## Jean Drèze and Amartya Sen

Print publication date: 1991

Print ISBN-13: 9780198286370

Published to Oxford Scholarship Online: January 2008

DOI: 10.1093/acprof:oso/9780198286370.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2020. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 03 August 2020

# 4 Malnutrition and Poverty in Latin America

Chapter:
(p.119) 4 Malnutrition and Poverty in Latin America
Source:
The Political Economy of Hunger: Volume 3: Endemic Hunger
Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780198286370.003.0005

# Abstract and Keywords

This chapter reviews the state of malnutrition and poverty in Latin America, and attempts to quantify the magnitudes of the problem and identify the poor and the malnourished. Growth plays a minor role in eradicating malnutrition within a generation, and there remains a significant role for redistribution. With regards to the targeting of redistribution, socio-economic groups with a high incidence of poverty should be given priority, and the identification of the socio-economic groups in terms of vulnerability to deprivation is crucial. The chapter suggests that the analysis of the patterns of poverty and malnutrition should be carried out at country-specific level.

# 4.1. Introduction and summary

The object of this chapter is to review the state of malnutrition and poverty in Latin America. We will be particularly interested in quantifying the magnitudes of the problem, and in identifying who the poor and malnourished are. The former provides the reason for policy action while the latter provides the basis for policy strategy. Our focus throughout will be on food‐related issues. When we discuss poverty we will be restricting attention to an inability to purchase a minimally nutritious diet, and when we discuss malnutrition we will consider the physical consequences of prolonged dietary deficiency. We will not, therefore, spend much time on the satisfaction of other, non‐food, basic needs, although this is clearly an area of great policy interest too.

A recent study, World Bank (1985), estimates that in 1980 almost 13 per cent of the population of 24 Latin American and Caribbean countries had energy‐deficient diets falling below 90 per cent of the FAO/WHO (1973) norms. This compared with incidences of food poverty of 44 per cent in sub‐Saharan Africa, 50 per cent in South Asia, and 14 per cent in East Asia. According to these figures, therefore, the problem of food poverty in Latin America, while severe, is still not as bad as in most other areas of the developing world. However, as we shall see in section 4.2, there is considerable variation among Latin American countries in terms of food poverty, ranging from lows of 1 per cent to highs of almost 50 per cent. It is the object of section 4.2 to present and evaluate these country‐specific estimates of food poverty in Latin America. We will also consider direct measures of the effects of malnourishment, in terms of the growth of children, for example. There are wide variations in the estimates available, since methodologies and measures differ from study to study. However, some patterns do emerge, among which is the uniformly higher incidence of poverty in rural areas when compared to urban areas. Poverty in Latin America is still very much a rural problem. In this the countries of Latin America are indeed rather like their counterparts in Asia and in Africa.

In section 4.3 we move from a discussion of the magnitudes of the problem of malnutrition and poverty to considering who the poor are. In particular, we are interested in obtaining incidences of poverty and malnutrition across different socio‐economic groups. But caution must be exercised in inferring continent‐wide patterns. At a sufficiently fine level of disaggregation no general inferences can be drawn, important as it is to have such disaggregation for country‐specific analysis. However, at a somewhat more general level of aggregation patterns which are not too dissimilar to those found in South Asia (p.120) are discernible. Within the rural sector, size of landholding is a significant influence on the risk of malnutrition, and agricultural workers and their families are particularly prone to being malnourished or below the food poverty line. Within the urban sector, slum dwellers have higher incidences of food poverty, as do families headed by manual workers. Also, at the national level, large families have a significantly high incidence of poverty, as do children between the ages of 5 and 14. These characterizations of the poor and malnourished prove useful in the policy discussions of section 4.4, to which we now turn.

Section 4.4 takes up two aspects of policy—the efficacy of growth in reducing malnutrition and poverty in Latin America, and the use of targeted redistributive measures. For the growth part of the story we calculate ‘cross‐over times’—the number of years required for the average poor person to cross the poverty line if income grows at the average rate of growth of per capita GNP of the past twenty years. We find that for many Latin American countries these times are inordinately high—in excess of 30 to 40 years—which may indicate the need for purposive redistribution. Distribution‐neutral growth can be seen as one extreme of targeting—the increases in income are not targeted at all. But the other extreme, of 100 per cent targeting where every person below the poverty line gets just enough to bring him up to the poverty line (no more, no less), is not feasible in most LDCs. The question then arises as to whether policy can be directed at broadly defined groups, and whether these groups can be ranked according to targeting priority. A model developed in Kanbur (1986) is used to show that if the object is to minimize the national poverty gap, groups should be ranked according to their incidences of poverty. The empirical findings of section 4.3, on the incidence of poverty and malnutrition by socio‐economic groups, thus turn out to be relevant in designing a well‐targeted policy for alleviating poverty and malnutrition.

Section 4.5 concludes the chapter with some topics for further research. At the level of Latin America as a whole, there is some payoff to reconciling the different estimates of the magnitudes of poverty to be found in different studies. However, in our view the really high payoff is doing country‐specific studies of the pattern of poverty and malnutrition by socio‐economic groupings. At the level of disaggregation where these become useful to national policy makers, they will cease to be useful in drawing continent‐wide inferences on the pattern of poverty. But that is how it has to be, and we would argue for more detailed country‐level research.

# 4.2. Magnitudes of the problem

## (a) Consumption‐based measures

The best‐known and most widely referred to measures of poverty for Latin America are those of Altimir (1982). The work on which this paper is based was (p.121) done in the late 1970s and the data refer to years around 1970. Altimir estimated the incidence of poverty for a number of countries in Latin America based on two poverty lines—one which was constructed as the cost of a diet satisfying minimum nutritional requirements, and another which included non‐food considerations as well. We shall start with a discussion of the methodology underlying the estimates, reproduced in Table 4.1.

Table 4.1  Estimates of the incidence of poverty in Latin America around 1970

% of households below the poverty line

% of households below the food poverty line

Urban

Rural

National

Urban

Rural

National

Argentina

5

19

8

1

1

1

Brazil

35

73

49

15

42

25

Chile

12

25

17

3

11

6

Colombia

38

54

45

14

23

18

Costa Rica

15

30

24

5

7

6

Honduras

40

75

65

15

57

45

Mexico

20

49

34

6

18

12

Peru

28

68

50

8

39

25

Uruguay

10

4

Venezuela

20

36

25

6

19

10

Latin America

26

62

40

10

34

19

Source: Reproduced from Altimir (1982: Table 12).

The food poverty line (or destitution line, as Altimir terms it) is constructed separately for each country. The starting‐point is the minimum energy and protein standards recommended by FAO/WHO (1973), for moderate activity and for specified weights (65 kg for males, 55 kg for females). Based on this, per capita calorie and protein requirements were calculated using sex and age composition of each country ‘around 1970’. The range of per head calorie requirements so calculated is between 2,260 and 2,350 Kcals per day, while that of protein is between 40.2 and 43.3 grams per day. The average requirement is then applied to the intake distribution, but indirectly via the cost of a diet which would supply these requirements in each country. Those house‐holds whose per capita income is below the cost of such a minimum requirements diet are then classified as being in ‘destitution’.

How is the minimum requirements food basket arrived at? Altimir is concerned to take into account the pattern of consumption in a country, arguing that if minimum requirements can be met with a basket that bears no relation to the foods consumed in that country, it would be unreasonable to choose that as a poverty line. Moreover, in order to be sure that we are within striking distance of the consumption pattern of the poor, it is their consumption pattern which has to be chosen as the reference. But how can we know who the (p.122) poor are when it is the object of the exercise to identify the poor? Altimir chooses to focus on the food consumption patterns of ‘the lower income strata, by which is meant ‘the lowest 50% plus one half of the next 30%’ (footnote (a) to Annex C) as revealed by surveys for Brazil (Rio de Janeiro), Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua, and Peru (Lima). One question that arises immediately is why, if this information was available, it was not converted directly into calorie and protein intake and the numbers of those with intake below FAO/WHO requirements read off the intake distribution. Also, it is not clear how the consumption patterns for the lower strata were estimated for countries not covered in Annex C, although Altimir does note that ‘Account was also taken of available nutritional recommendations concerning specific foods in the exceptional case of countries in which such recommendations have been issued and were applicable to present purposes.’

The Altimir method of arriving at minimum food baskets is thus not fully documented, and there appears to be a problem in using low‐income consumption patterns to generate poverty lines that are to be used to cut off low incomes. However, accepting these as given there is next the problem of pricing these baskets. This is done mainly for ‘the capital cities of the countries under consideration’. Adjustments to construct poverty lines in rural areas are largely ad hoc. For example ‘it is reasonable to regard the cost of standard minimum food baskets in rural areas as 25 per cent below the budget estimated for the capital city of the particular country’; and ‘there would therefore seem to be general justification for assuming minimum food budgets for non‐capital urban areas to be five per cent lower than those estimated for the capital city of each country’.

