CHAPTER 5. - Southwestern Niger

This chapter explores growth linkages between the farm and nonfarm sectors in Niger's rural economy, an economy characterized by households that conduct a number of highly diversified nonfarm activities. As in many other parts of the semi-arid tropics, the natural resource base in Niger is extremely fragile and rainfall is highly variable across and within years. Yet population densities on arable land are rising. Niger is also subject to a strong commercializing influence from extensive cross-border trade with Nigeria directly and through Benin (Hopkins and Reardon 1993). The combined impact of these forces creates pressure on a variable, at-risk agricultural resource base.

Given these circumstances, nonfarm income plays an important role in Niger in sustaining the livelihoods of rural farmers, assuring household food security, and providing liquidity for productive investments. Nonfarm activity not only provides income alternatives to households that are currently farming, it may alleviate pressure on an already fragile and deteriorating resource base. Therefore, understanding the nature of the nonfarm economy, its role within the rural economy, and the conditions necessary to stimulate its growth is of strategic importance. In particular, given the prevalence of cross-border trade, development policy cannot ignore the impact of major changes in export and import incentives on the rural nonfarm economy.

This chapter addresses these issues by examining the implications of current consumption patterns for growth, using ABSs and MBSs estimated from comprehensive household survey data. After classifying goods and services by sector (farm or non-farm) and degree of tradability, growth multipliers are estimated to explore the impact of an increase in tradables sector income on the demand for nontradables. Results provide evidence of large multipliers and strong nonfarm linkages.

An example of the importance of classifying goods by farm or nonfarm sector, rather than as agricultural or nonagricultural or rural or urban, is provided by comparing the conclusions of the present study with those of a 1989 study by Doan and Lewis. The Doan and Lewis study focused on rural market towns in western Niger to assess the extent to which they could serve as growth poles. Doan and Lewis concluded that the potential for income multipliers in market towns and, by extension, in rural areas of the zone is low, based entirely on demand considerations (they did not address or assume supply conditions). More specifically, they concluded that increases in agricultural incomes are not likely to lead to more diversified economic activities in rural areas, because of the low income base and low effective demand for "nonfarm" (that is, urban) goods among the rural population. Contrary to Doan and Lewis, the present study finds high value-added multipliers with great potential for stimulating more diversified income sources, provided the right kind of initial income stimulus can be achieved. Development strategies need to focus both on creating the initial catalyst for agricultural exports to coastal countries and on increasing the stimulative impact of the income generated from these exports.

Background and Policy Context

Niger is a vast landlocked country with a 1990 population of 7.7 million and per capita GNP of US$320 a year (World Bank 1996). It is one of the poorest countries in the world. Only 12 percent of the country is arable, and nearly 70 percent of the arable land receives no more than 300 to 400 millimeters of rain, the minimum for rainfed cultivation (irrigated land accounts for less than 1 percent of arable land). More than 90 percent of the population of Niger lives either along the Niger river or along the border with Nigeria. Despite its poor resource base, Niger has been largely self-sufficient in cereal production except during drought years.

The economy is dominated by the rural smallholder sector and uranium mining. Agriculture (including livestock) is the largest sector of Niger's economy. In 1990 it employed 90 percent of the country's labor force and accounted for 36 percent of GDP (World Bank 1996). In terms of both production and domestic consumption, millet and sorghum are the principal crops. They account for about 85 percent of total production during the main agricultural season and about 80 percent of national caloric intake. The dominant cropping system involves the intercropping of millet, sorghum, and cowpeas. Livestock (which are now owned primarily by sedentary farmers, rather than pastoralists) are an important component of the overall farming system.

Until the early 1970s, livestock and groundnuts were the main agricultural export products. Groundnuts have since declined in importance, and cowpeas are now Niger's leading cash crop with Nigeria being the major market outlet. Niger's agriculture has stagnated over the last few decades, owing largely to a combination of climatic, economic, and policy factors: recurring droughts, the impact of major policy changes in neighboring Nigeria, the commodity boom effects of the exploitation of uranium deposits in the 1970s and the decline in uranium prices after 1983, and the general impact of an overvalued CFA franc vis-a-vis a devalued Nigerian naira after 1986.

Uranium has been the principal foreign exchange earner since the mid-1970s. Since the collapse of the uranium market in the early 1980s, the government of Niger's stated policy has been to replace lost uranium revenue with an increase in agricultural productivity and exports (1983-87 five-year plan). To move the economy in this direction, the government and its donors embarked on a program of structural adjustment and agricultural price policy reform in 1983. The main goal of this effort, which has met with only partial success, was to reestablish the competitiveness of agricultural tradables in export markets. These principally include livestock and cowpea exports to Nigeria.⁹ The nominal devaluation of the CFA franc (FCFA) that occurred in January 1994 had a major stimulative impact on the domestic prices of exports in Niger, particularly livestock (Institut du Sahel 1997).

⁹It is worth noting in this context that Nigeria's recorded imports of livestock products in 1985, mostly from Europe, exceeded estimates of Niger's total farm sector GDP in the same year (see Delgado 1991).

Study Regions and Sample Characteristics

This study uses detailed household expenditure and income data from a survey conducted in Niger between September 1989 and December 1990. The data were collected for a USAID/Niger-funded project called "Household-Income Generating Strategies and Agricultural Price Policy Impacts in Niger," which was carried out in collaboration with the Institut National de Recherche Agricole du Niger (INRAN) and the ICRISAT Sahelian Center. The objective of the study was to examine the nature and determinants of household income-generating strategies and how they condition the effects of policy changes on production, consumption, and marketing of farm products, as well as investment in farm and nonfarm enterprises.

Fortnightly and monthly interviews enumerated crop and livestock transactions (purchases, sales, gifts), food and nonfood consumption, crop production, and non-farm income. The sample consisted of 135 randomly chosen households from five regions in the Sudano-Sahelian and Sudano-Guinean zones of western Niger. Fifteen villages were chosen to reflect the diversity of the regions in access to markets, infrastructure, size, and so forth. The survey methodology is described in more detail in Hopkins and Reardon 1989.

The regions included in this analysis fall administratively within the Dosso drtement (state) of western Niger and span two agroecological zones - the Sudano-Sahelian zone and the Sudano-Guinean zone. The Sudano-Sahelian zone (the northern and southern Boboye survey regions) has an average annual rainfall of 400-600 millimeters. This zone is moderately poor agroclimatically, with highly variable rainfall both within and between years. Millet, cowpeas, and some groundnuts are produced, with low yields per hectare. Livestock husbandry is widely practiced, but degradation has led to rapid reduction in grazing areas.

