheat, soybean or

pork, is genetically determined, so that all crop and animal

commodities show a specific composition (kg macronutrient per

kg dry matter) (Penning de Vries et al., 1989; Schmidt-Nielsen,

1988). Based on composition, commodities form four categories:

(i) starchy staples, crops that mainly provide carbohydrates with

few proteins; (ii) protein-rich crops, which provide proteins as well

as carbohydrates; (iii) oil crops, providing plant-based fats for the

production of oil, and carbohydrates and proteins for feed (Penning

de Vries et al., 1989); and (iv) animal commodities, which provide

high quality proteins and fats (Whitney & Rolfes, 1999). The

composition of a commodity determines its suitability for a food

item (Whitney & Rolfes, 1999). For example, starchy staples, such

as wheat, can be used for bread or pasta, while oil crops, such as oil

seed rape, provide oil (Voedingscentrum, 1998a, 1998b) for

margarines. Consumption changes can cause shifts in the

macronutrient composition of food consumption patterns includ-

ing a demand for different commodities.

The food system and natural resources

Several studies have shown that natural resource use varies

greatly between food items and food consumption patterns (e.g.

Engelenburg van, Rossum van, Blok, & Vringer, 1994; Hoekstra &

Chapagain, 2008; Kok, Biesiot, & Wilting, 1993; Kramer & Moll,

1995; Tukker & Jansen, 2006). Meat, fats, and drinks especially have

relatively large requirements for energy, land and freshwater

(MJ kg�1, m2 kg�1, m3 kg�1). One kilogram of pork providing

2000 kilocalories (Voedingscentrum, 1998a, 1998b), for example,

requires 86 MJ of energy (Kramer & Moll, 1995), 9 m2 of land

(Gerbens-Leenes & Nonhebel, 2005) and 4850 L of freshwater

(Hoekstra & Chapagain, 2008) for its production. In comparison, 1 kg

of paddy rice providing even more nutritional energy (3500 kilo-

calories) requires 22 MJ energy (Kramer & Moll, 1995), 3 m2 land

(Gerbens-Leenes, 2006) and 2300 L of water (Hoekstra & Chapagain,

2008). The example shows that changes in food consumption

patterns can have a considerable impact on natural resource use.

Methods and data

To analyze the relationship between food supply, food

consumption, and the contribution of animal foods to supply on

the one hand and income on the other, this study assesses cross-

sectional and time series relationships.

Units of calculation

The study does not express food supply and consumption in

terms of foods, but simplifies supply and consumption and uses

macronutrient composition (fats, carbohydrates and proteins) as

units of calculation. It indicates per capita food supply in the

fraction of nutritional energy provided by the macronutrients, the

macronutrient energy percentage (E%), as is common in nutrition

research (Whitney and Rolfes, 1999). It shows the contribution of

animal foods to supply in the fraction of nutritional energy derived

from animal sources (A%), and total food supply as availability of

nutritional energy (kilocalories per capita per day). E% and A% are

calculated by

protein E% ¼ P � kcal: p

E � 100% (1)

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608 599

fat E% ¼ F � kcal: f

E � 100% (2)

carbohydrate E% ¼ E � ððP � kcal: pÞ þ ðF � kcal: fÞÞ

E � 100% (3)

A% ¼ A

E � 100% (4)

where P is the average daily supply of protein (grams), kcal.p the

nutritional energy supply of protein (4 kilocalories per gram), E the

average daily per capita supply of nutritional energy (kilocalories),

F the average daily supply of fat (grams); kcal.f the nutritional

energy supply of fat (9 kilocalories per gram) and A the average

daily per capita supply of nutritional energy from animal sources

(kilocalories). The study derives data on per capita supply from

FAO food balance sheets (FAO, 2010) and values on nutritional

energy for protein and fat from the Dutch Nutrition Council

(Voedingscentrum, 1998a, 1998b).

Average per capita income depends, among other things, on the

development status of an economy, the size of households and

income distribution in a country. The World Bank (2005) calculates

economic output as gross domestic product (GDP). Information on

average per capita GDP is available for most countries from various

sources. The only database, however, that covers recent informa-

tion on GDP as well as historical economic developments is that of

Maddison (2003) which provides information on the economic

development status of almost all countries in the world on a

national and per capita basis from the Middle Ages onwards. It

expresses average GDP in 1990 International Geary-Khamis dollars

(G-K dollars). The Geary-Khamis method is an aggregation method

in which international prices and a countries purchasing power

parity, depicting relative country price levels, are estimated

simultaneously from a system of linear equations and expressed

in G-K dollars (United Nations Statistics Division, 2006). We derive

data on per capita GDP from Maddison (2003) as an indicator for

income (expressed as dollars per capita per year) for the cross-

sectional and time series relationships.

Cross-sectional relationships

The study assesses cross-sectional relationships between

income and food supply, the contribution of animal foods and

the composition of supply for 57 countries in 2001. Appendix A

gives an overview of these countries. Countries from Africa, Asia,

Eastern Europe, Latin America, the Middle East and the OECD, in

different stages of development and with more than five million

inhabitants, have been selected. These countries form two clusters

of developed and developing countries, with relatively high and

relatively low GDPs, however. To also cover countries with average

incomes, three small transition countries with GDPs in between

the two extremes, the United Arab Emirates, Estonia and Slovenia,

are added, clustered into a small country group.

Time series relationships

Over the last millennium Europe has shown continuous

economic growth (Maddison, 2003). Between 1700 and 2000,

for example, per capita GDP in France increased from 900 to 21,000

dollars and in Great Britain from 1250 to 20,000 dollars. Between

1961 and 2001, Italy, Greece, Spain and Portugal showed a three- to

fourfold increase of per capita GDP. These periods were accompa-

nied by large changes in food consumption patterns (Jobse-van

Putten, 1995; FAO, 2010; Fogel & Helmchen, 2002). Most studies of

historical food consumption describe changes in a qualitative way

(e.g. Jobse-van Putten, 1995; Mennell et al., 1992) and do not

provide quantitative data. An exception is the analysis by Fogel and

Helmchen (2002), which has quantified nutritional energy supply

for France and Great Britain between 1700 and 2000 (kilocalories

per capita per day). To evaluate per capita food supply over time,

this study first assesses a time series relationship between supply

and income in France and Great Britain over a period of three

centuries. It combines data from Fogel and Helmchen (2002) with

GDP data from Maddison (2003).

Secondly, the study evaluates a four-decade time series

relationship in southern Europe. For Italy, Spain, Portugal and

Greece, it assesses the relationship between the increase of per

capita supply, changes in the composition of food consumption

and changes in the contribution of animal foods on the one hand

and income on the other over the period 1961–2001. It applies

Eqs. (1)–(4) and derives data from the FAO (2010) and combines

this with data on GDP from Maddison (2003).

