Previous Page Table of Contents Next Page


Household income and expenditure studies


Introduction
Collection of households income and expenditure data
Analysis of data
References
Etudes sur les revenus et les dépenses monétaires des ménages

Solomon Bekure
Team leader, Arid Zones (Eastern and Southern Africa) Programme, ILCA, Kenya

Introduction

Very little is known about the economics of household level activities in pastoral production systems (Etcher and Baker, 1982). Despite this, interventions are frequently proposed which call for increased cash expenditures. The implications of this for different groups of households (poor, rich) needs to be assessed. An intervention may require increased labour input. Will there be enough labour and if so will sufficient food be available to sustain the energy requirements of increased effort? An intervention may call for culling of nonproductive and old animals from herds and flocks. What is the implication of this for the security and viability of different groups of pastoral households (rich, poor)? What needs of pastoral households can be manipulated to provide incentives for culling? It is in such a context that information on household income and expenditures is required.

Testing and evaluating the impact of interventions as well as assessing the welfare of pastoralists require benchmark data on income and expenditure patterns of different groups of pastoral households .

Unlike agriculturalists, pastoral households depend more on market transactions to satisfy their subsistence needs. Our own work in Kenya shows that even during the wet season the Maasai obtain up to 50% of their calorie intake from purchased food. This figure can increase to 70% during the dry season. Household income and expenditure data are useful for determining the demand of pastoralists for purchased goods and social services and for assessing the teens of trade between pastoralists and the rest of the economy. A very good example is given by Swift (1979) for Somali pastoralists. He analysed the barter terms of trade between pastoral products sold by pastoralists and those they purchase by constructing a pastoral cost of living index. He concluded that by the early 1970s their terms of trade had deteriorated and had "led the pastoral economy into a precarious position".

Collection of households income and expenditure data

In addition to producing milk and meat for their own consumption, pastoral households engage in a variety of transactions involving livestock, livestock products, cash and other items (such as crops, handicrafts etc.), to fulfill different goals. Animals and their products are sold to provide cash. They are given or lent to kin and friends to strengthen social ties and ensure long-term security. They may be similarly received. Animals may be exchanged for social reasons or to increase the productive capacity of herds and flocks.

In order to determine the entire household budget, income and expenditure studies should be designed in such a way as to include not only cash income and expenditures but also these important transactions. Usually, the quantities of these transactions are known and their values can be determined by using prices the items would have attracted had they been sold.

Sampling the target population

In any society the most important factor that influences patterns of household income and expenditure is the wealth status of the household. It is self-evident that the consumption of poor and rich households is markedly different. It is therefore essential that study samples adequately represent the gradient of wealth observed in the pastoral group under study. This may not present a problem in situations where a whole village or encampment or target population is studied as done by the ILCA teams in Mali and Kaduna. Even then villages and camps may be formed on the basis of social classes. Care should be taken in selecting sample villages and camps so that results can be generalisable to a known population type. In situations where the coverage in area is more extensive one has to resort to sampling the population as done by the ILCA teams in Kenya and Ethiopia. In that case a stratification of the population into wealth categories is essential (Bee Grandin (1983) for a detailed discussion of wealth effects and a rapid method of wealth ranking).

Given such a stratification of the target population, the available resources for collecting data and the time-frame and nature of the study, standard sampling procedures can be employed to determine the size of the sample and choosing them (Cochran, 1963).

Types of data required

Inventory of resources

Once the sample households in the study have been determined, an initial inventory of the human and livestock population needs to be made. Here care should be taken so that animals owned by the household but which are away from the main herd or flock at the time of the inventory are included. Animals not owned by the household but borrowed from others should be identified and recorded as such. Similarly members of the household who are away at the time of the census should be included and temporary visitors excluded. An inventory of major household goods also gives a good indication of investment and consumption patterns.