The Sudano-Guinean zone has an average annual rainfall of 600-850 millimeters. It is bordered to the south and east by Nigeria and to the west by Benin. This zone is moderately good agroclimatically and considered "high potential," although current performance is modest. The variation in its rainfall from year to year is lower than that of the Sudano-Sahelian zone. Millet, sorghum, some maize, cowpeas, bambara nuts, and groundnuts are produced. Yields are higher and animal traction is used to a greater extent than in the Sudano-Sahelian zone. Land constraints are less severe than in the northern zone. Livestock husbandry is an important part of the income-generating strategies of households.

Average household size, income levels, and asset holdings are fairly similar across agroclimatic zones, although there is variation among individual regions (Table 13). The northernmost region (northern Boboye) has the most degraded resource base and the lowest annual income level (US$103 per adult equivalent), while the southernmost region (Gaya River) has the highest income level (US$153 per adult equivalent).

Rural households in western Niger have incomes that are well diversified into nonfarm sources, even by Sahelian standards (Hopkins and Reardon 1993). Income from activities other than crop and livestock production makes up 52 percent of the average Sudano-Sahelian household's income and 43 percent of the average Sudano-Guinean household's income.

Household income diversification varies in nature by agroecological zone. In the lower-potential Sudano-Sahelian zone of Niger, diversification strategies are outward oriented, with migration playing a large role in household income-generating strategies. Nonfarm activities tend to be better linked to the economies of local towns and to migration than to the local crop and livestock economy. Diversification in this zone helps offset local crop risks and compensates for harvest shortfalls. In the higher-potential Sudano-Guinean zone, which receives higher and more stable rainfall, diversification is inward-oriented. Nonfarm activities tend to be linked to local crop and livestock activities and to proliferate where infrastructure is adequate.

Historically, trade has flourished between the coastal and Sahelian economies. Across Niger's border to the south is an economy with 4 times the number of consumers in all of the Sahel and 15 times the number in Niger. The importance of this cross-border trade, especially in the southern zone, is evident from the substantial share of household sales and purchases that occurred directly across the border during the 1989-90 survey period (Table 13). In the Sudano-Guinean zone, 20 to 40 percent of pulse sales and 30 to 40 percent of livestock sales by households took place directly in Nigerian and Benin markets. In addition, 15 to 30 percent of the zone's cereal is sold directly in cross-border markets. These magnitudes, although not inclusive of all cross-border trade, give an indication of the minimum amount of cross-border trade in the sample zones and in Niger's rural economy more broadly (Hopkins and Reardon 1992).

Classification of Household Expenditures

The detailed nature of the Niger data and the authors' knowledge of the commodities, households, and regions where the data were collected permit a rigorous and accurate categorization of individual goods. For example, millet flour (a nonfarm nontradable) is distinguished from unthreshed (with bran) millet (a farm tradable) and from millet cakes (a nonfarm nontradable). Likewise, traditionally processed peanut oil (a nonfarm nontradable) is distinguished from imported palm oil (a nonfarm tradable). Similarly, locally crafted nonfarm tradables, such as palm-frond woven mats, are distinguished from locally crafted nonfarm nontradables, such as calebasses (gourds used for utensils). Expenditures on 200 individual food items and 750 nonfood goods and services are classified: (1) by functional (commodity) category, (2) by sector (farm or nonfarm), and (3) by tradability.

Table 13 - Selected zone, sample, and household characteristics, Dosso, Niger, 1989/90

			Sudano-Sahelian zone		Sudano-Guinean zone
Characteristics			Northern Boboye	Southern Boboye	Dallol Maouri	Gaya Plateau	Gaya River
Long-term annual rainfall
	(millimeters)		450-500	500-550	600-700	700-750	750-800
Sample size (number of households)			23	23	23	20	24
Household size
	Average number of persons		7.2	9.0	8.7	7.6	8.7
	Average number of adult equivalents		5.4	6.7	6.7	6.0	6.5
Househood income
	Average income level
		(US$ per adult equivalent)	103	150	110	123	157
	Nonfarm income (percent)		36	70	57	34	43
Household assets
	Land (hectares)		5.7	6.3	6.4	5.6	5.0
	Land cultivated per adult equivalent (hectares)		1.1	0.9	1.0	0.9	0.8
Household cross-border purchases as share of total purchases (percent)
	Cereals		2	1	31	1	15
	Pulses		0	0	3	0	0
	Livestock		0	0	26	0	53
Household cross-border sales as share of total sales (percent)
	Cereals		0	0	3	0	18
	Pulses		0	0	19	3	42
	Livestock		0	0	40	0	31

Source: Hopkins and Reardon 1993.

Functional Classification

Food and nonfood goods were then classified into 6 broad functional categories: crops and crop products, livestock and livestock products, stimulants and beverages, consumption nondurables, consumption durables, and services. These groups are further subdivided into 20 disaggregated categories:

1. Crop and crop products are divided into 8 subcategories: millet, sorghum, and fonio (a local grain); maize; rice; pulses; processed staples; by-products; vegetables; and other foods.

2. Livestock and livestock products are divided into 2 subcategories: meat, fish, poultry, and eggs, and fresh milk and butter. Meat includes both live animals (small ruminants and fowl) purchased to slaughter for immediate consumption, as well as cooked and uncooked meat from a local butcher.

3. Stimulants and beverages are not subdivided further. This category includes cola nuts, cigarettes, tobacco, coffee, tea, and soft drinks.

4. Consumption nondurables (frequent purchases) are divided into 3 subcategories: fuelwood and kerosene, toiletries and soap, and miscellaneous (matches, batteries, and medicine such as aspirin or Nivaquine).

5. Consumption durables (infrequent purchases) are divided into 3 subcategories: clothing, furniture and kitchen utensils, and miscellaneous (for example, jewelry, flashlights, kerosene lamps, bedding material, and local construction materials).

6. Services are divided into 3 subcategories: social obligations, transportation (for example, cart or taxi fares), and miscellaneous (for example, repair services, utility [water and wood carrying] services, healer or mystic services, and cereal milling services).

Farm-Nonfarm Classification

Earlier growth linkages work in Africa tended to equate locally produced food with the farm sector, following Asian precedents and the tendency to distinguish sectors by location, that is, urban versus rural (King and Byerlee 1977; Haggblade and Hazell 1989; Doan and Lewis 1989). The farm or nonfarm distinction is perhaps more relevant for rural African economies since it allows processed food items (for example, processed cereals such as flour, breads, and cakes, processed vegetables, and processed meat such as butchered, dried, smoked, and grilled meat) to be appropriately placed in the nonfarm sector. Given that food processing is one nonfarm activity that is expected to grow most rapidly during a structural transformation (Haggblade, Hazell, and Brown 1987; Hazell and Haggblade 1989), it is particularly important to make this distinction.