The resulting food poverty lines, converted to 1970 US dollars using purchasing power parities, are shown in Table 4.2. Also shown in Table 4.2 are overall poverty lines, which are obtained by modifying the food poverty lines to take account of non‐food needs. Since our focus is on food we will not comment on these lines, except to note that ‘urban‐area poverty lines [were] taken as corresponding to private consumption budgets amounting to double the corresponding minimum food budget’, and that ‘the norm adopted in estimating rural‐area poverty lines is that expenditure on non‐food items represents only 75 per cent of the value of the corresponding minimum food budgets’. In contrast to this way of estimating poverty lines we have the method of Ahluwalia et al. (1979), who chose as their reference poverty line the 46th percentile of the Indian income distribution in 1975, and applied this to all countries after adjusting for purchasing power parities. These lines, for Latin American countries, are shown in Table 4.3. Before moving on to our main focus, on food poverty, let us dwell for a moment on a comparison of Tables 4.2 and 4.3. As can be seen the poverty lines used by Ahluwalia et al. are on the whole lower than the overall poverty lines used by Altimir. In fact, they are more in keeping with his food poverty lines. Not surprisingly, the incidences of poverty obtained by Ahluwalia et al. (1979) are more comparable to the (p.123) incidences of food poverty obtained by Altimir. But even with this comparison there are some big differences. The Ahluwalia et al. estimate of the incidence of poverty in Argentina exceeds the Altimir estimate by 4 percentage points while their estimate for Brazil is less than the Altimir estimate by a full 10 percentage points. Now the methods and data underlying these estimates differ greatly so it should not be surprising that estimated incidences of poverty differ (for a (p.124) critique of the Ahluwalia methodology, see Anand and Kanbur 1986). What is important to realize is that there are different methodologies, and widely differing estimates can arise (a sensitivity analysis for the specific case of Brazil is to be found in Hicks and Vetter 1983).

Table 4.2  Poverty and food poverty lines: annual per capita budgets, Latin America (1970 $US) Poverty lines Food poverty lines Metro‐politan area Urban average Rural average National average Metro‐politan area Urban average Rural average National average Argentina 319 319 210 296 160 160 120 151 Brazil 215 215 142 177 107 107 81 93 Chile 269 261 176 226 134 131 101 122 Colombia 302 291 199 252 151 146 113 132 Costa Rica 255 248 167 198 127 124 95 106 Ecuador 301 291 198 237 151 146 113 127 Honduras 217 209 142 162 109 105 81 87 Mexico 261 252 171 221 130 126 98 115 Peru 228 222 150 186 114 111 86 98 Uruguay 287 287 188 263 143 143 108 135 Venezuela 326 315 214 287 163 158 122 147 Source: Reproduced from Altimir (1982). Table 4.3 Purchasing power parity: poverty lines of individuals in poverty, Latin America, 1975 Poverty lines ($US)

% in poverty

Argentina

130

5

Brazil

152

15

Chile

141

11

Colombia

165

19

Costa Rica

Honduras

Mexico

128

14

Peru

161

18

Uruguay

Venezuela

121

9

Source: Ahluwalia et al. (1979).

Turning now specifically to food poverty, let us consider Altimir's estimates in Table 4.1. We see here a varied pattern across Latin America, from an incidence of 1 per cent for Argentina to a staggering incidence of 45 per cent in Honduras. For the two largest Latin American countries, the incidences are 25 per cent in Brazil and 12 per cent in Mexico. Comparing the rural and urban sectors, we see that the incidence of food poverty is greater in the rural sector than in the urban sector for every country except for Argentina, where it is equal to the urban incidence of food poverty. This pattern is of course similar to the one found in Asia. However, the much greater degree of urbanization in Latin America means that a greater proportion of the national poor are to be found in urban areas when compared to Asia.

The Altimir estimates of poverty, while widely cited, are not the only ones available. Table 4.4 presents estimates of food poverty for six Central American countries around 1980. In all, CEPAL estimates that in these six countries 13.6 million people, 60.4 per cent of the total population, were in overall poverty, and of these, 8.5 million, or 37.7 per cent of the total population, were in extreme or food poverty. We cannot, of course, directly compare the estimates in Tables 4.1 and 4.4. The time periods differ as do the methodologies. Focusing on Table 4.4, we see that within Central America itself the incidence of food poverty is below 20 per cent only for Costa Rica; for four out of the six countries it exceeds 30 per cent. In keeping with the pattern observed in Table 4.1, the incidence of rural poverty exceeds urban poverty by a large margin. However, as noted earlier, rates of urbanization in Latin America are high by the standards of developing countries in other continents. Central American urbanization rates are lower than for other Latin American countries (p.125) (ranging from 39 per cent for Honduras to 56 per cent in Nicaragua, compared to 72 per cent in Brazil and 84 per cent in Argentina, for example—figures for 1984, World Bank 1986 b), with the result that the majority of the poor do in fact reside in rural areas. In Central America, at any rate, poverty alleviation strategy will have to have a large rural component, and this is taken up later in this chapter.

Table 4.4  Estimates of households below the food poverty and overall poverty lines in Central America, 1980

% of households below the poverty line

% of households below the food poverty line

Urban

Rural

Total

Urban

Rural

Total

Costa Rica

13.6

34.2

24.8

7.4

18.7

13.6

57.6

76.4

68.1

44.5

55.4

50.6

Guatemala

58.1

66.2

63.4

22.8

36.2

31.6

Honduras

43.9

80.2

68.2

30.6

69.7

56.7

Nicaragua

45.6

80.0

61.5

21.6

50.0

34.7

Panama

42.9

67.3

53.9

11.8

38.3

23.7

Source: CEPAL (1983: Table 2).

Following up on the greater degree of urbanization in Latin America when compared to Asia or Africa, it would be interesting to get a more detailed account of the extent of urban poverty in Latin America. The results in Table 4.1 suggest an incidence of urban food poverty in 1970 ranging from 1 per cent for Argentina to 15 per cent in Brazil and Honduras (with Colombia close behind at 14 per cent). In Central America in 1980, Table 4.4 suggests a range of incidence from 7.4 per cent in Costa Rica to 44.5 per cent in El Salvador. However, these are broad aggregates for all urban areas in these countries. An interesting study by Musgrove (1985) focuses on ten cities in Colombia, Chile, Ecuador, Peru, and Venezuela (the cities are: Bogotá, Barranquilla, Cali, Medellín, Santiago, Quito, Guayaquil, Lima, Caracas, and Maracaibo). The study is based on household budget data collected between 1966 and 1969 (see Musgrove 1978). However, its detail makes this a valuable study for us in this section, and in subsequent sections which look more closely at the characteristics of the poor.

Musgrove's (1985) study relies on estimates of the costs of minimum diets in the five countries by Arellano (1975, 1977). Arellano's methodology is similar to that of Altimir (1982). He uses the FAO/WHO guidelines supplemented by recommendations by the Chilean National Health Service (Servicio nacional de salud (Chile) 1974) and calculates minimum costs by using consumption patterns and prices from Salazar‐Carillo (1978). Given the cost of the minimum requirements diet, which can be calculated once the family's age and sex composition is known, the distribution of families or individuals according to food expenditure as a percentage of the minimum cost can be constructed. Musgrove's results are reproduced in our Table 4.5. As can be seen, the incidence of food poverty amongst individuals varies from a low of 22.6 per cent in Barranquilla to a high of 65.7 per cent in Quito. The incidence among families is lower in every city—the figures for Barranquilla and Quito are 18.0 per cent and 56.4 per cent. This is related to the fact that families in food deficit tend to be larger (see Musgrove's Table 5). However, the extent of severe food deprivation in Latin American cities also stands out from Table 4.5. In every city except Caracas and Lima the percentage of individuals with food expenditure below 65 per cent of the minimum exceeds 10 per cent—in Caracas it is 9.9 per cent and in Lima, the best out of the ten cities, it is 4.5 per cent.

Table 4.5  Distributions of population by food expenditure relative to cost of minimum diet for ten cities in Latin America

Food expenditure as a percentage cost of minimum diet

Incidence of absolute poverty

0–64

65–99

100–149

150–499

≥500

Barranquilla

Share of population (families)

7.1

10.9

28.2

50.4

3.4

18.0

Share of population (individuals)

10.2

12.4

30.1

44.4

3.0

22.6

Bogotá

Share of population (families)

11.8

17.0

27.5

39.1

4.6

28.9

Share of population (individuals)

16.0

19.3

27.3

33.8

3.6

35.3

Cali

Share of population (families)

11.1

19.2

22.7

43.2

3.9

30.3

Share of population (individuals)

14.4

22.9

22.1

37.2

3.3

37.3

Caracas

Share of population (families)

6.4

11.6

21.8

57.7

2.4

18.1

Share of population (individuals)

9.9

16.6

25.0

47.1

1.4

26.5

Guayaquil

Share of population (families)

25.8

26.7

23.1

24.0

0.4

52.5

Share of population (individuals)

32.8

29.5

19.5

18.0

0.3

62.3

Lima

Share of population (families)

3.9

16.2

31.7

46.1

2.2

20.1

Share of population (individuals)

4.5

20.3

34.2

39.5

1.1

24.8

Maracaibo

Share of population (families)

11.9

19.3

28.9

39.1

0.8

31.1

Share of population (individuals)

13.5

21.9

31.1

33.3

0.2

35.4

Medellín

Share of population (families)

24.3

21.1

19.7

32.4

2.6

45.4

Share of population (individuals)

32.8

23.0

17.1

25.4

1.6

61.8

Quito

Share of population (families)

31.0

25.5

20.3

23.0

0.3

56.4

Share of population (individuals)

40.2

25.5

18.5

15.7

0.1

65.7

Santiago

Share of population (families)

8.9

16.6

24.8

45.4

4.3

25.5

Share of population (individuals)

13.0

20.9

26.6

37.3

2.1

33.9

Source: Reproduced from Musgrove (1985: Table 3).