Confirmation of trends and the gap between supply and consumption

This study firstly assesses the relationship between GDP and

national per capita food supply. Secondly, to evaluate whether

information from food surveys can confirm trends found, the study

assesses the relationship between consumption, expressed as

nutritional energy intake (kilocalories per capita per day), and

annual per capita GDP. A number of surveys have been done in

developing countries (FAO, 2005) and two time series are available

for developed countries, the Netherlands (Voedingscentrum/TNO,

1998) and the United States (USDA, 2005). The study combines

data from 31 food surveys from 26 countries (see Appendix B) with

information on GDP from Maddison (2003). Thirdly, it evaluates

the size of the gap between national per capita supply and

consumption. This provides information on food losses in the

supply chain. To confirm trends, the study also compares the fat E%

of urban and rural consumption. Data for this were derived from 11

surveys in developing countries that made a distinction between

urban and rural patterns (see Appendix B).

Results and discussion

Per capita income and food supply

The cross-sectional analysis indicates a relationship between

per capita food supply and income (GDP) following a power-law

dependency E = 850 � GDP0.14, thus showing an income elasticity

of about 0.14. Figure 2 shows that supply varies between 1600

kilocalories per capita per day for low GDPs and 3800 kilocalories

for high GDPs, a difference of a factor of almost two and a half. The[()TD$FIG]

Fig. 2. Relationship between annual per capita GDP (dollars) and nutritional energy

supply (kilocalories per capita per day) based on data from 57 countries in 2001. The

solid line shows the power-law regression (income elasticity 0.14, R2 = 0.71), the

shaded zone is the 90% confidence band.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608600

figure also shows that for low GDPs increases in food supply per

unit of GDP are large, while for high GDPs increases are much

smaller and income does not seem to affect supply any longer.

The power-law relationship results in an R2 of 0.71, the

elasticity 0.14 inhibits a standard error of 0.013 (t-stat 11, p value

3.8 � 10�15), while the logarithm of the constant is 2.93 with

standard error 0.048 (t-stat 60.5, p value 3.7 � 10�49). Data

analysis suggests that income elasticity decreases with per capita

income, but this does not result in a more significant relationship.

The figure also shows an appreciable scatter around the identified

relationship; the grey zone indicates the 90% confidence band. This

scatter may be due to socio-economic factors, such as the share of

income spendable on food, income distribution, or cultural

differences in food consumption patterns.

Figure 3 shows that in France and Great Britain increasing per

capita GDP parallels greater food supply, which doubles over the

three centuries considered from 1700 kilocalories per capita per

day in 1700–3500 kilocalories in 2000. The largest increase per

unit of GDP occurs for incomes below 5000 dollars, while above

this level the increase gradually slows down. The figure also shows

the relationship identified in the cross-sectional analysis and the

confidence band around it. Results for food supply in France and

Great Britain are within the confidence band and qualitatively

similar to trends in that analysis and show an income elasticity of

0.23: E = 395GDP0.23. The R2 is 0.88, the elasticity 0.23 inhibits a

standard error of 0.024 (t-stat 9.6, p value 5.6 � 10�7), while the

logarithm of the constant is 2.6 with standard error 0.087 (t-stat

30, p value 1.3 � 10�12). Quantitative differences may be due to,

among other factors, limitations of the spatio-temporal analogue

of comparing spatial differences between countries to temporal

changes within countries.

Figure 4a–c depicts the results for southern Europe in the period

1961–2001. Figure 4a shows that nutritional energy supply

increased from 2500 kilocalories per capita per day for a GDP of

3000 dollars (Portugal, 1961) to 3700 for a GDP of 12,500 dollars

(Greece, 2001). Again, results are qualitatively consistent with the

cross-sectional results (power-law relationship and confidence

band); the income elasticity found here is 0.21. The relation reads

E = 475GDP0.21. The R2 is 0.78, the elasticity 0.21 inhibits a standard

error of 0.026 (t-stat 8, p value 2.6 � 10�7), while the logarithm of

the constant is 2.7 with standard error 0.10 (t-stat 26, p value

9.3 � 10�16). Moreover, Fig. 4c shows that the fraction of food

supply from animal sources is explained well by per capita income

following a power-law with income elasticity of 0.43:

A% = 0.41 � GDP0.43, R2 is 0.88, the elasticity 0.43 inhibits a

standard error of 0.037 (t-stat 11.5, p value 9.9 � 10�10), while

the logarithm of the constant is �0.39 with standard error 0.15 (t-

stat �2.6, p value 0.016).

Per capita income and composition of food consumption

Figure 5a and b shows the relationship between the macronu-

trient composition of consumption and annual per capita GDP for

the cross-sectional analysis. The fraction of nutritional energy

provided by proteins does not change with income and is between

9 and 18 E%. The carbohydrate and fat E%, however, show a

connection with GDP. It can be estimated from Fig. 5a that in

countries with low GDPs, below 5000 dollars, people derive

[()TD$FIG]

Fig. 3. Relationship between annual per capita GDP (dollars) and nutritional energy

supply (kilocalories per capita per day) for France and Great Britain between 1700

and 2000 also showing the relationship of the cross-sectional analysis of Fig. 2. The

solid line and shaded zone denote the relation identified in the cross-sectional

analyses and the confidence band around it.

[()TD$FIG]

Fig. 4. (a) shows the relationship between annual per capita GDP (dollars) and

nutritional energy supply (kilocalories per capita per day) for southern Europe

between 1961 and 2001. The solid line and shaded zone denote the relationship

identified in the cross-sectional analyses and the confidence band around it. (b)

shows the relationship between annual per capita GDP and the composition of food

consumption patterns in terms of the fraction of nutritional energy derived from fat

(fat E%), protein (protein E%), and carbohydrate (carbohydrate E%) for southern

Europe between 1961 and 2001. (c) shows the relationship between annual per

capita GDP and the composition of food consumption patterns in terms of the

fraction of nutritional energy from animal sources (%) for southern Europe between

1961 and 2001, and the relationship based on the cross-sectional analysis.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608 601

nutritional energy mainly from carbohydrates, with a small

fraction from fats. In Bangladesh, for example, the country with

the lowest GDP in the analysis, people derive 80% of nutritional

energy from carbohydrates and 11% from fats. In consumption in

countries with high GDPs, carbohydrates are less important and

more energy is provided by fats. The average consumer in the US,

France and Denmark, for instance, derives 45–50 E% from

carbohydrates and 40 E% from fats. The figure also shows that

for countries with low GDPs, small changes in income cause large

changes in the composition of consumption, while for countries

with high GDPs, small income changes do not affect composition

because saturation has already occurred.