Consumption durables (kitchen utensils, furniture, and clothing), nondurables (fuelwood, kerosene, and soap), and services are classified as nonfarm goods and services. In addition, food items that originate off-farm (while using farm sector inputs) are also classified as being in the nonfarm sector (flour, cakes, breads, tomato paste, canned milk, cooked tubers, spices, grilled meat, and smoked fish). This is true regardless of whether they are imported or of local origin. Goods that originate on the farm (unprocessed cereals and pulses, fresh vegetables and fruits, by-products, and live animals) are classified as farm sector goods.

Tradability Assumptions

For western Niger, the external reference market at the national level of tradability (the national catchment) is the West African regional market, particularly Nigeria. As in other chapters, the local catchment area refers to tradability within the immediate geographic region (roughly a 100 kilometer radius from the study villages). Nontradables at the local level are hypothesized to be traded freely within this catchment, but are not traded outside it. Similarly, nontradables with respect to world markets are assumed to circulate freely within the West African regional catchment.

The detailed nature of the Niger data base and the authors' knowledge of the country allow a different technique to be used in this chapter to get around the problem of aggregating tradables and nontradables. Each of the 950 individual consumption items in the survey, as opposed to categories of goods and services, are classified as tradable or nontradable. Thus, a given category of goods (for example, millet, sorghum, and fonio) can be comprised of both tradable and nontradable items, with shares attributed to each. Table 14 presents the share of nontradable expenditures in each category of goods.

For example, at the national market level, only 5 percent of expenditures in the millet, sorghum, and fonio category were on nontradable items (fonio), while 95 percent were on tradable items. Millet and sorghum, although clearly nontradables on the world market, were considered tradable at the national level of tradability because of frequent commerce with Nigeria, which served as a much larger reference market. Pulses such as cowpeas, groundnuts, and bambara nuts are tradable within the local catchment area, but only groundnuts and cowpeas are tradable at the national level, and only groundnuts at the regional level. Hence the share of nontradable pulse expenditures increases from 0 to 58 percent as the market reference area increases.

Table 14 - Share of nontradables in consumption expenditures by commodity group for alternative tradability assumptions, Dosso, Niger, 1989/90

		Catchment area
Commodity group		Local	National	Regional
		(percent)
Crop and crop products		11	19	84
	Millet, sorghum, and fonio	5	5	100
	Maize	0	0	0
	Rice	0	0	0
	Pulses	0	21	58
	Processed staples	53	70	100
	By-products	45	99	99
	Vegetables and spices	26	63	74
	Other	12	34	65
Stimulants and beverages		1	1	70
Livestock and livestock products		88	90	100
	Meat, fish, poultry, and eggs	86	88	100
	Fresh milk and butter	100	100	100
Consumption nondurables		0	5	45
	Fuelwood and kerosene	0	14	14
	Toiletries and soap	0	0	100
	Other nondurables	0	0	12
Consumption durables		12	18	66
	Clothing	0	0	66
	Furniture and kitchen utensils	27	29	73
	Other durables	24	47	47
Services		100	100	100
	Social obligations	100	100	100
	Transport	100	100	100
	Other	100	100	100

Source: IFPRI/INRAN survey data, 1989/90.

Notes: Since each category of goods is the aggregate of a large number of individual items, the numbers in this table represent the percent of each category that is classified as nontradable. For example, the processed staple category includes processed coarse grains (millet flour and millet cakes), processed pulses (cowpea cakes), processed tubers (manioc flour and boiled yams), and bread. Although these items are in the same commodity grouping, some of the items are tradable (bread and manioc flour) while others are nontradable (millet cakes and boiled tubers). As another example, the pulses category includes cowpeas, groundnuts, and bambara nuts. All are tradable at the local level; only groundnuts and cow-peas are tradable at the national level, and only groundnuts are tradable at the regional (world market) level. For an explanation of the individual items included in each category of goods, see the discussion in the text.

Estimation of Multiplier Model Parameters

Following the methodology described in Chapter 3, the fixed-price value-added multiplier model requires estimation of the following parameters: MBS, value-added to gross output ratios, intermediate delivery coefficients, and the marginal savings ratio. However, the sensitivity analysis conducted in Chapter 4 indicates that the multipliers depend overwhelmingly on the tradability assumptions and the MBS. This section discusses how each group of parameters was obtained.

Marginal Budget Shares

MBSs are computed using parameters estimated through the Working-Leser model of consumption described in Chapter 3 (see equation 4). The explanatory variables used in estimation of the Working-Leser equations are presented in Table 15. Household size and composition (age and gender distribution) are key variables influencing household expenditure patterns. An important factor affecting household expenditure patterns is the liquidity of the household. Two measures of liquidity are incorporated into the model. Net total credit as a share of total expenditures reflects the ability of households to obtain credit from moneylenders and shopkeepers.¹⁰ Earned nonfarm income as a share of total income by contrast reflects the ability of households to generate cash income from nonfarm sources.

¹⁰Total credit is the sum of net consumption and net production credit and includes both cash and in-kind credit. Total credit was used because consumption and production credit are highly fungible.

In rural African households, consumption from own production is a large share of household expenditures. The effect of the household's ability to feed itself for the duration of the year is captured by including the production sufficiency ratio (PSR) as an explanatory variable. The PSR indicates the ability of a household to use the factors of production at its command, such as farm size and land quality, to meet the Food and Agriculture Organization of the United Nations' (FAO's) minimum caloric requirements of 2,280 kilocalories per adult equivalent per day.

The market village dummy variable is included to capture the effect of higher levels of infrastructure (for example, a weekly market) on consumption. Lastly, dummy variables for each region and agroclimatic zone are incorporated to reflect geographic differences, such as overall level of infrastructure and proximity to borders, and agroclimatic differences (in soils and rainfall, for example).