There are some interesting discrepancies between the picture in Table 4.5 and that painted by Altimir's figures in Table 4.1. For example, according to Altimir only 3 per cent of urban households were below the poverty line in Chile. However, Musgrove's calculations show that in Santiago 25.5 per cent (p.126) (p.127) of all families had an inadequate food intake, i.e. were below the poverty line. Similarly, Altimir's calculations show that only 6 per cent of urban households in Venezuela were below the poverty line, while Musgrove's figures show that in Caracas 18.1 per cent of families were below the poverty line. These are large differences. How can they be resolved? The obvious possibility is that the two sets of poverty lines are different. However, as Musgrove's (1985) Table 2 demonstrates, Arellano's estimates of poverty lines for these two cities at 1968 purchasing power parity are almost identical to the corresponding estimates by Altimir at 1970 purchasing power parity (for metropolitan areas in Chile and for urban areas as a whole in Venezuela). Could Altimir's broader coverage in urban areas account for the difference? This is possible, given that in Chile the percentage of urban population living in the largest city is 44 per cent while in Venezuela it is 26 per cent. However, given that Altimir only makes a 5 per cent adjustment to the poverty line in going from metropolitan to non‐metropolitan areas the distribution of income in the two groups would have to be markedly different to generate such large differences. In any case, for Venezuela the other city of Maracaibo shows an incidence of food poverty of 31.1 per cent in Musgrove's figures. In the absence of any obvious reconciling factors, we leave these wide discrepancies as a caution to those seeking to arrive at definitive statements of Latin American poverty on the basis of the current literature—the poverty is undoubtedly severe, but there are some differences in the exact quantitative magnitudes of the estimates.

## (b) The effects of malnourishment

The consequences of prolonged deficiencies in dietary intake are manifold. As well as greater susceptibility to disease, malnutrition affects the normal growth pattern of children, and malnutrition in pregnant women has severe implications for the survival and health of the newborn child. The effects on the growth patterns of children have been discussed extensively in the nutritional literature, and the well‐known Gomez (1955) classification is used to distinguish the extent to which children have low weights for age when compared to a reference population. World Bank (1983) provides data on extent of (p.128) malnourishment for Brazil, based on the 1974/5 Estudo nacional da despesa familiar (ENDEF), or National Household Expenditure Study conducted by Fundação Instituto brasileiro de geografio e estatística (FIBGE). These are reproduced in Table 4.6. As can be seen, over 37 per cent of Brazilian children suffer from first‐degree malnutrition (i.e. 76–90 per cent of normal weight for age). The incidence is highest among 5–10‐year‐olds, with over 43 per cent of this age group being in the first‐degree malnourishment category. The lowest incidence of first‐degree malnourishment is amongst infants. The picture is reversed for the most extreme form of malnutrition in the Gómez classification (third degree—60 per cent or less of normal weight for age). The highest incidence of this is amongst infants, at 2.5 per cent. Overall, in Brazil 0.9 per cent of children were suffering from the most severe manifestations of malnutrition. If one took an intermediate position between these two extremes and focused on second‐degree malnutrition (61–75 per cent of normal weight for age), then around 20 per cent of Brazilian children would be classified as malnourished.

Table 4.6  Different degrees of malnourishment in children, Brazil, 1975

Age group

First degree

Second degree

Third degree

Birth–5.99 months

17.4

8.9

2.5

6.00–11.99 months

22.3

11.0

1.4

1.00–1.99 years

32.2

11.1

0.6

2.00–4.99 years

38.1

14.8

0.2

5.00–9.99 years

43.1

25.9

0.4

10.00–14.99 years

36.8

27.5

1.6

15.00–17.99 years

37.1

13.4

0.9

TOTAL

37.2

20.2

0.9

Source: Reproduced from World Bank (1983: vol. ii, Tables 17, 18, and 19).

Data on malnourishment of children disaggregated by age are not, unfortunately, available for the whole of Latin America. However, Table 4.7 presents data on incidence of Grades I, II, and III malnourishment in children under the age of 5 for 26 countries in Latin America and the Caribbean. Comparing across Latin America, we see that Guatemala, Dominican Republic, El Salvador, Panama, Nicaragua, Guyana, Costa Rica, Colombia, Brazil, Belize, and Honduras all had incidence of Grade I malnourishment in the under‐5s running at over 40 per cent. Guatemala had the highest incidence of extreme malnourishment at 5.9 per cent.

Table 4.7  Malnutrition in children under 5, Latin America and the Caribbean

Country

Year

Incidence of malnourishment in sample

I

II

III

Antigua

1975

35.5

6.8

0.8

Bahamas

1974

14.6

0.6

0.9

1969

39.0

11.0

1.2

Belize

1973

40.0

18.0

1.2

Bolivia

1966–9

29.0

10.2

0.7

Brazil

1968

48.4

17.2

2.7

Chile

1975

13.7

3.2

0.9

Colombia

1966

45.6

19.3

1.7

Costa Rica

1966

43.7

12.2

1.5

Dominica

1970

19.7

5.1

3.4

Dominican Republic

1969

49.0

23.0

4.0

1965–9

28.9

9.6

1.2

1965

48.5

22.9

3.1

Guatemala

1965

49.0

26.5

5.9

Guyana

1971

43.0

16.0

1.7

Honduras

1966

43.0

27.2

2.3

Jamaica

1970

39.0

9.4

1.4

Montserrat

1971

28.0

3.5

0.0

Nicaragua

1966

41.8

13.2

1.8

Panama

1967

48.8

10.8

1.1

Paraguay

1973

4.9

2.2

0.7

Peru

1965–71

32.8

10.9

0.8

St Kitts‐Nevis and Anguilla

1974

33.3

5.4

0.1

St Lucia

1974

33.0

9.0

1.9

St Vincent

1967

47.0

14.0

1.5

Venezuela

1974

35.3

12.2

1.4

Source: Daza (1979), compiled from Pan American Health Organization data.

More recent cross‐sectional information on the effects of malnourishment among children in Latin America is available from UNICEF (1986), reproduced in Table 4.8. This information is not directly comparable to that in Table 4.7, since it is restricted to children between the ages of 12 and 23 months, and defines the incidence of acute malnourishment not according to (p.129) the Gomez classification but as ‘the percentage of children with greater than minus two standard deviations from the 50th percentile of the weight‐for‐height reference population, i.e. roughly less than 77% of the median weight‐for‐height of the United States National Center for Health Statistics reference population’. However, the information once again raises the question of the appropriate standard of comparison since not only are the absolute magnitudes of incidence very different (acute malnourishment in Brazil, for example, is reported as being less than 2 per cent on the Gomez classification in Table 4.6 while it is 6 per cent according to the UNICEF standards) but the relative rankings of countries can change too.

Table 4.8  Incidence of acute malnourishment among children aged 12–23 months, based on weight for height comparisons, Latin America and the Caribbean, 1980

Country

Incidence (%)

Bolivia

1

Brazil

6

Chile

11

Colombia

10

Dominican Republic

4

Jamaica

14

Panama

8

Source: Compiled from UNICEF (1986).

(p.130)

So far we have focused on the effects of malnutrition on the growth of children in Latin America. However, malnutrition can have other effects too. It can, for example, influence cognitive development (see Berg 1981). Pollitt (1980) surveys some of the relevant evidence for Latin America. Citing the work of Brockman and Ricciuti (1971), and Pollitt (1974) in Peru, and Monckeberg (1968) in Chile, on nutritionally rehabilitated marasmic children, he argues that their performance on intelligence and developmental tests was significantly below the average. But for children with kwashiorkor the work of Barreda‐Moncada (1963), Birch et al. (1971), and Cravioto and Robles (1965), in Venezuela, Jamaica, and Mexico respectively, did not show up any permanent effects.

The most extreme form of relationship between malnourishment and its consequences is the role of malnourishment in infant mortality. Table 4.9 presents World Bank data on infant mortality rates for Latin American countries. As they stand the data show remarkable declines in infant mortality rates and child death rates in Latin America over the past 20 years. However, the death rates are still high compared to industrial market economies (9 per 1,000 live births in 1984 for the infant mortality rate, almost zero for the child death rate) or some other developing economies in Asia (for Korea the two figures were 28 and 2 in 1984; for Sri Lanka they were 37 and 2 in the same year).

Table 4.9  Infant mortality rates and child death rates (age 1–4), Latin America, 1965–1984

Country

Infant mortality rate

Child death rate

(aged under 1)

(aged 1–4)

(per 1,000 live births)

(per 1,000 children in the age group)

1965

1984

1965

1984

Argentina

59

34

4

1

Bolivia

161

118

37

20

Brazil

104

68

14

6

Chile

110

22

14

1

Colombia

99

48

8

3

Costa Rica

72

19

8

113

67

22

5

120

66

20

5

Guatemala

114

66

16

5

Honduras

131

77

24

7

Mexico

84

51

9

3

Nicaragua

123

70

24

6

Panama

59

25

4

1

Paraguay

74

44

7

2

Peru

131

95

24

11

Uruguay

47

29

3

1

Venezuela

67

38

6

2

Source: World Bank (1986a, Table 27).

Of course, the figures in Table 4.9 establish only the broadest of trends and comparisons; they cannot, in particular, shed light on the role of malnourishment in child mortality. For this more detailed investigations are needed, and Table 4.10 summarizes the results of a number of studies from Puffer and Serrano (1973). They show the uniformly high percentage of total deaths of children under the age of 5 that can be attributed to nutritional deficiency, the lowest figure being 20.2 per cent in San Juan city in Argentina, the largest (p.131) being 46.9 per cent in rural municipios in El Salvador. While open to interpretation because of the exact definition of nutritional deficiency, these figures do suggest that a major requirement in the reduction of child mortality will continue to be the reduction of malnourishment in children.

Table 4.10  Nutritional deficiency and immaturity associated with death among children under 5, selected areas in Latin America and the Caribbean

Countries, project sites, and areas studied

Total deaths

% of total deaths

No.

