Previous Page Table of Contents Next Page

Micro- and macroevidence on the
impact of undernourishment

Sumiter Broca and Kostas Stamoulis


This paper presents arguments in favour of a focus on fighting hunger as a complement to fighting poverty. Hunger is a violation of a basic human right, and also imposes significant economic costs on society. Hence its reduction and ultimate eradication are the most urgent tasks facing national governments, civil society and the international community.

The plan of this paper is as follows. The first section documents the serious economic costs of widespread hunger for individuals and nations. This is followed by a discussion of the relationship between household income growth and nutrition, showing that income growth alone, at the rates observed in the recent past, cannot be expected to remove hunger in a reasonably short period. It follows that direct public interventions are also required. However, rapid income growth is needed, since it is a necessary condition for poverty and hunger alleviation, and would make it possible to finance public action against hunger. In the last section, it is argued that agricultural development, coupled with the development of rural non-farm (RNF) activities, is one of the most effective means of promoting income growth.


It is necessary to define some frequently used terms. The definitions used here correspond to those used in FAO (1999). The term “under-nourishment” is used to describe the status of people whose food intake does not provide enough calories to meet their physiological requirements on a continuing basis.

An alternative approach is to assess nutritional status through the physiological outcomes of poor nutrition. The term “undernutrition” is used here to describe the status of persons whose heights and weights lie below the lower limits of the ranges established for healthy people. It is critical to note that poor anthropometric status is the outcome not only of insufficient food intake but also of sickness spells. Infectious diseases tend to raise nutritional needs and lower the capacity to absorb nutrients. Food intakes that are adequate for a healthy person may be inadequate for someone in poor health, leading to weight loss in adults and children and growth retardation in growing children. Thus anthropometric measures incorporate information about food consumption as well as health inputs.

Among adults a commonly used measure is the body mass index (BMI), defined as the ratio of bodyweight in kg to the square of height in metres. A BMI in the range of 18.5 to 25 is considered to be healthy for adults, as recommended by FAO, WHO and others. BMI can clearly vary over an adult’s lifetime, but physical stature is determined by the time an individual reaches adulthood.

A natural question at this point is whether it makes sense to collect information on both food intake and stature. The answer is yes, because data on heights and weights are expensive to collect and for that reason are not collected on a regular basis. This makes it hard to use them for regular monitoring.[2] Data on dietary energy supply (DES), using FAO’s methodology, do have the merit of being relatively easy to collect and can be used for monitoring purposes. Another reason for collecting both types of information is that they are needed to guide public policy. It is not usually clear simply by looking at a country’s undernutrition record whether this results from inadequate food intakes or from frequent sickness spells, despite calorie intake levels that would otherwise be sufficient. The implications for public policy are quite different in these two cases. In the former case the aim should be to increase food intakes, while in the latter the emphasis should be on public health, sanitation and the provision of clean drinking-water, etc.

What then are the economic costs of undernourishment and undernutrition? First, at the most basic level a person requires adequate nutrition in order to perform labour. If this nutrition is not forthcoming, or if the person lives in an unhealthy environment, the result is poor nutritional status and the person's ability to do sustained work is reduced. Furthermore, if the person is shorter or has a smaller body frame as a result of past nutritional deprivation, he or she may lack the strength to perform physically demanding but also better rewarded tasks. Thus one would expect to find poor nutritional status, as measured by height, associated with lower wages and earnings.

Second, there is evidence that poor nutritional status leaves people more susceptible to illness. Thus a vicious cycle may exist whereby inadequate food intakes combined with frequent sickness spells result in poor nutritional status, which in turn creates an increased susceptibility to illness. Evidence on this point is presented below.

Third, there is a risk of intergenerational transmission of poor nutritional status. For example, women who suffer from poor nutrition are more likely to give birth to underweight babies. These babies thus start out with a nutritional handicap.

Fourth, there is evidence that poor nutrition is associated with poor school performance in children of school age. At its simplest, this is expressed as “a hungry child cannot learn”. This would not necessarily imply any impairment in the child’s cognitive ability, but merely that because of hunger the child is listless or tired and inattentive and cannot participate in learning activities. Unfortunately, it may also be the case that cognitive ability itself is impaired as a result of prolonged and severe malnutrition. In either case, the upshot is that children do poorly at school, thereby damaging their future economic prospects.

Fifth, people who live on the edge of starvation can be expected to follow a policy of safety first with respect to investments. They will avoid taking risks since the consequences for short-term survival of a downward fluctuation in income will be catastrophic. But less risky investments also tend to have lower rewards. Once again, the tendency is for poor nutrition to be associated with lower income.

Finally, there is some evidence that the macroeconomic performance of the whole economy may suffer as a result of the cumulative impact of these effects. It has been shown recently that the overall effect may be to reduce a country’s rate of economic growth.

Nutrition and productivity

What is the mechanism by which poor nutrition affects productivity? Dasgupta (1997, p. 15) explains that a person’s physical work capacity can be measured by his or her maximal oxygen uptake. The higher its value, the greater the capacity of the body to convert energy in the tissues into work. Here is the crux of the matter: clinical tests suggest that the maximal oxygen uptake per unit of muscle cell mass is more or less constant in well-nourished and mildly undernourished people. Since lean body mass is related to muscle cell mass, it follows that a higher BMI implies a higher maximal oxygen uptake and hence greater work capacity. Also, if two people have the same BMI, the taller of the two has more lean body mass, and hence higher maximal oxygen uptake and work capacity. Studies also suggest that maximal oxygen uptake depends on the concentration of haemoglobin in the blood. Since that depends on iron intakes, the connection between iron-deficiency anaemia and low productivity is also explained.

Empirical studies are now available on the impact of poor nutrition on individual productivity and wages in ten developing countries in Asia, Africa and Latin America. The evidence convincingly bears out the hypotheses advanced above. Table 1 summarizes the main features of these studies.

It is useful to distinguish between the effect of current undernutrition, as expressed in BMI, from the crystallized effect of past undernutrition, as expressed in adult height.

As far as the former is concerned, all the studies referred in Table 1 found that increased BMI had a significant impact on output and wages. For example, Croppenstedt and Muller (2000) found that in rural Ethiopia an increase of 1 percent in BMI increased farm output by about 2.3 percent and wages by 2.7 percent. Thomas and Strauss (1997) found that a 1 percent increase in BMI in their sample from urban Brazil was associated with a 2.2 percent increase in wages.

Another possibility is to measure current nutrition through calorie intakes. Here also Strauss (1986) and Thomas and Strauss (1997) reported significant impacts of increased calorie consumption on farm output and wages. For example, the latter study found that an increase of 1 percent in calorie intakes increased wages by about 1.6 percent at calorie intake levels of around 1 700 calories per day, but that this effect ceased to operate after calorie consumption levels reached around 1 950 calories per day.

The role of micronutrient deficiencies in reducing work capacity has also received increased attention lately. Horton (1999) states that “Studies suggest that iron deficiency anaemia is associated with a 17 percent loss of productivity in heavy manual labour, and 5 percent in light blue-collar work (studies cited in Ross and Horton [1998])”.

Summary of studies on the productivity impact of poor nutrition

Study authors


Group studied

Main findings

Croppenstedt and Muller (2000)


Rural households, mainly agricultural

Output and wages rise with BMI and WfH. Adult height has positive impact on wages

Bhargava* (1997)


Rural households, mainly agricultural

BMI, energy intake have positive impact on time spent on heavy activities for men, but not women

Strauss (1986)

Sierra Leone

Rural households, mainly agricultural

Calorie intake has positive impact on productivity

Satyanarayana et al. (1977)


Indian factory workers

WfH is significant determinant of productivity

Deolalikar (1988)


Southern Indian agricultural workers

Significant effect of WfH on farm output and wages

Alderman et al. (1996)


Rural households, mainly agricultural"

Adult height is significant determinant of rural wages

Haddad and Bouis (1991)


Sugar-cane growers

Adult height is significant determinant of rural wage

Strauss and Thomas (1998)

Brazil and United States

Adult male population, Brazil, United States

Adult stature, BMI have positive impact on wages in Brazil. Only stature has positive effect on wages in the United States

Thomas and Strauss (1997)


Urban population sample

BMI, adult height have strong, positive impacts on market wages

Spurr (1990)


Sugar-cane cutters and loaders

Weight, height are significant determinants of productivity

Immink et al. (1984)


Coffee and sugar-cane growers

Adult height has positive impact on productivity

Notes: BMI=body mass index, i.e. weight (kg)/square of height (metre). WfH=weight for height.
"* This study is not directly relevant since it focuses on time allocation decisions. However, it could be argued that ceteris paribus the ability to spend more time on heavy activities enhances one’s productivity and earning capacity in agriculture.

Strauss and Thomas (1998) present a succinct and illuminating review of the impact of adult stature and BMI on productivity through an analysis of two data sets from the United States and Brazil. They found that adult stature is positively correlated with wages in both countries, but the effect is strong in Brazil and weak in the United States. However, stature is also positively correlated with education. The suspicion naturally arises as to whether the seeming effect of stature on wages is simply a reflection of the fact that taller people are also better educated. Since it is widely accepted that better education does lead to higher wages, perhaps that is the underlying cause of the dependence of wages on stature. However, Strauss and Thomas (1998) show that this cannot be the explanation since in Brazil the impact of stature on wages was strong even in those adults who had no education.

Low stature and low BMI are also associated with lower labour force participation - not only do people with lower stature or BMIs earn less, but they are less likely to be in a position to earn wages at all. The probable reason for this is that people with low BMIs and low stature are also more likely to fall sick.

This evidence supports the hypothesis that higher stature and BMI are associated with higher wages because of their impact on maximal oxygencarrying capacity, not because they are proxies for otherwise unobserved qualities that are attractive to employers. Even if one were to argue that stature captures other unobserved investments in human capital in childhood, it is hard to explain the finding that people with no education are likely to earn higher wages if they are taller. Since manual jobs done by uneducated people typically involve heavy labour and do not require much initiative, but rather a willingness to carry out instructions, it is difficult to see what employers could be looking for in a tall person other than sheer physical strength. More evidence for this hypothesis is provided by the finding that the impact of higher stature on wages is weaker in the United States, where mechanization is more prevalent, than in Brazil where mechanization is less prevalent and thus physical strength matters more. Also note that even in Brazil, the better educated, who presumably do more sedentary labour, cannot expect to get higher wages if their BMI is 26 or higher, while in the United States, obesity actually lowers one’s chances of getting a higher wage.

For adults, the arrow of causation seems to point from stature to wages and not the other way round. Stature does not vary over a person’s adult lifetime, so when a correlation between stature and wages is found, it can safely be assumed that a change in wages cannot cause stature to vary, yet variation in stature can cause wages to vary. The implications of these findings are profound. The loss of income to those suffering from undernutrition can be large. Thus it seems that in Brazil people with BMIs of 26 earn wages that are considerably higher than wages earned by those with a BMI of 22 (both BMIs are well within the range of good nutritional status).Furthermore, people with BMIs of 26 are far more likely to find work than people with BMIs of 22.

Nutrition and health

Inadequate consumption of protein and energy as well as deficiencies in key micronutrients such as iodine, vitamin A and iron are key factors in the morbidity and mortality of children and adults. It is estimated that 55 percent of the nearly 12 million deaths each year among under five-year-old children in the developing world are associated with malnutrition (UNICEF, 1998). Similarly, it has been estimated that 45 percent of all deaths in developing economies in 1985 can be attributed to infectious and parasitic diseases such as diarrhoea and malaria, while these diseases account for about 4.5 percent of all deaths in industrial countries (Strauss and Thomas, 1998, p. 767). Based on research on the European past, Fogel (1994) finds that improvements in stature and BMI explained “over 80 percent of the decline in mortality rates in England, France and Sweden between the last quarter of the eighteenth century and the third quarter of the nineteenth”.

Modern evidence from a number of Asian countries is presented in Horton (1999). As many as 2.8 million children and close to 300 000 women die needless deaths every year because of malnutrition in these countries. Also noteworthy is the fact that anaemia is responsible for 20 to 25 percent of maternal deaths in most of these countries. This last observation points to the importance of micronutrient deficiencies in malnutrition. Iron deficiency is also associated with malaria, intestinal parasitic infestations and chronic infections. Chronic iodine deficiency causes goitre in adults and children, besides having an impact on mental health. The importance of subclinical vitamin A deficiency in child mortality has only recently been recognized through meta-analysis of clinical studies (Horton, 1999, p. 249). The relative risk of mortality for a child with subclinical vitamin A deficiency is 1.75 times that for a child who does not suffer from this deficiency. Thus in the Asian region, if about 10 percent of children suffer from subclinical deficiency, a conservative estimate, then 300 000 child deaths could be prevented through a successful vitamin A supplementation programme.

Nutrition and school performance

Considering the importance of nutrition in human development, there is a relative dearth of studies focusing on the role of the different malnutrition aspects on cognitive achievement among children in developing countries. Nevertheless, there is sufficient empirical evidence to indicate that early childhood nutrition plays a key role in cognitive achievement, learning capacity and ultimately household welfare. Available studies have shown that low birth weight, protein energy malnutrition in childhood, childhood iron-deficiency anaemia and iodine deficiency (e.g. being born to a mother with goitre) are all linked to cognitive deficiencies and the effects are more or less irreversible by the time a child is ready to go to school (Horton, 1999, p. 249). Childhood anaemia is associated with a decrease in score of one standard deviation in cognitive tests. Children are most vulnerable to malnutrition in utero and before they are three years of age as growth rates are fastest and they are most dependent on others for care.

Nutrition and macroeconomic performance

Horton (1999) provides a rough measure of the overall economic costs of malnutrition as a percentage of GDP for selected Asian countries. The author presents evidence for India, Pakistan and Viet Nam on the losses of adult productivity as a proportion of GDP resulting from stunting, iodine deficiency and iron deficiency. Estimates are also presented for losses, including childhood cognitive impairment associated with iron deficiency, for Bangladesh, India and Pakistan. When a significant proportion of the population is undernourished, potential rates of GDP growth can be curtailed. Adult productivity losses arising from the combined effect of stunting, iodine deficiency and iron deficiency are equivalent to 2 to 4 percent of GDP every year in these countries. These are very large totals indeed. It should be noted that these estimates were produced under conservative assumptions.

Thus there is evidence that there are considerable losses at the national level from malnutrition and that these losses accumulate over time. A recent FAO study (Arcand, 2001) indicates a strong relationship between economic growth and nutritional factors, as measured either by the prevalence of food inadequacy (PFI) or the dietary energy supply (DES) per capita. The impact of nutrition on economic growth appears to be strong and to operate directly, through the impact of nutrition on labour productivity, and indirectly through improvements in life expectancy.

Based on historical studies of single countries, Fogel shows that improvements in nutrition and health explain half the economic growth in the United Kingdom and France in the eighteenth and nineteenth centuries (Fogel, 1994). Using an accounting approach with concepts from demography, nutrition and health sciences, he stresses the physiological contribution to economic growth over the long term. Reductions in the incidence of infectious diseases, together with changes in diet, clothing and shelter, increased the efficiency with which food energy was converted into work output and translated into higher economic growth.


Not only is undernutrition an unacceptable violation of human rights, it also imposes a heavy economic burden on nations.

Given that the reduction of undernutrition is vital, how can it best be tackled? Can income growth among poor households take care of the problem on its own, or does it need a helping hand in the form of direct public interventions? Smith and Haddad (1999) used data from 63 countries in five regions, covering 88 percent of the developing world’s population over the period from 1970-95, to analyse the determinants of child malnutrition, as measured by the percentage of underweight children under five. One of their principal findings is that growth in per capita national income contributed to half the reduction in child malnutrition over this period.

However, Alderman et al. (2001) show that the WFS target is unlikely to be met without robust income growth, and not through income growth alone. They assert that “a combination of growth and specific nutrition programs will be needed”.

It is reasonable to say that while income growth has a substantial impact on undernutrition, taken alone it will not take care of the problem. The reasons for this are as follows. Nutritional status is the outcome of food intakes as well as health inputs. Therefore the solution to undernutrition is increased intakes of calories and micronutrients or better health and sanitation, safe drinking-water, etc. or both. Private income growth is not guaranteed to improve nutrition for several reasons. First, household income growth does not necessarily lead to increased calorie intakes. Second, some inputs into nutrition are public goods. Better health requires public investments. Third, since private investments in nutrition have a long-term payoff, private capital markets are unlikely to finance this investment if collateral cannot be provided. Fourth, parents are likely to underinvest in nutrition of girls, particularly in those countries in Asia where sons are more highly valued.

Although income growth, certainly at low levels of per capita income, will lead to growth in calorie consumption, the magnitude of this effect is unclear. A vast number of studies have attempted to measure the elasticity of demand for calories,[3] i.e. the percentage increase in calorie consumption associated with a 1 percent increase in income. In a seminal study, Reutlinger and Selowsky (1976) came up with estimates of the income elasticity of demand for calories that ranged from 0.15 to 0.30 for households at the calorie requirement level. Subsequent studies produced elasticity estimates ranging from 1.2 to as low as 0.01.

There are at least two reasons why this range is so large. One is the degree of aggregation. Some degree of aggregation is unavoidable in any household survey. For example, all cereals may be aggregated into one food group. If incomes rise, consumers are likely to substitute within the broad group to foods with a lower or higher calorie content than the average for the group. But the income elasticity of demand for calories is calculated under the assumption that there is no substitution within the food group. Hence it may be biased upwards or downwards depending on the foods to which consumers switch and the degree of aggregation. Another reason is the use of different functional forms and econometric estimation methods.

Some of the early calorie elasticity estimates tended to be on the high side - elasticities of 0.5 were not uncommon. With better recognition of the problems involved in estimating these elasticities, and improvements in survey techniques and econometric estimation techniques, elasticity estimates have generally decreased in size. In recent years, with the important exception of Subramanian and Deaton (1996) who obtained an estimate of about 0.45, most researchers have obtained low to very low elasticities, in the range of 0.01 to 0.15.

Behrman and Deolalikar (1989) provide an explanation for the finding that these elasticities are low. Their hypothesis is that there is a strong demand for more variety in foodstuffs and that this demand manifests itself even at relatively low income levels. This hypothesis was tested on the data set used for the University of Pennsylvania International Comparison Programme (ICP) project, which had data on prices, quantities and purchasing power parity incomes from 34 countries for 1975 and 60 countries for 1980. Nine food groups were covered, i.e. the degree of aggregation was quite high. As food budgets increased from very low levels, there was a very pronounced increase in the demand for food variety (Behrman and Deolalikar, 1989, p. 671). An important implication of this finding is that since the elasticity of substitution is higher among poor households, any increase in food prices will cause the poor to curtail their food consumption more dramatically than the rich. Hence food consumption by the poor will respond strongly to food subsidies that are sharply targeted on them.

Agricultural and rural non-farm growth

Pro-poor income growth is thus a necessary but often insufficient condition to reduce hunger within a reasonable time span. Without direct public measures to alleviate the most pressing and transient problems, income growth will only gradually solve the problem of hunger. But to finance direct public measures, income growth is needed. The question then becomes, under what circumstances is income growth pro-poor? This section attempts to answer the question.

A consensus has now emerged that the structure of growth matters for its impact on poverty and on human development generally. The 1996 UNDP Human Development Report shows clearly that economic growth, as measured by growth in per capita GDP, is associated with better human development. The relationship is quite strong: countries that achieved higher per capita GDP growth rates over the period from 1960 to 1992 also generally achieved higher values of the HDI, restricted to those components that do not rise automatically with income.

The report also shows that for the same growth rate, some countries managed to improve the HDI more than others. For example, why did Egypt not succeed in increasing its HDI strongly despite enjoying fast growth in per capita GDP? Why did other countries with the same per capita GDP growth rate, such as Lesotho, Indonesia and Malaysia, do much better at improving their HDI performance?

The answer lies in the fact that economic growth, reduction in poverty and inequality reduction are all outcomes of the same deeper processes (Srinivasan, 2000, makes this point forcefully). If these are such as to increase the returns to the assets possessed by the poor then economic growth and poverty reduction will be seen to go together. On the other hand if the process favours assets possessed by the wealthy then they will not. Hence the sectoral composition of growth is important; it matters greatly for poverty and hunger alleviation, in which sector overall economic growth originates.

This section argues that economic growth originating in agriculture, when coupled with growth in RNF incomes, is likely to be strongly poverty reducing, provided that it does not occur against a backdrop of extreme inequality in asset ownership, especially of land. Timmer (1997) found that in countries with highly skewed income distribution, growth reaches the poor with difficulty, whether it originates from increases in agricultural or nonagricultural productivity. According to some estimates, high-inequality countries would need twice as much growth as low-inequality countries to achieve the same reduction in poverty levels (Hammer, Healey and Naschold, 2000).

Why does economic growth originating in the agricultural sector matter for poverty reduction? The majority of the world’s poor still live in rural areas and depend crucially on agriculture for their livelihoods. Hence an increase in agricultural productivity should raise incomes in agriculture. This alone will not necessarily help the poor. The next step is to ask where the extra income is spent. There is some evidence that in many countries this income increment is largely spent on goods such as the services of merchants, artisans, mechanics, etc. and on simple agricultural and household goods. The defining characteristic of most of these goods and services is that they are effectively non-tradable. Furthermore, these commodities generally require low inputs of capital and skills to supply and are ideally suited to the capabilities of the rural poor. But because they are effectively non-tradable, their growth is constrained by the growth of demand in the local rural market, which is stagnant. Hence, if this barrier could be removed, these commodities could grow and help the poor escape poverty. That is precisely what the extra income from agricultural growth does: it creates demand for these locally non-tradable goods, provided this extra income is not hoarded or spent outside but is spent locally, which is more likely in a society of smallholders than in one of large landlords. If all goes well, a virtuous cycle could be created, with agricultural and RNF income growing and helping each other to grow. Four important pieces of evidence are needed to validate this hypothesis. First, incremental budget shares of non-tradables out of agricultural income have to be large; second, income from non-tradables should be important for the poor; third, poverty reduction should follow agricultural growth with a lag; and finally, high initial inequality will short-circuit this process.

The argument presented above will now be discussed in more detail. The majority of the world’s poor still live in rural areas and depend crucially on agriculture for their livelihoods (IFAD, 2001). It seems probable then that raising the profitability of agriculture will be helpful to the poor. This involves taking steps to increase agricultural productivity per hectare, or encouraging a switch to higher-valued crops.

The initial impact of increased profitability in farming is to raise the incomes of those who own land. This in itself may help reduce poverty if the poor also own some land and participate in the productivity increase, but obviously not if the very poor do not generally own land. But there may also be an increase in demand for labour because agriculture itself, and the construction of the infrastructure needed to support agricultural development, are both very labour intensive. Those who were earlier unemployed may thereby find work while those already employed may find themselves working more hours. Either way their incomes go up. However, for poverty reduction, it is not the initial rise in incomes that matters. What does matter is what incomes are spent on.

It is well known that, as incomes rise, the proportion spent on staples declines, while the proportion spent on superior foods such as superior grains, vegetables, fruit, milk, meat, etc. increases. These commodities are effectively nontradable and more likely to be purchased locally, because they are bulky or perishable. At the same time the proportion of income spent on the services of merchants, artisans, mechanics, etc. is likely to go up. This is partly because agricultural growth creates a demand for agricultural implements, but also because rural consumers start to demand goods such as bicycles, which need repairs, or start to eat outside the home so creating a demand for food stalls. Services are by definition also non-tradable. Finally, there is a third category of effectively non-tradable goods comprising simple agricultural inputs such as hoes, rakes, spades and so forth, which may be bought and sold locally, but which do not have much of a market outside rural areas.

The combined effect of these patterns of rural spending can be large. Using household consumption data from 1980s surveys in Burkina Faso, the Niger, Senegal and Zambia (with additional data from Zimbabwe), Delgado, Hopkins and Kelly (1998) show that the share of additional income spent on non-tradables ranges from 32 percent in Senegal to 67 percent in Burkina Faso and Zambia. This spending had multiplier effects that were also calculated. The combined impact on household incomes turns out to be surprisingly large. For example, in Burkina Faso, a US$1 increase in income from farm tradables led to an increase of US$1.88 in income from non-tradables, while in Zambia a US$1 increase led to an increase of US$1.57 in income from non-tradables.

The next step is to show that income from non-tradables looms large in the incomes of the poor. Since livestock can easily be raised at little cost on smallholdings, small animals such as sheep and goats are often kept by the poor, and livestock income is generally of more importance to them[4] (see Adams and He, 1995, for an example). Services such as running a food stall or setting up a simple repair shop do not require much in the way of either skills or capital. Neither does the manufacture of simple agricultural implements. Hence it is precisely in the provision of these goods and services that the poor can find gainful employment and thus raise their incomes. The labour required to supply these goods and services does not have to be withdrawn from some other activity, since there is often a great deal of unemployment, disguised or open in the rural areas of developing countries. Tables 2 to 4 provide evidence as to the importance of RNF incomes to the poor.

From Table 2 it is clear that the mean share of non-farm income in household income is nowhere less than about 30 percent and is as high as 45 percent in eastern/southern Africa. Shares in employment are equally high, ranging from 25 percent in Latin America to almost 45 percent in parts of Asia.

Non-farm shares in total rural income and employment

Regions and subregions





per capita
GNP 1995
(US$) (2)

Mean (%) (1)

Coefficient of variation

Mean (%) (1)

Coefficient of variation

Sub-Saharan Africa






Eastern/southern Africa






West Africa












East Asia






South Asia






Latin America






(1) Mean refers to the mean over the case studies considered for each region and subregion.
(2) Average per capita GNP is calculated as the simple average over the countries covered by the case studies and is based on estimates from the World Bank, World Development Report 1997.
Note: The income shares represent the share of non-farm income in total income of households that are mainly farm households (and the rural landless). The employment shares represent the share of households in the rural population (both in rural areas and small rural towns) with non-farm activity as the primary occupation.
Source: Special chapter on rural non-farm activities in FAO (1998).

Income sources in rural India by expenditure quintile, 1994

Per capita consumption expenditure quintile








Total farm income







Total off-farm income














Agricultural wages







Non-agricultural wages














Regular employment







Other income







Source: Lanjouw and Shariff (2001).
Note: Subtotals do not necessarily add up to totals because of rounding.

Sources of income in the Mexican ejido sector by farm size, 1997

Number of households

Farm size in rainfed equivalent hectares













Shares in total income


Total farm income







Total off-farm income














Agricultural wages







Non-agricultural wages





















Other income







Source: de Janvry and Sadoulet (2000).

Table 3 provides evidence that in one large poor country, India, the share of income from non-farm sources is highest, at 60 percent, for households in the bottom expenditure quintile. It declines as income increases, down to only 33 percent for the richest quintile. A broadly similar trend holds in Mexico, as Table 4 shows, although households are classified by the amount of land farmed, rather than expenditure as in the case of India. To sum up, non-farm income is important to all rural households, but is particularly important to poor rural households.

Hence, for the poor, the RNF sector offers a relatively easy escape route from poverty. But anyone who thinks of supplying these goods and services runs into a demand bottleneck. Because they are effectively non-tradable in most circumstances, they can only be sold locally. There is not much local demand for them in a stagnant rural economy and, until the economy is created, there is no point in expanding output. But if agricultural productivity and hence the incomes of those who own land can be increased - and if they spend this extra income on goods and services provided by the RNF sector - then the bottleneck to the RNF sector’s expansion can be cleared and it can grow and provide important benefits for the poor. Even landless agricultural labourers and others not directly employed in this sector benefit since their power to bargain for higher wages goes up if alternative sources of employment are available.

A critical implication is that the impact on poverty occurs with a lag. Growth in agricultural income will not initially reduce poverty and may not at first have any impact even on the wages of unskilled agricultural labour. It is only later, after incomes have been generated in the RNF sector, that poverty should begin to decline. Once it does, however, it should decline very quickly.

Good econometric evidence of a positive relationship between agricultural growth and poverty alleviation is available from India, which has had a long period of sustained agricultural growth starting from the early 1970s. The most detailed study is by Datt and Ravallion (1998), who relate differences in poverty reduction to differences in agricultural growth rates for different Indian states. Since macroeconomic, trade, sectoral and social policies apply to the whole of India and so are all held fixed, the “pure” effect of agricultural growth on poverty reduction can be isolated.

The main point of the Datt and Ravallion (1998) paper is the following. From the early 1970s, when growth in agricultural yields in India became strong, poverty as measured by the squared poverty gap index began to decline. The squared poverty gap index does not merely count the number of people whose incomes are below the poverty line. It also measures how far below the poverty line their incomes are, and gives progressively higher weights to incomes the further they are below the poverty line. Not only did the number of people in poverty decline, as measured by the headcount index, but poverty also became less severe, i.e. the consumption of the poorest of the poor also increased. The claim that agricultural yield growth bypassed the poorest cannot be supported on the basis of this finding.

What were the channels through which agricultural growth helped the poor? An important finding is that rural wages increased, but with a lag: “Higher average farm-yields benefited poor people both directly and via higher real wages... The benefits to the poor from agricultural growth were not confined to those near the poverty line” (Datt and Ravallion, 1998).

The fact that wages do respond to agricultural growth but with a lag is a critical piece of evidence, showing that time is required for the RNF sector to grow after the initial impetus from growth in agriculture. When the RNF grows, the demand for labour goes up. Agricultural workers find that their bargaining power has gone up and they can start demanding higher wages. Therefore agricultural growth should cause wages to go up, but with a lag. Thus far the discussion has concentrated on how the process works if everything goes well. Under what conditions would the process not benefit the poor?

Agricultural growth puts money initially into the hands of those who own land. Its impact on poverty depends on whether this income is spent on goods and services that are supplied locally, or on imports. The poor will not benefit if it is not spent locally, on goods and services provided by the RNF sector. But this is what may happen when there are marked inequalities in landownership and the initial increase in agricultural income is concentrated in a few hands. Wealthy landowners may have metropolitan tastes and the wealth to indulge them, and will be unlikely to patronize local suppliers. There may be gains to the poor arising out of extra agricultural employment and possibly lower food prices, but the add-on effect on local employment and industry arising out of expenditure by farmers on locally made products will be lacking.

Bautista's (1995) case study in the Philippines illuminates these issues. He points out that over the period 1965-80, crop production in the Philippines grew at a rate of 5.2 percent p.a. and livestock at a rate of 6.4 percent, among the highest growth rates in Asia. The growth of crop production was evenly shared between rice and non-traditional export crops. These high growth rates were at least partly a result of a sevenfold increase in real government expenditure on agriculture, the bulk of which was devoted to irrigation that took half of all agricultural investment by 1980. This was at the cost of investments in rural roads whose share dropped to barely 2 percent of agricultural expenditure. At the same time, human development was exceptionally good in the Philippines, with rates of literacy, infant mortality and life expectancy all either better or comparable with its neighbours in Southeast Asia. Despite all this, there was no significant reduction in poverty.

The primary reason was that the income gains from agricultural growth were highly concentrated. First, where rice farmers were concerned, only those who had access to irrigation could benefit. Despite all the investment in irrigation, only 18 percent of arable land was irrigated by 1980. Second, subsidies on credit and fertilizers were pocketed by large farmers who also enjoyed better access to infrastructure. Large farmers also enjoyed implicit subsidies - through low tariffs, an overvalued exchange rate and a low interest rate - on imported farm machinery that displaced landless agricultural labourers. The consequences were clear: “Income gains were concentrated in the already more affluent segment of the rural population. As a result, rural consumption favoured capital-intensive products and imported goods rather than labour-intensive, locally produced goods... Accordingly the rate of labour absorption in both agriculture and industry was very low, and given the rapid expansion of the labour force, it prevented real wage rates from moving upward. As a result the incidence of poverty increased over the period” (Bautista, 1995, p. 144).

A similar situation can arise in countries where governments place legal and administrative hurdles in the path of smallholders who wish to grow commercial crops. The cultivation of these crops is then in the hands of wealthy farmers who are likely to spend any increments in their income on imported goods while spending little on locally produced goods. Allowing small farmers to share in the profits from commercial crop cultivation would have increased the likelihood of these profits being spent locally, thus creating income-earning opportunities for others. But if this is not the government's policy, the result is that the agriculture sector registers growth but this growth has little or no impact on poverty.

Thus, agricultural growth provides opportunities for the poor to increase their incomes. Whether the poor can seize these opportunities depends on their education and health, on their access to credit and savings services, and on whether they are excluded by social custom or government fiat from income-earning activities (such as women shut out from credit markets). Measures to increase the capital available to the poor (human, financial, physical, natural and social) are therefore likely to pay big dividends in terms of their ability to lift themselves out of poverty.

To conclude, the key point is that growth in agricultural incomes, by creating demand for the output of the RNF non-tradable sector, makes it possible for that sector to grow. Since the capital and skill requirements of the sector are well suited to the capabilities of the poor, its rapid growth can help eliminate poverty. Thus agricultural growth ultimately reduces poverty and does so with a lag. But this benign process cannot work if there are marked initial inequalities in the agricultural sector since these act to prevent agricultural incomes from being spent locally and therefore do not create the multipliers needed.


Adams, R. & He, J. 1995. Sources of income inequality and poverty in rural Pakistan. IFPRI Research Report No. 102. Washington, DC, IFPRI.

Alderman, H., Appleton, S, Haddad, L., Song, L. & Yohannes, Y. 2001. Reducing child malnutrition: how far does income growth take us? World Bank. (processed)

Alderman, H., Behrman, J., Ross, D. & Sabot, R. 1996. The returns to endogenous human capital in Pakistan’s rural wage labour market. Oxford Bull. Economics and Statistics, 58(1): 30-55.

Arcand, J. 2001. Undernourishment and economic growth: the efficiency cost of hunger. FAO Economic and Social Development Paper No. 147. Rome, FAO.

Bautista, R. 1995. Rapid agricultural growth is not enough: the Philippines, 1965-1980. In J. Mellor, ed. Agriculture on the road to industrialization. Baltimore and London, Johns Hopkins Press.

Behrman, J. & Deolalikar, A. 1989. Is variety the spice of life? Implications for calorie intake. Rev. Economics and Statistics, 71(4): 666-672.

Bhargava, A. 1997. Nutritional status and the allocation of time in Rwandese households. J. Econometrics, 77: 277-295.

Bouis, H. 1994. The effect of income on demand for food in poor countries: are our databases giving us reliable estimates? J. Development Economics, 44(1): 199-226.

Croppenstedt, A. & Muller, C. 2000. The impact of farmers' health and nutritional status on their productivity and efficiency: evidence from Ethiopia. Economic Development and Cultural Change, 48 (3): 475-502.

Dasgupta, P. 1997. Nutritional status, the capacity for work and poverty traps. J. Econometrics, 77: 5-37.

Datt, G. & Ravallion, M. 1998. Farm productivity and rural poverty in India. J. Development Studies, 34: 62-85.

De Janvry, A. & Sadoulet, E. 2000. Income strategies among rural households in Mexico: the role of off-farm activities. University of California at Berkeley and the World Bank. (processed)

Delgado, C., Hopkins, J. & Kelly, V. 1998. Agricultural growth linkages in sub-Saharan Africa. IFPRI Research Report No. 107. Washington, DC, IFPRI.

Deolalikar, A. 1988. Nutrition and labour productivity in agriculture: estimates for rural south India. Rev. Economics and Statistics, 70: 406-413.

FAO. 1998. The State of Food and Agriculture 1998. Rome.

FAO. 1999. The State of Food Insecurity in the World 1999. Rome.

Fogel, R. 1994. Economic growth, population theory and physiology: the bearing of long-term processes on the making of economic policy. American Economic Rev., 84(3): 369-95.

Haddad, L. & Bouis, H. 1991. The impact of nutritional status on agricultural productivity: wage evidence from the Philippines. Oxford Bull. Economics and Statistics, 53(1): 45-68.

Hammer L., Healey J. &. Naschold, F. 2000. Will growth halve global poverty by 2015? 8 July, ODI Poverty Briefing.

Horton, S. 1999. Opportunities for investments in nutrition in low-income Asia. Asian Development Rev., 17(1,2): 246-273.

IFAD. 2001. Rural poverty report 2001 - the challenge of ending rural poverty. Oxford, UK, Oxford University Press.

Immink, M., Viteri, F., Flores, R. &. Torun, B. 1984. Microeconomic consequences of energy deficiency in rural populations in developing countries. In E. Pollit & P. Amante, eds. Energy intake and activity. New York, Alan R. Liss.

Lanjouw, P. & Shariff, A. 2001. Rural non-farm employment in India: access, incomes and poverty. Paper prepared for the Programme of Research on Human Development of the National Council of Applied Economic Research, sponsored by UNDP, New Delhi, India.

Reutlinger, S., & Selowsky, M. 1976. Malnutrition and poverty: magnitude and policy options. World Bank Occasional Paper No. 23. Baltimore, USA, Johns Hopkins University Press for the World Bank.

Ross, J. & Horton, S. 1998. Economic consequences of iron deficiency. Ottawa, Micronutrient Initiative.

Satyanarayana, K., Naidu, A., Chatterjee B. & Rao, B. 1977. Body size work output. American J. Clinical Nutrition, 30: 322-325.

Smith, L. & Haddad, L. 1999. Explaining child malnutrition in developing countries. A cross-country analysis. FCND Discussion Paper No. 60. Washington DC, IFPRI.

Spurr, G. 1990. The impact of chronic undernourishment on physical work capacity and daily energy expenditure. In G. Harrison & J. Waterlow, eds. Diet and disease in traditional and developing countries. Cambridge, UK, Cambridge University Press.

Srinivasan, T. 2000. Growth, poverty reduction and inequality. Department of Economics, Yale University, USA. (mimeo)

Strauss, J. 1986. Does better nutrition raise farm productivity? J. Political Economy, 94(2): 297-320.

Strauss, J. & Thomas, D. 1995. Human resources: empirical modeling of household and family decisions. In J. Behrman & T. Srinivasan, eds. Handbook of development economics. Vol. 3A. Amsterdam, Elsevier.

Strauss, J. &. Thomas, D. 1998. Health, nutrition, and economic development. J. Economic Literature, 36(2): 766-817.

Subramanian, S. & Deaton, A. 1996. The demand for food and calories. J. Political Economy, 104: 133-162.

Thomas, D. & Strauss, J. 1997. Health and wages: evidence on men and women in urban Brazil. J. Econometrics, 77: 159-185.

Timmer, C. 1997. How well do the poor connect to the growth process? Discussion Paper No. 178. Cambridge, MA, Harvard Institute for International Development (HIID).

UNICEF. 1998. The state of the world’s children 1998: focus on nutrition. Oxford, UK, Oxford University Press.

[1] Kostas Stamoulis and Sumiter Broca are chief and economist, respectively, with the Agricultural and Development Economics Division of the FAO.
[2] On the other hand, since anthropometric data do not change rapidly (except during wars or famines) it is not necessary to collect these data more often than once every five years.
[3] See Bouis (1994) and Strauss and Thomas (1995) for details.
[4] Another consideration is that livestock are often kept on marginal or degraded land that would otherwise not contribute to income.

Previous Page Top of Page Next Page