Table 15 - Independent variables included in the Engel function regressions, Niger

Description	Name	Unit
Intercept	INTERCEPT	FCFA
Reciprocal of per capita total expenditure	I/TEXPPC	FCFA
Log of per capita total expenditure	TEXPPCLN
Household size	POP	Persons
Household size/per capita expenditure	POP_XC
Number of adult females (over 15) as portion of family size	SHAF	Percent
SHAF/per capita expenditure	SAF_XC
Number of female children (5-14 years) as portion of family size	SHEF	Percent
SHEF/per capita expenditure	SEH_XC
Number of infants (under 5) as portion of family size	SHI	Percent
SHI/per capita expenditure	SI_XC
Dummy variable for market village: market village = 1; nonmarket village = 0	MKT2
Market village/per capita expenditure	MKT2_XC
Production sufficiency ratio	PSR	Percent
PSR/per capita expenditure	PSC_XC
Total credit as share of total expenditures	STCX	Percent
STCX/per capita expenditure	STCX_XC
Nonfarm earned income as a share of total income	SH_NONFE	Percent
SH_NONFE/per capita expenditure	SHFE_XC
Dummy variable for region two: Region 2 = 1; otherwise = 0	DUMREG2
DUMREG2/per capita expenditure	REG2_XC
Dummy variable for region four: Region 4 = 1; otherwise = 0	DUMREG4
DUMREG4/per capita expenditure	REG4_XC
Dummy variable for region five: Region 5=1; otherwise = 0	DUMREG5
DUMREG5/per capita expenditure	REG5_XC
Dummy variable for market group: Sudano-Guinean = 1; Sudano-Sahelian = 0	DUM_MG3
DUM_MG3/per capita expenditure	MG_XC

Note: Region 2 is Southern Boboye; region 4 is Gaya Platea; region 5 is Gaya River. The region 1 dummy (Northern Boboye) was not significant. Region 3 (Dallol Maouri) is represented by the intercept.

Technical Parameters

Value added and intermediate delivery parameters to the farm sector were calculated using the detailed input/output farm budget data from the IFPRI/INRAN Niger survey. Parameters for the nonfarm sector were based on data from a SAM for Niger (Dorosh and Nssah 1991).¹¹ The savings ratio is the one derived from the Burkina Faso data and considered a reasonable estimate for the Sahel. The technical parameters used in the model are given in Table 16. The numbers reflect the fact that more hired labor was used in crop production in Niger than in Burkina Faso, for example, as evidenced by the higher coefficients for nonfarm, nontradable intermediate deliveries in Niger.

¹¹The sensitivity analysis of the estimated multipliers to parameter assumptions in Chapter 4 found that a 1 percent change in intermediate demand coefficients was only associated with a 0.1 percent change in the multiplier. Given the range within which multipliers are interpreted, it can be assumed that this simplification has little effect on results.

Table 16 - Technical parameters used in growth multiplier calculations, Dosso, Niger, 1989/90

	Farm sector		Nonfarm sector
Coefficients	Tradable	Nontradable	Tradable	Nontradable
Intermediate deliveries from nontradable farm sector	0.00	0.12	0.06	0,03
Intermediate deliveries from nontradable nonfarm sector	0.09	0.12	0.10	0.20
Value-added shares	0.51	0.75	0.49	0.69
Savings ratio	0.06	0.06	0.06	0.06

Sources: For the farm sector, intermediate delivery and value-added parameters were derived from IFPRI/INRAN input-output farm budget data, 1989/90. For the nonfarm sector, these parameters were calculated using secondary data from a social accounting matrix for Niger (Dorosh and Nssah 1991). The savings ratio is based on Burkina Faso estimates (see Chapter 4 for details).

Note: Regional definitions of tradability were used to compute the above coefficients.

Household Expenditure Patterns

Growth linkages analysis largely concerns the strategic implications of consumption patterns. Once the tradability of goods and services is established, the estimated MBSs by sector are central to the multipliers obtained. Interpreting results often requires a more disaggregated view of consumption response to income changes than categories such as "farm nontradables," for example, allow. Therefore the present section will discuss expenditure patterns in considerable detail, disaggregated by the goods and services group, agroecological zone, and income group.

Table 17 summarizes the expenditure behavior of rural Niger's households for the overall sample average and by agroecological zone. ABSs provide a view of the relative magnitude of different products in the household's current budget. Marginal expenditure shares, on the other hand, provide a view of how households will allocate increments to income. In both cases, they are expressed in percentages and sum to 100 across goods and services groups.

Overall Expenditure Patterns

Food¹² accounts for nearly three-quarters of the average expenditure of sample households. Although, as Engel's Law predicts, the budget share of food decreases as incomes increase, nearly 60 percent of any additional expenditure will still be spent on food, while 40 percent will be allocated to nonfood goods and services. The importance of food in both average and incremental expenditures indicates that the capacity of the farm sector to respond to increased demand as incomes rise will be a key factor in the success of rural growth strategies.

¹²Food includes crop and livestock products in both processed and unprocessed form.

Table 17 - Expenditure patterns for Sudano-Sahelian and Sudano-Guinean zones, Dosso, Niger, 1989/90

		Sudano-Sahelian zone			Sudano-Guinean zone			Overall sample
Commodity group		ABS	MBS	Expenditure elasticity	ABS	MBS	Expenditure elasticity	ABS	MBS	Expenditure elasticity
		(percent)
Crop and crop products		64.45	40.81	0.63	60.80	48.42	0.80	62.29	46.27	0.74
	Millet, sorghum, and fonio	39.79	14.16	0.36	38.61	24.45	0.63	39.09	17.69	0.45
	Maize	5.03	9.05	.80	3.05	1.40	0.46	3.86	4.71	1.22
	Rice and wheat	1.34	2.87	2.14	0.88	0.37	0.42	1.07	1.04	0.98
	Pulses	2.76	-1.50	-0.54	6.07	9.10	1.50	4.72	8.15	1.73
	Processed staples	4.61	3.56	0.77	4.05	2.47	0.61	4.28	3.50	0.82
	By-products	0.89	1.10	1.23	0.66	0.98	1.49	0.75	1.04	1.38
	Vegetables	7.68	8.12	1.06	6.12	7.07	1.16	6.75	7.69	1.14
	Other	2.35	3.44	1.46	1.37	2.58	1.89	1.77	2.45	1.39
Stimulants and beverages		2.62	4.12	1.58	5.65	5.02	0.89	4.42	3.70	0.84
Livestock and livestock products		9.47	15.19	1.60	8.50	15.53	1.83	8.90	13.20	1.48
	Meat, fish, poultry, and eggs	7.26	10.84	1.49	7.33	13.41	1.83	7.30	10.13	1.39
	Fresh milk and butter	2.21	4.36	1.98	1.18	2.12	1.81	1.60	3.07	1.92
Consumption nondurables		6.54	7.53	1.15	5.50	4.67	0.85	5.92	6.59	1.11
	Fuelwood and kerosene	2.14	2.60	1.21	1.25	1.82	1.46	1.61	2.08	1.29
	Toiletries and soap	2.26	1.98	0.88	1.83	1.44	0.79	2.00	2.08	1.04
	Other nondurables	2.14	2.95	1.38	2.43	1.37	0.57	2.31	2.43	1.05
Consumption durables		6.35	16.19	2.55	9.24	15.77	1.71	8.06	13.95	1.73
	Clothing	3.28	8.15	2.48	6.04	10.28	1.70	4.92	8.02	1.63
	Furniture and kitchen utensils	1.15	1.30	1.12	1.70	2.22	1.31	1.48	2.63	1.78
	Other durables	1.92	6.75	3.52	1.50	3.27	2.18	1.67	3.30	1.98
Services		10.58	16.15	1.53	10.31	10.59	1.03	10.42	16.28	1.56
	Social obligations	3.88	8.09	2.08	4.81	7.29	1.52	4,43	10.16	2.29
	Transport	2.68	3.86	1.44	2.12	2.11	0.99	2.35	2.85	1.22
	Other	4.02	4.21	1.05	3.38	1.20	0.35	3.64	3.27	0.90

Source: IFPRI/INRAN survey data, 1989/90.
Notes: ABS is average budget share; MBS is marginal budget share.

