Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Household response to the impact of drought in Kenya

Charlotte Neumann, Richard Trostle, Michael Baksh, Duncan Ngare, and Nimrod Bwibo


The issue of food security in sub-Saharan African countries is of growing concern to international, national, and local policy makers. Problems with food supply range from intermittent local shortages to regional famines.

Kenya is largely representative of other African states with regard to this issue. Agricultural output increased steadily after independence until the mid 1970s. Since then production of the major food crops has not kept up with increased demand because of dramatic population growth rates, climatic variability, and problems with the organizational structures of food production, storage, and distribution [1]. Per capita production in 1980 was 82% of what it was ten years before, and food aid increased from 2,000 metric tons in 1974/75 to 115,000 in 1981/82 [2].

To deal with this growing disparity between production and consumption, the Kenyan Fifth Development Plan defines a two-pronged strategy. It envisions national self-sufficiency through increased production of both staple foods and export crops (to protect the nation's foreign exchange income). Production is to be increased by introducing higher-yield grain varieties, intensifying cultivation on land already being cultivated, and cultivating more land, mainly in the country's drier zones [3].

The plan estimates that 30% of the expected increase in production will come from expansion into lands of low potential. These drier zones experience high evaporation rates and are subject to erratic rain fall, and farmers in the regions will be subject to frequent production shortfalls. Even high-potential agricultural zones suffer dramatic crop failures, resulting in conditions that require outside assistance to maintain adequate food-intake levels. [4]. Those who seek a livelihood from semi-arid lands will be at even greater risk of serious food-intake disruptions.

To understand better how development policies should be formulated, it is important to understand what happens to households during a food crisis. In 1984 Kenya experienced an extensive drought that placed many areas under severe food stress. Production was substantially interrupted, and many households had to turn to alternative sources of food. The nationwide production of maize was 50% below normal, and that of wheat and potatoes was 70% below normal [5]; and food-aid imports climbed to 425,000 metric tons in 1984-85 [6].

This unanticipated and unfortunate event presented to the Nutrition Collaborative Research Support Program (CRSP) in Kenya a unique opportunity to understand the impact of drought on rural households.* Detailed data were collected on food intake, food sources (including market availability), food prices, nutritional status, and health status before, during, and after the drought. These data provided the basis for an analysis of drought impact and household response to reduced food supply. This paper documents the impact of the drought and the strategies employed by households to cope with the situation, and examines how development policies might reduce household vulnerability to future disruptions in the food supply.

The study site

The study was conducted in Embu District in central Kenya, on the southeastern slopes of Mount Kenya, in the country's highlands region, an area with fertile volcanic soil. The study area spanned two agroclimatic zones, a lower zone at an altitude of 3,0004,000 feet and an upper zone at 4,000-5,000 feet. The region is populated by small-holder agriculturalists producing both food and cash crops. The principal food crops are white maize and kidney beans; many also grow English potatoes, sweet potatoes, millet, sorghum, arrowroot, bananas, and cassava as secondary crops. The main cash crop is coffee; a few households cultivate small amounts of tobacco and cotton.

Most Embu households also keep small numbers of cattle, goats, and chickens, and engage in temporary or permanent remunerative labour. Most cash work is in the fields of larger landowners, in local small industries (e.g., carpentry, blacksmithing, tailoring), or in the rural service sector of shops, hotels, and restaurants. Other cashearning opportunities are the production and sale of commodities such as charcoal, milk, wood, and honey.

Several economic institutions organize the production and distribution activities of Embu households. Of paramount importance are the local marketplaces, which regularly serve as places to exchange food and material goods. Government and corporate institutions responsible for the organization and marketing of coffee, cotton, and tobacco also play a significant role. Because most Embu farmers (90% ) grow coffee, the New Kyeni Co-operative Farmer's Society is of extreme importance. This co-operative purchases locally produced coffee and provides credit and technical assistance to all coffee growers.