Rate per 100,000 of population

Nutritional deficiency

Immaturity

Both causes

Argentina

Choco Province

Resistencia

804

2,070

40.0

22.1

62.2

Rural departments

837

2,357

37.6

13.6

51.3

San Juan Province

San Juan (city)

326

1,291

20.2

33.1

53.4

Suburban departments

780

2,194

27.4

30.4

57.8

Rural departments

1,050

2,403

31.1

23.7

54.9

Bolivia

La Paz

4,115

2,660

36.0

11.5

47.6

Viacha

161

4,806

30.4

10.6

41.0

Brazil

Recife

3,635

2,933

46.2

20.2

66.4

Ribeirão Preto

Ribeirão Preto (city)

464

1,088

34.5

35.3

69.8

Franca

434

1,942

36.4

27.6

64.1

Communities

228

1,300

38.2

28.5

66.7

São Paulo

4,312

1,769

30.4

28.4

58.8

Chile

Santiago

2,489

1,298

23.7

31.8

55.5

Comunas

225

1,395

35.6

17.8

53.3

Colombia

Cali

1,627

1,607

36.4

19.7

56.2

Cartagena

1,255

1,459

44.7

20.2

64.9

Medellín

1,348

1,444

42.3

19.7

61.9

2,738

2,636

37.2

17.1

54.3

Rural municipios

1,082

5,049

46.9

7.9

54.8

Mexico

Monterrey

3,953

1,813

36.1

18.3

54.5

Source: Reproduced from Daza (1979: Table 2); data compiled from studies in Puffer and Serrano (1973).

So much for the effects of malnutrition on children. What about adults? Here too we could compare weights with reference groups, but more interesting are other aspects such as work capacity. Viteri (1971) argued, based on a sample of Guatemalan men, that those who are poorly nourished also have lower working capacity, the latter being measured in terms of the earlier appearance of oxygen debt and of decreased tolerance to oxygen debt, compared to a reference group of well‐nourished men. Similarly, he showed that sustained rehabilitation of nutrition in those who had previously been chronically undernourished did in fact increase their work capacity. In a follow‐up study in Guatemala, Flores et al. (1984) studied 58 sugarcane cutters and 56 coffee pickers. Their results are reproduced in Table 4.11, together with the comparative figures for 22 ‘well‐nourished’ men taken from Viteri (1971). (p.132) While these figures, too, have to be treated with care (for one thing the reference group is much younger than the experimental group) they do suggest a relationship between malnutrition (as measured by anthropometric measures such as body weight, height, weight‐height, etc.) and maximal work capacity. The economic consequences of this relationship between malnourishment and (p.133) productivity have been explored by Bliss and Stern (1978) and by Dasgupta and Ray (1986). But there is the further question of how malnutrition among men compares with that among women in Latin America. The evidence on this is scanty, but in a study of agricultural migrant workers in southern Brazil, Desai et al. (1980) found that low weight for height is less common among women than among men, although in weight, triceps skinfold, and arm circumference women were much below the standard on average. The results are summarized in Hamilton et al. (1984), and are reproduced in Table 4.12.

Table 4.11 Age, anthropometric measurements, and physical working capacity of two samples of agricultural workers and well‐nourished men in Guatemala ( ± SE)

Variable

Sugarcane cutters

Coffee pickers

Well‐nourished men

Age (years)

35.1

±1.2

38.1

±1.4

18.8

±0.4

Body weight (kg)

52.6

±0.7

54.9

±0.8

60.8

±1.1

Height (cm)

159.3

±0.7

163.9

±0.9

166

±1.0

Weight–height (kg/cm)

33.0

±0.4

33.4

±0.4

36.6

±0.6

Mid‐upper arm circumference (mm)

251

±2

254

±2

281

±3

Leg circumference (mm)

315

±2

328

±3

352

±3

Triceps skinfold (mm)

4.5

±0.2

4.9

±0.2

8.7

±0.5

Arm muscle circumference (mm)

236

±1.8

238

±1.7

253

±3

Body surface (m2)

1.53

±0.01

1.58

±0.01

1.67

±0.02

Oxygen uptake at 150 beats/min. (ml/min.)

1,761

±38

1,968

±35

Maximal work capacity (ml/min.)

2,341

±58

2,685

±54

2,800

±124

Maximal work capacity per kg body weight (ml/kg/min.)

44.5

±0.9

49.1

±0.8

46.3

±0.8

Source: Reproduced from Flores et al. (1984).

Table 4.12  Anthropometric comparisons: men and women, Brazil

No. in sample

Mean weight as % of standard

Mean triceps skinfolds as % of standard

Mean arm circumference as % of standard

Mean arm muscle circumference as % of standard

Weight for height relative to standard (% in sample)

>90%

81–90%

<80%

Females

85

110

87

97

100

80

14.1

5.9

Males

39

95

40

93

101

66.7

23.1

10.2

Source: Desai et al. (1980), summarized in Hamilton et al. (1984).

To conclude this section, we note that the problem of malnutrition is severe in Latin America. This is so whether malnutrition is measured by inability to purchase a minimally nutritious diet, or whether it is measured by its consequences in children or in adults. However, a note of caution is needed—while the weight of the evidence does suggest an overall severity in the problem, there are discrepancies between different studies. These arise largely because they adopt different methodologies. Altimir (1982) calculates food poverty lines for each country separately, using the prices and composition patterns in that country, while Ahluwalia et al. (1979) simply translate the Indian poverty line to Latin American (and all other) countries. Similarly, while World Bank (1983) uses weight for age to classify malnutrition, (p.134) (p.135) UNICEF (1986) uses weight for height. Thus the rankings of countries within Latin America, as well as comparisons of Latin American countries with other LDCs, may differ from study to study. This is a problem we will have to live with given the literature as it now stands. But it does indicate that care is needed in drawing inferences and that a fruitful direction for further research is to reconcile these differences.

# 4.3. Characteristics of the poor and malnourished

Who are the poor and malnourished in Latin America? An answer to this question is obviously important in terms both of beginning an analysis of the causes of poverty and malnourishment in Latin America, as well as of designing policy to aid the malnourished. The policy aspects of the question will be taken up in the next section. In what follows we concentrate on the characteristics of malnourished and poor households. We are interested primarily in the incidence of poverty and malnourishment in groups defined by geographical and socio‐economic characteristics, since we will use this to derive rankings of groups towards which policy should be targeted.

Let us start with the basic distinction between rural and urban poor which we have already touched on in the previous section. As is clear from Tables 4.1 and 4.4, the incidence of poverty in the rural areas of Latin America is uniformly, and in many cases dramatically, higher than the incidence of poverty in urban areas. However, the extent of urbanization is much higher in Latin America than in comparable LDCs. Thus in Brazil the rate of urbanization was 51 per cent in 1965 and 72 per cent in 1984. In Colombia the figures are 40 per cent and 54 per cent. Hence while it is still true that both in incidence and in absolute numbers poverty is a rural phenomenon in Latin America as a whole, it is much less so than in poor Asian countries, which combine high incidences of rural poverty with low rates of urbanization.

If we were considering policies at the broadest level, designed to raise incomes in rural areas as a whole or urban areas as a whole, then it could be argued that the high incidences of rural poverty indicated a targeting towards policies that benefit rural incomes (see Kanbur 1986). However, most often the policy discussion is at a finer level of disaggregation, and we need to characterize poverty within the urban and the rural sector. Starting with the urban sector, consider Table 4.13, which reports the results of the study of poverty in ten Latin American cities by Musgrove (1985). It will be recalled from the discussion in section 4.2 that poverty is defined in terms of the cost of a minimum diet, so that we are focusing attention on food poverty. Table 4.13 shows incidence of poverty by occupation of family head, and the results are striking though perhaps not surprising. The incidence of poverty is by and large highest among families headed by manual workers. In those cases where this is not true, the leading occupational group in terms of poverty incidence is (p.136) personal services. Office workers have relatively low incidences of poverty. The implications of this structure of incidence for the impact of recession on urban food poverty in Latin America is clear—high unemployment rates will inevitably hit manual workers, and will translate themselves into poverty and malnourishment with maximum effect (see World Bank 1986 b). The counteracting policy is the mirror image of this—actions to protect the incomes of manual workers will be the best targeted strategy from the point of view of alleviating poverty. Another striking result from Musgrove's (1985) study (see his Table 6) is that the incidence of poverty is much higher among large families than among small families. From the point of view of targeting, therefore, measures which condition benefits on number of children in a family are likely to be well targeted (see also Ferber and Musgrove (1983) for more detailed analysis for Bogotá, Medellín, and Lima on characteristics of poor households).

Table 4.13  Percentage of families in absolute poverty (food expenditure less than norm) by city and occupation of family head, Latin America

City

Office workers

Vendors

Drivers

Manual workers

Personal services

All families

Barranquilla

0.0

14.6

19.8

18.9

42.8

18.0

Bogotá

18.3

24.1

34.5

41.3

46.5

28.9

Cali

9.1

25.2

31.8

39.8

58.1

30.3

Caracas

12.9

12.0

16.4

32.1

34.1

18.1

Guayaquil

41.1

51.5

61.1

69.3

60.9

52.5

Lima

15.0

21.6

16.1

31.4

22.6

20.1

Maracaibo

22.6

23.2

24.5

35.1

36.8

31.1

Medellín

31.7

37.2

47.5

66.9

49.9

45.4

Quito

31.4

46.2

59.6

79.4

70.9

56.4

Santiago

11.1

23.0

21.4

36.6

27.5

25.5

Source: Reproduced from Musgrove (1985: Table 11).

A study by Garcia et al. (1981) provides some further information on incidence of poverty and characteristics of the poor in Bogotá and in Cali. While these results are not directly comparable to those of Musgrove since the methodology for identifying households suffering from malnutrition differ—in fact Garcia et al. use two different methods—it is interesting to note that the conclusions on incidence of poverty and family size are unchanged. Table 4.14 presents the results and it is seen that for each of two poverty standards (labelled 0.8R and 1.0R) the incidence of malnutrition increases with family size, and dramatically so. Households with 10 members or more have a better than 60 per cent chance of suffering from malnutrition, which has obvious implications for use of observable indicators in targeting expenditures and in policy design (subject of course to the incentive effects of such targeting).