Locally produced coarse grains (millet, sorghum, and fonio) account for the largest single share of both average expenditures (39 percent) and incremental expenditures (18 percent). This suggests a strong rural demand for locally produced coarse grains as incomes rise. Livestock and livestock products account for 9 percent of total household expenditures, with 13 percent of any increase in income spent on these products. Of this 13 percent, 10 percent of incremental income will be spent on meat and 3 percent on milk.

There are a number of specific food items whose budget shares will rise with increases in income. They include in ascending order of their MBSs: meat, pulses, vegetables (including onions, fresh and dried gumbo, fresh tomatoes, tomato concentrate, cooked leaves, and some spices), maize, milk, other foods (for example, fruits, tubers, oils, canned and powdered milk, and sugar), and by-products (for example, cowpea fodder and millet stalks). To the extent that these food items are nontradable, rural growth can be further stimulated through an income increase in the farm tradables sector that is spent on them. In value terms, 85 percent of total expenditures on meat are on nontradables at the local level (that is, they represent expenditures on raw and cooked meat rather than live animals for slaughter), 100 percent of total expenditures on milk are on nontradables, 45 percent of expenditures on by-products, and 26 percent of expenditures on vegetables and spices.

Nonfood goods account for 14 percent of total expenditures and services account for 10 percent. Spending on these broad categories will increase as incomes increase - 25 percent of additional income will be spent on nonfood goods, while 16 percent of any increase in income will be spent on services. Thus, farm sector growth has the potential to increase the demand for nonfood goods and services. Again, whether this translates to a stimulus for growth in the local economy depends on the tradability of the nonfood items demanded. To the extent that the increase in income is spent on local nontradables (for example, locally made furniture and kitchen utensils, baskets, and all services), the local nonfarm economy will get a boost from an increase in farm income. About 25 percent of furniture and kitchen utensils and miscellaneous durables are categorized as nontradables at the local level.

A larger share of nonfarm goods are classified as nontradables with respect to world markets. For example, 100 percent of all toiletries and soap, 66 percent of clothing (regionally fabricated cloth), 73 percent of furniture and kitchen utensils (metal beds, plates, and pots), and 47 percent of other durables (metal trunks and sheets) are all nontradables. Most of these products are locally manufactured in neighboring coastal countries, especially Nigeria. Thus, a boost to the farm tradables sector in Niger, due to devaluation of the FCFA for example, is likely to increase demand for cross-border products manufactured in Nigeria on the income side, although the price effect will be to discourage consumption of imports (Hopkins and Reardon 1992).

Expenditure Patterns by Agroecological Zone

Disaggregation of sample expenditure patterns and multiplier analysis by agroecological zones helps identify differences in the type of multiplier growth that can be achieved in each of them and those sectors that are most likely to be stimulated by the spending of increased household incomes.

Coarse grains (millet, sorghum, fonio, and maize) account for 45 percent of the Sudano-Sahelian household budget and 42 percent of the Sudano-Guinean budget (Table 17). MBSs are also similar across zones - 23 versus 25 percent. However, the composition of the incremental coarse grain budget (between the locally produced and the imported coarse grains) differs greatly. The MBS for millet, sorghum, and fonio is 70 percent greater in the higher potential Sudano-Guinean zone, compared with the Sudano-Sahelian zone. Nearly a quarter of any increase in income will be spent on millet, sorghum, and fonio in the Sudano-Guinean zone - only 1 percent will be spent on maize.

In the Sudano-Sahelian zone, on the other hand, only 14 percent of any increase in income will be spent on locally produced coarse grains; 9 percent will be spent on maize. Virtually no maize is produced in this zone. The demand for maize is filled largely by imports from Benin, Ghana, and Nigeria. Maize produced in the Sudano-Guinean zone (largely in the Fadama areas along the Niger river) is mostly for home consumption; little enters commercial channels.

Sudano-Guinean households have larger coarse grain stocks from own production than do Sudano-Sahelian households.¹³ During the hungry season, when local coarse grain stocks are low and relative millet prices are at their highest levels, maize accounts for nearly a quarter of total cereal calories in the Sudano-Sahelian zone (Hopkins and Reardon 1993). The households in the Sudano-Sahelian zone substitute commercially available imported maize for locally produced coarse grains to fill their cereal needs during the hungry season.

¹³ For the 1989 and 1990 harvest years, households in the Sudano-Guinean zone, on average, had a three-months greater supply of grain from own production in stock than did farmers in the Sudano-Sahelian zone.

Maize is often substituted for local coarse grains in western Niger if its price is low enough relative to millet and sorghum. This lends support to the classification of millet and sorghum as tradables in western Niger at the local and national levels. Furthermore, if upward pressure is placed on millet and sorghum prices, as a result of increased demand induced by linkages, maize imports from Benin, Ghana, and Nigeria will increase.

Expenditure patterns (average and marginal) for livestock and livestock products are similar across agroecological zones (the ABS is 9 percent and the MBS is 15 percent). In the Sudano-Sahelian zone, livestock and livestock products have the same relative importance as locally produced coarse grains. Income growth will put considerable pressure on local milk and meat supplies.

Expenditure Patterns by Income Tercile

When expenditure patterns are summarized across income terciles (Table 18), food, as expected, represents a smaller share of total expenditures in the top third of sample households than in the bottom third of households (two-thirds versus three-quarters).