Although over the period 1961–2001 there were no average

annual per capita GDPs of southern Europe countries below 3500

and above 19,000 dollars, when comparing Figs. 4b–5a, results

show similar trends. Figs. 4c and 5b show the relationship between

the fraction of nutritional energy derived from animal sources (A%)

and GDP. For countries with low GDPs, A% is almost negligible; for

countries with high GDPs, the fraction is about 25–40%. In

Bangladesh, for example, A% is only 3%, while in Denmark A% is

40%. The figures show that for low GDPs, differences per unit of

GDP are large; for high GDPs, differences per unit GDP are much

smaller. It should be stressed, however, that A% indicates the

fraction of energy derived from animal foods and not the amount

consumed. Some countries with a high GDP, for example Canada

and the US, show relatively small consumption of animal foods, 27

and 30 E% respectively. In absolute numbers, however, for the

average Canadian the annual supply was 101 kg of meat and

204 kg of milk, while for a US citizen it was 121 kg of meat and

262 kg of milk. In an OECD country with a lower GDP than the US

and Canada, the Netherlands, the relative consumption of animal

foods is larger than in the US and Canada, 34 E%, but actual meat

supply is less with 90 kg per capita per year and milk supply larger

with 336 kg (FAO, 2010). The example also shows that consump-

tion of animal foods does not increase indefinitely. Jobse-van

Putten (1995), for instance, has also shown that in the western

world high income groups consume less meat that low income

groups. This trend has also been observed in Portugal (Rodrigues,

Caraher, Trichopoulou, & De Almeida, 2008). Meat and milk require

relatively large natural resource use (Wirsenius, 2003). Therefore

this result is important from an environmental perspective.

In general, animal protein is of better quality than plant-based

protein (Whitney & Rolfes, 1999). For most countries the protein E%

does not show great differences. An increase in the fraction of

nutritional energy derived from animal foods, therefore, does not

imply an increase in the fraction of protein, but rather an

improvement in protein quality. Especially for developing countries

this trend is important, because it improves the quality of the

consumption pattern.

Figure 5a and b also show that in some countries consumers

deviate from trends. In Japan, for example, a high GDP is combined

with a relatively small per capita food supply, while the

composition of consumption also resembles a pattern related to

a lower GDP. This indicates that factors other than GDP, such as

culture, also affect consumption.

Figure 6 shows the fat E% of per capita consumption derived

from 11 surveys in developing countries that make a distinction

between urban and rural patterns. Apart from Egypt, urban

consumption has a greater fat E% than rural consumption. The

surveys were done in countries with relatively low GDPs, i.e.

within the range where large differences in composition of food

consumption occur per unit of GDP. It is likely that per capita GDP

was higher for urban populations, which would explain the

difference in fat E%. The result confirms the other trends.

Trends

Results of the cross-sectional and time series relationships all

show that large changes in food supply and composition of

consumption occur for relatively low annual per capita GDPs,

below 5000 dollars, while for a GDP between 5000 and 12,5000

dollars changes are relatively small and above a GDP of 12,500

dollars food supply and the composition of consumption become

quite stable. This is in accordance with many detailed studies of

specific consumer groups, which have shown that increasing

societal affluence causes shifts in the consumption of specific foods

[()TD$FIG]

Fig. 5. (a)showsthe relationshipbetween annual percapita GDP andthe composition

of food consumption patterns in terms of the fraction of nutritional energy derived

from fat (fat E%), protein (protein E%) and carbohydrate (carbohydrate E%). (b) shows

the relationship between annual per capita GDP and the composition of food

consumption patterns in terms of the fraction of nutritional energy from animal

sources (A%); the solid line denotes the power-law function with income elasticity

0.52, R2 = 0.73. The relationships were based on data from 57 countries in 2001.

[()TD$FIG]

Fig. 6. Fraction of nutritional energy derived from fat (fat E%) for urban and rural

populations in nine countries based on data from 11 food surveys in developing

countries (see Appendix B).

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608602

and commodities. It is also in accordance with Sen (1981) the Nobel

prize winner who found that food shortages do not result from a lack

of food but from a lack of access to food. By simplifying per capita

consumption further than existing studies, we identify strong

similar trends and shows general patterns in nutritional changes.

WhenFigs.2 and 5a are compared,animportantresult isthat for low

income countries the increase in supply happens faster than the

change in composition. This is relevant for environmental studies.

An increase in low incomes means initially that people buy more of

the same foods. When this is the case the use of natural resources is

linear to supply in terms of food calories. Next, people shift towards

consumingmore fatsandanimal foods and this changecontinues for

longer. This requires the production of commodities that entail a

different and possibly increased use of natural resources.

Per capita income and nutritional energy intake

Most of the available food surveys used in this study (see

Appendix B) were done in countries with large differences in per

capita income. Sixteen surveys indicate nutritional energy intakes

between 2000 and 2500 kilocalories per capita per day, which is in

the range of actual physical requirements (Whitney & Rolfes,

1999). Eight surveys show intakes between 1700 and 2000

kilocalories, while five studies report intakes of over 2500

kilocalories per capita per day. We find no relationship between

nutritional energy intake and annual individual per capita GDP,

however. For the Netherlands and the US, for example, two

countries with a high GDP, energy intakes are between 2000 and

2500 kilocalories per capita per day. This is similar to intakes in

countries with low GDPs. Twelve food surveys make a distinction

between rural and urban consumption, but do not find substantial

differences between energy intakes. The general impression is that

in western countries per capita food consumption, expressed as

nutritional energy intake, has increased over the last few decades.

In science and anecdote it is often assumed that obesity is partly

driven by eating specific foods or by eating in general (De Graaf,

2006; Mela, 2006; Van den Bos & De Ridder, 2006), while most

people consider dieting to be a solution to the problem of obesity

(Polivy & Herman, 2006). We show that nutritional energy intake is

far more constant than food supply or the composition of

consumption.

The gap between supply and consumption

Results show that when income increases the availability of

food also increases (expressed as per capita supply in kilocalories),

but actual consumption remains the same. This means that with

increasing income, the size of the gap between actual per capita

consumption and supply grows. For low GDPs the study finds a

ratio of supply to actual consumption of about 1.0. For high GDPs,

however, this ratio is higher, about 1.8, which indicates that about

half of supply is not eaten at all. This gap is larger than estimates

showing that in 1995 total losses in the food chain in the US were

27% of total supply (Scott Kantor, Lipton, Manchester, & Oliveira,

1997). That study, however, excludes weight reductions that occur

when commodities are processed into final food products, so

producing estimates lower than our results. A possible explanation

is that affluent countries with a high GDP also have more

industrialized food industries with longer food chains that are

probably less efficient, than countries with low GDPs. Another

explanation might be that in countries with a low GDP, where food

is more scarce, people prepare and consume food more efficiently

and so generate fewer and smaller waste streams. The FAO food

balance sheets do not provide information to confirm this

hypothesis. The evaluation of the increasing gap between supply

and per capita consumption that coincides with increasing GDP

requires further research. The result is important for environmen-

tal sciences, because it means that rising incomes are accompanied

by a less efficient use of natural resources.

Uncertainty and inaccuracy of results

Four factors cause uncertainty and inaccuracy of results. These

are: (i) data quality; (ii) the use of average data; (iii) the use of

supply data and (iv) the use of inhomogeneous data. A fifth factor

adds to uncertainty in interpreting the results: (v) uncertainty in

spatio-temporal analogues.