This information is vital for two reasons. First, it quantifies the wealth status of the sample household. Second, it provides the basic population data to perform per capita computations without which meaningful comparative analysis cannot be made.

Household income and expenditure items

For designing the data collection formats background information is required on the nature of items that form the income and consumption baskets of the pastoral households to be studies as these vary from culture to culture. A comprehensive list of these items should be established from the researcher's personal knowledge or from informal surveys involving a few pastoralists and shopkeepers in the area or a combination of these. In addition to standardising the format for enumeration, it is also a good device to facilitate recall by respondents.

The income items include :

- livestock and livestock products (animals. milk, ghee, hides and skins, manure etc.)
- agroforestry products (crops, wood, charcoal, honey etc.)
- cottage industry products (handicraft, beer, medicinal herbs etc.)
- other forms of employment (trade)
- other cash inflows (remittance, borrowing)

Cash expenditure items can be grouped as:

- food
- health and hygiene
- clothing transport
- livestock
- livestock inputs durable household goods
- others (cash outflows such as loans given).

Frequency of data collection

Extracting information on household budgets, especially expenditures, is extremely difficult because one has to rely on the memory of the respondents to recall such data. Information on pastoral households' income is by far easier to get because most of it is derived from the sale of animals, which they remember very well. The feet that such sales happen very infrequently facilitates recall. On the other hand expenditures, especially on food items, occur so frequently in irregular amounts that recall becomes difficult.

In collecting household income and expenditure data, the shorter the time span the respondent is requested to recall the more accurate is the information obtained. Researchers have used different frequencies of collecting such data ranging from one-shot surveys asking questions to estimate income and expenditure for a specified period of time (e.g. per month or per week etc.), to continuous daily recording for a long period of time (a year or more). Within ILCA, the Kenya team has used a frequency of once a month, the Kaduna and Mali teams twice weekly. Of course the latter is more reliable but is, however, very expensive. Unless it is combined with the conduct of other research requiring daily observations, it may not be solely justified on the basis of the precision gained. Even after using this method in Mali, Swift (1983) writes:

'It is likely that enumerators were only partially sucessful in recording these details of household daily transactions. There was inevitably some resistance to such detailed questioning and at times clearly false information was given or important transactions were forgotten or concealed.'

Our own experience in Kenya using a once-a-month recall method proved that respondents could account only for about 70% of their cash income. In conjunction with a nutrition study, which required food intake data on a daily basis, household expenditures were monitored for a month on a daily basis on selected households. A comparison of the two methods showed that the monthly recall accounted for 73% of the total expenditures recorded on a daily basis.

Table 1. Comparison of monthly and daily recalls on four households in Merueshi Group Ranch, Kenya (October, 1982).

ITEM

CASH EXPENDITURES FOR ONE MONTH ON

Food

Livestock

Others

Total

Monthly recall

Daily recall

Monthly recall

Daily recall

Monthly recall

Monthly recall

Daily recall

Mon.rec.

KShs

1308

3047

1905

1980

1188

1015

4401

6042

Proportion of total

0.30

0.50

0.43

0.33

0.27

0.17

1.0

1.0

Ratio of monthly recall to dally recall

0.43

0.96

1.17

0.73

Source: ILCA/Kenya (unpublished data).

However, this percentage varied significantly across different expenditure categories as shown in Table 1. The monthly recall method underestimates food expenditures substantially, proving that frequent but irregular flows are hard to recall during a long period of time. Researchers with wide experience in this field recommend that a three-day span is the most that respondents can recall with a high degree of accuracy (Nester, pers. comm.)

The frequency- of collecting household and income expenditure data should depend on the nature of the research. If a complete understanding of the process is required the highly expensive more frequent methods need to be used. If general orders of magnitude are required the less frequent methods may be used. One may even sample seasonally by concentrating interviews during specific months or weeks in dry and wet seasons. Unfortunately, one may not encounter normal wet and dry seasons, especially in East Africa. One must, therefore, take into account the degree of representativeness of such samples in the analysis of the data (see below).

Quality control in enumerator data collection

Invariably, enumerators are used to advantage to obtain income and expenditure data covering a large number of households in one or more geographic locations simultaneously. The use of enumerators, however, introduces biases in the data collected as noted above. Careful training and supervision is essential.

Analysis of data

Table 2 shows a summary of income and expenditure patterns of Tuareg households in Mali. The dominant source of income is the sale of animals, representing 42% of total income. Salaries earned are a close second accounting for 35%. Income from cereals, other food and consumption items and clothing is not from the sale of these items as such, but represents the value of these items obtained in exchange. The purchase of food items accounts for 49% of total expenditures, the most prominent item being cereals (22%). Expenditure on the purchase of livestock and livestock inputs is also high (32%) followed by expenditures on clothing (10%).

Table 2. Mean annual household income and expenditure of 16 Tuareg households in Mali.

 

Income

Expenditure

Amount (MF)

%

Amount (MF)

%

Animals

92,350

42

66,194

32

Milk

1,069

-

6,600

3

Cereals

16,038

7

46,919

22

Other food and consumption

15,963

7

51,319

24

Clothing

1,813

1

21,506

10

Equipment and artisan work

15,150

7

9,556

5

Salarie etc.

78,250

35

9,425

4

TOTAL

220,656

100

209,519

100

Source: Swift (1982).

Tables 3 and 4 give a summary of per capita income and expenditures of Maasai households in Mbirikani Group Ranch in Kenya. In this case opportunities to -earn income from employment outside the pastoral sector are restricted. Livestock sales (mainly cattle) represent a very high proportion (83%) of total cash income. The only other significant source of cash income is trading in livestock, which provides 11% of total cash income.

Table 3. Sources of cash income of Maasai pastoralists in Mbirikani Group Ranch (July 1981 to June 1982)


cash income
Mean annual per capita

Source

KShs

%

Livestock sales

1523

83


cattle

1474

81


smallstock

49

2


milk sales

0

0


hides & skins

9

1

Trading

205

11

Remittance

9

1

Borrowing

42

2


OTHERS

38

2

TOTAL

1821

100

Source: ILCA, Kenya 1982 (unpublished data).

Table 4 gives a more disaggregated picture of mean per capita cash expenditure for food and non food items excluding purchases of livestock and livestock inputs. Cereals, sugar (consumed in tea with milk) and fats/oils, which provide a substantial portion of the calorie requirements of the Maasai, are the major items of cash expenditure, followed by clothing.

From the above it is clear that the pastoral households studied both in Mali and Kenya spend a very high proportion of their income on food and livestock. The income elasticity of food expenditure is very low.

Table 4. Annual per capita expenditure on food and non-food items by Maasai households in Mbirikani Group Ranch (July 1981 to June 1982).

Food items

Expenditure

KShs

% of total food items

Maize

222*

49

Wheat

20

4

Sugar

55

14

Tea

30

7

Fat/oils

29

6

Potatoes

7

2

Vegetables

2

0

Meat

6

2

Other food

6

2


Subtotal

387

86

Hotel food

28

6

Hotel drink

36

8

Subtotal food items

451

100

Non-food items


% of total food items

Tabacco

3

4

Kerosine

2

2

Soap/Omo

5

6

Transport

13

15

Medical

9

11

Clothing

47

56

Household items

3

4

Beads

2

2


Subtotal non-food items

84

100

Source: ILCA, Kenya 1982 (unpublished data).
*74 kg per capita per annum.

Since the consumption of other items is almost negligible, the implication is that any increased income will be largely spent on the purchase of breeding stock unless attempts are made to increase pastoralists' demands for other consumption and investment items.

A more sophisticated analysis of household income and expenditure data can be performed using regression techniques. A common representation of this is given as:

Ei = 0 + 1 1 +2 2 +......+ m m

where

Ei= Expenditure on the ith commodity or group of commodities

0 least square mean

(j = 1.2........,m) = coefficients of the independent variables
(j = 1,2.......,m) = independent variables specified as determinants of expenditure e.g.

X1 = household size
X2 = wealth or income group
X3 = season of the year etc.

Usually, the log form of the above function is preferred as the coefficients specified directly produce elasticities of expenditure.

The main determinants of expenditures of pastoral households are: (a) household size, (b) income or wealth status and (c) seasons of the year.

Ceteris paribus, the larger the household size the higher the expenditure, the higher the wealth status of the household the bigger the expenditure, and the drier the season the larger the expenditure, especially on food. Table 5 shows how the cash income and expenditures on cereals of the Tuareg in Mali is affected by the different seasons.

Table 5. Seasonal cash income and expenditures on cereals of 16 Tuareg households in Mali (1371-1981).

Season

Cash income Excluding salaries

Cash expenditure on cereals

MF

%

%

Hot (March-June)

39,100

38

23,806

53

Rains (July-October)

44,206

42

12,513

27

Cold November-February)

20

9,300

20


TOTAL

103,394

100

45,619

100

Source: Swift, 1982.

The use of expenditure baskets of pastoralists to calculate terms of trade for pastoral products as calculated by Swift (1979) is shown in Table 6.

Table 6. Barter terms of trade for pastoral products in southern Somalia

Camels

1847

1891

1951-53

1956-58

1971-72

1974-75

Barter value of 1 adult male camel

rice (kg)

140

204

71

108

159

192

maize (kg)

978

380

472

165

421


sorghum (kg)

1110

978

490

460

284

1070

sugar (kg)

22

81

88

88

111

156

tea (kg)



25

22

27

49

dates (kg)

265

341





local cloth (yd)

83

25





American grey sheeting (yd)


138





Indian printed cloth (yd)


576





cotton futa cloth (yd)



39

43



Cattle

Barter value of 1 adult ox or bull

rice (kg)

140

184

44

61

86

95

maize (kg)


880

235

267

164

418

sorghum (kg)

1110

880

305

260

154

648

sugar (kg)

22

73

55

50

60

156

tea (kg)



16

12

15

30

dates (kg)

265

306





local cloth (yd)

83

22





American grey sheeting (yd)


124





Indian printed cloth (yd)


518





cotton futa cloth (yd)



24

24



Sheep

Barter value of 1 adult sheep

rice (kg)

21

10

12

16

41

25

maize (kg)


49

63

69

78

108

sorghum (kg)

167

49

81

68

73

167

sugar (kg)

3.2

4.1

15

13

29

24

tea (kg)



4.2

3.2

7

8

dates (kg)

40

17





local cloth (yd)

13

1.2





American grey sheeting (yd)


7





Indian printed cloth (yd)


29





cotton futa cloth (yd)



6.4

6.3



Goats

Barter value of 1 adult goat

rice (kg)

14

5

9

12

32

15

maize (kg)


24

49

54

61

68

sorghum (kg)

111

24

59

53

57

105

sugar (kg)

2

2

11

10

22

15

tea (kg)



3

3

6

5

dates (kg)

27

8.4





local cloth (yd)

8

0.6





American grey sheeting (yd)


3.4





Indian printed cloth (yd)


14





cotton futa cloth (yd)



5

5



Ghee

Barter value of 1 kg ghee

rice (kg)

4

5

2

4

7

5

maize (kg)


23

11

16

13

23

sorghum (kg)

29

23

14

15

12

36

sugar (kg)

0.6

2

3

3

5

5

tea (kg)



0.7

0.7

1

2

dates (kg)

7

8





local cloth (yd)

2

0.6





American grey sheeting: (yd)


3





Indian printed cloth (yd)


13





cotton futa cloth (yd)



1

1



Source: Swift, 1979.

References

Eicher, C.K. and Baker, D.C. 1982. Research on agricultural development in sub-Saharan Africa: A critical survey. Michigan State University, East Lansing.

Grandin, B.E. 1983. The importance of wealth effects on pastoral production: A rapid method for wealth ranking. Paper presented at the Workshop on Pastoral Systems Research in sub-Saharan Africa' ILCA, Addis Ababa.

Cochran, W.G. 1963. Sampling techniques. John Wiley & Sons, Inc. New York

Swift, J. 1979. The development of livestock trading in a nomad pastoral economy: The Somali case. In: Proceedings of the international meeting on nomadic pastoralism, pastoral production and society. Cambridge University Press.

Swift, J. 1982. In Les systèmes de production pastoraux au Mali. First draft. ILCA, Addis Ababa.

Etudes sur les revenus et les dépenses monétaires des ménages

Résumé

On dispose de peu de renseignements sur l'économie des ménages dans les systèmes de production pastorale. En dépit de cela, des interventions qui impliquent l'accroissement des dépenses monétaires sont fréquemment proposées. L'expérimentation et l'évaluation des i interventions et de leurs effets de même que l'évaluation de la qualité de la vie des éleveurs nécessitent des données de base sur la structure du revenu et des dépenses de différents groupes de ménages pastoraux.

Outre la production de lait et de viande auto-consomée, les ménages pastoraux participent à diverses transactions de bétail, de produits animaux, de cultures commerciales, d'objets artisanaux etc., en vue de réaliser leurs différents objectifs. Pour déterminer la totalité du budget du ménage, les études sur le revenu et les dépenses doivent être conçues de sorte à refléter non seulement le revenu et les dépenses monétaires mais aussi d'autres transactions importantes.

Dans toute société, le facteur le plus important qui influence la structure des revenus et des dépenses du ménage est la richesse du ménage. Il est évident que la consommation est différente selon qu'il s'agit d'un menage riche ou pauvre. Il est par conséquent essentiel que les échantillons de l'étude représentent adéquatement le gradient de richesse observé dans le groupe pastoral étudié. Une fois que les mènages-échantillons de l'étude auront été déterminés, il faudra entreprendre un inventaire initial de la population humaine et animale. Un inventaire des biens essentiels du ménage donne également une image fidèle de la structure des investissements et de la consommation. Cette information est vitale pour deux raisons: elle quantifie la richesse du ménage - échantillon et fournit des données de base sur la production pour effectuer les calculs par tête d'habitant sans lesquels aucune analyse comparative significative ne peut être effectuée.

Pour la conception des formulaires de collecte de données, il convient de rassembler des informations sur la nature des éléments qui forment le revenu et la consommation des ménages pastoraux à étudier, étant donné que ceux-ci varient d'un cadre culturel à l'autre.

Dans la collecte des données sur le revenu et les dépenses du ménage, l'information recueillie est d'autant plus exacte que la période de rappel est courte. La fréquence de la collecte des données sur les dépenses et les revenus des ménages devrait être fonction de la nature de la recherche entreprise. Si une compréhension approfondie du processus est requise, ce sont les méthodes fréquentes, très coûteuses qui doivent être utilisées. Par contre, si les ordres de grandeur requis sont généraux, il convient d'utiliser les méthodes les moins fréquentes.

Les enquêtes effectuées au Kenya et au Mali ont clairement montré que les ménages pastoraux observés dépensaient une très forte proportion de leur revenu pour acquérir des produits alimentaires et du bétail. Puisque la consommation des autres produits était presque insignifiante, on peut en déduire que tout accroissement du revenu sera en grande partie consacrée à l'acquisition de reproductrices, à moins que des efforts ne soient déployés pour accroître la demande pour d'autres biens de consommation et d'investissement.

Une analyse plus sophistiquée des données sur les dépenses et les revenus des ménages peut être effectuée en utilisant les techniques de régression


Previous Page Top of Page Next Page