Cover Image
close this bookAgricultural Growth Linkages in Sub-Saharan Africa - Research Report 107 (IFPRI, 1998, 152 p.)
View the document(introduction...)
View the documentForeword
View the documentAcknowledgments
View the documentSummary
View the documentCHAPTER 1. - Introduction
View the documentCHAPTER 2. - Concepts, Prior Work, and Issues Pertaining to Agricultural Growth Linkages
View the documentCHAPTER 3. - Methodology and Overview of Case Studies
View the documentCHAPTER 4. - North to South in Burkina Faso
View the documentCHAPTER 5. - Southwestern Niger
View the documentCHAPTER 6. - The Senegalese Groundnut Basin
View the documentCHAPTER 7. - Eastern Province, Zambia and Gazaland District, Zimbabwe
View the documentCHAPTER 8. - Conclusions
View the documentBibliography

CHAPTER 7. - Eastern Province, Zambia and Gazaland District, Zimbabwe

As in previous chapters, this chapter uses available household survey data collected by IFPRI and its collaborators for other purposes to examine the relations between farm and nonfarm growth linkages.22 The data set for Zambia permits empirical estimation of consumption parameters and growth multipliers from initial growth in the tradables sectors in two distinct agroecological regions, with detailed investigation of differences observed across the distribution of farm sizes. Although the Zimbabwe data are more qualitative in nature, they permit a view of significant differences in household demand patterns between larger commercial farm households and more traditional, smaller communal sector farms.

22A shorter form of the present chapter covering Zambia only and omitting disaggregated results for the expenditure analysis by total expenditure decile was published in Hazell and Hojjati 1995.

The Study Regions

While data on the importance of the rural nonfarm economy in Zambia and Zimbabwe are fragmentary, they generally support Hazell and Haggblade's (1989) figures for Sub-Saharan Africa as a whole. Census data from Zimbabwe show that, in 1982, about 20 percent of total full-time employment in rural areas and rural towns was in rural nonfarm economic activity (Zimbabwe, Central Office 1985). In Zambia, Headland and Lundahl (1983) found that the rural nonfarm economy accounted for 30 percent of secondary rural employment in two study regions. Due and Mindenda (1985) found that the rural nonfarm economy contributed 24 percent of total rural cash income in three provinces in Zambia, and Marter and Honeybone (1976) reported that about 90 percent of rural households received some income from nonfarm activity. Most rural nonfarm firms are small, averaging only 1.6 workers in Zambia. Women play an important role in rural nonfarm activities, owning some 60 percent of all rural nonfarm firms in Zambia (Milimo and Fisseha 1986).

Zambia

The Zambian data were collected in a study region comprising nine agricultural districts in the Eastern Province. The survey was conducted in 1985/86 by a joint team from IFPRI, the Rural Development Studies Bureau of the University of Zambia, and the Eastern Province Agricultural Development Project. A sample of 330 farm households were selected on a regionally representative basis, and interviews were conducted monthly. Detailed information was collected on all aspects of the household economy, including sources of income and employment, farm inputs and outputs, and household consumption of self-produced foods and purchased goods and services of all kinds. The survey did not include any nonfarm households in local towns (Celis, Milimo, and Wanmali 1991).

The study region has two distinct agroecological zones - the Eastern Plateau and the Luangwa Valley. Annual rainfall ranges from 850 millimeters to 1,050 millimeters in the higher-altitude plateau; it is concentrated between the months of November and April. Agricultural activities are confined to this period. Sowing starts with the onset of rains and harvesting is completed by the end of May. More than 80 percent of the human population and all of the cattle are concentrated in the plateau region, and oxen cultivation is becoming widespread. Farms are larger in this region (2.74 hectares on average) than in the valley.

Maize is the dominant crop on the plateau, accounting for more than 80 percent of the cultivated area. Both traditional and hybrid varieties of maize are grown. Hybrid maize is cultivated exclusively as a market crop. Groundnuts are the other major crop in the plateau region. About 67 percent of the sample households used fertilizers in the survey year, mainly on maize, and more than 55 percent of the cropped area was fertilized (Jha and Hojjati 1993).

The valley is thinly populated, receives less rainfall, has higher temperatures, and is heavily infected with tsetse flies, carriers of trypanosomiasis. Maize and groundnuts are important in this zone too, but crops like sorghum, rice, millet, and cotton also occupy significant areas. Groundnuts and cotton are the only cash crops in this zone. Use of hybrid maize is practically nonexistent, and fertilizer use is negligible. Thus, all three major technological options - hybrid maize, fertilizer, and animal traction - are absent from the valley. Farms are smaller in this zone (0.97 hectares on average) and hoe cultivation prevails. The valley region is also lacking in infrastructure and support systems. Cash income is mostly from nonfarm sources.

Zimbabwe

The study region is Gazaland District in Manicaland Province. There is considerable diversity in the agroecological conditions and farming systems in the region. It is useful to distinguish among communal farming areas and the Middle Sabi and Chipinge areas where commercial farming is concentrated. The communal areas are characterized by poor land and small farm sizes (0.91 hectares on average), whereas Middle Sabi and Chipinge have rich soils and much larger farms (averaging 159 and 169 hectares, respectively). Households in the communal farming areas grow hybrid maize, local maize, millet, vegetables, fruits, tobacco, and wheat. Cotton, fruits, and vegetables are dominant in Middle Sabi, and fruits, vegetables, tobacco, and wheat in Chipinge. Maize is a much less important crop in the commercial farming areas.

The survey was conducted in 1987/88 in Gazaland District by IFPRI, Zimbabwe's Department of Physical Planning in the Ministry of Local Government, Rural and Urban Development (Wanmali and Zamchiya 1992). A sample of 297 farm households were selected to represent the different subregions. Data were collected on income, employment, and the amounts of money spent on different goods and services for consumption and farm business purposes. Again, the survey did not include nonfarm households in the local towns. The data are not nearly as complete as those for Zambia, and the scope of the analysis that could be undertaken for this study was therefore restricted.