Data quality

The first factor that contributes to uncertainty and inaccuracy is

data quality. This study derives data on food supply from FAO food

balance sheets (FAO, 2010), for which the FAO obtains data from

national datasets. However, data from different countries probably

are not of equal quality, as this depends on the degree of

development of national statistical organizations. Within coun-

tries, data quality varies between years. Major events, such as

political instability, or improvements in the methods of statistical

organizations affect data quality. Even in countries with high-

standard statistical organizations, different sources provide

different data. For the Netherlands in 2000, for example, per

capita butter consumption varies by a factor of three among

datasets. According to the FAO the Dutch consume 2.1 kg of butter

per capita per year (FAO, 2010), according to the Statistics

Netherlands (CBS) 3.3 kg (LEI-DLO/CBS, 2002) and Eurostat

estimates 6.8 kg (LEI-DLO/CBS, 2002). The FAO adjusts basic data

and estimation/imputation of the missing data is necessary in

order to maintain a certain degree of consistency, completeness

and reliability in the food balance sheets (FAO Statistics Division,

2008). Although for some countries data quality might be poor, the

FAO food balance sheets are the only source of information

available to perform an analysis of the type presented here.

The study derives data on GDP from Maddison (2003), who has

expressed GDP in 1990 International Geary-Khamis dollars. The

Geary-Khamis system is an aggregation method in which

international prices and a country’s Purchasing Power Parity,

depicting relative country price levels, are estimated from a system

of linear equations and expressed in G-K dollars (United Nations

Statistics Division, 2006). Individual country GDP values can be

substantially different depending on the PPP methodology used,

however. To compare cross-sectional and time series relationships

this study prefers to apply only one source of GDP data and

therefore uses the database of Maddison (2003), since it covers

both historical and recent global information.

Data on historical food supply are obtained from the historical

analysis of Fogel and Helmchen (2002). That study reported

nutritional energy intakes below physiological requirements. In

general, nutritional energy requirements are constant per unit of

body mass (Whitney & Rolfes, 1999). Average energy intakes below

the physiological requirement might be possible, though, if the fact

that three centuries ago people were smaller and had more

children, with less body mass, is taken into account.

The use of average data

The second factor that contributes to uncertainty and inaccu-

racy is the use of average data. Per capita data are derived from

information on a national level and are therefore average numbers.

In some countries, disparity in income distribution is large and

differences in food consumption occur among population groups.

These differences are not reflected in national data, which means

that the use of average data underestimates trends found here.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608 603

The use of supply data

The third factor that contributes to uncertainty and inaccuracy

is the use of FAO supply data that exclude non-market production.

Sometimes people produce food outside the market, for example in

their gardens (Fernandes & Nair, 1986; Pallot & Nefedova, 2003).

The FAO food balance sheets give supply data on a national level

and do not take non-market production into account (FAO, 2010).

By thus excluding non-market production, we probably underes-

timate supply.

The composition of supply and consumption

Before food is available for consumption, commodities and

foods go through complete food chains, from farm to fork, in which

processes are used to produce the final foods. In all chain links and

in transportation between links losses occur. Neglecting losses and

excluding non-market production could have an effect on the

assessments. To analyze the use of supply rather than consumption

data in the analysis of trends in the composition of national food

supply, this study compares differences between the composition

of per capita consumption and related supply. Every commodity

has a specific composition in terms of macronutrients, such as the

fat E%. Results have already shown that the contribution of protein

to nutritional energy of consumption, the protein E%, is stable, but

that fat and carbohydrate E% vary, reflecting a difference in the

composition of consumption. We assume that a difference in fat E%

between per capita consumption and related supply reflects a

difference in composition. For the comparison the study uses

information on the fat E% obtained from 18 food surveys, marked

with an asterisk (*) in Appendix B. The study calculates the fat E% of

related supply using Eq. (2), deriving data from the FAO (2010).

Figure 7 shows that the fat E% of consumption and related supply

are similar, indicating that the macronutrient compositions are the

same. Eventual losses or non-market production do not cause shifts

in the composition. This justifies the use of FAO data for the analysis

of trends in the composition of national food supply.

The use of food survey data

The fourth factor for uncertainty and inaccuracy is that the

study derives information from food surveys that were probably all

performed in different ways, generating different types of

inaccuracies and uncertainties. For example, people tend to

underreport consumption (CBS, 1994; Kok et al., 1993). The way

food surveys have addressed this problem has probably varied

among surveys, causing diverse inaccuracies. The lowest value is

for the Philippines at 1800 kilocalories per capita per day and the

highest for Jordan, 3200 kilocalories per capita per day. These

examples show that methods used for surveys have probably

differed, generating under- and overestimations, an unquantified

inhomogeneity that the present analysis cannot compensate for.

The use of spatio-temporal analogues

An additional fifth factor of uncertainty in interpreting the

results concerns the use of results from cross-sectional analyses for

drawing temporal inferences. Interrelations between countries

and globalization impact on developments in the economy,

agricultural technology and cultural preferences in, especially,

developing countries in a way that may significantly deviate from

historical and present trends in developed countries. Illustrations

of this are the quantitative differences between the cross-sectional

results and the results from the long and medium-term

longitudinal analyses of European countries. Deviations found,

however, are restricted to quantitative results, qualitative findings

were found to be robust. Projections based on the identified

relations are therefore quantitatively uncertain.

Future changes

Although there are many uncertainties and despite the use of

rough estimates, differences among countries, developments in

time and differences between urban and rural populations, all

results show similar changes in direction. It is stressed, however,

that results obtained here cannot be taken at face value. They give

an indication of the direction of changes in food supply, the

composition of consumption and the contribution of animal foods

and of their magnitude. Combined with estimates of increases in

GDP, the study provides a tool to quantify these changes and

indicate where and when they will probably take place.

The most important finding of this paper is that the main

changes occur for per capita annual incomes below 12,500 dollars.

If trends found here are also valid for the future, this has important

consequences in the coming decade not only for food security, but

also for the use of natural resources such as arable land and

freshwater. Currently, about 85% of the world population lives in

six regions: (i) the OECD countries, (ii) Latin America, (iii) Africa,

(iv) China, (v) India and (vi) the rest of Asia. Table 1 shows the

nutrition, GDP and population characteristics of these regions.

In four regions per capita income levels are below 5000 dollars

per year, i.e. within the range where the largest changes occur.

China, India and the rest of Asia combine low GDPs with large

growth rates. This means that in the next 10 years considerable

changes are likely to occur in Asia. If the Asian countries maintain

economic growth along existing lines, the next decade might show

a substantial increase in per capita food supply, while the

composition of consumption might shift towards the affluent

patterns of western countries, characterized by substantial

consumption of fats and animal foods and limited consumption

of starchy staples. Latin America and Africa will probably see little

economic growth. There, population growth will be the main

driver in increasing total food demand. For the OECD no substantial

changes are likely, because food consumption in these countries

has already reached saturation level and population size is more or

less stable.

To estimate food demand for the period 2003–2030, the FAO

(2003) has indicated that developing regions will show a shift

[()TD$FIG]

0 10 403020

Fat E% food surveys

F a t

E %

f o

o d

b a la

n c e s

h e e ts

Fig. 7. Comparison between the fraction of nutritional energy derived from fat (fat

E%) of actual consumption and of related supply. Data on fat E% of consumption

were derived from food surveys, the fat E% of related supply was calculated from

FAO food balance sheets.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608604

towards increased food supply, as well as greater consumption of

specific commodities such as cereals, sugar, oils and animal foods,

while consumption of pulses, roots and tubers will decrease. When

information from our study is combined with estimates of GDP

growth, results can contribute to the scenario analysis of the FAO

and provide additional information on when and where changes

are likely to occur.

The impact on natural resources

Increased supply or a shift in consumption towards foods with

greater requirements for natural resources will have impacts on

the production system and the pressure on these natural resources.

The European transition towards affluent food consumption

patterns was a gradual process, taking place over centuries. The

economies developed step by step and agriculture could keep pace

with the growth in demand. Today, economic growth occurs at a

much faster rate (Maddison, 2003). Especially for developing

countries with a low GDP, this process contributes to pressure on

agriculture to produce sufficient food of a required quality in the

coming decade.

At present 38% of the global land area is in use for food

production (FAO, 2003) and sustainable options to increase this are

few. Moreover, in the last 10 years a global deceleration of yield

growth occurred (FAO, 2003). The pressure on freshwater is also

great. Humans already use 86% of freshwater, mainly for

agriculture. This water can for a large part be attributed to

the consumption of animal foods (Hoekstra & Chapagain, 2007).

The expected global increase in consumption of these foods,

therefore, will put additional pressure on the availability of

freshwater. While land and freshwater are mainly needed in

agriculture, energy requirements occur in all links of a food chain.

In the past decades energy requirements for food increased not

only due to consumption of different foods, but also due to the use

of other production and transportation methods (Gerbens-Leenes,

2006). It can therefore be expected that the general trends of

consumption found in this study will also have a considerable

impact on energy use. Knowledge of the impact of different food

items and categories on the use of resources provides a tool to

indicate pathways towards more sustainable consumption, for

example by increased efficiency, prevention or substitution.

Conclusions

The study confirms that throughout the world a nutrition

transition is taking place, in which people shift towards more

affluent food consumption patterns. This transition is taking place

at different stages and at different paces. The cross-sectional and

time series relationships show similar patterns of change. For low

income countries, an increase in per capita GDP is accompanied by

changes towards the food consumption patterns of western

countries, characterized by a large gap between supply and actual

consumption. Whereas actual consumption remains stable, total

supply (kilocalories per capita per day) differs by a factor of two

between low and high income countries. In this way economic

growth also causes a shift towards a more inefficient food system,

with greater use of natural resources. A second characteristic of

changes in consumption is the switch in the fraction of nutritional

energy from carbohydrates to fats and to animal foods, while the

protein fraction remains stable. People with low incomes derive

nutritional energy mainly from carbohydrates; the contribution of

fats to nutritional energy is small, that of protein the same as for

high incomes and that of animal sources negligible. People with

high incomes derive nutritional energy mainly from carbohydrates

and fats and the contribution of animal sources is substantial. In

general, whenever and wherever economic growth occurs, per

capita food supply and the composition of supply and consumption

show the same change of direction. The results of the study

are based on a simplified food system using rough estimates. In

reality, the food system is far more complex and there are

many factors, such as culture, that influence food consumption

patterns. By simplifying the system, the study shows general

trends that would not have been found in a more detailed analysis.

For specific situations, however, results might deviate from trends

found in here.

The importance for environmental studies is that results show

that the largest changes in food consumption patterns, and thus

the largest increase in the use of natural resources, occurs in the

range of incomes below 5000 dollars per year, i.e. in developing

countries. With an income of above 12,500 dollars saturation has

occurred and per capita use of natural resources for food does not

necessarily increase any further. In the coming 10 years large

changes in food consumption patterns are likely to occur in Asia

and especially in China and India, two countries that combine great

economic growth, low income levels and poor food consumption

patterns. The European transition occurred gradually, enabling

agriculture and trade to keep pace with the growth in demand.

Changes in economic circumstances change the demand for food.

A continuation of present economic trends might cause a

considerable pressure on the food system, because changes are

occurring much faster than they did in Europe and causing

additional pressure on finite natural resources.

Table 1

Per capita nutrition characteristics in 2001, GDP characteristics (dollars), expected national GDP growth and population characteristics for six regions (85% of the global

population).

Region Nutrition characteristics 2001a GDP characteristicsb Population characteristics

Energy

supplyc Fat

Protein

Energy from

animal sources (%)

Annual per

capita GDP 2001

National

GDP growthd Estimated annual

per capita GDP 2015

Size 2001

(billion)e Annual

growthf Size 2015

(billion)

China 2953 26 11 20 3800 8% 11,600 1.29 0.7% 1.42

India 2385 19 9 8 1926 6% 4,300 1.03 1.4% 1.25

OECD 3493 36 12 27 21538 2% 29,500 0.89 0.4% 0.94

Asiag 2540 18 9 9 2760 7% 7,000 0.76 1.3% 1.20

Africa 2519 18 10 7 1615 4% 2,900 0.52 2.6% 0.74

Latin America 2905 26 11 20 6174 2% 8,400 0.45 1.3% 0.54

a Source: FAO (2010). b Source: Maddison (2003). c Kilocalories per capita per day. d Based on data from the International Monetary Fund, 2010IMF (2010) for 2001–2005. e Source: FAO (2005). f Source: FAO (2003). g Without China and India.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608 605

References

Boom-Binkhorst van der, F. H., Winkelman, M. L. J., Lith van, A., von Lossonczy von Losoncz, T. O., Amesz, M. F., & Schure-Remijn, P. J. M. (1997). Voedingsmiddelen en dranken in het Nederlandse voedingspatroon [Food items and beverages in the Dutch food package]. In A. Stafleu, J. M. Veen, & W. H. Vredebregt-Lagas (Eds.), Mens en Voeding (pp. 233–234). People and foodBaarn, the Netherlands: Nijgh en van Ditmar.

Braun von, J. (1988). Effects of technological change in agriculture on food consump- tion and nutrition. Rice in a West African Setting. World Development, 9, 1083– 1098.

Braun von, J., & Paulino, L. (1990). Food in sub-Saharan Africa, Trends and policy challenges for the 1990s. Food Policy, 505–517.

Carlsson-Kanyama, A., Engström, R., & Kok, R. (2005). Indirect and direct energy requirements of city households in Sweden. Options for reduction, lessons for modeling. Journal of Industrial Ecology, 9(1–2), 221–235.

Catsberg, C. M. E., & Kempen-van Dommelen, 1997. The knowledge of Foods. Intro, Baarn, the Netherland (in Dutch).

Centraal bureau voor de statistiek (CBS), hoofdafdeling statistieken van inkomen en consumptie. (1994) [Central Bureau of statistics on income and consumption. (1994)]. Gebruikershandboek budgetonderzoek 1992 [User manual expenditure sur- vey 1992]. Voorburg, the Netherlands: CBS.

De Graaf, C. (2006). Effects of snacks on energy intake. An evolutionary perspective. Appetite, 47, 18–23.

