Household strategies and rural livelihood diversification.

I. INTRODUCTION

The orthodox view of processes of economic change is that they involve transitions between states of economic structure or performance that can be distinguished from each other. An important component of this view is that sectors are compartmentalised. Rural is contrasted with urban, agriculture is contrasted with industry, and people are interpreted as making complete transitions from one type of activity to another as development proceeds.(1) This view is buttressed by the not always explicitly stated idea that specialisation and division of labour are essential ingredients for the transformation of national economies.(2)

Diversification as an individual or household level survival strategy does not fit well into the conventional picture. Many of its attributes stand in contrast to straitened notions of sectors, specialisation and transition. The received wisdom is that diversification is merely a transient phenomenon [e.g., Saith, 1992] or one associated with the desperate struggle for survival in declining economies. Yet diversification may not be so transient, and it may be associated with success at achieving livelihood security under improving economic conditions as well as with livelihood distress in deteriorating conditions [Collier, 1988; Preston, 1989].

This article undertakes a review of ideas, propositions and policy inferences surrounding diversification as a livelihood strategy of rural families in developing countries. A number of recent articles in development studies journals have examined aspects of diversification, utilising case study evidence [Reardon et al., 1992; Chandrasekhar, 1993; Adams, 1994; Bigsten, 1996; Dercon and Krishnan, 1996; Taylor and Wyatt, 1996; Carter, 1997], or have put forward more general interpretations of rural change in which diversification is regarded as a central feature [Bernstein et al., 1992; Bryceson, 1996; Heyer, 1996]. In addition, several researchers have undertaken comparative reviews of empirical evidence on rural household income portfolios, derived from household surveys [von Braun and Pandya-Lorch, 1991; Sahn and Sarris, 1991; Sahn, 1994; Reardon, 1997].

Rural livelihood diversification cuts across a number of typically self-bounded arenas of policy discussion in development studies including rural poverty [Jazairy et al., 1992], household risk strategies [Carter, 1997], household coping strategies [Davies, 1996]; intrahousehold relations [Hart, 1995], rural growth linkages [Hazell and Haggblade, 1993], rural non-farm activity [Fisher et al., 1997], and rural-urban migration [Stark, 1991]. While overlaps occur between these arenas, they each tend to bring rather partial insights to bear on the causes, opportunities, effects and policy implications of diversification.

The fragmentation of insights into diversification means that the literature abounds with conflicting propositions about its meaning. Diversification may occur both as a deliberate household strategy [Stark, 1991] or as an involuntary response to crisis [Davies, 1996]. It is found both to diminish [Adams, 1994] and to accentuate [Evans and Ngau, 1991] rural inequality. It can act both as a safety valve for the rural poor [Zoomers and Kleinpenning, 1996] and as a means of accumulation for the rural rich [Hart, 1994]. It can benefit farm investment and productivity [Carter, 1997] or impoverish agriculture by withdrawing critical resources [Low, 1986].

It is not the intention of this article to attempt to resolve conflicting interpretations of diversification. There is no need to do so because the causes and consequences of diversification are differentiated in practice by location, assets, income, opportunity and social relations; and it is not therefore surprising that these manifest themselves in different ways under differing circumstances. Nevertheless a useful purpose is served by disentangling different strands in the arguments and evidence concerning diversification. In many sub-Saharan African countries structural adjustment has initiated the widespread dismantling of the sectoral layer of state intervention (for example, agricultural policies) that used to act as an interface between the macro-economy and rural household decisionmaking. With the phasing out of these so-called meso-level policies, the only way of connecting macroeconomic policies with micro-level interventions is through a better understanding of household strategies [Berry, 1986; Lipton and Ravallion, 1995].

The article has several aims. The first aim is to pull together into a single discussion a moderately inclusive cross-section of the literature that examines aspects of diversification as a central or major theme. The let-out clause 'moderately inclusive' is inserted here because livelihood diversification crosses so many boundaries of development discussion that its pursuit at the margins could make its treatment quite intractable.(1) The second aim is to set out the different and sometimes contradictory interpretations of livelihood diversification in the areas of its determinants and its implications for poverty, income distribution, agricultural performance and gender relations. The third is to summarise the implications for policy that have been advanced in studies of livelihood diversification.

It is appropriate to remark in passing that livelihood diversification is neither just a rural nor only a developing country phenomenon. It is well-documented as a survival strategy of urban dwellers in developing countries [Baker and Pedersen, 1992; Maxwell, 1995; Rakodi, 1995; de Haan, 1997; Moser, 1998]; and is becoming increasingly prevalent amongst farm families in developed countries as agricultural price and other supports to farming are removed [Benjamin, 1994; Kelly and Ilbery, 1995; Hearn et al., 1996]. It is also an emerging feature of labour markets more generally in the industrialised countries, being associated with the rise of part-time and home-based working patterns. This article restricts its scope, however; to diversification in the rural economy of developing countries, and particularly to the experience of sub-Saharan Africa on which much of the recent literature is focused.

The article proceeds in the following way. The next section elaborates the concept of livelihood diversification, its social dimensions, and the income categories that are utilised in economic studies of income portfolios. The third section briefly reviews the empirical investigation of livelihood diversification, including sources that contain evidence on rural income portfolios in sub-Saharan Africa in the 1980s and 1990s. The fourth section considers approaches to, and explanations of, the determinants of livelihood diversification. The fifth, sixth, and seventh sections examine, respectively, the relationships of livelihood diversification to poverty and income distribution; farm technology and agricultural performance; and gender dimensions. The eighth section briefly reviews some policy implications of diversification.

II. CONCEPTS OF LIVELIHOOD DIVERSIFICATION

In this article livelihood diversification is defined as the process by which rural families construct a diverse portfolio of activities and social support capabilities in their struggle for survival and in order to improve their standards of living.

Several components of this definition merit elaboration. First, a livelihood is more than just income [Lipton and Maxwell, 1992]. Income refers to the cash earnings of the household plus payments in kind that can be valued at market prices. The cash earnings component of income include items like crop or livestock sales, wages, rents, and remittances. The in-kind component of income refers to consumption of own farm produce, payments in kind (for example, in food), and transfers or exchanges of consumption items that occur between households in rural communities.

A livelihood encompasses income, both cash and in kind, as well as the social institutions (kin, family, compound, village and so on), gender relations, and property rights required to support and to sustain a given standard of living. Social and kinship networks are important for facilitating and sustaining diverse income portfolios [Berry, 1989; 1993: Ch. 7; Hart, 1995; Bryceson, 1996]. Social institutions are also critical for interpreting the constraints and options of individuals and families distinguished by gender, income, wealth, access and assets. For example, differential access rights to land are often the key determinant of distinct livelihood strategies pursued by poor compared to better-off rural households. Likewise, social proscriptions on permissible courses of action of women can make big differences to the livelihood options available for women compared to men [Dwyer and Bruce, 1988; Davies and Hossain, 1997].

A livelihood also includes access to, and benefits derived from, social and public services provided by the state such as education, health services, roads, water supplies and so on [Lipton and van der Gaag, 1993; Blackwood and Lynch, 1994]. An oft-stated finding of poverty research is the tendency for public service provision to be biased towards the better off and more accessible locations, communities, and social groups, thus exacerbating the material deprivation already experienced by the poor as a result of inadequate levels of assets and income [World Bank, 1990; Lipton, 1996].

Livelihood diversification is therefore not synonymous with income diversification.(4) Nevertheless, many; but not all, economic studies of diversification focus on different income sources and their relationship to income levels, income distribution, assets, farm output and other variables [e.g., Reardon et al., 1992; Adams and He, 1995]. The term income diversity refers to the composition of household incomes at a given instant in time; diversification, on the other hand, interprets this as an active social process whereby households are observed to engage in increasingly intricate portfolios of activities over' time. Lack of comparable evidence across intervals of time means that it is rarely possible to state firmly whether household livelihoods are more diverse now than they were, say, ten or twenty years ago [Heyer, 1996; Reardon, 1997]; nevertheless, and certainly for sub-Saharan Africa, there seems to be an informed consensus that diversity has been increasing in recent history [Bryceson, 1996], and a few intertemporal case studies seem to bear out this contention [Valentine, 1993].

Economic studies distinguish between several different categories and sub-categories of income source when referring to diverse income portfolios. Such distinctions are not arbitrary; they refer to different labour markets with different features of seasonality, sustainability, barriers to entry, location, potential income growth and so on [Reardon, 1997]. Different individuals and families are likely to possess different potential access to different income sources, and, therefore, participation in these sources will have different impacts on poverty and income distribution.

The primary categories are farm, off-farm, and non-farm income sources [Saith, 1992]. Farm income includes livestock as well as crop income and comprises both consumption-in-kind of own farm output and cash income from output sold.(5) Off-farm income typically refers to wage or exchange labour on other farms (i.e. within agriculture). It also includes labour payments in kind, such as the harvest share systems and other non-wage labour contracts that remain prevalent in many parts of the developing world.(6) Non-farm income refers to non-agricultural income sources. Several secondary categories of non-farm income are commonly identified. These are (i) non-farm rural wage employment, (ii) non-farm rural self-employment,(7) (iii) property income (rents, etc.), (iv) urban-to-rural remittances arising from within national boundaries, and (v) international remittances arising from cross-border and overseas migration.

Not all investigators follow the same conventions for categorising income sources. In a detailed study in rural Pakistan, for example, Adams and He [1995] have the six major income categories of agricultural, non-farm, livestock, rental, domestic remittances and international remittances. In this scheme, wage labour in agriculture is included with own-farm income in the agricultural category; livestock is taken separately from crop income; and non-farm income covers wages, salaries and self-employment outside agriculture, but excludes rents and remittances that are assigned their own categories.(8)

The social and economic characteristics of non-farm income sources tend to be examined in separate literatures. Thus the rural non-farm literature [e.g., Saith, 1992; Fisher et al., 1997] has different preoccupations from the migration and remittance literature [e.g., Stark, 1991]. The former is concerned with stimulating non-agricultural activities in rural areas with a variety of social and economic goals in view. The latter tends to be preoccupied with the motives underlying migration, the rate of urbanisation, and the amount and effects of the remittances sent home.

Most research into income diversification utilises the household as the unit for empirical investigation. The household may be conceived as the social group which resides in the same place, shares the same meals, and makes joint or coordinated decisions over resource allocation and income pooling [Meillassoux, 1981; Ellis, 1993: Ch. 1]. Alternatively, some version of the idea that the household represents a coalition of players committed by choice or custom to act as one unit vis-a-vis the rest of the world can be invoked [Stark, 1991; Preston, 1994]. There are many instances where these definitions fail to capture important attributes of individual and collective welfare, including spheres of individual decision-making, power relations in the social unit, and constraints on permissible courses of action by gender [Folbre, 1986; Guyer and Peters, 1987; Bruce, 1989].

The farm household economic model treats the household as a single decision-making unit maximising its welfare subject to a range of income-earning opportunities, and a set of resource constraints [Hymer and Resnick, 1969; Nakajima, 1970; 1986; Singh et al., 1986; Ellis, 1993]. This yields a baseline economic proposition, ignoring social institutions and risk factors, that the household will allocate its labour time so that the marginal returns per unit of labour are the same across different activities, whether on-farm, off-farm or non-farm. Intra-household economic approaches based on bargaining theory do not necessarily yield different predictions about patterns of engagement by household members in different labour markets [Haddad et al., 1997: Ch. 1].

Many researchers recognise that the study of livelihood diversification requires a more spatially extended understanding of the household than the conventional definition. The role of non-resident family members in contributing to the well-being of the resident group requires explicit recognition. Households with members working away in urban centres or abroad are often referred to as 'split families', and their livelihood strategies are described as 'straddling' the rural and urban sectors [Rempel and Lobdell, 1978; Murray, 1981; Stichter, 1982; