CCP:GR-RI/04/4


Item II.D of the Provisional Agenda

COMMITTEE ON COMMODITY PROBLEMS

JOINT MEETING OF THE
INTERGOVERNMENTAL GROUP ON GRAINS
(30TH SESSION) AND THE
INTERGOVERNMENTAL GROUP ON RICE
(41ST SESSION)

Rome, Italy, 10-11 February 2004

CEREALS AND OTHER STARCH-BASED STAPLES:
ARE CONSUMPTION PATTERNS CHANGING?

Table of Contents



I. ABSTRACT

1. Dietary patterns of cereals, root crops and other starchy staples around the globe are constantly evolving. Using descriptive analysis of time series data in conjunction with non-parametric techniques, this paper explores the nature of this evolution and attempts to elicit the forces driving change. It is found that as overall calorie intake has grown over time, dietary shares of carbohydrate-based staples have declined. The falling importance of carbohydrate-based commodities in national diets has masked considerable intra-variability in the consumption of these foodstuffs. Such changes have been pervasive, and appear to be dominated by a shift away from lower-valued staples. In addition, a more diversified structure of starchy food intake has emerged in the majority of countries, while at the same time, consumption patterns in the different regions appear to be converging, especially towards wheat and rice and away from “minor” coarse grains, plantains and roots and tubers. The analysis confirms that urbanisation, income and preference shifts have played an important role in determining dietary intake of starchy staples. These three variables, which are seemingly interrelated, are likely to continue shaping consumption patterns in the future.

II. INTRODUCTION

2. Throughout the world, growth in per capita food consumption has been accompanied by substantial change in the commodity composition of diets. This paper reviews and analyses evolving patterns of food consumption over time with particular emphasis on the consumption of carbohydrate-based staples, namely, cereals, starchy roots, bananas and plantains (CRBP). This study will seek to address the following issues:

3. Owing to space limitations, the implications of the findings are not fully explored in this study, but needless to say, they could provide an important contribution in assessing policies towards nutrition, trade and food security, especially for predicting future food needs.

4. The consumption data employed in this study are drawn from FAO supply utilization accounts, which cover the annual period from 1961 to 2001. A total of 126 developing countries, plus major sub-regional groupings and a single group representing all developed countries, are included in the analysis.

5. A combination of descriptive analysis of time series data and non-parametric techniques conducted on the dietary shares of food commodities, expressed in calories per person per day, forms the methodological basis for this paper. The use of “shares” has the advantage of revealing the relative importance of food types in overall consumption. Since this approach is unit-free and invariant to potential inconsistencies in measurement, it also permits the construction of aggregates and importantly, it allows for comparisons of dietary structures.1

6. Generality is adopted at the expense of precision, since a rigorous, parametric analysis at the country level, such as standard demand models, would require a considerable amount of data, which in many cases do not exist or are sparse. Moreover, a central aim of this paper is to make inference about the broader factors affecting change across a large number of countries, and the techniques that are employed, fit neatly into this appeal.

III. BACKGROUND

A. CEREALS AND STARCH BASED STAPLES IN OVERALL FOOD CONSUMPTION

7. At the global level, CRBP outweigh all other foodstuffs as sources of daily dietary energy. According to the latest FAO estimates (2001), these foods provide 53 percent of the world’s average daily calorie intake. At the country level, however, there is wide diversity, for instance, CRBP commodities in 2001 accounted for approximately 26 percent of daily calories consumed in the United States, but in Bangladesh, this was 83 percent.

8. Despite growth in most countries in the absolute level of consumption of CRBP, the relative importance of this food group as a staple is in long term decline. As calorie intake has increased over time, the share of CRBP in daily diets has fallen. The decline is pervasive: out of the 181 countries for which consumption data are reported by FAO, 147 experienced negative growth in the share of CRBP over the period 1961-2001. Figure 1 illustrates this tendency, and also contrasts the dietary importance of CRBP foodstuffs in developing and developed countries. It is seen that while a considerable gap exists between both country groups, CRBP dietary shares over the past decade appear to have somewhat narrowed.

9. Falls in shares of CRBP in diets have been accompanied by rapid increases in the consumption of oils and fats, meat and fish, dairy products and to a lesser extent, sugar. As can be seen from Figure 2, these four food groups together now constitute more than half of the total daily calories available for food in the developed countries and almost a third in developing countries.