Duchin, F. (2005). Sustainable consumption of food. A framework for analyzing scenarios about changes in diets. Journal of Industrial Ecology, 9(1–2), 99– 114.

Engelenburg van, B. C. W., Rossum van, T. F. M., Blok, K., & Vringer, K. (1994). Calculating the energy requirements of household purchases, a practical step by step method. Energy Policy, 22, 648–656.

Falkenmark, M. (1989). Water scarcity and food production. In D. Pimentel & C. W. Hall (Eds.), Food and natural resources (pp. 164–191). San Diego, CA: Academic Press.

FAO. (2003). In J. Bruinsma (Ed.), World Agriculture. Towards 2015/2030. An FAO Perspective. London: Earthscan Publications.

FAO. (2010). Food Balance Sheets. http://www.fao.org FAO. (2008). The state of food and agriculture 2008. Biofuels. Prospects, risks and

opportunities. Rome. Fernandes, E. C. M., & Nair, P. K. R. (1986). An evaluation of the structure and function of

tropical homegardens. Agricultural systems, 21, 279–310.

Fischer, G., Van Velthuizen, H., Shah, M., & Nachtergaele, F. O., 2002. Global Agro- ecological Assessment for Agriculture in the 21st century: Methodology and Results. International Institute for Applied Systems Analysis, Laxenburg, Austria/Food and Agriculture Organization of the United Nations, Rome, Italy.

Fogel, R. W., & Helmchen, L. A. (2002). Economic and technological development and their relationships to body size and productivity. In B. Caballero and B. M. Popkin (Eds.), The nutrition transition, diet and disease in the developing world (pp. 9–24). Food Science and Technology International Series. London, San Diego: Academic Press.

Gerbens-Leenes, P. W. (2006). Natural resource use for food. Land, water and energy in production and consumption systems. Thesis University of Groningen, Groningen, the Netherlands.

Gerbens-Leenes, P. W., & Hoekstra, A. Y. (2007). The water footprint of the Netherlands. Analysis of international impacts and options for change. CE&M research report 2007R-008/WEM-005. ISSN 1568-4652. Enschede, the Netherlands: University of Twente.

Gerbens-Leenes, P. W., & Nonhebel, S. (2002). Consumption patterns and their effects on land required for food. Ecological Economics, 42, 185–199.

Gerbens-Leenes, P. W., & Nonhebel, S. (2005). Food and land use. The influence of consumption patterns on the use of agricultural resources. Appetite, 45, 21–31.

Grigg, D. (1995). The nutrition transition in Western Europe. Journal of Historical Geography, 22(1), 247–261.

Hertwich, E. G. (2005). Consumption and industrial ecology. Journal of Industrial Ecology, 9(1–2), 1–6.

Hoekstra, A. Y., & Chapagain, A. K. (2007). Water footprints of nations. Water use by people as a function of their consumption pattern. Water Resource Management, 21, 35–48.

Hoekstra, A. Y., & Chapagain, A. K. (2008). Globalization of water. Sharing the planet’s freshwater resources. London: Blackwell Publishing, London.

International Monetary Fund (IMF). (2010). World Economic Database, September 2004. http://www.imf.org/external/pubs/ft/weo/2004/02/data/

Ivens, W. P. M. F. , Dankert, G., Eng van der, P. J., Faber, D. C., van Keulen, H., Klaver, W., et al. (1992). World food production. Heerlen, the Netherlands: Open universiteit.

Jobse-van Putten, J. (1995). Eenvoudig Maar Voedzaam [Simple but nutritious]. Nijme- gen/Amsterdam, the Netherlands: SUN/P.J. Meertens-Instituut.

Karlsson, R. (2001). Svinn i storhushåll. mängder, sammansättning och möjligheter att minska [Losses in food service institutions. Options for reductions]. Master’s Thesis. Stockholm, Sweden: Department of Systems Ecology, Stockholm University.

Kok, R., Biesiot, W., & Wilting, H. C. (1993). Energie-intensiteiten van voedingsmiddelen [Energy intensities of food items]. IVEM-report 59. Groningen, the Netherlands: Center for Energy and Environmental Studies (IVEM).

Kramer, K. J. (2000). Food matters. Thesis University of Groningen, the Netherlands. Kramer, K. J., & Moll, H. C. (1995). Energie voedt, nadere analyses van het indirecte

energieverbruik van voedingsmiddelen [Energy feeds, analysis of indirect energy use of food items]. IVEM-report 77. Groningen, the Netherlands: Center for Energy and Environmental Studies (IVEM).

Landbouw-Economisch Instituut (LEI-DLO), Centraal Bureau voor de Statistiek (CBS) [Agricultural Economic Institute (LEI-DLO), Central Bureau of Statistics (CBS)]. (2002). Landbouwcijfers 2002 [Agricultural data 2002]. ‘s-Gravenhage, the Nether- lands: LEI/CBS.

Lang, T., 2002. Can the Challenges of poverty, sustainable consumption and good health governance be addressed in an era of globalization? In B. Caballero and B. M. Popkin (Eds.), The nutrition transition. diet and disease in the developing world (pp. 51–71). Food Science and Technology International Series. London: Academic Press.

Latham, J. R. (2000). There’s enough food for everyone, but the poor can’t afford to buy it. Nature, 404, 222.

Maddison, S. (2003). The world economy. Historical statistics. Paris, France: OECD. Mela, P. J. (2006). Eating for pleasure or just wanting to eat. Reconsidering sensory

hedonic responses as a driver for obesity. Appetite, 47, 3–9.

Mennell, S., Murcott, A., & Otterloo van, H. A. (1992). Sociology of food, eating diet and culture. London: International Sociological Association SAGE Publications.

Millenium Ecosystems Assesment, 2005. Millennium Ecosystem Assessment Synthesis Report. Pre-publication Final Draft Approved by MA Board on March 23, 2005. Coordinated by the United Nations Environment Programme.

Musaiger, A. O. (1989). Changes in food consumption patterns in Bahrein. Nutrition and Health, 6, 183–188.

Nonhebel, S. (2004). On resource use in food production systems. The value of livestock as ‘rest-stream upgrading system’. Ecological Economics, 48(2), 221–230.

Pallot, J., & Nefedova, T. (2003). Trajectories in people’s farming in Moscow oblast during the post-socialist transformation. Journal of Rural Studies, 19, 345–362.

Penning de Vries, F. W. T., Jansen, D. M., Ten Berge, H. F. M., & Bakema, A. L. (1989). Simulation of ecophysiological processes of growth in several annual crops (pp. 63– 64). Wageningen, the Netherlands: Centre for Agricultural Publishing and Docu- mentation (Pudoc).

Penning de Vries, F. W. T., Van Keulen, H., & Rabbinge, R. (1995). Natural resources and limits of food production in 2040. In J. Bouma (Ed.), Eco-regional approaches for sustainable land use and food production. The Netherlands: Kluwer Academic Publishers.

Pindyck, R. S., & Rubinfeld, D. L. (2005). Microeconomics (sixth edition). Upper Saddle River, New Yersey, 07458, USA: Pearson Education, Inc.