Analysis of Growth Linkages in Zambia

Sources of Farm-Nonfarm Linkages

Conceptually, five different linkages might be important in Zambia, two in factor markets and three in product markets. The factor market linkages involve direct investment and labor flows among farm and nonfarm enterprises. Product markets include backward production linkages from agriculture to rural input suppliers and forward production linkages from agriculture to processors and distributors, and consumer demand linkages generated as a result of increasing farm incomes. Growth multipliers of the sort estimated in the present report specifically examine backward production linkages and consumer demand linkages, which are thought to account for most of the intersectoral linkages in zones such as Eastern Province.

While farm activity receipts probably finance investment in village-level nonfarm activity, there is less evidence to suggest any significant capital transfers to specialized nonfarm businesses in the local towns. This is partly because many of these firms are owned by whites and Asians, but also because the dominance of public marketing agencies undermines the traditional role of the trader in mobilizing surpluses within the regional economy. The three principal sources of rural household income are own agriculture, wage agriculture, wage nonagriculture, and local nonfarm business. Migration, whether long-term or seasonal, is an important factor flow but outside the domain of this study. The distribution across the year of the other three principal sources of income (other than migration) in the sample is shown in Table 33.

For production linkages, farmers purchase inputs for agriculture (backward linkages) and require agroprocessing and marketing services (forward linkages) for their products. Table 34 shows the average per hectare costs of these items for the valley and plateau, together with a breakdown by farm-size quartile. Farmers in the plateau region purchase twice the value of farm inputs and agroprocessing services per hectare as farmers in the valley. The dominant cost in both regions is fertilizers followed by milling and, in the plateau, oxen hire. Total per hectare expenditure on local inputs is only 50 percent larger in the plateau, so the strength of the total demand linkages to the regional economy is not as large as the initial differences in total expenditures would suggest. Fertilizer is imported into the region (and country), and hence it represents a demand leakage as far as the local (national) economy is concerned.

Table 33 - Seasonably in wage and nonfarm business earnings of the average farm household in Eastern Province, Zambia


Valley

Plateau


Wage earnings

Nonfarm business

Wage earnings

Nonfarm business

Month

Agriculture

Nonagriculture


Agriculture

Nonagriculture



(percent annual earnings in category)

January

8.9

2.3

2.9

12.3

5.9

9.9

February

11.2

3.8

6.7

11.1

13.5

4.4

March

9.0

13.1

4.0

7.1

11.1

9.7

April

7.0

15.4

5.5

7.1

9.1

6.9

May

3.9

4.6

0.2

4.2

8.2

0.0

June

6.1

7.1

8.4

7.9

5.3

12.8

July

1.7

4.6

10.8

10.3

4.9

6.7

August

3.1

7.7

7.4

5.2

7.4

6.7

September

6.2

27.7

10.6

10.0

10.3

7.6

October

3.5

8.5

19.2

7.8

12.5

9.3

November

20.1

1.4

12.1

11.2

8.3

8.8

December

19.3

3.4

12.2

5.8

3.5

17.3

Source: IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

The smallest farms use inputs and milling most intensively in both regions. This is also true for local inputs, so agricultural growth focused on small farms can be expected to lead to the strongest production linkages within the regional economy.

In addition to recurrent input costs, farmers also make longer-term, on-farm investments that lead to additional demand linkages to the nonfarm economy. However, as Table 35 shows, these investment costs are small and are used almost exclusively for livestock in the plateau region; hence they represent demand linkages to the farm sector itself. Investment costs are almost nonexistent among the valley farmers.

Household expenditures for consumption purposes are the dominant type of demand linkages in Eastern Province. The average household spends 1,058 kwacha (K) on goods and services for consumption each year, compared to K346 for farm inputs, and K15 for on-farm investment. Table 36 provides additional details about household consumption. The average household consumes goods and services valued at K3,191 each year, but since K2,133 are homegrown foods, purchased items only amount to K1,058 (not shown in the table). Food, alcohol, and tobacco account for 85 percent of the total value of consumption and for 47 percent of total purchases. Of the nonfoods consumed, clothing and footwear and consumer nondurables (such as fuel and soap) are the most important.

As household incomes increase, the demands for farm inputs and agroprocessing and marketing services typically increase in direct proportion to farm output. But household consumption demands are more complex, with varying income elasticities of demand for individual commodities. The next section is therefore devoted to an analysis of how consumer expenditure patterns in the study region respond to income increases.

Table 34 - Mean annual farm input expenditure per hectare by farm size quartile in the plateau and valley regions, Eastern Province, Zambia, 1986

Input

Plateau

Valley


All farms

Farm size quartile

All farms

Farm size quartile



1

2

3

4


1

2

3

4


(kwacha)