Table 4.14  Incidence of malnutrition in Bogotá and Cali by family size, 1978, for two requirement levels (%)

Household size

Bogotá

Cali

0.8R

1.0R

% population in category

0.8R

1.0R

% population in category

1

1.8

3.2

1.1

5.4

5.4

1.3

2

3.7

5.7

4.7

4.3

10.3

4.8

3,4

6.9

13.8

28.2

6.3

14.3

25.3

5,6

18.2

30.1

31.2

20.0

31.7

31.9

7–9

33.0

49.0

24.8

36.1

58.9

27.2

10+

50.9

73.4

10.0

41.6

63.0

9.6

TOTAL

21.1

33.1

100.0

22.1

36.4

100.0

Source: Reproduced from Garcia et al. (1981).

Garcia et al. (1981) have some further interesting characterizations of the (p.137) malnourished in urban Colombia. The first concerns the incidence of malnourishment by age. They find that the peak incidence occurs in the age range 5–14 whether they look at Bogotá or Cali, 1973 or 1978, using either one of their two methods for identifying malnourishment, or either one of their two standards for minimal requirements (Garcia et al. 1981: Tables 9a and 9b). This is a very strong result. It matches up with the results on large families, and with the ENDEF results for Brazil (Table 4.6) and confirms the focus that policy must have on children, or at least on households with children.

Turning from the urban sector specifically to the more general national picture, a number of studies exist which give more information on incidence of malnutrition by socio‐economic groupings. Table 4.15 reports a disaggregation of incidence of malnutrition in children under 5 by father's occupation, for Guatemala in 1980. The figures in Table 4.15 present a natural ranking of target groups in Guatemala. Overall they reflect the higher incidence of poverty in the rural sector than in the urban sector that we have already commented upon—the agricultural occupations are mostly the high malnutrition incidence categories. Within agriculture, the role of size of landholding is seen very clearly. For children whose father is a farmer with less than one manzana the chance of having height below the standard norm is well in excess of a half, while the same chance for those whose fathers have 5 or more manzanas is below two‐fifths. Policy towards small holdings is clearly indicated, and here the picture is similar to that in Asia (we have deliberately excluded issues of within family distribution—see Harbert and Scandizzo 1982).

Table 4.15  Percentage of children less than 5 years of age below standard height, by father's occupation, Guatemala, 1980

Father's occupation

% of children less than 5 years of age below standard height

Farmer with less than 1 manzana a

57.1

Farmer with 1–1.9 manzanas a

56.5

Agricultural wage worker (coffee and sugar)

54.0

Farmer with 2–4.9 manzanasa

51.6

Agricultural wage worker (food crops)

51.6

Unpaid family worker

44.8

Unskilled urban wage worker

39.7

Farmer with 5 and more manzanasa

37.7

Handicraft sales

35.8

Skilled urban worker

34.7

Professional or administrative worker

25.8

(a) 1 Manzana = 0.7 ha.

Source: Reproduced from Hintermeister (1984: Table 3.3), compiled from INCAP (1980).

Agricultural wage‐worker households fall somewhere in between small landholdings and larger landholdings in their propensity to contain malnourished children. But differences of 2 or 3 percentage points in incidence should not, perhaps, be significant in determining policy stance given the (p.138) errors inherent in the data. For similar sorts of reasons the fact that incidence of child malnutrition is 2.4 percentage points higher in households where the father is an agricultural wage worker in coffee and sugar than in households where he works in food crops is perhaps not sufficient to dictate a major reorientation of policy towards wages in these two sectors unless further research confirms the differential as being significant and persistent. What is clear is that favourable policies towards agricultural wage workers in general vis‐à‐vis their urban counterparts is indicated. Here the poverty incidences differ by 10 percentage points or more (this pattern is also documented for Mexico, see Bergsman 1980). Thus while within the urban sector policy should be directed towards raising the incomes of unskilled workers, as between urban and rural workers in general the latter should have priority. These implications for targeting are unlikely to be overturned by further careful research (see also Peek and Raabe 1984).

A similar breakdown of incidence of malnutrition by occupational groups is given in Table 4.16 for Costa Rica. The table is based on a Family and Community Questionnaire of the Ministry of Health, 1980, data from which is reported in UNICEF (1981). While not strictly comparable with Table 4.15 because of different criteria used for identifying malnutrition (Table 4.16 uses the Gomez classification, Grades II and III), it should be clear that what we (p.139) have is once again a hierarchy with implications for targeting. There are some interesting anomalies, such as the fact that farmers with less than 1 hectare show a lower incidence of poverty than farmers with 1 to 3 hectares—but this is probably because the former group includes some rich urban households. Leaving this aside, a contrasting pattern to that in Guatemala emerges. In Costa Rica agricultural workers by and large have higher incidences of malnutrition than smallholders, although there is of course some overlap. But on the other hand there are similarities—urban workers are less likely to suffer from malnutrition than rural workers, and (after 1 hectare) big farmers are less likely to suffer from malnutrition than small farmers. This seems to be in keeping with the pattern in Asia.

Table 4.16  Malnutrition by functional groups, Costa Rica

Functional groups

% with malnutrition Grades II and III

AW vegetables and fruits

9.3

AW unknown products

9.3

AW basic grains

9.1

AW coffee

9.0

Unskilled workers and artisans

8.9

AW beef cattle

8.5

F from 1 to 3 ha

8.1

F from 5 to 10 ha

7.8

F from 3 to 5 ha

7.3

AW cattle not specified

6.7

Workers in the transitory informal economy sector

6.6

Pensioners

6.5

AW sugarcane

6.2

Qualified operators and artisans

6.2

All types of management

6.0

Office employees and salesmen

5.8

F less than 1 ha

5.5

F from 25 to 50 ha

5.0

AW banana and African palm

4.6

Service workers

4.4

F 50 or more ha

4.4

AW dairy cattle

3.5

F from 10 to 25 ha

3.4

Professionals and technicians

1.8

TOTAL

6.6

Note: AW = Agricultural workers; F = Farmers.

Source: Reproduced from Pacey and Payne (1985: Table 6.1).

(p.140) Finally in this discussion of the occupational determinants of the incidence of poverty, we consider the results of Valverde et al. (1980). These are based on a broader definition of occupation, grouping together all urban slum households into one category, for example, but they provide a finer breakdown by extent of malnutrition. As can be seen in Table 4.17, so far as first‐degree malnutrition is concerned all four groupings are pretty similar in their incidences. The largest difference is one of 5 percentage points (on an average incidence of more than 50 per cent), between workers living on coffee plantations and sugar plantations. The two features that emerge are, firstly, that urban slum dwellers have similar incidences of malnutrition to poor agricultural workers and, secondly, that workers living on coffee plantations have the highest incidence out of the four groups for first‐degree malnutrition. The implication of the first feature is that if policies can be targeted towards urban slum dwellers then the choice between those and policies which transfer an equal amount of purchasing power to certain types of agricultural workers will not much affect the overall impact on poverty. However, within El Salvador the wages of coffee plantation workers are clearly a priority if the objective is to reduce poverty.

Table 4.17  Nutritional status of children 6–59 months (Gomez classification) for four occupational groups, El Salvador, 1976

Regions

No. of cases

% children

First degree

Second degree

Third degree

Total

Smallholding, semiproletariat farmers

1,447

53.8

21.0

1.7

100.0

Workers living on coffee plantations

1,043

54.3

20.4

2.2

100.0

Workers living on coastal cotton and sugar plantations

1,489

48.8

12.5

1.5

100.0

Urban slums

1,369

51.1

14.0

1.3

100.0

Source: Reproduced from Deere and Diskin (1984: Table 5), based on Valverde et al. (1980).

It has been argued by some that the regional divide is more important as a discriminator of poverty incidences than the rural–urban divide. Thomas (1982) recognizes the importance of the rural poor in Brazil, and yet argues for the importance of regional divisions:

The more interesting outcome of the poverty measures undoubtedly points out that considerably more poverty exists in the nation's rural areas than in urban areas. For instance, a poverty index for the nation's rural areas is approximately 20% higher than (p.141) the national average, while that of the metropolitan areas is roughly 50% below national averages. Compared to such urban–rural differences, however, some regional differences, particularly between the Northeast and the Southeast, are more striking. The sharpest contrast is provided by a comparison of the state of São Paulo with the Northeast region. While a poverty index for São Paulo state is roughly 2½ times below the national average, that for the Northeast is 1⅔ as high as this average.

The figures on which Thomas bases his argument are reproduced in Table 4.18. While using the same ENDEF survey that we have been relying on for information on Brazil in this paper, they are based on poverty lines which are food poverty lines augmented to take account of non‐food needs and are hence not directly comparable to our discussions on malnutrition. However, it is interesting to note that if we take the figures for incidence of malnutrition in children given in Table 4.19, then for the case of Grade III, the national incidence is 4.5 times the incidence in the South‐East, while the incidence in the North‐East is almost 2.5 times the national incidence. Of course, this happens with much smaller percentages, but Thomas's general point is well taken. Regional variations in the incidence of malnutrition in Mexico are detailed in Table 4.20, and similarly large differences are clear (see also Ferrari 1980, for Peru, and Herrick and Hudson 1981, for Costa Rica).