Poleman, Th. T. , & Thomas, L. T. (1995). Income and dietary change. Food Policy, 20(2), 149–159.

Polivy, J., & Herman, C. P. (2006). An evolutionary perspective on dieting. Appetite, 47, 30–35.

Popkin, B. M. (2002). The dynamics of the dietary transition in the developing world. In B. Caballero and B. M. Popkin (Eds.), The nutrition transition. diet and disease in the developing world (pp. 111–129). Food Science and Technology International Series. London: Academic Press.

Receveur, O., Boulay, M., & Kuhnlein, H. V. (1997). Decreasing traditional food use affects diet quality for adult dene/métis in 16 communities of the Canadian Northwest territories. Journal of Nutrition, 127, 2179–2186.

Rockstrom, J. (1999). On-farm green water estimates as a tool for increased food production in water scarce regions. Physics and Chemistry of the Earth, Part B, 24(4), 375–383.

Rodrigues, S. S. P., Caraher, M., Trichopoulou, A., & De Almeida, M. D. V. (2008). Portuguese households’ diet quality (adherence to Mediterranean food pattern and compliance with WHO population dietary goals). Trends, regional disparities and socioeconomic determinants. European Journal of Clinical Nutrition, 62, 1263–1272.

Rosegrant, M. W., & Ringler, C. (1998). Impact of food security and rural development of transferring water out of agriculture. Water Policy, 1, 567–586.

Schmidt-Nielsen, K. (1988). Animal physiology. Adaptation and environment (third ed.). New York: Press Syndicate of the University of Cambridge.

Scott Kantor, L., Lipton, K., Manchester, A., & Oliveira, V. (1997). Estimating and addressing America’s Food Losses. Food review, 20(1), 2–12.

Sen, A. (1981). Poverty and famines. An essay on entitlement and deprivation. Oxford Clarendon Press.

Tukker, A., & Jansen, B. (2006). Environmental impacts of products. A detailed review of studies. Journal of Industrial Ecology, 10(3), 159–182.

United Nations Statistics Division. (2006). http://unstats.un.org/unsd/methods/icp/ gdp/gdp06_htm.htm

United States Department of Agriculture. (USDA) Agricultural Research Service 2005. Food and Nutrient Database for Dietary Studies, 1.0. http://www.ars.usda.gov/ Services/docs.htm?docid=7637

Van den Bos, R., & De Ridder, D. (2006). Evolved to satisfy our immediate needs. Self control and the rewarding properties of food. Appetite, 47, 24–29.

Voedingscentrum. (1998a). In B. C. Breedveld, J. Hammink, & J. H. M. van Oosten (Eds.), Nederlandse Voedingsmiddelentabel. Dutch food itemsDen Haag, the Netherlands: Voedingscentrum[[nl]]Food center.

Voedingscentrum, & TNO. (1998). Zo eet Nederland 1998 [This is how the Netherlands eats 1998]. Den Haag, the Netherlands: Voedingscentrum [Food Centre].

Vringer, K., & Blok, K. (1995). The direct and indirect energy requirements of house- holds in the Netherlands. Energy Policy, 23(10), 893–910.

Wandel, M. (1988). Household food consumption and seasonal variations in food availability in Sri Lanka. Ecology of Food and Nutrition, 22, 169–182.

Whitney, E. N., & Rolfes, S. R. (1999). Understanding nutrition (8th ed., pp. 3–40). Belmont, USA: Wadsworth Publishing Company.

Wijn de, J. F., & Weits, J. (1971). Steensma’s Voedingsleer. Steensma’s Food Science]. Amsterdam/Haarlem, the Netherlands: Scheltema en Holkema.

Wirsenius, S. (2003). The biomass metabolism of the food system. A model-based survey of the global and regional turnover of food biomass. Journal of Industrial Ecology, 7(1), 47–80.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608606

http://www.fao.org/

http://www.imf.org/external/pubs/ft/weo/2004/02/data/

http://unstats.un.org/unsd/methods/icp/gdp/gdp06_htm.htm

http://www.ars.usda.gov/Services/docs.htm%3Fdocid=7637

World Bank. (2005). East Asia’s Dollar Influx-Signal for Change. Special focus. Gender Equality in East Asia. https://publications.worldbank.org/subscriptions/WDI

WWF, 2007. Allocating scarce water. A WWF primer on water allocation, water rights and watr markets. WWF-UK, Godalming, UK.

Appendix A

Overview of the 57 countries for which this study performed

the cross-sectional analysis.

Africa: Algeria, the Democratic Republic of Congo, Côte d’ Ivoire,

Egypt, Ethiopia, Ghana, Kenya, Morocco, Nigeria, Sudan, Tanzania,

and South Africa

Asia: Bangladesh, China, India, Indonesia, Malaysia, Pakistan,

the Philippines, Sri Lanka, Thailand, and Vietnam

Eastern Europe: Poland

Latin America: Argentina, Brazil, Chile, Colombia, Ecuador,

Guatemala, Mexico, Peru, and Venezuela

Middle East: Israel, and Syria

OECD: Austria, Belgium, Canada, Denmark, Finland, France,

Germany, Greece, Iceland, Ireland, Italy, Japan, the Netherlands,

Portugal, Spain, Sweden, Turkey, United Kingdom, United States

Additional, small countries: The United Arab Emirates, Estonia,

Slovenia.

Appendix B

Overview of countries and national food surveys used in this

study. The 18 food surveys that provide data on fat E% are marked

with an asterisk *.

Argentina

Britos, S., & Scacchia, S. (1998). Disponibilidad y consumo de

alimentos en Argentina. Escuela de Nutrición [Food availability and

consumption in Argentina. School of Nutrition]. Argentina: Uni-

versidad Nacional de Buenos Aires [National University of Buenos

Aires].

Bangladesh*

Jahan, & Hossein. (1998). Malnutrition in Bangladesh: Bangladesh

National Nutrition Survey, 1995–96. Bangladesh: Institute of

Nutrition and Food Science, Dhaka University.

Brazil

Galeazzi, M. A. M., & Falchoni Jr., P. (1998). Inquérito de Consumo

Alimentar da Area Metropolitana de Brası́la-Relatório [Nutrition survey

in the area of Brasilia-Relatorio]. Brası́lia: Técnico-Secretaria de Saúde

de Brası́lia.

Cambodja

National Institute of Statistics (NIS/MOP). (1996). Socio-

Economic Survey of Cambodja. Data from the Multi-Indicator Cluster

Survey (MICS) of the Socio-Economic Survey of Cambodja (SESC)

sponsored by the Asian Development Bank in collaboration with the

UNICEF/UNDP/CARERE and ILO. Cambodja: Royal Government of

Cambodja.

China*

Ge, K., Zhai, F., & Yan, H. (1996). Institute of Nutrition and Food

Hygiene (INFH) 1985. Summary Report of the 2nd National Nutrition

Survey in 1982. Beijing, China: Institute of Nutrition and Food

Hygiene.