Local inputs

66.46

140.36

56.74

44.72

26.66

42.38

44.26

43.63

46.01

36.50


Milling

40.01

87.52

35.68

24.27

14.46

29.29

37.03

32.93

29.82

20.73


Veterinarian

0.08

0.00

0.04

0.25

0.02

0.00

0.00

0.00

0.00

0.00


Abattoir

0.75

1.01

0.51

0.92

0.55

4.49

0.98

8.70

1.77

5.64


Hired oxen

10.49

23.83

8.02

7.04

3.50

0.00

0.00

0.00

0.00

0.00


Hired tractor

0.33

0.00

0.00

0.00

1.32

0.00

0.00

0.00

0.00

0.00


Hired truck

0.98

0.00

3.95

0.00

0.13

0.00

0.00

0.00

0.00

0.00


Hired other machinery

0.25

0.00

0.00

0.09

0.09

0.50

0.00

0.00

0.00

1.76


Repair of tractor or machinery

9.05

13.50

5.88

11.71

4.88

4.63

6.25

1.90

6.89

3.74


Cooperative fees

0.58

0.35

1.98

0.01

0.06

0.00

0.00

0.00

0.00

0.00


Other

3.95

14.16

0.68

0.44

0.84

3.47

0.00

0.00

7.53

4.63

Other inputs

152.79

297.34

94.95

113.27

109.04

61.52

137.75

7.76

106.42

16.02


Fertilizer

139.78

283.71

86.63

98.86

93.64

48.50

114.55

0.00

87.28

10.87


Pesticides

0.15

0.18

0.11

0.00

0.31

0.73

0.00

0.00

1.98

0.59


Seeds

4.80

1.52

4.86

3.70

9.14

3.20

2.54

3.84

5.26

1.00


Fuel, oil, and lubricants

0.70

0.57

0.00

0.96

1.21

2.39

11.32

0.00

1.19

0.10


Bags purchased

7.37

11.37

3.34

9.75

4.74

6.70

9.34

3.92

10.71

3.46

Total

219.25

437.70

151.69

157.99

135.70

103.90

182.01

51.39

152.43

52.52

Source: IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Note: Farms were ranked by size, then divided into four groups, each having the same number of farms. The group with the smallest average size is the fourth quartile; that with the largest is the first quartile.

Household Expenditure Analysis

The measure of total consumption expenditure used in the regressions includes the value of all foods grown and consumed by the households. These foods were valued at retail market prices. Estimation followed the procedures outlined in Chapter 3.

The explanatory variables selected for estimation of the budget share equations are presented in Table 37. It is assumed that households with larger farms will have access to larger amounts of homegrown foods. Since the Zj variables are expressed in per capita terms, family size has been included so that the model permits this variable to influence both the intercept and the slope of the individual Engel functions.

Table 35 - Mean annual farm investment expenditures by farm size quartile in the plateau and the valley regions of Eastern Province, Zambia, 1986

Investment

Plateau

Valley


All farms

Farm size quartile

All farms

Farm size quartile



1

2

3

4


1

2

3

4


(kwacha)

Livestock expenditure

18.28

34.30

12.61

23.49

3.07

1.22

2.06

2.22

0.00

0.00


Cattle

15.78

30.74

10.57

22.12

0.00

0.00

0.00

0.00

0.00

0.00


Goats

0.44

1.76

0.00

0.00

0.38

0.07

0.00

0.25

0.00

0.00


Pigs

0.53

1.67

0.00

0.00

1.04

0.00

0.00

0.00

0.00

0.00


Doves

0.02

0.00

0.00

0.00

0.00

0.09

0.00

0.31

0.00

0.00


Poultry

1.26

0.13

2.04

1.38

1.65

1.06

2.06

1.66

0.00

0.00

Machinery expenditure

0.06

0.15

0.01

0.00

0.15

0.00

0.00

0.00

0.00

0.00

Source: IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Note: Farms were ranked by size, then divided into four groups, each having the same number of farms. The group with the smallest average size is the fourth quartile; that with the largest is the first quartile.

Table 36 - Annual consumption expenditure by the average farm household, Eastern Province, Zambia, 1986

Commodity group

Plateau

Valley

Total sample


(kwacha per household)

Food, alcohol, and tobaccoa

2,495.05

3,128.07

2,634.00

Cereals and cereal products

836.13

942.61

859.50

Fruits, vegetables, and legumes

1,030.77

1,398.05

1,111.39

Meat and fish

323.72

520.00

366.81

All other food (nontobacco)

149.56

100.46

138.78

Alcohol

146.84

152.43

148.07

Cigarettes and tobacco

7.86

14.53

9.30

Clothing and footwear

226.87

113.69

202.02

Consumer nondurables

168.34

93.90

152.00

Durables and housing

98.06

48.23

87.12

Transport

50.12

34.83

46.76

Health and education

41.92

45.92

42.80

Social obligations

23.58

34.53

25.98

Total expenditures

3,103.94

3,499.17

3,190.68

Source: IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

aIncludes the value of home-produced foods consumed by the household.

Table 37 - Independent variables included in Zambia regressions

Variable

Unit

Intercept

Kwacha

Reciprocal of per capita expenditure

Kwacha

Log of per capita expenditure

...

Log of family size

Log of people

Log of family size/per capita expenditure

...

Farm size per capita

Hectare

Farm size/total expenditure

...

Number of adult females (over 11 years) as proportion of family size

Percent

Number of women/per capita expenditure

...

Number of adult males (over 11 years) as proportion of family size

Percent

Number of men/per capita expenditure

...

Age of household head

Years

Age of household head/per capita expenditure

...

Education of household head

Years

Education of household head/per capita expenditure

...

Dummy for household head:

...


male = 1; female = 0

...

Dummy for agricultural districts:

...


Chiwizi = 1; otherwise = 0

...


Nkhoka = 1; otherwise = 0

...


Mphata = 1; otherwise = 0

...


Chipili = 1; otherwise = 0

...


Sinda = 1; otherwise = 0

...


Chaweya = 1; otherwise = 0

...


Kasendek = 1; otherwise = 0

...


Makangil = 1; otherwise = 0

...

Cash income from off-farm wage and other sources as proportion of total expenditure

Percent

Eight dummy variables are used to capture the influence of location on household expenditure behavior. These variables summarize the combined effects of differences in infrastructure, distance to nearest town, and other location-specific characteristics. They are delineated on the basis of local government branches. The ordinary least squares (OLS) results were statistically satisfactory, and most of the explanatory variables were significant and of the sign expected.

There are 197 food items and 58 nonfood items included in the survey data. Where relevant, items are also subdivided into those that are locally produced, homegrown, and imported. Although this amount of detail is helpful, some aggregation is desirable for the Engel curve estimation because some commodities are strong substitutes for others, and an expenditure on one is not independent of the other. Also, where expenditure observations are few or the budget share is tiny, individual Engel curves would be difficult to estimate.

All food and nonfood goods and services are classified into 12 basic groups: cereals and cereal products; fruits, vegetables, and legumes; meat and fish; all other food; alcohol; cigarettes and tobacco; clothing and footwear; consumer expendables; durables and housing; transport; health and education; and social obligations.

Table 38 summarizes the expenditure behavior of the average farm household. These results were obtained by evaluating the MBSs and the expenditure elasticities in equations (5) and (7) at the sample mean values for all independent variables.