Figure 1: Dietary shares of CRBP staples in developed and developing countries

Undisplayed Graphic

10. Growth in per capita food intake has been accompanied by substantial variation in commodity composition, particularly in developing countries. These tendencies are further evidenced in Table 1 and Figure 3: with the former tabulating changes in the structure of food consumption by the major country aggregates, while the latter reporting on the variation2 in dietary structures over time for these country groups.

11. Of the groups reported in Figure 3, developed countries exhibit the lowest variation over time in consumption patterns: a part of this due to the fact that food consumption in these countries has reached saturation levels, leading to declining substitution among food types. By contrast, product variability in developing countries has been pronounced (see Annex 1). Among them, countries located in South Asia have experienced the largest changes in dietary patterns. Despite only a moderate increase in regional calorie intake, dietary variability has arisen from strong growth in fats and oils consumption and a marked fall in pulse demand have been the main reasons for the observed increased variability. Food consumption in East and South East Asia has increased more than in any other region (43 percent over 1961-2001) and has been accompanied by large variation in food patterns, through higher meat, sugar and oils and fats shares and rapidly declining consumption of CRBP foodstuffs.

Figure 2: Comparison of Dietary Structures in Developed and Developing Countries

Undisplayed Graphic

 

Figure 3: Variation in Dietary Patterns, 1961-2001

Undisplayed Graphic

 

Table 1: Changes in the Commodity Composition of Food Consumption,
Major Country Groups

 

12. Growth in total food consumption in Sub-Saharan Africa has been the lowest of any region, and substitution between major food groups has been limited, with CRBP foods continuing to dominate diets. As a result, dietary patterns have been relatively stable over time for the region. Likewise, regional food patterns have varied little in North Africa, Near East, Oceania, Latin America and the Caribbean, even though they have experienced moderate to large growth in overall calorie intake. An exception to variability has been rapidly rising oils and fats consumption – a feature in all three regions - and a fast declining dietary share of fruits and vegetables in Oceania.

13. It is clear that dietary patterns in most regions have undergone varying degrees of change, but at any point in time, there appears to be a marked relationship between the level of total calorie intake and dietary shares of particular food commodities. For instance, developed countries consume more calories than any other group, but their share of CRBP is the lowest among all. At the other extreme are countries situated in Central Africa, a region which ranks first in terms of under-nourishment and where total calorie intake is around half of developed country intake, but where CRBP food shares are double those of developed countries. A similar negative relationship holds for pulses.3 By contrast, for all other commodities, except for fruit and vegetables, dietary shares appear positively related to the level of calories consumed.

14. At the disaggregated level, Table 2 presents the results of a regression analysis, which estimates the relationship between dietary shares and total calories consumed across all countries for the major food groups. The elasticities, which measure the responsiveness of dietary shares to changes in total calorie intake, reported in the table, tend to confirm the descriptive evidence at the regional level, that is, that the share of CRBP (and pulses) in total food consumption tends to decline as overall calorie intake rises, contrary to what happens for the other major food groups.

Table 2: Relationship between dietary shares and total calorie intake

Commodity

Elasticity

CRBP

-0.743

Dairy

0.713

Fruit and vegetables

0.027

Meat and fish

0.604

Oils and Fats

0.656

Pulses

-0.546

Sugar

0.695

15. In summary, these findings suggest that when consumers have improved access to foodstuffs, whether through their increased affordability or through their greater availability, they have tended to vary their diets with foods other than the traditional staples of CRBP and pulses.

16. The aggregates reported in Table 1 and Annex 1 are not particularly insightful in revealing the full nature of changes in food consumption patterns, particularly when there is large heterogeneity within regional diets or when large changes in major consuming countries have a distorting effect on the aggregates. For these reasons, the selected distributions in Annex 3 illustrate compactly, the number of countries that have experienced shifts in dietary shares, and provide information about the nature of these shifts.4

17. For example, over the period 1961-2001, the share of CRBP foodstuffs in the diets has fallen in around 80 percent of all countries, with one-in-three countries experiencing falls of 10 percent or more. Changes in CRBP dietary shares,5 have been large and negative over time. This is not the case for meat, oils and fats, dairy and fruit and vegetables. Globally, shifts6 in these commodities have been for the most part positive, but small and gradual over time, with the exception of several countries that have experienced very large dietary share increases of these foodstuffs.