Ge, K., Zhai, F., & Yan, H. (1996). The dietary and nutritional

status of Chinese population. 3rd National Nutrition Survey, 1992.

Beijing, China: People’s Med. Pub. House.

Colombia

Ministerio de Agricultura DANE-DRI-PAN. (1984). Encuesta

Nacional de Alimentatión, Nutrición y Vivienda DANE-PAN-DRI 1981

[Ministry of Agriculture DANE-DRI-PAN 1984. National Feeding,

Nutrition and Housing Survey DANE-PAN-DRI 1981]. Bogotá:

Franza Pardo T-Bogotá (Mimeógrafo).

Costa Rica

Ministerio de Salud. (1996). Ministerio de Salud 1996. Encuesta

Nacional de Nutrición. Fasciculo No 1: Consumo Aparente [Ministry of

Health 1996. National Nutrition Survey. Fascicle No 1: Apparent

Consumption]. San José, Costa Rica.

Egypt*

Hassanyn, A. S. (2000). Food Consumption Pattern and Nutrients

Intake Among Different Population Groups in Egypt. Final Report

(Part 1). Egypt: Nutrition Institute, WHO/EMRO.

El Salvador

Asociación Demográfica Salvadorĕna (ADS), Ministerio de Salud

Pública y Asistencia Social (MSPAS), & Instituto de Nutrición de

Centro América y Panamá (INCAP) [Salvadoran Demographic

Association, Ministry of Public Health and Social Assistance, &

Institute of Nutrition of Central America and Panama (INCAP)].

(1990). Evaluación de la Situación Alimentaria Nutricional en El

Salvador [Evaluation of the nutritional situation in El Salvador]. El

Salvador: ESA NES-88.

Equador

Freire, W. (1988). Diagnóstico de la situación alimentaria y

nutricional y de salud de la población ecuatoriana menor de cinco años

– DANS -1986 [Diagnosis of the alimentary, nutritional and health

state of the Ecuadorian population less than five years – DANS -1986].

Quito, Equador: CONADE, MSP.

Iran*

Djazayery, A., & Samimi, B. (1996). (Surveys for 1983 and 1992)

Food consumption and energy intake patterns in the rural and

urban areas of Iran, 1983–1992. Agricultural Economics and

Development, 4, 218–248.

Jamaica

Simeon, D. T., & Patterson, A. W. (1994). Energy and protein

accessibility at the household level in Jamaica: Results from a national

survey 1989. Jamaica: CFNI.

Jordan*

Department of Statistics (DOS). (1997). Household Income and

Expenditure Survey. Amman, Jordan.

Madagaskar

FAO. (2004). L’état de l’insécurité alimentaire 2001 dans le monde

[The state of food insecurity in the world]. Rome, Italy: Organisation

des Nations Unies pour l’alimentation et l’agriculture [Food and

Agriculture Organisation of the United Nations], http://

www.fao.org.

Mali*

FAO. (2005). Profiles nutritionnels par pays [Nutritional profiles

per country]. Mali: Departement Economique et Social, Alimenta-

tion et nutrition [Department of Economic and social affairs, food

and nutrition]. http://www.fao.org/es/nutrition/mli-f.stm.

Mexico*

INNSZ. (1990). Encuesta Nacional de Alimentación en el Medio

Rural ENAL 1989 [National Feeding Survey in Rural Areas ENAL 1989].

México: INCMNSZ.

Avila, A., Shamah, T., & Chavez, A. (1997). Enquestas de

Alimentación y Nutrición en el Medio Rural, 1996. Resultados

por entidad [Feeding and Nutrition Surveys in Rural Areas,

1996. Results by organization]. INNSZ, DEDESOL, DIF, SSA, Golernos

de los Estados [Governments of the States]. Mexico: IMSS, INI, Unicef.

The Netherlands*

Voedingscentrum [Food Center], & TNO. (1998). Zo eet Neder-

land 1998 [This is how the Netherlands eats 1998]. Den Haag, the

Netherlands: Voedingscentrum [Food Center].

Panama

Ministerio de Salud. (1992). Ministerio de Salud 1992. Encuesta

Nacional de Consumo de Alimentos. Panamá: Departamento de

Nutrición y Dietética Panamá [Ministry of Health 1992. National

Food Consumption Survey. Panamá: Dietetic and Nutrition Depart-

ment Panamá].

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608 607

https://publications.worldbank.org/subscriptions/WDI

http://www.fao.org/

http://www.fao.org/es/nutrition/mli-f.stm

Instituto de Nutrición de Centro América y Panamá (INCAP),

Oficina de Investigaciones Internationales de Salud, & Ministerio

de Salud Pública y Asistencia Social (MSPAS) [Institute of Nutrition

of Central America and Panama (INCAP), International Health

Research Office, & Ministry of Public Health and Social Assistance

(MSPAS)]. (2000). Evaluación nutricional de El Salvador 1969

[Nutritional Assessment of El Salvador 1969].

Peru

Amat, C., & Curonisy, P. (1981). La alimentación en el Perú

[Feeding in Peru]. Lima, Perú: Centro de Investigación [Research

Center University of the Pacific].

Philippines*

Food and Nutrition Research Institute of the Department of

Science and Technology (FNRI-DOST) of the Philippines. (2000).

National Survey of 1993: Final Results.

Sri Lanka*

Department of Census and Statistics. (1993). Household Income

and Expenditure Survey 1990/91, Final Report. Department of Census

and Statistics. Sri Lanka: Ministry of Policy Planning and

Implementation.

Turkey*

Hundd, & Moh. (1997). Food consumption survey in 7 provinces,

Project Report. Ankara, Turkey: Hacettepe University, Department

of Nutrition and Dietetics, Ministry of Health.

United States*

United States Department of Agriculture (USDA) Agricultural

Research Service. (2005). Food and Nutrient Database for Dietary

Studies, 1.0. http://www.ars.usda.gov/Services/docs.htm?docid=

7637.

Venezuela*

Luna Bazó, P., & Bracho, M. (1987). Encuesta Nacional de

Nutrición. Area Socio Alimentaria ‘‘Encuesta de Consumo’’. Mimeo-

grafiado [National Nutrition Survey. Socio Alimentary Field ‘‘Con-

sumption Survey’’. Mimeografiado]. Caracas, Venezuela: Instituto

Nacional de Nutrición, Direccion Téchnica [National Nutrition

Institute, Technical Direction].

Vietnam*

Tu Giay, & Chu Quoc Lap. (1990). Final report on the subject

64D.01.01 of the National Research Programme National General

Survey 1989. Hanoi, Vietnam: The Governmental Science and

Technology Committee, NIN.

National Institute of Nutrition (NIN). (1995). Sentinel food and

nutrition surveillance system data. Hanoi, Vietnam: NIN.

Zimbabwe

Bursztijn, P. G. (1985). A diet survey in Zimbabwe. Human Nutr.

Appl. Nutr. 39 (5), 376–388.

P.W. Gerbens-Leenes et al. / Appetite 55 (2010) 597–608608

http://www.ars.usda.gov/Services/docs.htm%3Fdocid=7637