The study area experiences two rainy seasons a year. The "long rains" fall during March, April, and May, and the "short rains" during October, November, and December, with the heaviest rainfalls occurring in April and November. Much of the agricultural cycle is synchronized with these seasons; maize and beans, for example, are planted at the beginning of a rainy season and harvested three to four months later. Rainfall averaged 1,133 mm per year from 1951 to 1980 for Embu District [7] but varies locally depending on altitude and aspect (exposure to moisture).

Droughts are not uncommon in this region. The probabilities of different levels of drought occurring in any given year, calculated on the basis of observed rainfall over the past 76 years, are shown in table 1 [7]. The designation of the levels of drought rests on the assumption that the minimum rainfall required for dryland agriculture is 300 mm per growing season.

TABLE 1. Drought probabilities for Embu District, Kenya


Level of drought

Mild Moderate Severe
Long rains .36 .22 .08
Short rains .41 .29 .18

The 1984 drought and associated food crisis in Embu motivated the Nutrition CRSP to conduct a food relief effort. Relief assistance was obtained from the World Food Programme, UNICEF, the Netherlands embassy, CARE, and World Vision International and distributed by the CRSP with relatively little delay. The food aid mobilized by the CRSP staff was channeled through local community famine relief committees formed throughout the area by the Office of the President, Government of Kenya. All study households received donated food, as did other households identified by the famine relief committees, the local chief and sub-chiefs, and the project staff as being in need of assistance. The relief effort was conducted from September 1984 through April 1985; the primary objectives were to provide households with emergency subsistence rations and seeds for planting.

Sampling and methods Study sample

The sampling universe for the project comprised the sublocations of Kathanjure, Kathunguri, and Karurumo in Kyeni South Location, Embu District (Eastern Province). * According to the 1979 Kenyan national census, these sublocations included 2,059 households and 11,810 individuals [9]. A total of 247 households participated in the study for a minimum of one year. These households were not selected at random; rather, they were selected on the basis of the presence of a school-age child, toddler, and/or pregnant female head of household.

Comparison of the selected sample to the whole population of the study area (using Kenyan census data) indicates significant differences only in age distribution and household size. The sample population was younger than the overall population, and the size of the average study household (7.3 members) was larger than that reported in the census (5.7 members). These differences in household size are attributed, in part, to our exclusion of single-parent and childless households and those headed by elderly individuals beyond reproductive age.

Data collection

Daily rainfall data for the first six months of the study were obtained from the Embu Agricultural Research Station. Thereafter, the CRSP measured rainfall daily (to the nearest 0.1 mm) at four locations in the study area.

The food intake of each household was measured on two successive days each month for a minimum of one year. Most such measurements were actual weights of food portions served to and consumed by household members; recall was used to calculate all food eaten at night (when field workers were absent) or away from the household compound (e.g., children fed at school). The records also identified each item of food consumed as home grown, purchased, or a gift.

Conversion of food intake to nutrient intake was accomplished with a nutrient data base constructed specifically for Kenyan foods. Because standard food composition tables [10; 11] do not provide the nutritional composition of many Kenyan foods, 45 foods from Embu, including staples, main dishes, and uncommon items, were analysed by a commercial laboratory (Medallion Labs, Minneapolis, Minn., USA).

As a technique to detect intake fluctuations, a monthly household kilocalorie intake ratio (HCIR) was calculated for each household. A daily HCIR was obtained by dividing the total household kilocalorie intake by the total basal metabolic rate (BMR)-the sum of the individual BMRs of the household members determined from age, sex, and weight data and assumptions about activity levels using FAD/WHO/ UNU equations [12]. The monthly HCIR was then arrived at by calculating the mean of the daily HCIRs.

Heights, weights, and other anthropometric measurements were taken monthly for male and female household heads, schoolchildren, toddlers, and infants. The techniques used to perform the measurements were highly standardized, following procedures defined previously [13-15]. All measurements were conducted by a pair of field workers, each of whom measured and recorded data independently. If the differences between readings exceeded pre-set limits, the measurements were taken again.

Household agricultural production data were collected monthly for major food crops (e.g., maize, beans). The amounts of crops that were harvested, stored, sold, purchased, and/or planted every month were determined through interviews with household heads. In addition, surveys of food availability and prices were made periodically in the five local markets. For each market visit, amounts and prices of all available foods were estimated and recorded.

Study households were surveyed to determine changes in food production, distribution, and consumption that occurred as a consequence of the drought-induced food shortage. Recall information was solicited on foods eaten, the average number of meals consumed per day, the relative size of portions, whether wage labour was performed or new businesses started, and whether land, animals, or other possessions were sold in order to buy food.



Comparison of rainfall totals by season for 1984 and 1985 with data from 1951-1980 indicate that the long rains of 1984 were unusually low (table 2). Figure 1 compares the rainfall totals by month with the normal or expected amounts. Not only were the long rains of 1984 only 52% of normal, but 46% of this rain fell during four consecutive days in April. Furthermore, of the 92 days in this season, 73 (79%) saw no rainfall or less than 1 mm. The result was complete failure of the long-rain harvest (June-August) of 1984.

Food intake

Total household food intake declined significantly from April through September 1984; the period of decline from June through September was clearly a result of the poor harvest after the long rains. Average household food intake over this period is shown in figure 2.

The lowest HCIRs were reached during September, October, and November of 1984 (fig. 3), when the ratios were below 1.0 in 30% of all households (meaning that these households were not even fulfilling their basal metabolic needs), compared to 14% for the same three months in 1985.

TABLE 2. Rainfall in Embu District (millimetres)

  Long rains (Mar.-May) Short rains (Oct.-Dec.) Total year
minimum 231 138 676
mean 549 401 1,132
maximum 937 1,280 1,995
1984 283 602 930
1985 682 355 1,147

Source: Ref. 7.

The 1951- 1980 data are for the entire district; the 1984 and 1985 data are only for the study area.

The comparison of the same time period for 1984 and 1085 is intended to control for seasonal variations in household food intake, with 1985 intake taken to represent the more usual conditions. Food crop production in Embu is bimodal, and therefore intake is usually less subject to extreme seasonal fluctuations than is the case in other ecological settings [16]. Seasonal variations of some food items did occur in 1985, but overall consumption levels were relatively stable during this `'normal" year. Indications are, however, that 1985 production levels were not as high as expected due to inadequate rainfall during that year's long rains and to latent effects of the drought (e.g., consumption of maize and bean seeds normally reserved for planting).

In addition to analysing kilocalorie consumption, the monthly intake of specific nutrients was compared to identify changes in diet quality. Consumption of protein, fat, carbohydrates, and iron followed the same pattern as kilocalorie consumption; intakes reached their lowest point during September 1984.

Finally, a survey of study households revealed that 63% ate fewer meals (1.1) per day during the drought than before (2.2). Equally important, 70% of all households reported eating fewer portions per meal during the drought.

Nutritional status

Weight loss of household members occurred almost simultaneously with the onset of the food crisis. Most striking was an increase in the prevalence of severe protein-energy malnutrition among children from 2% to 6%, with obvious clinical cases of kwashiorkor with oedema and wasting.

Adults also showed considerable weight loss during the drought months. The decline in body weight in men and nonpregnant, non-lactating women that occurred during October, November, and December 1984 is illustrated in figure 4. Among school-age children, 65% did not gain any weight and 14% lost weight from August to December. Monthly scores of mean percentage of median weight for age (fig. 5) showed a dramatic decline in weight through November, when the food shortage was at its most severe. There were also increases in the percentage of stunting among schoolchildren as linear growth declined.

The impact of the food crisis on toddlers was less a matter of severe weight loss than one of little or no growth. Weight gain was interrupted, and height increased very little. That toddlers' nutritional status was not worse during the drought suggests that their food intake may have been somewhat protected by the family. Nevertheless, several toddlers required emergency treatment at a local hospital rehabilitation unit.

Fig.1. Comparison of normal and actual monthly rainfall in the study area

Consumption of local food crops

After each harvest farmers attempt to store enough maize and beans to provide for adequate intake until the next harvest. The balance is sold either to the National Cereals and Produce Board (NCPB) depots or in the local markets to provide a source of cash income to meet financial obligations [3].

The usual consumption pattern of white maize corresponds to the timing of the harvests. Thus, consumption is usually the highest during January-March and June-August, and then trails off until the next harvest. Because of the drought, the long-rains harvest of 1984 (June-August) contributed very little home-grown white maize to household consumption (fig. 6). Average daily household consumption of home-grown white maize for July-August in 1984 was 872 g, compared with 2,152 g for the same time period in 1985. By the end of 1984, consumption of white maize had dropped to nearly zero for many households.

FIG. 2. Mean daily household kilocalorie intake, by month, February 1984-November 1985

FIG. 3. Mean daily household kilocalorie intake ratio (HCIR), by month, 1984-1985

The consumption pattern for kidney beans is very similar to that for maize. Consumption following the long-rains harvest of 1984 was less than half that in 1985 (fig. 7). Average daily household consumption of home-grown kidney beans during July and August of 1984 was 358 g, compared with 1,075 g for the same period in 1985.

Finally, the impact of staple food shortage is dramatized by the fact that 93% of all households reported eating foods during the drought that they normally did not eat. The most common alternative items were yellow maize (94%), sukumawiki (Brassica oleracea) (79%), cassava (55%), green pawpaws (55%), kiambuu bananas (52%), wheat flour (46%), sweet potatoes (40%), and arrowroot (Colocasia antiquorum) leaves (40%). In addition, insects and grubs, which are not normally eaten, reportedly were eaten by some people.

Changes in food sources Of critical importance to understanding how households responded to the drought is an examination of changes in their food sources. Analysis of the sources of the twelve foods that constitute the bulk of household intake indicates that 57.7% of the kilocalories derived from this "food basket" in September November 1984 were purchased, compared with 27.4% in 1985 (fig. 8). The 1984 household was able to grow only 12% of the kilocalories in this food basket, while the 1985 household grew nearly 66%. Gift foods (from food relief and other households) were 4.5 times as high in 1984 as in 1985.

FIG. 4. Mean monthly weight of male and non-pregnant, non-lactating female heads of households, January 1984November 1985

FIG. 5. Mean monthly weight of schoolchildren as percentage of median weight for age, January 1984-May 1985

FIG. 6. Average household consumption of white maize by source, 1984-1985 (No data are available on the sources for the first four months of 1984)

FIG. 7. Average household consumption of kidney beans by source, 1984-1985 (No data are available on the sources for the first four months of 1984)

The three foods of greatest importance were white maize, beans, and potatoes. Figure 9 shows the percentage of each of these crops that was purchased, grown, and received as a gift during the months of the food crisis compared with the same period in 1985. For all three foods, a higher percentage was purchased in the 1984 season: 58% versus 24% for white maize, 45% versus 21% for kidney beans, and 67% versus 32% for English potatoes. Consumption of yellow maize, a very undesirable food eaten only during times of hardship, was heavy in 1984 (497 g per house hold per day, of which two-thirds was donated), but was negligible in 1985 (2.5 g, all of which was donated).

The need to purchase food during the shortage posed a severe financial burden for most households; virtually all resorted to some means of generating income that they normally would not have pursued.

FIG. 8. Household consumption of 12 selected food items by source during September November 1984 and the same period in 1985

(Food items include yellow maize, white maize. sugar, English potatoes, sweet potatoes, rice, yellow-maize flour, white-maize flour, sukuma leaves, cow's milk, meat, and kidney beans )

FIG. 9. Sources of major foods (white maize, kidney beans, and English potatoes), September-November 1984 and 1985

Thus, 69% of the households reported that the male and/or female head engaged in work for wages in such jobs as farm labour (46.7%), house construction (4.8%), and rural access road work (2.4%). Also, 58.1 % of the households reported that they sold animals to obtain money to buy food (36.6% sold chickens, 30.0% cattle, and 24.6% goats), and 14% engaged in temporary businesses by selling garden produce, meat, firewood, and charcoal. More drastically, two households sold land to buy food.

The second major source of food during the drought was donations. Aside from receiving CRSP-initiated food aid, 41% of the study households also received aid from friends and relatives, 24% from the government, and 15% from local churches. The CRSP-donated foods totaled 193,524 kg of maize, 35,946 kg of beans, 4,092 litres of oil, and 5,696 kg of dehydrated milk. These amounts, however, did not provide for all the households' energy requirements. Table 3 gives the kilocalorie equivalent of food aid distributed by CRSP by month, and figure 10 shows the amount of all donated foods received by the study households by source and month. As discussed below, these amounts do not represent actual consumption; only a portion of the donated food was consumed by the study households.

TABLE 3. Food relief distributed by CRSP, October 1984 April 1985 (kilocalorie equivalents)

  No. of households Persons/household kcal/household/day kcal/person/day
Oct. 1984 253 7.7 1,832 238
Nov. 1984 246 7.8 3,550 455
Dec. 1984 248 7.9 3,345 423
Jan. 1985 247 8.0 4,164 521
Feb. 1985 178 8.0 6,314 789
Mar. 1985 152 8.0 5,430 679
Apr. 1985 154 8.0 1,477 185

Because many households ate seeds during the food crisis that normally would have been reserved for planting, CRSP recognized that this could prolong the shortage by inhibiting the next harvest and also distributed seeds as part of its relief effort. A total of 5,000 kg of Katumani hybrid maize seed and 15,900 kg of bean seed were distributed in March and October of 1985, just prior to the rainy seasons.

Impact of food assistance

Household energy intake increased somewhat in December 1984 and January 1985 despite the fact that crops were not harvested until late February 1985. This was due in part to CRSP's food assistance programme. Figure 11 shows actual household intake and an estimate of the intake with the kilocalories derived from CRSP food aid excluded. This estimate suggests that, had food aid not been available, the crisis would have been more severe; the most serious impact would have been in January 1985 with a daily household intake of only 7,227 kcal (1,163 kcal below household BMR).

This estimated view of the crisis seems reasonable. Consumption of locally produced food began to show an increase by January 1985, when harvests reflected the return of the 1984 short rains. By March, 88% of the total kilocalories was from home-grown maize and beans and only 11% was from donated foods.

FIG. 10. Average daily amount of food donated to the study households, by source and month, September 1984 December 1985

FIG. 11. Mean household kilocalorie intake, estimated intake with kilocalories from CRSP food aid excluded, and household basal metabolic rate, February 1984-November 1985

TABLE 4. Household receipt and consumption of donated foods



kg %
Maize 148.9 75.8 50.9
Beans 62.9 36.5 58.0
Milk 6.4 1.9 29.7
Corn oil 3.0 0.9 30.0

This is in contrast to November 1984, when 10% was from home-grown foods and 40% from donated foods.

The estimated view of food intake is subject, however, to two major sources of error. First, it assumes that the households would have permitted intake to drop to the estimated level rather than taking action to defend a higher level. Had no relief been available, there might have been even more reliance on such measures as selling of livestock or other possessions to purchase food, out-migration, or greater consumption of less desirable foods to maintain intake levels. It is difficult to say what level of consumption would have been tolerated before other coping strategies might have been implemented, or what impact they might have had on intake.

The second source of error comes from the fact that not all the food aid was consumed directly by the households which received it. As indicated in table 4, the average household received more donated food than it consumed. One explanation is that households sold some of the donated food; many families reported purchasing yellow maize in the market, where it is normally not available, indicating that other households may have been selling their allocations. If donated food was sold and the earnings were used to buy other foods, the kilocalories derived from those purchased foods would also have to be deducted from total intake to provide an accurate picture of the impact food aid had on household consumption. Unfortunately, data are not available to indicate this more accurately.

Another explanation for the discrepancy between the amounts of donated and consumed food is inter household sharing. Many study households had friends and relatives nearby who did not receive aid, and it is probable that some food distributed to study households was subsequently passed along to others who were undergoing shortage.

Food prices and food availability

Because purchasing food was a primary coping strategy during the drought, the issues of food price and availability become extremely important. The prices of selected items in local markets reached their highest level in November 1984 (table 5), precisely when many households were heavily dependent on purchased foods. Given these increases, a household had to commit substantially more of its financial resources to maintain the same level of intake. In other words, whereas a Kenyan shilling purchased 1,113 kcal of white maize in May 1984, it purchased only 747 kcal in November 1984 (1 g of white maize is equivalent to 3.6 kcal). The November prices of finger millet, bulrush millet, and sorghum all rose well over 100% from their May prices.

It is usually assumed that during a food crisis little food is available in the markets to purchase [17].

TABLE 5. Food prices for selected items in local markets (Kenyan shillings per kilogram)

  May1984 Aug.1984 Nov.1984 Jan.1985
White maize 3.19 3.41 4.82 4.51
Finger millet 9.79 10.47 20.71 18.82
Bulrush millet 6.15 10.21 17.27 6.55
Sorghum 6.93 10.04 18.97 7.14
Kidney beans 11.14 11.69 13.16 9.73
Cowpeas 16.94 19.35 22.10 15.73

TABLE 6. Daily availability of selected foods in local markets (kilograms)

  May1984 Aug.1984 Nov.1984 Jan.1985
White maize 152 163 224 171
Kidney beans 93 24 114 64
English potatoes 123 93 122 123

The CRSP found, however, that the availability of three major foods was greatest in November 1984 (table 6). This was well into the period of the decline in household intake, and also the time of the highest prices for these items.

Discussion and policy implications

The findings in this study clearly show the severe impact that drought and local crop failure can impose on small scale agricultural households. Major disruption of such a population's supply of subsistence foods leaves its members extraordinarily vulnerable to economic disaster and acute malnutrition.

Specific consequences in Embu due to the 1984 drought included a marked decline of food intake; sharp declines in body weight or, at best, slowed growth rates; high dependency on purchased foods; the need to sell assets and perform increased wage labour to generate critical cash income; increased market food prices; and an apparently high availability of foods on the market. These observations require further elaboration, particularly with regard to their implications for appropriate development policy.

The failure of the 1984 long-rains harvest (June August) led to a severe decline of household energy intake. A major consequence of this was a decline in nutritional status characterized by an increase in the prevalence of protein-energy malnutrition among children, severe weight loss among adults, weight loss and increased stunting among schoolchildren, and slowed growth among toddlers.

The immediate response to hunger and the primary coping strategy of households during the drought was to purchase food. In fact, well over twice as much food was purchased in the latter part of 1984 as in the same period in 1985. This response to extensive subsistence crop failure has been observed elsewhere [18-20]. In Kenya, purchasing food has been the primary "modern" household response to shortages since the 1940s, with the development of a rural transportation and market system [20]. Unfortunately, this does not automatically substitute for growing food in the garden.

As production failed and dependence shifted toward market foods, the possession of cash became indispensable for farmers. However, cash acquisition was a significant problem since employment opportunities were scarce. A lack of employment opportunities meant that income had to be generated in some other way, the most common of which was the sale of farm animals. Unfortunately, the value of these animals decreased dramatically, partly because they too had lost weight, but also because many were on the market and few people could afford them. Thus, fawners were not only forced to pay more for food, but their potential purchasing power declined.

To worsen the already severe financial burden, the prices of staple foods reached their highest points precisely at the time of the greatest dependence on purchased foods. Unusual cost increases were clearly seen in Embu: the average seasonal price variation for all crops is approximately 50%, but during 1984 the variation averaged 76% [21]. These increases effectively discounted the value of household assets and inhibited the effectiveness of the purchasing strategy. Therefore, households had to sacrifice proportionately more of their assets to make the food purchases necessary for survival; two project households made the ultimate sacrifice of selling their land to survive.

The market-place changes were therefore as dramatic a blow to household food security as the drought itself. Farmers were not only being forced to pay more for food, but the value of their assets was discounted so that they had less to draw on to make their purchases. Households were thus subjected to a two-pronged attack on their intake levels.

For populations already economically marginal, the decision to sell assets (for low returns) for short-term sustenance at the expense of long-term stability is painful and distressing. Sen has characterized this process as a "food battle" in which success depends on what people own, and whether or not assets can be exchanged for enough food. He cautions, "These serious conflicts and their terrible outcomes have the appearance of order and legitimacy precisely because we refuse to see them as what they are, namely, unequal struggles to capture enough food to eat and to survive" [22]. Liquidation of assets under unfavourable exchange conditions is clearly not an effective strategy for dealing with drought. It only serves to intensify the level of poverty that presently exists and enhances households' vulnerability to future climatic perturbations.

Although it is commonly assumed that there is little market food available to purchase during food crises, we found, on the contrary, that the availability of major foods was greatest in November 1984. Ironically, this was well into the period of lowest food intake and highest prices. It seems to indicate that it was not a lack of food that produced the sharp decline in intake, but rather how accessible that food was to the average household. Similarly, in an analysis of the great Bengal famine of 1943, Sen found that food availability was highest during the period of greatest suffering; in part this was because inflated prices and lack of income reduced household "exchange entitlement," or the ability to exchange production or labour for adequate amounts of food [17]. Sen argues that such crises occur not necessarily because of shortage but from the inability of people to obtain food, and this appears to have been the case in Embu.

At the same time, however, we recognize that a true shortage existed in the regional food supply. Thus, the high availability of foods at market-places may have been illusory in the sense that it may not have been a true reflection of availability in the market system. That is, although foods were highly available on market day at a given site, this may have been not because there was a surplus in the storehouse but rather because few could afford the steep prices. Had prices been at normal levels, foods might have been considerably less available. We suggest that prices were high because some could afford them (or had to give in to them), and that this retarded the volume of exchange, leaving an apparent but somewhat deceptive impression of plenty.

To the extent that this hypothesis is correct (and additional economic research should yield insights on foods available in the market and the reasons for their high prices), government policies need to consider the impact of unfavourable terms of exchange. Specifically, opportunistic price increases during crises, particularly by longdistance merchants who profit from localized shortages, need to be prevented. Close government supervision of inter-regional food distribution, with nominal, controlled price increases, would presumably go far towards minimizing problems such as were seen in Embu. This underscores a major difficulty of the purchasing strategy: it allows market dynamics to set prices, which in turn become the rationing mechanism. This naturally works against the poor.

In any case, food was available on the market for those with cash to purchase it. Because many households were in danger of starving, their ability to generate income is a key issue that demands attention. Government policies should be oriented towards periods of local or regional food shortage and should focus on providing employment opportunities or some other means of ensuring that farmers have purchasing power when crops fail. Such an objective is in the interest of governments as it would provide longterm security. When households are forced to sell assets for short-term survival, their recovery after the crisis is that much harder, if possible at all, leaving them less productive and perhaps more dependent on others or national social services.

While the Kenyan government's plan to decrease the disparity between food production and consumption by intensifying production of subsistence and export crops is commendable, our findings point to the need for African food policies to recognize the critical role that household survival strategies play during shortages. Because it is probable that many droughts and other causes of severe shortages will occur throughout Africa and elsewhere, it is urgent that the shortcoming of current systems of food production and exchange be uncovered and minimized. Because much effort is going into the intensification of production (e.g., farming of marginal land, reliance on new crop varieties), which inevitably is associated with increased risk of failure, the development of appropriate exchange policies and programmes is now more important than ever.

Finally, we urge that population growth be slowed and education (especially agricultural) be increased, and that more research and development be pursued along the lines of home and community food storage, the production of more drought-resistant crops, and the implementation of rotating credit schemes. Households must be able to defend an adequate level of intake without having to engage in destructive measures that threaten their future productive capacity.

All families should be entitled to a defensible food intake level (i.e., one that can be maintained under adverse conditions, against the insults of climatic variability or any other source of disturbance). This means dealing with households' vulnerability through expanding their resources, which, in turn, means confronting the issue of poverty. As long as poverty exists, households will be unable to defend their level of food intake during a crisis. They will be dependent on outside assistance to survive, assistance that usually only comes once severe damage has been done to the household as a productive unit. We must strive to enhance their endurance, for this is the only rational solution to the ravages posed by drought.


Data collection was supported by USAID Contract Grant No DAN 1309 -G-.1070.

Data analysis was supported in part by the UCLA Office of International Studies and Overseas Programs.

Numerous people contributed to the collection of the data presented in this paper. We especially wish to thank Susan Weinberg, A. A. J. Jansen, and Eric Carter.


1. Wisner B, Mbithi PM. Drought in eastern Kenya: comparative observations of nutritional status and farmer activity at 17 sites. Nairobi: Institute for Development Studies, 1973.

2. World Bank. World development report. Washington, DC: Oxford University Press, 1984.

3. Kliest T. Regional and seasonal food problems in Kenya. Nairobi: Food and Nutrition Planning Unit, Ministry of Finance and Planning, 1985.

4. Wisner B. The human ecology of drought in eastern Kenya. Doctoral dissertation, Clark University, Worcester, Mass, USA, 1978.

5. Cohen JM, Lewis DB. Role of government in combatting food shortages: lessons from Kenya 1984-1985. In: Glantz MH, ed. Drought and hunger in Africa. Cam" bridge: Cambridge University Press, 1987:269-96.

6. Odiko G. Kenya's food crisis over, report says. Daily Nation (Nairobi) 9 Jul 1985:p.17.

7. Downing TE, Mungai DN, Muturi HR. Drought climatology of central and eastern Kenya. In: Climatic variability and agricultural production in central and eastern Kenya. Nairobi: National Environment Secretariat, Ministry of Environment and Natural Resources, 1985.

8. Neumann CG, Bwibo N. The collaborative research support program on food intake and human function, Kenya project. Final report. Los Angeles: University of California School of Public Health, 1987.

9. Republic of Kenya. Kenya population census, 1979. Nairobi: Central Bureau of Statistics, Ministry of Finance and Planning, 1981.

10. Watt BK, Merrill AL. Composition of foods. Agriculture handbook 8. Washington, DC: US Department of Agriculture, 1975.

11. FAO. Food composition tables for use in Africa. Rome: Food Consumption and Planning Branch, Nutrition Division, FAO, 1968.

12. FAO/WHO/UNU. Energy and protein requirements. Report of a joint expert consultation. WHO technical report series, no. 724. Geneva: WHO, 1985.

13. Jelliffe DB. Assessment of nutritional status of the community. Geneva: WHO, 1966.

14. Weiner JS, Lourie AJ. Human biology: a guide to field methods. Oxford: Blackwell Press, 1969.

15. Zerfas AJ. Anthropometric field methods: general. In: Jelliffe DB, Jelliffe EFP, eds. Human nutrition: a comprehensive treatise. New York: Plenum Press, 1975.

16. Hussain MA. Seasonal variation and nutrition in developing countries. Food Nutr 1985;11(2):23-36.

17. Sen AK. Poverty and famine: an essay on entitlement and deprivation. Oxford: Clarendon Press, 1981.

18. Kennedy ET. Effects of sugarcane production in southwestern Kenya on income and nutrition. Final report. Washington, DC: International Food Policy Research Institute, 1988.

19. World Bank. Poverty and hunger: issues and options for food security in developing countries. Washington, DC: World Bank, 1986.

20. Herlehy TJ. Historical dimensions of the food crisis in Africa: surviving famines along the Kenya coast 18801980. African Studies Center working paper no. 87. Boston: Boston University, 1984.

21. Olsen W. Kenya's dual grain market: the effects of state intervention. Quoted in Meilink HA, ed. Food consumption and food prices in Kenya. Food and Nutrition Planning Unit report no. 21. Nairobi: Ministry of Finance and Planning, 1987.

22. Sen AK. Food battles: conflict in the access to food. Food Nutr 1984;10(1):81-89.

Contents - Previous - Next