B. CONSUMPTION PATTERNS WITHIN THE CEREALS AND STARCH-BASED STAPLE GROUP

18. Consumption shares of CRBP foodstuffs have exhibited the largest variability among all food products, but this particular attribute could well be a product of the considerable intra substitution that has taken place within this food group. Indeed, dietary patterns of within CRBP foodstuffs are undergoing a constant process of change, just as they are for the major food types. The following summarises the main characteristics of the individual products, in terms of their geographical concentration of consumption, their major end usage and the evolution of dietary patterns. The relevant data for the discussion are contained in Table 3 and Annexes 2 and 3.

19. Most countries produce or import a variety of CRBP foodstuffs. The predominant type varies from region to region, but it is almost always supplemented by other starch-based staples. Beginning with the grains, wheat is currently the primary staple food for almost one-third of the world’s population (including most developed countries). Among the developing countries, wheat ranks first in dietary shares in countries in the Near East and North Africa, in many localities in Latin America, Pakistan and the North of India. It is also the secondary staple for over 3.3 billion people in developing regions (or 70 percent of this population). The availability of locally grown wheat is limited by climate, since its cultivation is possible only in temperate zones or the higher elevations of tropical and sub-tropical zones. Consequently it is also a widely imported foodstuff. Moreover, wheat is not a homogeneous grain. For instance, in the Near East and North Africa, and in parts of South Asia (India and Pakistan), wheat is processed into an unleavened bread; in East and South East Asia, it is eaten in the form of noodles; and in North Africa, as ‘couscous’. Other wheat preparations featuring in many diets include bread, biscuits, and bakery products.

20. Over the past four decades, shares of wheat in national diets of developing countries have more than doubled, as evidenced in Annex 2. The prevalence of this growth is depicted in Annex 3. As much as 80 percent of all developing countries have experienced increasing shares of wheat consumption, with one-in-three countries experiencing shifts in shares of 10 percent or more. Much of the expansion took place before the 1980s, reflecting inter alia the rapid increase in imports, with countries in Asia and sub-Saharan Africa being the main engine of growth. The 1980s witnessed a sharp slowdown in the growth of wheat shares, but there has been a resurgence in the past decade with around 60 percent of all developing countries recording increased shares of this commodity in CRBP consumption.

Table 3: Changes in the Commodity Composition of CRBP Consumption, Major Regions

Undisplayed Graphic

 

21. Maize represents either the major staple or main supplementary staple for 1 billion people in developing regions, mostly in Sub-Saharan Africa, which is largely attributed to the fact that it can be cultivated in most tropical and sub-tropical areas, that it is widely traded, and also owing to its end-use versatility. In Latin America, white and yellow maize is extensively used to make unleavened bread (‘tortilla’), and also eaten ‘on the cob’, while in Sub-Saharan Africa white maize is processed into various products, but popular forms include starchy pastes such as porridge. In the Near East, maize flour is commonly used to make bread, while in South and South East Asia (notably Indonesia and the Philippines) it is consumed in a number of diverse ways. At the global level, maize has held its share as a staple in diets over time, with moderate increases in shares of Central, Southern and East African countries. However, this could be associated with larger aid flows of maize into these regions, which have sustained consumption, following internal conflicts and climatic crises that have been rife in these sub regions.

22. Millet and sorghum are grown chiefly in areas that are able to withstand drought and heat Both are major foods for around 60 million people concentrated in the inland areas of tropical Africa, who consume them mostly in the form fermented or unfermented preparations These grains also appear in diets in large parts of India and Pakistan, where they are consumed predominantly as unleavened bread. Among the other coarse grains, barley is popular in North Africa and the Near East, where it is used to make bread and as an ingredient added to soup dishes. It is also widely eaten in parts of East Asia (Republic of Korea and Japan particularly), where it is added to rice. Rye is commonly used to make bread in many developed countries, particularly those in Northern and Eastern Europe, but also in areas of North Africa and the Near East. Throughout most regions, the shares of these coarse grains in diets have undergone pronounced falls. As a combined group they once constituted the major staple in much of Sub-Saharan Africa, notably in countries in West Africa, but the popularity of the more refined cereals, particularly rice and wheat and to some extent maize has led to their substitution.

23. Rice is grown in many regions, under a wider variety of climatic and soil conditions than any other crop. It is nearly always eaten in boiled form, without further processing other than milling, which is in contrast with most other cereals. However, in some countries in the Far East, manufactured rice products appear in diets, mainly in the form of rice noodles, which compete to some extent with wheat noodles. Rice is the major cereal for roughly 3.4 billion people in developing countries. Almost nine tenths is produced and consumed in the Far East, where in many countries, notably China, Thailand, Myanmar, Cambodia, Viet Nam, Malaysia and Sri Lanka, it represents the main staple, far outweighing other CRBP staples. In other regions, rice is rarely the predominant food (a handful of Sub-Saharan African countries are an exception), but as can be seen in Annex 2, it is a highly esteemed supplementary staple, growing in popularity in much of Africa, Latin America and the Caribbean, Oceania, and in the Near East.7

24. As for the non-cereal staples, starchy roots and bananas and plantains, consumption is mostly confined to the source of production, given their perishable nature and also the costliness of transportation. This is somewhat reflected in the small share of just 10 percent of consumers in developing countries that depend on starchy roots and tubers for primary or secondary sources of dietary energy. Cassava and potatoes are the dominant food types in this group. Over time, potato-based products have registered some growth, while cassava and other starchy roots have largely maintained their share in consumption in developing regions. This trend is not reflected in the commodity aggregates, where the drastic decline in food consumption of sweet potatoes in China has had a decisive influence on the overall trend.

25. Cassava is widely cultivated in tropical regions. Its popularity derives mainly from its easy cultivation in warm humid climates and on impoverished soils, and to its resistance to drought and pests. In many cases it is planted as a reserve against famine and harvested at the convenience of the farmer over a period of many months. Cassava provides for most of the carbohydrates in many areas of tropical Africa (notably, the Democratic Republic of Congo, Ghana, Mozambique and Angola) and is also the main staple in Paraguay, in parts of Indonesia, Philippines, Sri Lanka and India. The main cassava root product is a coarse meal or a flour, which can then be made into bread (especially in Latin America and the Caribbean) but sometimes it is consumed as a vegetable (e.g. Oceania). Some countries, such as Thailand and Indonesia, manufacture cassava (tapioca) flour, which is subsequently processed for direct consumption.

26. Potatoes represent the secondary staple in many developed countries, but its consumption in developing countries is less widespread and is heavily concentrated among a few countries in the Near East (Turkey, Tunisia ) and South America (Argentina and Peru) and Malawi. Sweet potatoes and yams are the primary staple in several Sub Saharan African countries and numerous island states in Oceania, and the third or fourth staple in many other developing countries.

27. Among all the staple carbohydrate food groups, bananas and plantains rank the smallest in terms of global consumption; nevertheless, these foodstuffs are the primary staple for over 25 million people in developing countries. Cultivation and consumption of bananas and plantains is mainly confined to tropical zones, for instance, East Africa (notably Uganda), Central Africa, the Caribbean and Oceania.

28. In summary, the most conspicuous development in the dietary structures of CRBP staple foods, has been the trend towards wheat and rice at the expense of the lower valued, coarse grains. In many areas, especially where total calorie intake has been low and where wheat and rice initially have accounted for a small proportion of the diet, the increased consumption of both cereals has tended to supplement rather than displace traditional staples. West Africa is an exception, where rice has superseded the other individual coarse grains, and now forms the region’s primary staple. Most of the larger changes in CRBP consumption were concentrated in the 1960s and 1970s, and dietary patterns of these foodstuffs have since shown less marked changes

IV. IDENTIFYING THE COMMON DETERMINANTS OF EVOLVING CRBP DIETARY PATTERNS

A. AN OVERVIEW OF THE FACTORS DRIVING CHANGE IN FOOD CONSUMPTION PATTERNS

29. The evolution of dietary patterns is regarded as the result of changes in main economic determinants of consumption, namely income and relative prices. However, consumption patterns also respond to other important factors such as urbanisation, immigration, population age, as well as government policies and consumer preference shifts. A priori there are reasons to suggest that some forces will be more pertinent than others in explaining changes in CRBP patterns.

Income and relative prices

30. There is a substantial body of evidence suggesting that the proportion of budgets spent on food decreases as income rises.8 Several studies report that developing countries spend as much as 50 percent of their budgets on food: a great deal higher than the 15 percent estimated for developed countries, where such expenditures are much less responsive to changes in income and food prices than they are in developing countries.