Together, food, alcohol, and tobacco account for 85 percent of total household expenditures, leaving only a small share of the budget for nonfoods. This is not unusual in poor agricultural regions. Hazell and R (1983), for example, report an ABS for food of 81 percent for farm households in the Gusau region of northern Nigeria. However, the expenditure elasticity for food, alcohol, and tobacco is less than unity, implying that its budget share would decline as total incomes increased. This is also reflected in the MBS; only K75 of an additional K100 of total expenditure would be allocated to food, alcohol, and tobacco, while K25 would go to nonfoods. Clearly, farm sector growth has the potential to strengthen the local demand for nonfoods in the Eastern Province region.

Fruits, vegetables, and legumes account for 35 percent of the ABS and 37 percent of any increment to total expenditure. These shares are unusually large; Hazell and R (1983) report an ABS of only 7.5 percent and an MBS of 8.7 percent for fruits, vegetables, and legumes. In Eastern Province, dominant foods in this subgroup are pumpkins and mangos.

Table 38 - Expenditure behavior of the average farm household, Eastern Province, Zambia, 1986

Commodity and locational group

ABS

MBS

Expenditure elasticity


(percent)


Commodity group


Food, alcohol, and tobacco

84.53

74.55

0.88



Cereals and cereal products

28.96

16.71

0.58



Fruits, vegetables, and legumes

34.74

37.25

1.07



Meat and fish

11.65

11.02

0.95



All other food except alcohol

4.31

4.36

1.01



Alcohol

4.59

5.01

1.10



Cigarettes and tobacco

0.29

0.18

0.64


Clothing and footwear

5.63

8.62

1.53


Consumer expendables

4.63

5.02

1.07


Durables and housing

2.17

4.87

2.25


Transport

1.17

3.28

2.81


Health and education

1.11

1.94

1.74


Social obligations

0.70

1.72

2.45

Locational group


Food



Locally produced and purchased

12.82

13.08

1.02



Homegrown

68.97

58.05

0.84



Imported

2.68

3.42

1.31


Nonfood



Locally produced

2.90

6.74

2.31



Imported

12.56

18.71

1.49


Total nontradables

75.87

66.27

0.87



Food

72.96

59.54

0.82



Nonfood

2.90

6.74

2.31


Total tradables

24.13

33.73

1.40

Source: Estimated from the Working-Leser model in Chapter 3 with data from IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Cereals and cereal products also account for a substantial share of the base budget (29 percent), but their importance declines quickly as incomes rise. The expenditure elasticity is only 0.58, and the MBS (17 percent) is about half the ABS.

All the nonfood groups have expenditure elasticities greater than unity, implying that they would all increase in importance in the budget if incomes rose. The relative increases would be greatest for transport, social obligations, and durables and housing, while the largest absolute increases would be for clothing and footwear, consumer expendables, and durables and housing.

Locational Linkages

To capture the locational linkages inherent in the expenditure data for Zambia, a second classification of all goods and services was undertaken. For Zambia, it was not possible to repeat reliably the classification of goods and services by tradability categories at three levels as was done in the West African cases, where data availability and knowledge of local trade patterns permitted such a direct classification. Rather, an older approach to classifying goods was used, based on what is known about production patterns in the Zambian study areas, with somewhat looser classification at the national level into tradables and nontradables. In Zambia five production groups were defined: homegrown foods, locally produced and purchased foods, imported foods, locally produced nonfoods, and imported nonfoods.

Homegrown foods are defined as all foods that are homegrown or collected from the bush. They include home prepared meal, ground and whole maize, sorghum flour, finger millet flour, rice, sweet potatoes, cassava, potatoes, beans, cowpeas, groundnuts, greengram, pumpkin, cabbage, lettuce, onions, tomatoes, banana, mango, orange, lemon, papaya, beef, buffalo, frog, goat, rabbit, mice, mutton, mole rat, bush pig, pork, chicken, duck, dove, organ meat, eggs, caterpillars, fish, fresh milk, milk powder, cheese, lard, honey, varieties of local beer, soda ash, and other fruits, vegetables, and meat.

Locally produced and purchased foods are taken to be all purchased foods that are produced within the region. They include rice, bread, buns, brown sugar, as well as all the homegrown foods listed above, when they are purchased by households.

Imported foods are not produced within the region. They include cooking oil, white sugar, salt, and when purchased from a shop, butter, margarine, and roller and breakfast meal.

Locally produced nonfoods: Of the nonfoods consumed by the sample households, the following are classified as locally produced: social obligations (ceremonies, bride price, gifts, and payments to relatives), schooling, traditional and modem medical care, repairs, improvements to and construction of houses, transportation, fuel (firewood and charcoal), timber and planks, tailoring, and shoe repair.

Imported nonfoods are cloth and sewing materials, shoes, soap and cleaning powder, razor blades, candles, paraffin, kitchen utensils and glass wear, bicycles, linens, blankets, electrical appliances, stoves, mattresses, watches and clocks, jewelry, cosmetics, stationery, and medicines.

Table 38 shows that 69 percent of the average household's budget is allocated to home-produced foods, 13 percent is spent on other locally produced foods, and 3 percent is spent on locally produced nonfoods. That is, about 85 percent of the total budget is allocated to goods and services produced within the region, and only 15 percent is allocated to regional imports. A very high proportion of any increase in total expenditure also goes to items typically produced within the region; the MBS for all foods and nonfoods produced within the region is 78 percent. This demonstrates strong household demand linkages to the local economy, but linkages that are predominantly of benefit to the farm sector rather than to the local nonfarm economy. Indeed, only 7 percent of the marginal budget is allocated to locally produced nonfoods. But with an elasticity of 2.3, local nonfoods will likely become more important in the budget as incomes increase.

An important implication of these results is that increases in the region's main export crops (maize, groundnuts, and cotton) could, by increasing farm incomes, generate strong growth in the local demand for a wide range of farm products. Many of these - some fruits, vegetables, meats, and fish - are relatively high-value products and, to the extent that their supply could be increased locally, this would lead to additional rounds of increased local farm incomes.