Table 18 - Spending patterns for lower and upper expenditure terciles, Dosso, Niger, 1989/90

		Lower expenditure tercile			Upper expenditure tercile			Overall sample
Commodity group		ABS	MBS	Expenditure elasticity	ABS	MBS	Expenditure elasticity	ABS	MBS	Expenditure elasticity
		(percent)
Crop and crop products		66.08	51.31	0.78	58.34	39.56	0.68	62.29	46.27	0.74
	Millet, sorghum, and fonio	44.13	19.72	0.45	34.36	13.90	0.40	39.09	17.69	0.45
	Maize	3.51	5.72	1.63	3.78	3.28	0.87	3.86	4.71	1.22
	Rice and wheat	1.02	2.01	1.97	1.25	-0.13	-0.11	1.07	1.04	0.98
	Pulses	4.32	10.83	2.51	4.39	5.15	1.17	4.72	8.15	1.73
	Processed staples	4.30	2.85	0.66	4.48	4.77	1.06	4.28	3.50	0.82
	By-products	0.70	1.28	1.83	0.97	0.78	0.80	0.75	1.04	1.38
	Vegetables	6.52	6.57	1.01	7.00	9.17	1.31	6.75	7.69	1.14
	Other	1.58	2.32	1.47	2.11	2.64	1.25	1.77	2.45	1.39
Stimulants and beverages		4.18	-1.66	-0.40	4.96	9.02	1.82	4.42	3.70	0.84
Livestock and livestock products		8.51	10.75	1.26	9.56	16.93	1.77	8.90	13.20	1.48
	Meat, fish, poultry, and eggs	7.33	7.86	1.07	7.45	13.08	1.75	7.30	10.13	1.39
	Fresh milk and butter	1.18	2.87	2.44	2.11	3.85	1.83	1.60	3.07	1.92
Consumption nondurables		5.55	6.56	1.18	6.28	7.08	1.13	5.92	6.59	1.11
	Fuelwood and kerosene	1.44	1.75	1.21	1.83	2.36	1.29	1.61	2.08	1.29
	Toiletries and soap	2.10	1.74	0.83	1.96	2.71	1.38	2.00	2.08	1.04
	Other nondurables	2.00	3.07	1.53	2.50	2.01	0.80	2.31	2.43	1.05
Consumption durables		6.49	12.62	1.94	9.52	15.21	1.60	8.06	13.95	1.73
	Clothing	3.95	7.30	1.85	5.65	8.86	1.57	4.92	8.02	1.63
	Furniture and kitchen utensils	1.19	2.71	2.28	1.55	2.49	1.60	1.48	2.63	1.78
	Other durables	1.35	2.62	1.94	2.32	3.86	1.66	1.67	3.30	1.98
Services		9.20	20.44	2.22	11.33	12.21	1.08	10.42	16.28	1.56
	Social obligations	3.91	12.78	3.27	5.10	7.84	1.54	4.43	10.16	2.29
	Transport	1.52	4.01	2.64	2.41	1.51	0.63	2.35	2.85	1.22
	Other	3.78	3.65	0.97	3.83	2.86	0.75	3.64	3.27	0.90

Source: IFPRI/INRAN survey data, 1989/90.

Note: ABS is average budget shares; MBS is marginal budget shares. Expenditure terciles are used as proxies for the lower, middle, and upper thirds of the income distribution. They are based on total household expenditures per adult equivalent, including income in kind valued at market prices per adult equivalent.

In terms of increments to income, the poor spend 62 percent of additional income on food and the rich, 56 percent. Although these MBSs are quite similar, the composition of the incremental food basket is different. The rich will spend more of any additional income on livestock products (17 versus 11 percent) while the poor will spend more on crop and crop products (51 versus 40 percent).

The poor allocate a larger share of their budget to locally produced coarse grains than the rich do (45 versus 34 percent). And, the poor spend more of any increments to income on local coarse grains than the rich do (20 versus 14 percent). The ABSs for maize, on the other hand, are similar between lower and upper tercile households (4 percent), but the MBSs are two times greater in the lower tercile households (6 versus 3 percent). Thus, millet and sorghum are preferred coarse grains in the diets of rural people in Niger; expenditure elasticities for maize decrease rapidly with increases in income, whereas millet and sorghum expenditure elasticities are constant across income terciles. At 1.6, the expenditure elasticity for maize is highly elastic in lower-tercile households and barely inelastic (0.9) in upper-tercile households, while the expenditure elasticity for millet and sorghum are both near 0.4.¹⁴

¹⁴ Also Chapter 4 on "Food Consumption" in Hopkins and Reardon 1993 for additional support of this point.

The ABS for livestock and livestock products is similar across terciles (9 percent) but the MBS is 55 percent greater in the upper tercile (17 versus 11 percent). Relatively wealthier households will spend more of any increments to income on livestock products than local coarse grains (17 versus 14 percent), whereas poorer households will spend a higher share on local coarse grains than on livestock products (20 versus 11 percent).

Compared with poorer households, wealthier households have slightly higher ABSs (16 versus 12 percent) and MBSs (22 versus 19 percent) for nonfood goods. The upper income tercile also spends 11 percent of its income on services, while the lower tercile spends 9 percent. Interestingly, and contrary to expectations, the poor will spend a much higher proportion of any increment to income (20 percent) on services than the rich (only 12 percent). To put this in context, the poor will spend the same share of any increment to income on services as on locally produced coarse grains (20 percent).

The largest category of service expenditures are "social obligations." This category includes contributions made to other households for various ceremonies (such as baptisms or marriages) but does not include religious expenditures, which are placed in the "other" category.

One possible explanation for the high share of incremental income that the poor allocate to services, given that expenditures on services are largely items related to the creation or alleviation of social obligations, is the "social security" factor. Poorer households tend to be more dependant on the goodwill of others, and therefore they spend a larger share of increments to income to cement social relations. In the higher-potential Sudano-Guinean zone, the MBS for services is greater for the rich than for the poor (13 versus 5 percent), while in the lower-potential Sudano-Sahelian zone, the MBS for services is greater for the poor (21 versus 13 percent). The finding that growth multipliers are larger for lower-tercile households is in part driven by the magnitude of the MBS for services in poor households.

Expenditure Patterns by Sector

The tradability assumptions underlying Table 14 can be combined with the expenditure pattern data summarized in Tables 17 and 18 to estimate the shares of total household income and increments to household income that are allocated to each of four sectors: farm tradables, farm nontradables, nonfarm tradables, and nonfarm nontradables. These are basic data for assessing whether household income growth will stimulate production of demand-constrained goods (nontradables), and whether growth will be on the farm or nonfarm side.

As can be seen in Table 19, ABSs and MBSs for nontradables as a group are higher in the Sudano-Sahelian zone, whereas expenditures on tradables are higher in the higher potential Sudano-Guinean zone, with its proximity to the border. Thus, Sudano-Sahelian households consume more locally produced nonfood goods as incomes rise, while households in the Sudano-Guinean zone consume more imported goods. The sample zone with the easiest access to Nigerian markets (the Dallol Maouri Sudano-Guinean study region) has the lowest MBS for nontradables (41 percent at the preferred national definition of tradability), while the Sudano-Sahelian region further from the border has a higher share (45 percent). This illustrates the impact of proximity to the border. The MBS is substantially higher in the Northern Boboye subsample (not shown), the farthest region from the border.