31. Irrespective of level of development, estimated aggregate price and income elasticities tend to be much lower for staple commodities, typically CRBP, than other foodstuffs; while, income and price elasticities of demand for higher valued food categories tend to be inversely related to the level of income. Thus, given that consumer reactions to income and price changes differ across food types, rising income or variations in prices will change the composition of food demand and these changes will be more pronounced in developing countries. By contrast, consumers in developed countries are expected to make relatively small adjustments between food consumption groups when they experience changes in income or prices.

32. The increased diversification of developing country diets has occurred alongside growth in overall calorie intake; suggesting that changes in income might be an important factor in explaining differences in dietary patterns. However, much of the empirical evidence on dietary change looks at the substitution between the main food categories rather than the nature of adjustments within the same food groups. To put matters into context, among the different food groups, CRBP has exhibited the most variation, but also intra product variability has been exceptionally high. What then is the relevance of prices or income in explaining these variations? A plausible answer is provided by the theory of “two-stage budgeting”. According to the theory, consumers are assumed to allocate their income first to broad consumption groups, and then make purchasing decisions for items within each group typically based on their relative prices. For example, when the price of rice increases consumers may choose to consume more maize or sorghum, or if food budgetary outlays increase, consumers may wish to buy more meat, rather then buying more cereals. Ultimately, when per capita incomes reach a sufficiently high level, the influence of income and price on the choice of food becomes less significant and other factors (such as convenience) assume greater importance.

Urbanization

33. Research has shown that changes in demographic variables lay behind the evolution of food consumption in many countries.9 One of the most influential variables has been urbanisation, which is defined as the movement of populations from rural areas to urban centers. While most developed countries have, by and large, completed such transition, urbanisation is still an ongoing process in many developing countries.

34. Urbanisation is largely associated with economic growth, which can undergo large upswings and downturns, but there is something more of a rigidity and permanency with urbanisation. That is, when incomes do fall, consumers in cities do not necessarily ‘ruralise’,10 and when incomes rise over sufficiently large periods, rural populations might diminish to the extent that urbanisation would no longer be a factor in explaining consumption patterns.

35. Aside from urban incomes, which tend be, on average, much higher than rural incomes, there are several other reasons, widely cited in the literature, that explain why rural and urban diets differ:

36. The effects of urbanization on the composition of national food diets are not necessarily uniform and are largely dependent on the characteristics of existing dietary patterns. For example, referring to the research cited in note 13, increased urbanization has led to declining rice consumption in Asia, while it has boosted rice consumption in Sub-Saharan Africa. Nonetheless empirical evidence reveals there is some consistency in responses across countries. For instance, household survey data collected by FAO (1993) showed that in rural Indonesia, urbanisation led to a fall in starchy root consumption and increased consumption of higher valued staples. In India, while rice consumption was found to be fairly stable in both urban and rural areas, average per capita wheat consumption increased, displacing the consumption of lower-valued coarse grains, a pattern largely associated with urban population growth.

Government policies

37. Governments of many developing countries employ food security measures to ensure that adequate food supplies are available and to keep consumer prices within “reasonable” limits. The instruments used commonly involve price controls,12 the distribution at subsidized prices to low income segments of the population, regulations on marketing and trade etc. Insofar as such measures discriminate in favour of the dominant staple food, they indirectly, affect the pattern of consumption and alter the competitive position of one staple vis-à-vis another and tend to reinforce the position of the national staple.

Consumer preferences

38. Changes in consumer preferences play an important role in shaping patterns food consumption. Preferences are symbolised by the perceptions, tastes and attitudes that consumers hold towards food types. Given a fixed set of relative prices and a budget constraint, they largely determine the choice of foods that consumers purchase, and ultimately shape consumer behaviour.

39. Preferences, themselves, are determined by a host of social and demographic factors, including urbanisation. For example, exposure to a greater variety of foodstuffs and advertising enables urban consumers to make more informed choices, which then lead to realigning preferences. In this respect, several authors13 have shown that there is a tendency for dietary structures to become increasingly similar across countries that they attribute to the convergence of consumer preferences. Many of them argue that globalisation, through the presence of multinational food processing and distribution firms operating in many countries, reinforced by heavy advertising, has facilitated this process of convergence.

40. Preferences, however, are generally unobservable. All that can be learnt about them is from the behaviour that can be observed when consumers respond to changes in income, prices, and other factors.