The production classification of goods rather than the tradability classification in many cases still clearly indicates whether incremental expenditures are on demand-constrained goods or not. Household expenditures on imported goods represent a direct leakage from the local economy. To deal with this problem for products that were not clearly imported to Zambia, all major goods and services groups consumed were loosely classified into tradable and nontradable groups, using a definition of tradability roughly equivalent to the national definition in the West African cases.

There is little reason to believe that locally produced nonfood goods are exported from the study region. Many nonfood goods, such as items of clothing or household furnishings, are specifically tailored to local tastes and are not likely to be in great demand in urban areas. It is also unlikely that they could compete in other rural areas because of poor road connections and the probable availability of similar goods not burdened by interregional transportation costs. Most local nonfood expenditures are on services, which are nontradables by definition. Consequently, for the purposes of this chapter, it is assumed that all locally produced nonfood goods and services are nontradables. On the other hand, many foods are tradables, either as regional exports or imports. They include roller meal, breakfast meal, white maize, rice, dry groundnuts, beef, cattle, margarine, butter, cooking oil, white sugar, and salt.

The results of these assumptions (at the bottom of Table 38) show a high ABS and MBS for regional nontradables (76 percent and 66 percent, respectively). In contrast, the ABS was 25 percent and the MBS was 32 percent in Gusau, northern Nigeria (Hazell and R 1983). Such high budget shares imply strong demand linkages to the local economy. However, there is one cautionary note: the MBS for nontradables is less than the ABS, and the expenditure elasticity is only 0.87. This implies that the importance of nontradables in the budget will decline as income increases.

This last result is contrary to the result found in household expenditure studies for Nigeria (in Gusau, for example, Hazell and R (1983) report an expenditure elasticity for nontradables of 1.3) or for Niger in Chapter 5 of this report. However, it is similar to results for Burkina Faso in Chapter 4 and Senegal in Chapter 6. The low elasticity in the Eastern Province of Zambia arises from the dominance of nontradable foods that themselves have low expenditure elasticities. In contrast, at 2.3, the elasticity for nontradable nonfoods is quite high. However, with an MBS of only 7 percent, these demand linkages to the local nonfarm economy will remain relatively small until household incomes rise significantly.

Contrast across Farm Size and Expenditure Deciles

One objective of the expenditure analysis in each of the case studies is to see how changes in income distribution that accompany growth affect the aggregate demand for different goods and services, and particularly how these changes affect the strength of the aggregate demand linkages to the local economy. For this purpose, analysis of the expenditure patterns of households by different income or farm size groups provides especially pertinent results.

Per capita expenditure and farm size were used to classify households into different groups. Surprisingly, these two variables are not correlated (the correlation is only -0.03), so the ensuing two sets of farm classifications are quite different. Tables 39 and 40 present the MBSs for different commodity groups by per capita expenditure deciles and farm size deciles, respectively. To derive these results, all the household characteristic variables were evaluated at their decile means.

Nonfoods become more important as per capita expenditure or farm size increases. The biggest increases occur for clothing and footwear, durables and housing, transport, and social obligations. Many of these are locally produced and, in fact, the MBS for locally produced nonfood increases from 2.7 percent to 8.9 percent between the bottom and top per capita expenditure deciles, and from 6.2 percent to 8.4 percent between the bottom and top farm size deciles. These are still rather small MBSs in terms of supporting much local nonfarm activity. Nevertheless, income increases in the hands of the richer or larger farm households lead to strong consumer demand linkages to the local nonfarm economy. The opposite is true for farm production linkages (Table 34).

Growth Multipliers

The limitations of the fixed-price model, as discussed in Chapter 2, may not be a significant problem in Eastern Province. First, although it is a static equilibrium approach that ignores the growth effects of additional investment, the results in Table 35 suggest that these investment linkages are small in Eastern Province. Second, because the model does not incorporate any explicit specification of the labor market, it does not allow for inelasticity in the supply of nontradables. Although this is a potentially serious limitation in Eastern Province, given a very low population density and seasonal labor bottlenecks, there is evidence of countercyclical seasonal movements of labor between the farm and nonfarm sectors (Table 33). These movements suggest some degree of complementarity between the two sectors in labor use. Third, the model describes a self-contained regional economy and in doing so ignores spillovers to or from major urban areas or to other rural areas in Zambia. Here, as elsewhere, this criticism is correct, but only strengthens results that show intersectoral linkages to be high, even leaving some sources out.

Table 39 - Marginal budget shares by per capita expenditure decile, Eastern Province, Zambia, 1986


Per capita expenditure decile

Item

1st

2nd

3rd

4th

5th

6th

7th

8th

9th

10th


(percent)

Commodity group


Food, alcohol, and tobacco

83.07

78.04

77.27

76.61

74.25

73.43

72.75

70.41

71.17

69.26


Cereals and cereal products

26.31

22.30

22.83

23.17

15.07

15.31

11.47

11.30

11.29

8.51


Fruits, vegetables, and legumes

40.31

37.66

33.82

35.97

38.54

37.31

37.79

35.85

37.96

37.22


Meat and fish

7.34

8.88

9.60

8.37

12.09

10.82

13.77

13.22

12.08

14.20


All other food (except alcohol and tobacco)

5.46

5.04

5.20

4.44

4.06

4.19

4.03

3.87

4.06

3.18


Alcohol

3.68

4.11

5.70

4.55

4.38

5.59

5.46

5.88

5.50

5.68


Cigarettes and tobacco

-0.02

0.05

0.12

0.10

0.12

0.22

0.22

0.29

0.29

0.48


Clothing and footwear

5.33

7.32

7.97

8.62

8.41

9.13

8.60

9.86

10.09

10.47


Consumer expendables

5.33

5.41

4.41

4.61

5.66

5.19

4.41

5.51

4.32

4.98


Durables and housing

2.91

3.98

4.18

4.22

4.56

4.99

5.76

5.26

6.25

6.63


Transport

1.35

2.25

2.57

2.31

3.54

3.62

4.22

4.30

4.05

4.56


Health and education

1.52

2.05

2.13

2.08

1.96

1.71

2.08

2.28

1.79

1.56


Social obligations

0.49

0.94

1.47

1.52

1.62

1.93

2.18

2.37

2.33

2.53

Locational group


Food













Locally produced and purchased

6.04

9.98

12.73

12.06

13.48

14.84

15.31

18.26

16.09

17.96



Homegrown

73.73

64.46

61.17

61.95

57.30

55.47

53.98

48.39

52.15

48.06



Imported

3.30

3.60

3.37

2.60

3.47

3.12

3.46

3.76

2.93

3.24


Nonfood













Locally produced

2.67

4.78

5.90

6.17

6.71

6.95

8.15

8.79

8.38

8.85



Imported

14.26

17.18

16.82

17.22

19.04

19.61

19.11

20.80

20.45

21.88


Total nontradables

76.05

69.46

68.98

69.87

65.35

65.34

63.57

61.81

62.76

60.77



Food

73.38

64.68

63.07

63.69

58.64

58.38

55.42

53.01

54.39

51.91



Nonfood

2.67

4.78

5.90

6.17

6.71

6.95

8.15

8.79

8.38

8.55


Total tradables

23.96

30.54

31.02

30.14

34.66

34.66

36.43

38.19

37.24

39.26



Average farm size (hectares)

2.95

2.15

1.84

2.12

2.74

02.16

2.73

2.20

2.24

2.32



Average family size

9.39

7.62

7.25

6.03

5.76

5.43

5.86

4.53

4.44

3.43



Per capita expenditure (kwacha)

206.35

292.91

362.24

431.61

514.71

594.70

689.85

911.97

988.95

1,328.28

Source: Estimated from the Working-Leser model in Chapter 3 with data from IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Notes: All household characteristic variables are evaluated at decile means. The first decile is the poorest.

Table 40 - Marginal budget shares by farm size decile, Eastern Province, Zambia, 1986

Item

Farm size decile


1st

2nd

3rd

4th

5th

6th

7th

8th

9th

10th


(percent)

Commodity group


Food, alcohol, and tobacco

77.37

77.72

76.52

76.88

75.03

75.59

74.63

74.33

71.37

67.58


Cereals and cereal products

20.40

16.59

15.36

17.30

13.61

19.50

17.64

17.21

15.12

14.63


Fruits, vegetables, and legumes

34.74

40.42

39.76

39.26

40.40

39.64

38.40

34.98

35.39

29.59


Meat and fish

13.15

11.96

13.81

10.84

11.25

7.65

9.43

11.07

9.94

12.07


All other food (except alcohol and tobacco)

3.96

4.02

2.76

4.31

4.42

3.99

4.26

5.06

4.83

5.52


Alcohol

4.69

4.38

4.40

4.98

5.07

4.70

4.79

5.97

6.09

5.83


Cigarettes and tobacco

0.43

0.35

0.43

0.20

0.28

0.11

0.11

0.04

0.00

-0.06


Clothing and footwear

7.61

7.02

7.85

7.71

2.12

8.59

8.87

8.71

9.50

11.38


Consumer expendables

4.24

4.56

4.46

4.97

4.62

5.39

4.91

5.34

5.58

5.30


Durables and housing

4.97

4.77

5.00

4.28

5.19

4.38

5.01

4.08

5.25

6.00


Transport

2.42

2.73

3.02

2.88

3.48

2.84

3.09

3.53

4.23

4.86


Health and education

1.63

1.44

1.21

1.71

1.54

1.81

1.96

2.16

2.34

2.87


Social obligations

1.76

1.76

1.95

1.57

2.02

1.40

1.53

1.85

1.72

2.01

Locational group


Food













Locally produced and purchased

13.23

12.73

14.10

12.72

14.03

11.66

12.09

15.92

14.21

15.88



Homegrown

61.43

61.99

59.73

61.87

57.97

61.89

59.31

54.88

53.34

47.66



Imported

2.71

3.00

2.69

3.29

3.03

3.04

3.23

3.53

3.82

4.04


Nonfood













Locally produced

6.21

6.14

6.50

5.92

6.93

6.20

6.39

7.11

7.49

8.39



Imported

16.41

16.13

16.98

17.20

18.04

18.21

18.98

18.56

21.14

24.02


Total nontradables













Food

66.49

68.04

67.68

68.47

65.85

68.65

67.70

67.04

63.69

61.22



Nonfood

6.21

6.14

6.50

5.92

6.93

6.20

6.39

7.11

7.49

8.39


Total tradables

33.52

31.96

32.32

31.53

34.15

31.35

32.30

32.96

36.31

38.78



Average farm size (hectares)

0.33

0.64

0.84

1.16

1.46

1.79

2.28

3.07

4.12

7.79



Average family size

5.63

5.31

4.79

5.54

4.63

5.57

6.62

6.35

7.57

8.00



Per capita expenditure (kwacha)

641.67

616.89

732.33

546.07

751.24

581.78

567.88

533.90

567.66

686.53

Source: Estimated from the Working-Leser model in Chapter 3 with data from IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Notes: All household characteristic variables are evaluated at decile means; the first decile is the poorest. The marginal budget share for food, alcohol, and tobacco declines across both the per capita expenditure and the farm size, mostly because of a decline in the marginal budget shares for cereals and cereal products and fruits, vegetables, and legumes.

The Model

The model used on the Zambia data is identical to the one in Chapter 3, with two exceptions mandated by the local situation and data availability. First, tradable nonfarm activity is not separately considered. Second, the farm tradables sector is disaggregated for the plateau and the valley. In the terminology of Chapter 3, the four sectors are tradable farm production in the valley (T) and plateau (M) regions; nontradable farm items (A); and nontradable nonfarm items (N).

It is necessary to distinguish between tradable production in the valley and in the plateau in order to capture important differences in the technology used. Farmers in the valley use more traditional technologies and crop varieties, especially for maize, the main tradable crop, whereas many farmers in the plateau use modem maize varieties and fertilizers. Nontradable farm items comprise a mix of livestock products and fruits and vegetables, together with various foods that are gathered or hunted in the bush. They are classified as nontradables because of their perishability and the absence of interregional marketing channels. Their production technology is assumed to be similar in the valley and plateau regions. Nonfarm nontradables consist mostly of agroprocessing, artisan work, and wholesale and retail trading, and encompass both rural and urban activity. The region does not produce any significant amounts of non-farm goods for export.

Model Results

The model's coefficients are estimated using the available survey data described earlier, together with various sources of secondary information. The coefficients are summarized in Table 41 and the results in Table 42. Two scenarios are defined: a base model and a variant for sensitivity analysis in which fruits and vegetables are reclassified as tradables.

The base model has large value-added multipliers of 2.57 for the valley and 2.48 for the plateau. In other words, an additional kwacha (K.1.00) of value added, generated in tradable farm production in the valley through technological change, leads to another K1.57 of value added in the regional economy. Most of the indirect income is generated in the farm nontradables sector, whereas the nonfarm sector only increases its income by K0.20. This is because the region's households allocate the lion's share of their incremental expenditure to nontradable foods (especially fruits and vegetables), and because the farm sector requires relatively few nontradables as intermediate inputs to production. The dominance of the consumer demand linkages is also confirmed by calculating the multipliers under the assumption that all the MBSs are zero. This leads to multipliers of 1.02 and 1.05 in the valley and plateau, respectively.

Table 41 - Semi-input-output parameters for the Zambian study region


Tradable agriculture

Nontradable agriculture

Nontradable nonagriculture

Coefficient

Valley (T)

Plateau (M)

(A)

(N)

Input-output coefficients


Nontradable agriculture (aAi)

0.01

0.01

0.01

0.01


Nontradable nonagriculture (aNi)

0.02

0.03

0.02

0.10


Value added to gross output ratio (vi)

0.90

0.80

0.95

0.80


Rural



Valley (V)


Plateau (P)

Urban (U)

Household coefficients


Marginal budget shares







Nontradable agriculture (bAh)

0.61


0.58

0.44


(0.23)a


(0.22)a

(0.30)a



Nontradable nonagriculture (bNh)

0.073


0.068

0.15


Leakage ratio (sh)

0.05


0.10

0.15


Value-added shares







Nontradable agriculture (VAh)

0.475


0.475

...



Nontradable nonagriculture (VNh)

0.15


0.15

0.50

Source: Estimated from the IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

aAlternative coefficient for a model experiment in which fruits and vegetables are reclassified as tradables.

Table 42 - Regional income multipliers for valley and plateau agriculture


Base solution

Fruits and vegetables reclassified as tradables

Sector

Valley

Plateau

Valley

Plateau

Sector incomes


Tradable agriculture

1.00

1.00

1.00

1.00


Nontradable agriculture (A)

1.37

1.28

0.30

0.29


Nontradable nonagriculture (N)

0.20

0.20

0.10

0.12



Total

2.57

2.48

1.41

1.41

Household incomes


Valley, rural (V)

1.72

0.67

1.17

0.17


Plateau, rural (P)

0.72

1.68

0.17

1.17


Urban (U)

0.13

0.13

0.07

0.07



Total

2.57

2.48

1.41

1.41

Source: Results of the multiplier model in Chapter 3 using data from IFPRI. Rural Development Studies Bureau. Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986.

Note: All figures are the income increase induced by a K1.00 increase in the income of tradable agriculture in either the valley or plateau regions.

In keeping with the small income gains in the nonfood sector, urban households gain relatively little additional income from farm sector growth (K0.13 for each K.1.00 of additional value added in agricultural goods). Nearly all the multiplier gains are captured by the farm households themselves, again because of the importance of farm nontradables in the multiplier.

The large multipliers would have to be scaled down sharply if the supply response of the nonfarm nontradables sector were inelastic. This is shown by the sensitivity analysis in the last two columns of Table 42, which report the multiplier results when fruits and vegetables are reclassified as tradables (and thus are constrained by supply, rather than demand). In this case, the multiplier is the same for the valley and the plateau and, at only 1.41, it is now closer to the multiplier estimates for Sub-Saharan Africa given by Haggblade, Hazell, and Brown (1989) and Haggblade and Hazell (1989). Of the K0.41 of nontradable income generated by each additional K1.00 of income in farm tradables, 30 percent (K0.07) arises in the nonfarm (urban) sector. Consumption linkages to farm nontradables therefore continue to dominate, even though fruits and vegetables (which account for 37 percent of the marginal budget in Table 38) have been reclassified as tradables.

These multiplier results confirm weak linkages between the farm and nonfarm sectors in Eastern Province, both in the plateau and valley regions. As farm households gain more income, they prefer to spend that income on additional foods, particularly horticultural and livestock products. Potentially, this could generate large multipliers within the farm sector itself, but only if the supply of these kinds of perishable foods is elastic. Agricultural research and improved marketing channels could play an important role in fostering the needed supply response.

The dominance of demand for nontradable foods in marginal expenditure in Zambia is similar both to the West African country cases of the present report and to previous work. Hazell and R (1983) report a similar pattern of demand for Gusau in northern Nigeria. They postulate that poor roads and transport systems, together with long distances from villages to towns, discourage farm households from diversifying their consumption into nonfoods. A similar situation probably exists in Eastern Province, in which case stronger farm-nonfarm growth linkages may not emerge until the level of infrastructure has been significantly upgraded.

Analysis of Growth Linkages in Zimbabwe

The household survey data set from Gazaland in Zimbabwe is much more limited than the Zambian data. In the absence of data on consumption of home-produced foods, gifts, in-kind payments, and barter trade, the consumption analysis is restricted to cash expenditures. One attractive feature of the Gazaland data set is that it permits comparison of the cash expenditure behavior of smallholders in the communal areas with that of large-scale commercial farmers in the Middle Sabi and Chipinge areas.