When disaggregated by income terciles, demand for nontradables as a group is income-elastic (MBS > ABS), but significantly more so for the poor than for the rich at the national level of tradability. In Table 18, demand for services by the poor plays a preponderant role in explaining this result. Various farm nontradables such as certain pulses and by-products also have a higher income elasticity of demand. Using the alternative assumptions for a regional level of tradability (with world markets), demand for nontradables is income inelastic for both rich and poor, largely because Niger's millet and sorghum, which account for large expenditure shares, are nontradables with respect to the world market.

Growth Multipliers

In the summary of growth multipliers in Table 20, the numbers represent the total net additions to average household income in dollars that result from an initial income increase of $1.00 in the tradable farm or nonfarm sectors. The values depend on which of the mutually incompatible definitions of tradability is chosen: local, national, or regional. In keeping with the discussion in Chapter 3, the appropriate set of assumptions is embodied in the definition using the national level of tradability. "Local" and "regional" results are offered primarily to show the sensitivity of results to the set of assumptions chosen, particularly elasticity of supply. Sources of economic growth can be decomposed and attributed to new spending on demand-constrained, nontradable goods and services (including new intermediate demands) in either the farm or the nonfarm sector.

Table 19 - Average and marginal budget shares by sector for overall sample and sample subgroups, Dosso, Niger, 1989/90

Catchment area and sector			Overall sample		Sudano-Sahelian zone		Sudano-Guinean zone		Lower expenditure tercile		Upper expenditure tercile
			ABS	MBS	ABS	MBS	ABS	MBS	ABS	MBS	ABS	MBS
			(percent)
Local
	Tradables
		Farm	51	32	54	31	49	37	55	37	46	24
		Nonfarm	24	29	21	31	26	29	22	20	26	38
		Total	75	61	75	62	75	66	77	57	72	62
	Nontradables
		Farm	7	13	7	13	8	14	7	11	9	16
		Nonfarm	18	26	18	25	18	20	16	32	19	22
		Total	25	39	25	38	26	34	23	43	28	38
National
	Tradables
		Farm	48	28	51	29	46	34	53	34	44	20
		Nonfarm	20	25	17	26	23	25	19	17	22	33
		Total	68	53	68	55	69	59	72	51	66	53
	Nontradables
		Farm	10	17	10	15	10	18	9	14	11	20
		Nonfarm	21	30	22	30	21	23	19	35	23	26
		Total	31	47	32	45	31	41	28	59	34	46
Regional
	Tradables
		Farm	8	13	9	13	8	9	7	16	9	10
		Nonfarm	10	12	9	14	10	13	8	7	11	17
		Total	18	25	18	27	18	22	15	23	20	27
	Nontradables
		Farm	50	32	52	31	48	42	55	32	46	30
		Nonfarm	32	43	30	42	34	35	30	45	35	42
		Total	82	75	82	73	82	77	85	77	81	72

Source: IFPRI/INRAN survey data, 1989/90.

Note: ABS is average budget share; MBS is marginal budget share. Expenditure terciles are used as proxies for the lower, middle, and upper thirds of the income distribution. They are based on total household expenditures per adult equivalent, including income in kind valued at market prices.

Table 20 - Growth multipliers and decomposition of multipliers under alternative tradability assumptions for initial income shocks to tradable farm and nonfarm sectors, Dosso, Niger, 1989/90

		Tradable farm			Tradable nonfarm
Sample subgroup		Local	National	Regional	Local	National	Regional
Overall sample
	Tradables	1.00	1.00	1.00	1.00	1.00	1.00
	Farm nontradables	0.21	0.29	0.90	0.24	0.32	0.95
	Nonfarm nontradables	0.56	0.67	1.44	0.60	0.72	1.51
	Total multiplier	1.77	1.96	3.34	1.84	2.03	3.47
Sudano-Sahelian zone
	Tradables	1.00	1.00	1.00	1.00	1.00	1.00
	Farm nontradables	0.20	0.25	0.83	0.22	0.28	0.88
	Nonfarm nontradables	0.54	0.66	1.36	0.58	0.70	1.43
	Total multiplier	1.73	1.91	3.18	1.80	1.98	3.31
Sudano-Guinean zone
	Tradables	1.00	1.00	1.00	1.00	1.00	1.00
	Farm nontradables	0.21	0.27	1.28	0.23	0.30	1.35
	Nonfarm nontradables	0.45	0.53	1.38	0.48	0.57	1.45
	Total multiplier	1.65	1.80	3.67	1.72	1,87	3.81
Lower expenditure tercile
	Tradables	1.00	1.00	1.00	1.00	1.00	1.00
	Farm nontradables	0.19	0.26	0.98	0.22	0.29	1.04
	Nonfarm nontradables	0.65	0.76	1.58	0.70	0.81	1.66
	Total multiplier	1.84	2.03	3.57	1.91	2.11	3.70
Upper expenditure tercile
	Tradables	1.00	1.00	1.00	1.00	1.00	1.00
	Farm nontradables	0.25	0.34	0.82	0.27	0.37	0.87
	Nonfarm nontradables	0.49	0.62	1.35	0.53	0.66	1.43
	Total multiplier	1.74	1.96	3.18	1.81	2.02	3.30

Source: IFPRI/INRAN survey data, 1989/90.

For example, the second set of four numbers in the first column in the upper lefthand comer of Table 20 (the Sudano-Sahelian Zone) should be interpreted as follows: assuming local tradability, a $1.00 increase in household incomes from an outside event affecting tradables will lead to $0.20 of additional income from spending on farm nontradables, and to $0.54 of additional income from spending on nonfarm nontradables. Thus, the total multiplier is 1.73, and the net extra growth from spending on demand-constrained items (that is, from growth linkages) is $0.73.

Elastic supply of the nontradable items demanded is a reasonable assumption for the study zones, although less so in a few specific cases. Milk, for example, might be a case where the relative price would need to increase substantially and to stay high over a considerable period of time in rural Niger before producers would find a way of meeting increased demand through increased production.

Six points emerge from Table 20. First, regardless of the tradability assumption used, multipliers for Niger - as for the other countries - are larger than previous African linkage studies have shown. The national reference market definition yields a multiplier of 1.96.

Second, linkages with the nonfarm economy appear stronger than previous studies suggest, and in fact stronger than other countries in the present study. For western Niger overall, the additional income generated in the nonfarm sector is 2 to 2.5 times that of the farm sector. The additional income accruing to the nonfarm sector from a $1.00 shock to the farm tradables sector is $0.54, while that to the farm nontradables sector is only $0.20. Thus, of the $0.73 of indirect gain, 74 percent is generated in the nonfarm economy. Third, in western Niger, multipliers are larger for the poorest one-third of households than for the richest one-third (5 percent larger using the national definition).

Fourth, farm-nonfarm linkages appear to be stronger for poorer households: income increments to lower-income households stimulate the nonfarm sector more than income shocks to the upper-income households. A $ 1.00 increase in income from farm tradables for lower income households generates $0.76 of additional income in the nonfarm sector, versus only $0.62 for upper-income households. This result is driven by the fact that the poor spend $0.35 of each additional dollar on nonfarm nontradables, as opposed to $0.26 spent by the rich (Table 19). The richest one-third of households spend $0.33 of each additional $1.00 on nonfarm tradables, while the poor spend only $0.17. The consumption preference of the rich for tradable items represents a leakage from the local rural economy from the standpoint of net income generation.

Fifth, the size of the multiplier is driven largely by the consumption spending patterns of rural households. Consumption linkages account for 79 percent of multipliers in Niger. Sixth, multipliers for increased income in the nonfarm tradables sector, as would be the case if Niger exported rural handicrafts, are of the same order of magnitude as those for increments to income in the farm sector. The multipliers from stimulating farm or nonfarm income are not significantly different. However, the ability to promote sustained growth in the tradables sector, which is essential to jump-starting sustained growth overall, will depend on the comparative advantage of farm tradables versus nonfarm tradables, an issue not investigated in this report.

Conclusions and Policy Implications

Two strategic questions are addressed in this chapter. First, what should the thrust of policy be in order to maximize and sustain economic growth made possible by policy reforms that promote farm tradables in rural areas? Second, what policies can promote the development of nonfarm incomes and employment in western Niger, thereby alleviating some of the pressure on a fragile natural resource base to supply livelihoods from farming? There are five sets of conclusions of direct relevance to these two questions.

First, the consistently high farm multipliers under different assumptions (zone, tradability, and income group) suggest that rural demand-led growth is feasible under the variety of alternative hypotheses about supply and demand constraints embodied in the three alternative sets of assumptions about tradability. A considerable amount of extra growth can be achieved by boosting rural incomes, stimulating demand for nontradable goods, and bringing underemployed resources into production. In most cases the additional growth from linkages to the nontradables sector was at least as large as the initial income stimulus from the tradables sector.

Second, development strategies that boost rural incomes broadly, by putting money in the hands of many rural consumers, will have a large overall impact on growth, especially compared to policies that put money in the hands of a few large producers. The analysis indicates that 79 percent of additional growth from spending new income from tradables on demand-constrained items (that is, growth linkages) is attributable to consumption demand, and only 21 percent to intermediate demands for nontradable inputs used in production. The conclusion is that widespread stimulus to demand-constrained sectors will be primarily from the consumption side.

This insight is further supported by the multiplier analysis by income tercile, which shows that income in the hands of the poorest third of households stimulates more overall growth (has a higher multiplier), and more growth in the nonfarm sector (more of the multiplier comes from the nonfarm sector), than is the case for the richest one-third of households.

Third, the research shows that increases in farm income are an efficient way to stimulate growth in the rural nonfarm sector. Increases in cash crop or livestock incomes, for example, will lead to at least as much growth again in things like transportation services, processed food items, local handicrafts, and local nontradable foods. Two-thirds of the additional growth from an initial income stimulus is in the nonfarm sector.

Alternatives for promoting rural nonfarm employment are unlikely to achieve sustained or widespread success. The nonfarm tradables sector is not negligible, with an ABS of 20 percent and an MBS of 25 percent (assuming national tradability). Yet, few areas in the study zone appear to have a comparative advantage in exporting non-farm items; a few niche activities such as processed foods, salt extraction, and woven mats cannot provide a widespread boost to income. Most nonfarm tradables consumed in the zone are importables (clothing, kitchen utensils, assorted other consumer durables, fuel, toiletries, and other nondurables). Policies directed specifically toward producing these items locally are likely to fail.

However, nontradable nonfarm items, such as services, some processed foods, and local utensils and furniture, account for 47 percent of incremental household spending. These items cannot be competitively imported to the zone (or exported from it); stimulating widespread local demand for them will lead to widespread local growth in these activities, provided barriers to entry do not prevent local people from expanding production.

The conclusion is fundamental: finding a way to boost rural household incomes broadly and in a sustained fashion (that is, automatically, year after year) is the way to stimulate growth in nonfarm employment. Direct support to nontradable, nonfarm enterprises, in the absence of a sustained market for the product driven by another income source, cannot create growth in this sector. Since most viable nonfarm production activities are nontradables, they should be approached first on the demand side and then on the supply side.

Fourth, nonfarm linkages will amplify rural growth, but only the tradable (supply-constrained) sectors can serve as the stimulus. Which tradable activities to promote on the supply side is primarily a matter of comparative advantage. Although not explored in this study, the need for widespread income growth points to the farm tradables sector. Livestock, hides, onions, and pulses, for example, are the only exports widely produced in the study zones at present. To stimulate growth in the nontradables sector, new income must be continually infused from outside the zone, and attention must be paid to improving the elasticity of supply of the items demanded.

Calculations of returns to promoting the supply side of tradable activities should take into account not only the direct, but also the indirect benefits from linkages-induced growth. As the results indicate, these may be at least as large as the direct benefits themselves. Interventions on the supply side would include agricultural research, extension, and infrastructure.

Fifth, rural growth strategies also require enhancement of the supply-responsiveness of the goods and services demanded as rural incomes rise. Increases in local demand for tradables are (somewhat heroically, in the case of Niger) assumed to be automatically met by increased imports (or reduced exports) of the item in question, at a price determined by markets outside the zone of analysis. Increases in local demand for nontradables, the driving force of linkages, only serve to boost local incomes if they are translated directly into increased local production of the items in question.

Policy plays a big role both in ensuring that newly demanded tradables are in fact freely available in local markets, through lowering costs of distribution, and in directly facilitating the supply-response of local production of nontradables. Commodity groups with high MBSs such as millet and sorghum, milk, and meat, require particular attention. Since 13 percent of incremental income in rural areas is spent on livestock products, increased attention should be focused on the livestock sector to enable it to meet both growing domestic demand as incomes rise and export opportunities, as devaluation of the CFA makes the sector more competitive.

Government should pay particular attention to maize, an importable. Widespread income growth in rural Niger would put considerable pressure on local grain supplies, as shown by the high MBS for coarse grains. In zones such as the Sudano-Sahel, income growth will lead to demand for maize imports, at the same time that Niger's recent devaluation will make maize imports considerably more expensive. The resulting upward pressure on prices will stimulate local coarse grain production further. Unless the local supply of these grains is elastic, their price will rise relative to exportables, cutting into the profitability of the latter.