B. METHODOLOGY

41. The remainder of this section attempts to elicit the factors, and their relative importance, in explaining changes in CRBP consumption around the globe.

42. An alternative framework to traditional parametric analyses at the country level (e.g. regression-based demand models) is employed, since, standard models of demand require a considerable amount of data, which in many cases do not exist or are sparse, and, furthermore, cannot identify the broader factors affecting changes across a large number of countries, which constitutes a central aim of this paper.

43. The method employed in this section combines ‘cluster analysis’ to determine the group structure of countries based on variables identifying patterns of change in the dietary shares of CRBP foods14 with ‘discriminant function analysis’ to determine the economic and social characteristics of the countries falling into the different groups thus identified. These techniques allow the determination of those factors that differentiate between the country groupings that have different relative CRBP shares. While this framework is less demanding in terms of data requirements than others, it nevertheless can still provide some understanding about the nature of changing pattens of consumption across a large number of countries.

C. FINDINGS

44. Cluster analysis was applied to each year, over the period 1961-2001, to identify how CRBP consumption patterns have evolved over time. The results are summarised in Figure 4. The average shares of the primary and secondary commodities in CRBP consumption are used to label the identified clusters. When the average group share of a single CRBP commodity exceeded 75 percent, the dominant commodity appears in the label twice. In addition, the figure maps country movements between particular CRBP structures over the periods 1961-1981 and 1981-2001.

45. At any point in time, the clusters appear plausible on both casual and close examination. For instance, CRBP clusters appear largely dependent on geographical boundary, but also on the income status of the countries. Comparing the mappings over time, several prominent features emerge:

46. A small group of countries, (Cameroon, Haiti, Namibia, Tanzania and Togo), have maintained the same pattern of CRBP consumption (mostly coarse grains and roots) since the 1960s. In some instances this could be a result of either entrenched preferences for particular CRBP commodities or unavailability/unaffordability of alternatives for the consumers in those countries. All of the other countries have changed their dietary structure of CRBP foodstuffs at some point in time. Among the most marked changes has been the diversification towards wheat, which is reflected by the doubling of the number of countries relying on this commodity as a primary or secondary staple since 1961. Furthermore, almost two-thirds of all countries now have CRBP structures dominated by ‘rice-wheat’ or ‘wheat-rice’, and structures consisting predominantly of starchy roots and or coarse grains have largely diminished. These general movements to and away from different CRBP foodstuffs are supported by the earlier analysis.

47. Discriminant analysis was then undertaken to identify the forces that have shaped CRBP structures. Among the variables discussed in Section 2.3 that are amenable to measurement, urbanisation, income, imports of CRBP foodstuffs (an indicator of availability), were deemed likely candidates15. It is reasonable to regard the four-cluster structure as having evolved in response to changes in these variables over time.

48. The results from this analysis, suggest that the rate of urbanisation, on average, appears to account for the overwhelming proportion of the variation in CRBP cluster patterns, while income growth plays a minor, but still significant, role in explaining cluster formation. On the other hand, growth rates in the imports of CRBP foodstuffs provided little explanatory power.

49. An important caveat is that the findings relate principally to the average country response to changes in urbanisation and income. Therefore, a more in-depth analysis at the country level may bring forth other factors that have determined CRBP patterns.

50. To this end, tests for the stability of preferences16 were conducted for several countries in Sub-Saharan Africa. Urbanisation and income changes, identified as important forces driving change, while positive for the region, have been well below the global average.

51. Using annual real retail prices and corresponding quantities for the major CRBP foodstuffs in each country, the results from the tests showed that consumer preferences in the selected countries have (permanently) shifted away from less refined, lower-valued staples (namely millet, sorghum and root crops) to those lying at the other end of the spectrum (rice and wheat).

V. CONCLUSIONS

52. Throughout the world, dietary patterns of cereals, root crops and other starchy staples are constantly evolving. As overall calorie intake has grown over time, the shares of these staples have declined in favour of vegetable oils and animal products. The falling importance of carbohydrate-based commodities in national diets has masked considerable intra-variability among these staples, suggesting that consumers have altered their consumption of dietary staples. Such changes, which have been pervasive and affected all regions, appear to be dominated by a shift away from lower-valued, coarse grains, towards rice and wheat, casting aside the assertion that preferences towards staples are entrenched.

53. A more varied structure of starchy food intake has emerged in the majority of countries, as consumers moved away from a single starchy “staple” to a more diversified starchy food consumption. At the same time, consumption patterns in the different regions appear to converge, especially towards wheat and rice and away from “minor” coarse grains, plantains and roots and tubers.

54. It was found that urbanisation, income and preference shifts have played an important role in determining dietary intake of starchy staples. These three variables, which are seemingly interrelated, are likely to continue shaping consumption patterns in the future. On the contrary, the size of imports was not identified as a major factor influencing them.

55. Several policy implications can be drawn from those findings. First, as demand for the “finest” grain (wheat or rice) products increases in tandem with higher levels of economic development and urbanization, governments in the developing countries will either have to promote production of those grains or to accept increasing reliance on imports of these commodities. Agricultural development policies and trade facilities will have to be adjusted accordingly.

56. In many cases, traditional CRBP, notably millet, sorghum, cassava and sweet potatoes, are produced by small farmers. As domestic consumers move away from those products, the consequences for those producers might be severe, especially when few opportunities for shifting to alternative crops or activities exist. In order to alleviate their plight, governments might adopt a wide array of measures, including the promotion of processing of those agricultural products for non-food uses. In the case of cassava, for instance, there is wide scope for using the root as a raw material for feed, starch or ethanol production.

___________________________

1 However, there are limitations. Firstly, a declining dietary share of a foodstuff does not necessarily imply a fall in the absolute level of consumption. Secondly, the use of calories implies that biases in aggregation may arise, since greater weight will be accorded to those foods that yield the highest level of calories per unit.

2 Here, ‘variation’ in aggregate consumption patterns, is measured by the average of the coefficients of variation over time for each food item.

3 For example, consumption of this food type in developed countries is again the lowest, but the highest in East Africa, a region which ranks second in terms of under-nourishment.

4 That is, each distribution describes the central tendency (or median change), the dispersion, and the symmetry (or biasness) of changes, over a given period. Cumulative frequencies are also plotted which give an idea as to the percentage of countries that have undergone a particular level of shift.

5 The distributions appear alike across all decades, with each being centred negatively and symmetric

6 The distributions for meat, oils and fats, dairy and fruit and vegetables, while similar across decades, are heavily biased towards the right, and are dispersed narrow, relative to CRBP.

7 Indeed, the distributions of changes in rice shares, shown in Figure 3, exhibit a similar pattern to wheat. The frequencies that fall into the negative range, relate mainly to countries in Asia, where wheat has displaced “saturated” levels of rice consumption.

8 See Regmi, et al. (2002) for a survey of this evidence.

9 See Regmi and Dyck (1992) for a review of this research.

10 There is evidence that if economic depression is very severe and long lasting, then urban habitants may return to rural livelihoods. Examples are provided by the Asian financial crisis in the 1990s.

11 Several studies have concluded that greater participation in the urban labour force and the associated increase in the opportunity cost of time have increased the demand for foods that take less time to prepare than traditional staples. For example, Huang and Bouis (1996) found that the demand for rice has fallen significantly in the Chinese Province of Taiwan, chiefly at the expense of wheat. On the other hand Reardon (1993) and Kennedy & Reardon (1994) identified that demand for rice has increased significantly in urban areas of West Africa, as the processing and cooking costs of rice are lower than for the traditional coarse grain cereals. Similarly, the increased value of time appears to be an important factor raising demand for bread in quasi-urban households in Kenya (Kennedy and Reardon, 1994) and urban households in Sri Lanka (Senauer, Sahn, and Alderman 1986).

12 It is noted that the number of countries resorting to price controls is falling.

13 See for example Blandford (1984), Connor (1994), Herrmann and Roeder (1995) and Gil et al. (1995).

14 The idea behind cluster analysis is to sub-divide the sample into homogeneous groupings or clusters, each of which shares as many similarities as possible with each other, while being as ‘different’ to every other group as possible (Galbraith, 1998).

15 Since domestic price data were not readily available, changes in CRBP relative prices were not considered.

16 A nonparametric approach based on the weak form of revealed preference (WARP) theory was used to test for preference shifts. The essence of the test is that if a consumer can afford the same two bundles of goods at different times and does not display consistency in his or her preferences between the bundles at all times, then there is evidence of structural change in preferences, while the absence of any such inconsistency suggests stable preferences. Further details of the methodology and the findings from the analysis are contained in the fuller version of this paper, which again will be made available on request.