Nonfarm activity is important for the communal farmers; only 60 percent report farming as their primary occupation. Nonfarm occupations include trading, teaching, office work, extension work, driving, and personal services. In comparison, nearly 100 percent of the farmers in Middle Sabi and Chipinge report farming as their primary occupation.

Cash expenditures are available for household consumption and farm inputs (Table 43). These are expressed in three ways: as shares of total consumption or farm cash expenditure, as per capita costs, and as per hectare costs. The expenditure groups are largely self-explanatory. Food and personal services include fruits, vegetables, poultry, meat products, dairy products, tea, and coffee, beer, tobacco, photo services, general provisions, blacksmith, and tinsmith. Consumer durables include household utensils, china and glassware, watches, charcoal braziers, bicycles, wooden furniture, beds, mattresses, and linens.

Table 43 - Comparison of purchasing expenditure behavior of the average household in Zambia, 1986, and Zimbabwe, 1987/88

Item

Zimbabwe

Zambia


Communal

Middle Sabi

Chipinge

Plateau

Valley


(percent)

Consumption


Food and personal services

38

49

29

43

40


Clothing and footwear

22

22

2

22

20


Health and education

10

8

9

3

8


Consumer durables

21

7

4

22

22


Building and construction

3

5

19

1

0


Fuel and energy

3

8

34

3

3


Bus and road transport

2

0

3

6

7


Post and telecommunication

1

1

0

0

0


Total

100

100

100

100

100

Farm


Machinery and implements

21

89

92

...

0


Inputs

79

7

1

100

100


Veterinary and agricultural extension

...

4

7

...

0


Total

100

100

100

100

100


(Zimbabwe dollars)

(kwacha)

Per capita expenditure







Consumption

144.00

895.00

3,425.00

187.00

567.00


Farm

36.00

270.00

1,311.00

54.00

43.00

Per hectare expenditure







Consumption

385.00

64.00

85.00

666.00

762.00


Farm

68.00

9.00

96.00

163.00

53.00

Farm size (hectares)

0.91

158.06

169.18

2.74

0.97

Family size (persons)

5.82

3.64

3.94

6.01

5.94

Source: Estimated from the IFPRI, Rural Development Studies Bureau, Eastern Province Agricultural Development Project, and National Food and Nutrition Commission survey, Eastern Province, Zambia, 1986, and IFPRI, Department of Physical Planning, Government of Zimbabwe Survey, Gazaland, Zimbabwe, 1987/88.

The average smallholder in the communal farming areas has an annual per capita cash expenditure of Z$180 (in Zimbabwean dollars), of which 80 percent is allocated to household consumption. This is considerably less than the cash expenditure of commercial farmers; they spend Z$1,165 per capita in Middle Sabi and Z$4,736 in Chipinge. Again, about 80 percent of total cash expenditure is allocated to household consumption. However, one cannot conclude from this that the commercial farmers have stronger links to the local economy. First, commercial farmers use a lot more land to generate their larger per capita cash expenditures. When expressed on a per hectare basis, the smallholders actually outspend the commercial farmers by a large margin.

Second, the composition of their expenditure patterns is different. Commercial farmers spend much larger shares of their cash farm costs on machinery and implements. They also spend much more on fuels and energy; these account for 37 percent of cash consumption expenditure in Chipinge. All these items are imported into the region.

Food and personal services account for large shares of the cash consumption of all types of farms, but smallholders spend proportionally more on clothing and footwear, durables, and bus and road transport. In addition to fuels and energy, commercial farmers spend larger shares of cash expenditure on building and construction (especially in Chipinge).

These results are insufficient to determine whether the expenditure patterns of smallholders or commercial farmers generate stronger regional income multipliers. Although the necessary data on noncash expenditures are not available for estimating a model of the regional economy, the last two columns of Table 43 show how strikingly similar cash expenditure data for smallholders in Zimbabwe are to data for the valley and plateau farmers in Eastern Province, Zambia. Only the composition of their cash farm costs differ significantly: Zambian smallholders do not incur any expenditures on machinery and implements. On this basis, there is every reason to believe that consumption linkages in Zimbabwe are just as high as in Zambia, and production linkages may be even higher.

Insights from the Southern African Cases

This chapter has shown that the farm-nonfarm growth linkages are surprisingly strong in Eastern Province, Zambia, particularly if fruits and vegetables are counted as nontradables characterized by an elastic supply. Under that assumption, regional growth multipliers are estimated at about 2.5. That is, each K1.00 of additional value added generated in farm tradables as a result of technological change leads to another K1.5 of income in the regional economy. This is about three times the size of the multiplier estimated for typical regions in Sub-Saharan Africa by Haggblade, Hazell, and Brown (1989). The Zambian case shows a surprising similarity to the West African results in this regard.

Because Zambian farmers spend large shares of incremental income on nontradable foods, and because farm production and investment linkages are still very weak, most of the growth multiplier arises within the farm sector itself. Only K0.20 of income (or 13 percent of the multiplier) is generated in the local nonfarm economy. Households mainly producing nonfarm goods in the region's towns also gain little income from the multiplier; only K.0.13 compared with nearly K2.4 for rural households mainly producing farm goods.

These results imply that, at current per capita income levels, farm sector growth will lead to only modest levels of diversification out of farming in the regional economy. However, the farm-nonfarm linkages might be strengthened by (1) investments in rural infrastructure and transport systems that better link the villages and towns, and (2) continued policy reform to create a more enabling economic environment for the region's farmers and nonfarm entrepreneurs.

The strong household demand linkages for farm nontradables could be a powerful force for regional economic growth. This requires, however, that the supplies of many important nontradable foods, especially fruits and vegetables, be elastic. If they are inelastic, then the size of the multiplier shrinks dramatically from 2.5 to 1.4. Agricultural research and improved marketing channels, especially ones that draw more households into market participation, could play ah important role in promoting the needed supply response.