Table 4.18  Regional distribution of poor, Brazil, 1984

Regions

Poverty line (local currency)

% below

No. below (000)

Region I

2,585

16.35

1,419

Metropolitan Rio de Janeiro

2,824

13.88

991

Other urban

2,086

17.78

186

Rural

1,520

31.33

276

Region II

2,663

20.79

4,333

Metropolitan São Paulo

3,374

13.82

1,148

Other urban

2,199

13.28

939

Rural

1,753

20.59

697

Region III

1,885

18.96

3,364

Pôrto Alegre

2,893

14.93

237

Curitiba

2,245

10.54

85

Other urban

2,068

16.73

972

Rural

1,526

18.00

1,607

Region IV

1,631

29.39

3,809

Belo Horizonte

2,282

18.57

340

Other urban

1,775

21.14

1,093

Rural

1,541

40.83

2,431

Region V

1,620

48.01

13,903

2,603

27.55

340

Recife

2,152

31.55

612

Fortaleza

1,879

36.24

388

Other urban

1,795

38.13

3,189

Rural

1,377

54.78

8,967

Region VI

2,896

16.00

108

Region VII

1,730

14.35

526

Belém

2,144

20.75

154

Other urban

1,843

27.63

306

Other urban

1,794

12.22

259

Areas

Metropolitan

2,831

17.38

4,403

Other urban

1,942

22.63

6.944

Rural

1,562

39.36

13,978

National average

1,953

29.36

27,462

Source: Thomas (1982: Table 19).

Table 4.19  Incidence of malnutrition among children below the age of 18 in Brazil, by severity and region (%)

Region

Severity of malnutrition (Gomez classification)

I

II

III

North‐East

38.2

28.0

2.2

South‐East

36.2

15.4

0.2

Frontier

39.0

23.3

0.7

Brazil

37.2

20.2

0.9

Source: World Bank (1983)

Table 4.20  Malnutrition incidence in Mexico, by region (%)

1st degree

2nd degree

3rd degree

North

Sonora Agua Prieta

31.2

5.5

0.0

Sinaloa Concordia

46.8

12.5

0.0

Coahuila Finisterre

52.8

20.8

0.0

Nuevo León San José de Raices

56.0

24.0

0.0

Central west

Hidalgo Almoloya

41.1

12.8

3.1

Xochiocoatlan

51.5

23.9

1.3

El Nith

30.5

43.9

10.4

Puebla San Juan Epatlan

70.5

17.6

5.9

Tlaxcala Santa Cruz Aquiahuac

48.4

14.0

1.7

Gulf zone

Veracruz Paso del Toro

50.0

16.7

0.8

Metropolitan Mexico

Xochinilco Santa Cruz Xochitepec

49.8

9.6

0.4

Tlalpan San Muguel Xicalco

63.4

8.7

1.3

AVERAGE

49.3

17.5

1.1

Source: Austin (1979: Table 5).

What, then, do we know about the characteristics of the poor and the malnourished in Latin America? While it is dangerous to generalize from individual country studies, and we have ourselves raised objections to some of the methodologies pursued, some tentative suggestions might perhaps be entertained. Firstly, the incidence of poverty is significantly higher in rural areas than in urban areas. Secondly, while such differences in incidence indicate policy targeting, the differences in incidence between regions can be much larger. Thirdly, the incidence of malnutrition and poverty among large families is very high and, relatedly, children between 5 and 14 are particularly at risk. Fourthly, among rural farmers we have the expected relationship between size of holding and incidence of malnutrition and poverty. Fifthly, agricultural workers show higher incidences of poverty than their urban counterparts. These features of poverty and malnourishment in Latin America have implications for policy, and we have already touched on some of these. The next section continues the analysis in greater detail.

# 4.4. Aspects of policy: growth and redistribution

The magnitude of absolute poverty is a function of both the mean level of income and its distribution. If malnutrition is the specific focus, then its magnitude is a function of mean nutritional intake as well as the distribution of intake. Policies to alleviate poverty and malnutrition can therefore be thought of as either growth oriented or redistribution oriented. Of course in principle there need be no inherent conflict between the two—for example, at low levels (p.142) (p.143) (p.144) of nutritional intake an improvement in nutrition improves individual productivity and hence the productive capacity of the economy (see Dasgupta and Ray 1990). But in the literature, growth and redistribution are pitted as opposites, with difficult choices having to be made between the two. In what follows, therefore, we will divide our discussion of policy into the impact of growth and the impact of redistribution. In the latter category we will discuss targeting priorities in the Latin American context.

Consider first the effects of growth on poverty. In doing so we assume that growth is distribution neutral. It leaves relative inequality unchanged, i.e. all incomes grow at the same rate. What would be the impact on absolute poverty? If the poverty line is z and the mean income of the poor is $y ― p$, then with an annual growth rate of g it will take:

$Display mathematics$

years for the average poor person to cross the poverty line. This ‘crossover time’ (see Kanbur 1987) gives an indication of the efficacy of growth in alleviating absolute poverty. Using Altimir's (1982) data, Table 4.21 shows the crossover time for a selection of Latin American countries on the basis of their past growth rates in per capita GNP.

Table 4.21  Calculations of crossover times, Latin America

Country

$y ― p / z$

100g (% per year)

T (years)

(1)

(2)

(3)

Argentina

0.745

0.3

98

Brazil

0.538

4.6

14

Chile

0.567

−0.1

Colombia

0.512

3.0

23

Costa Rica

0.564

1.6

36

Honduras

0.491

0.5

143

Mexico

0.608

2.9

17

Peru

0.448

−0.1

$∝$

Venezuela

0.625

0.9

52

Sources: (1) calculated from Altimir (1982: Table 14); (2): average annual growth rates of GNP per capita, 1965–1984, taken from World Bank (1986: Annex Table 1); (3): calculated from (1) and (2) using formula in the text.

Now Altimir's figures for the ratio mean income of the poor to the poverty line are for the overall poverty line and not the food poverty line. If we can assume that this ratio is unchanged when using the food poverty line then the calculations in Table 4.21 are applicable to food poverty alleviation by means of growth as well. Crude as they are, the figures in Table 4.21 do give an indication of the magnitude of the problem in Latin America. For Peru and (p.145) Chile, which have recorded a negative average annual rate of growth over the past 20 years, the crossover time is of course infinite. Obviously these negative growth rates are influenced by the impact of the current recession. But even if this were taken out and a low positive growth rate were substituted, very long crossover times would still result. This is seen for the case of Honduras, where the average annual growth rate of income per head has been only 0.5 per cent. Combining this with the depth of poverty in the country, we get an incredible 143 years for the average poor person to cross the poverty line. Obviously the structure of the data is bound to change over such a long time—the calculation is merely intended to convey the small role that growth at historical rates can play in alleviating poverty. In fact, suppose that the Honduran growth rate of income per head doubled to 1 per cent a year as a result of growth‐oriented policies. The crossover time would then come down to around 70 years—still a very long time horizon for alleviating the poverty of only the average poor person.

However, there are some Latin American countries where, because of a combination of low poverty and/or high growth rates, crossover times are around 15 years. Mexico and Brazil are interesting cases in point. The latter has higher poverty and a higher growth rate, with the result that they have similar crossover times.

There are three major problems with the crossover time calculations. First, the crossover time focuses only on the income of the average poor person and says nothing about the numbers of the poor. There have been some attempt to project these numbers in the future, notably the projections by Ahluwalia et al. (1979) to the year 2000. These rely on the absolute poverty line discussed in section 4.2, and make an adjustment to the distribution of income based on the Kuznets curve estimates of Ahluwalia (1976). By and large these estimates show significant declines in the incidence of poverty in Latin American countries although (1) they are based on growth projections of the last decade and (2) the methodology underlying them has been criticized by Anand and Kanbur (1986). Secondly, the poverty calculations are not directly nutritionally focused. Thirdly, no account is taken of possible changes in the pattern of consumption as income grows. Of course in order to take into account these criticisms we would have to have detailed survey data and conduct careful simulations for particular countries. One such exercise is that conducted by the World Bank (1983) for Brazil (see also Knight and Moran 1981) and we now turn to those results.

Table 4.22 presents the situation as it was in 1975. We have focused attention on those whose calorie deficits exceed 400 calories (based on standards at the low end of the FAO/WHO guidelines). On the basis of ENDEF data, the table shows that in 1975, 17.3 per cent of Brazilians had a deficit in excess of 400 calories. In contrast to our earlier findings on the incidence of food poverty, we see that the incidence of large deficits is greater in urban areas than in rural areas. However, while the incidence is lower in rural areas, the (p.146) (p.147) average deficit of those below the chosen cut‐off is greater in the rural areas. Thus although fewer rural people suffer from the extreme of calorie deficit (both in terms of absolute numbers and in relation to the total rural population), those who are poor are, on average, more severely malnourished. Comparing across regions, we see that as before the South‐East has the lowest incidence of malnutrition. However, while the North‐East and frontier regions are about equal in their incidences, Frontier's incidence is marginally higher, and so is its mean deficit.

Table 4.22  Deficits exceeding 400 calories by region, Brazil, 1975

Total Population (000)

No. in deficit (000)

Incidence (%)

Average deficit (calories/person/day)

Total deficit (m. calories/day)

Total of all deficits (m. calories/day)

North‐East

Rural

17,739.8

2,428.9

13.7

540

1,310.4

3,268.8

Urban

14,291.7

6,952.6

48.7

529

3,678.4

5,151.3

Total

32,031.5

9,381.5

29.3

532

4,988.8

8,420.0

South‐East

Rural

20,046.2

321.7

1.6

476

153.3

1,096.3

Urban

44,524.8

5,471.1

12.3

527

2,882.6

7,592.5

Total

64,571.0

5,792.8

9.0

524

3,035.9

8,688.8

Frontier

Rural

5,268.7

1,763.4

33.5

593

1,087.2

1,717.3

Urban

5,274.0

1,633.4

31.0

617

968.2

1,624,2

Total

10,542.7

3,396.8

32.2

605

2,055.4

3,341.4

Brazil

Rural

43,054.7

4,514.0

10.5

565

2,550.8

6,082.3

Urban

64,090.5

14,057.1

21.9

536

7,529.1

14,367.6

Total

107,145.2

18,571.1

17.3

543

10,079.8

20,450.0

Source: reproduced from World Bank (1983: Table 15).

From this baseline, the World Bank simulated a pessimistic scenario of 4 per cent per year total income growth and an optimistic scenario of 7 per cent per year total income growth over 1980–2000. The growth is assumed to be distribution neutral and population projections are used to convert total growth into per capita income increases. Estimates of the elasticity of per capita calorie consumption with respect to family per capita expenditure are used to convert income growth into calorie consumption increases. The resulting configurations of extreme malnourishment are shown in Table 4.23. As can be seen, there is a dramatic decline in the incidence of extreme malnourishment for the whole of Brazil. Taking the pessimistic scenario the incidence drops from 17.3 per cent to 4.3 per cent, a factor of 4. However, because of population growth the absolute number of extremely malnourished only drops (p.148) from 18.6 million in 1975 to 8.7 million in 2000. Extreme malnourishment is eliminated entirely in the South‐East under the optimistic scenario, and an incidence of only 0.5 per cent persists over the whole of Brazil.

Table 4.23  Projected deficits exceeding 400 calories by region in the year 2000, Brazil

Total Population

Pessimistic

Optimistic

(000)

(000)

% of total

(000)

% of total

North‐East

Rural

24,891.8

Urban

31,634.8

4,720.0

14.9

156.0

0.5

Total

56,526.6

4,720.0

8.4

156.0

0.3

South‐East

Rural

25,165.9

Urban

96,365.9

664.5

0.7

Total

121,531.8

664.5

0.5

Frontier

Rural

10,061.4

1,522.8

15.1

453.4

4.5

Urban

15,100.5

1,761.3

11.7

416.9

2.8

Total

25,161.9

3,284.1

13.1

870.3

3.4

Brazil

Rural

60,119.1

1,522.8

2.5

453.4

0.8

Urban

143,101.2

7,145.8

5.0

572.9

0.4

Total

203,220.3

8,668.6

4.3

1,026.3

0.5

Source: Reproduced from World Bank (1983: Table 32).

The above figures do indicate that growth at historically observed rates can make an appreciable dent in extreme malnutrition. But this is just a reflection of the very low standard chosen. If we look at the consequences for the population with any degree of deficit at all, then the World Bank (1983) calculations show that under the pessimistic scenario the incidence of calorie deficit drops from 67.2 per cent in 1975 to 46.5 per cent in 2000, but because of population growth the absolute numbers increase from 72 million to 94 million. Thus the problem of malnutrition, not of the extreme variety but malnutrition nevertheless, will persist if growth takes place at the assumed rate. Even with the optimistic scenario in the year 2000 there will be over 46 million malnourished Brazilians (22.7 per cent of the total population).

It is with this consideration in mind that the World Bank (1983) analysed the costs of direct distribution of diet supplements to eliminate all calorie deficits, the assumption being of course that the distribution would be targeted 100 per cent efficiently—each individual would get just enough to eliminate his or her own deficit, no more and no less. The cost estimates are thus a lower bound. The estimates obtained, under different assumptions about growth rate, sequencing of implementation, and type of dietary supplements, range between 1 and 2 per cent of GDP in each of the years 1980–2000.

The calculations above indicate once again the importance of targeting. With 100 per cent targeting the burden of eliminating malnutrition is less than 2 per cent of GDP. With untargeted income increases, in the form of distribution‐neutral growth, dietary inadequacy would be the norm for almost 50 million Brazilians by the end of the century. Of course, 100 per cent targeting is very rarely available to policy makers, and we turn now to a discussion of the type of redistributive strategies open to them in a Latin American context.

While 100 per cent targeting is a policy maker's pipe dream, information on the poverty characteristics of socio‐economic and regional groupings can be used to sharpen the extent to which the benefits of expenditure reach the poor and malnourished (see Mateus 1983). The theory underlying such targeting is discussed in Kanbur (1986, 1987). If y is income, f(y) is the frequency density of income, and z is the poverty line in income space, then Foster et al. (1984) have suggested the following poverty index:

$Display mathematics$

where the parameter α‎ reflects the degree of aversion to the depth of poverty. The poverty measured can be the inability to afford the cost of a minimally nutritious diet (as calculated by Altimir 1982, for example). Or, as Kakwani (p.149) (1986) has suggested, if y is calorie intake then P α‎ could be a direct measure of the extent of malnourishment. Two special cases of P α‎ are worth noting. When α‎ = 0,

$Display mathematics$

which is simply the incidence of poverty. When α‎ = 1,

$Display mathematics$

which is simply the incidence of poverty times the average poverty gap as a ratio of the poverty line.

From our point of view the interesting feature of the P α‎ class of measures is that they are additively decomposable across population subgroups. Thus if x 1 is the fraction of population in group 1 and x 2 is the fraction in group 2 (x 1 + x 2 =1), then:

$Display mathematics$

Where P 1,α‎ is the group 1 specific measure and P 2,α‎ is the group 2 specific measure (the extension to many mutually exclusive and exhaustive groups is obvious). Consider now a redistribution which takes an amount Δ‎2 from every member of group 2 (whether rich or poor) and gives Δ‎1 to every member of group 1 (whether rich or poor). Budgetary balance requires:

$Display mathematics$

Should such a redistribution be undertaken? It is shown in Kanbur (1986) that:

$Display mathematics$

which is negative if:

$Display mathematics$

In other words, if the objective is to minimize P α‎ budgetary expenditure should be targeted towards groups with high P α‎−1.

This targeting rule has a particularly simple interpretation when α‎ = 1. Because then P α‎ = P 1 and P α‎−1 = P 0. Thus if the objective is to minimize P 1, groups with high P 0s, i.e. high incidences of poverty, should be the beneficiaries of redistribution. This is natural: if we are concerned to minimize the poverty gap, then we want to maximize the probability that a given (marginal) income transfer will reach a poor person—the extent of his or her own poverty gap is irrelevant. It should now be clear where our characterization of malnutrition and poverty in the previous section comes in. The incidences of poverty in different socio‐economic groupings provide a natural (p.150) ranking for targeting (the identification of ‘high risk’ groups is recommended as a research strategy by Pinstrup‐Andersen 1981). In Latin America, as in Asia, the incidence of poverty in the rural sector is higher than in the urban sector. Redistributive measures which operate at this level of aggregation should therefore he used to raise rural incomes. With more finely tuned instruments, the details of povery within the rural sector and the urban sector can be used to guide redistributive policy. Within the rural sector, helping smallholders and agricultural workers will be a well‐targeted policy. Within the urban sector, manual workers' incomes should be a prime focus of policy. The counterpart to this is the severe effect that recession is bound to have, via manual workers' earnings, on poverty (see ffrench‐Davis and Raczynski 1986 and Lustig 1986). In both rural and urban sectors, raising large families' incomes is an extremely well‐targeted policy for reducing poverty.

Apart from these general conclusions about regional and occupational groupings at the broadest level, once we get into finer classifications differences do begin to emerge between Latin American countries (cf. the contrasting patterns in Costa Rica and Guatemala in comparing incidences of poverty between smallholders and agricultural workers). But this is not surprising. At a certain level of detail the analysis has to become country specific, and moves beyond the scope of this region‐wide survey (for a similar view, see Peek 1984). It is to be hoped, however, that the method of identifying target groups will prove useful in the country analysis as well.

# 4.5. Further research

To end, we would like to pick out two topics which seem to be high on the research agenda for the analysis and alleviation of malnutrition and poverty in Latin America. Firstly, there is the task of reconciling the many different estimates of the magnitude of the problem. All estimates agree that the problem is severe and, furthermore, the patterns of malnourishment and poverty revealed are fairly consistent. But there are discrepancies in numbers of malnourished and poor as estimated by different authors. The discrepancies can sometimes be attributed to differences in the poverty line chosen, but in many cases the underlying methodologies differ and a systematic attempt at evaluating these methodologies is needed (for a start in this direction see Anand and Kanbur 1986).

However, even more important than resolving the discrepancies in estimates of numbers in poverty at the all Latin America level is the analysis of the patterns of poverty and malnutrition at the detailed, country‐specific level. Given the minor role that growth can play in eradicating malnutrition within a generation, there may well be a role for redistribution. But redistribution with 100 per cent targeting towards the poor is not feasible. Rather, we will have to rely on policy instruments which are much coarser, instruments which affect (p.151) the incomes of broad groups of individuals in the same way. But which of several broadly defined groups should be favoured in a targeted strategy of poverty alleviation? The answer depends on how we measure poverty and what instruments we have at our disposal. Following the arguments in Kanbur (1986), we showed in section 4.4 that if our object was to reduce the national poverty gap (or nutrition gap) then, under a particular model of the effects of policy on income (nutrition), socio‐economic groups with a high incidence of poverty (malnutrition) should be high on the list for targeting. If the object of policy is a different index of poverty and malnutrition, or if the effects of policy are modelled in a different way, then of course the targeting rule will be different. Some of these alternative targeting rules are derived in Kanbur (1986), but what is important is to apply them to particular countries.

What is required for such an application is a detailed profile of poverty by policy‐relevant socio‐economic groupings. It is probably true that the level of disaggregation necessary for policy analysis in each country will mean that no continent‐wide inferences can be drawn on the patterns of poverty. But the division of labour we are suggesting is a natural one—the use of broad regional and occupational categories to comment on the pattern in Latin America as a whole, and the use of much more finely disaggregated categories for national policy prescriptions. In our view it is the latter course of research and analysis that will yield the highest payoff.

(p.152) References

Bibliography references:

Ahluwalia, M. S. (1976), ‘Inequality, Poverty and Development’, Journal of Development Economics, 3.

———Carter, Nicholas G., and Chenery, Hollis B. (1979), ‘Growth and Poverty in Developing Countries’, Journal of Development Economics.

Altimir, Oscar (1982), ‘The Extent of Poverty in Latin America’, World Bank Staff Working Paper No. 522 (Washington, DC: World Bank).

Anand, S., and Kanbur, S. M. R. (1986), ‘International Poverty Projections’, mimeo (Helsinki: WIDER).

Arellano, Aquiles (1975), ‘Hacia una canasta de consumo mínimo’, mimeo (Santiago: Universidad de Chile).

———(1977), ‘La pobreza en diez ciudades sud‐americanas’, mimeo (Rio de Janeiro: ECIEL Program).

Austin, J. (1979), ‘The Role of Nutrition in Rural Development: The Mexican Experience’, in Inter‐American Development Bank (1979).

Barreda‐Moncada, G. (1963), Estudios sobre alteraciones del crecimiento y del desarrollo psicológico del sindrome pluricarencial (kwashiorkor) (Caracas: Editora Grafas).

Berg, Alan (1981), ‘Malnourished People: A Policy View’, World Bank Poverty and Basic Need Series (Washington, DC: World Bank).

Bergsman, Joel (1980), ‘Income Distribution and Poverty in Mexico’, World Bank Staff Working Paper No. 395 (Washington, DC: World Bank).

Birch, M. E., Pineiro, C., Adcalde, E., Toca, T., and Cravioto, J. (1971), ‘Relation of Kwashiorkor in Early Childhood and Intelligence at School Age’, Pediatric Research.

Bliss, C., and Stern, N. (1978), ‘Productivity, Wages and Nutrition: The Theory’, Journal of Development Economics.

Brockman, L., and Ricciuti, H. (1971), ‘Severe Protein‐Calorie Malnutrition and Cognitive Development in Infancy and Early Childhood’, Developmental Psychology.

CEPAL.(1983), ‘Satisfacción de las necesidades básicas de la población del istmo centroamericano’, E/CEPAL/MEX/1983/L.32.

Cravioto, J., and Robles, B. (1965), ‘Evolution of Adaptive and Motor Behavior during Rehabilitation from Kwashiorkor’, American Journal of Orthopsychiatry.

Dasgupta, P., and Ray, D. (1990), ‘Adapting to Undernourishment’, in the first volume of this book.

Daza, Carlos H. (1979), ‘Formulation and Implementation of Nutrition Policies: PAHO Strategies for Latin American and the Caribbean’, in Inter‐American Development Bank (1979).

Deere, Carmen Diana, and Diskin, Martin (1984), ‘Rural Poverty in El Salvador; Dimensions, Trends and Causes’, Working Paper WEP 10‐6/WP64 (Geneva: World Employment Programme Research, ILO).

Desai, I., et al. (1980), ‘Food Habits and Nutritional Status of Agricultural Migrant Workers in Southern Brazil’, American Journal of Clinical Nutrition.

FAO/WHO (1973), ‘Energy and Protein Requirements’, WHO Technical Report Series No. 552, Report No. 52 (Geneva: Report of a Joint FAO/WHO Ad Hoc Expert Committee FAO Nutrition Meetings).

(p.153) Ferber, Robert, and Musgrove, Philip (1983), ‘Identifying the Urban Poor: Characteristics of Poverty Households in Bogotá, Medellín, and Lima’, Latin American Research Review.

Ferrari, Marco A. (1980), ‘The Urban Bias of Peruvian Food Policy: Consequences and Alternatives’, Ph.D. thesis, Cornell University.

ffrench‐Davis, Ricardo, and Raczynski, Dagmar (1986), ‘The Impact of Global Recession on Living Standards: Chile’, mimeo (Helsinki: WIDER).

Flores, R., Immink, M. D. C., Torum, B., Diaz, E., and Viteri, F. E. (1984), ‘Functional Consequences of Marginal Malnutrition among Agricultural Workers in Guatemala’, Food and Nutrition Bulletin.

Foster, J., Greer, J., and Thorbecke, E. (1984), ‘On a Class of Decomposable Poverty Measures’, Econometrica.

Garcia, Jorge, Mohan, Rakesh, and Wagner, M. Wilhelm (1981), ‘Measuring Urban Malnutrition and Poverty: A Case Study of Bogotá and Cali, Colombia’, World Bank Staff Working Paper No. 447 (Washington, DC: World Bank).

Gomez, F., et al. (1955), ‘Malnutrition in Infancy and Childhood with Special Reference to Kwashiorkor’, in Levine, S. (ed.), Advances in Pediatrics (New York: Yearbook Publisher).

Hamilton, Sahni, Popkin, Barry, and Spicer, Deborah (1984), Women and Nutrition in Developing Countries (New York: Praeger).

Harbert, Lloyd, and Scandizzo, Pasquale L. (1982), ‘Food Distribution and Nutrition Intervention: The Case of Chile’, World Bank Staff Working Paper No. 512 (Washington, DC: World Bank).

Herrick, Bruce H., and Hudson, Barclay (1981), Urban Poverty and Economic Development: A Case Study of Costa Rica (Boston, Mass.: St Martins Press).

Hicks, James F., and Vetter, David Michael (1983), ‘Identifying the Urban Poor in Brazil’, World Bank Staff Working Paper No. 565 (Washington, DC: World Bank).

Hintermeister, Alberto (1984), ‘Rural Poverty and Export Farming in Guatemala’, Working Paper WEP10‐6/WP71 (Geneva: World Employment Programme Research, ILO).

INCAP (1980), Estudio de regionalización de problemas nutricionales en Guatemala (Guatemala).

Inter‐American Development Bank (1979), Nutrition and Socio‐economic Development of Latin America (Washington, DC).

Kakwani, N. C. (1986), ‘On Measuring Undernutrition’, mimeo (Helsinki: WIDER).

Kanbur, S. M. R. (1986), ‘Budgetary Rules for Poverty Alleviation’, mimeographed (Princeton University).

———(1987), ‘Measurement and Alleviation of Poverty’, IMF Staff Papers.

Knight, Peter T., and Moran, Ricardo (1981), Poverty and Basic Needs: Brazil (Washington, DC: World Bank).

Lustig, Nora (1986), ‘Economic Crisis and Living Standards in Mexico: 1982–1985’, mimeo (Helsinki: WIDER).

Mateus, Abel (1983), ‘Targeting Food Subsidies for the Needy: The Use of Cost‐Benefit Analysis and Institutional Design’, World Bank Staff Working Paper No. 617 (Washington, DC: World Bank).

Monckeberg, F. (1968), ‘Effect of Early Marasmic Malnutrition on Subsequent Physical and Psychological Development’, in Scrimshaw, N. S., and Gordon, J. (eds.), Malnutrition, Learning and Behavior (Cambridge, Mass.: MIT Press).

(p.154) Musgrove, Philip (1978), Consumer Behavior in Latin America (Washington, DC: The Brookings Institution).

———(1985), ‘Food Needs and Absolute Poverty in Urban South America’, Review of Income and Wealth.

Pacey, Arnold, and Payne, Philip (eds.) (1985), Agricultural Development and Nutrition, (London: Hutchinson).

Peek, Peter (1984), ‘Rural Poverty in Central America: Dimensions, Causes and Policy Alternatives’, Working Paper WEP10–6/WP70 (Geneva: World Employment Programme Research, ILO).

———and Raabe, Carlos (1984), ‘Rural Equity in Costa Rica: Myth or Reality?’ Working Paper WEP10–6/WP67 (Geneva: World Employment Programme Research, ILO).

Pinstrup‐Andersen, Per (1981), ‘Nutritional Consequences of Agricultural Projects: Conceptual Relationships and Assessment Approaches’, World Bank Staff Working Paper No. 456 (Washington, DC: World Bank).

Pollitt, E. (1974), Desnutrición, pobreza e inteligencia (Instituto nacional de investigación y desarrollo de la educación).

———(1980), Poverty and Malnutrition in Latin America: Early Childhood Intervention Programs (New York: Praeger).

Puffer, R. D., and Serrano, C. V. (1973), Patterns of Mortality in Childhood, Scientific Publication 267 (Washington, DC: Pan American Health Organization).

Salazar‐Carillo, Jorge (1978), Prices and Purchasing Power Parities in Latin America, 1960–1972 (Organization of American States, ECIEL Program).

Servicio nacional de salud, Chile (1974), Frecuencia y porciones de alimentos según recomendación (Sección nutrición, Sub‐departamento de fomento de salud).

Thomas, Vinod (1982), ‘Difference in Income, Nutrition and Poverty within Brazil’, World Bank Staff Working Paper No. 505 (Washington, DC: World Bank).

UNICEF (1981), Social Statistics Bulletin, 4/4.

———(1986), Statistics on Children in UNICEF Assisted Countries (New York: UNICEF).

Valverde, V., et al. (1980), ‘Life Styles and Nutritional Status of Children from Different Ecological Areas of El Salvador’, Ecology of Food and Nutrition.

Viteri, F. E. (1971), ‘Considerations of the Effect of Nutrition on the Body Composition and Physical Working Capacity of Young Guatemalan Adults’, in Scrimshaw, N. S., and Altschul, A. M. (eds.), Amino Acid Fortification of Protein Foods (Cambridge, Mass.: MIT Press).

World Bank (1983), ‘Brazil: Human Resources Special Report’, vols. i and ii, A World Bank Country Study (Washington, DC: World Bank).

———(1985), ‘Ensuring Food Security in the Developing World: Issues and Options’ (Washington, DC: Agriculture and Rural Development Department, World Bank).

———(1986a), World Development Report (Oxford: Oxford University Press).

———(1986b), ‘Poverty in Latin America: The Impact of Depression’, (Washington, DC: World Bank).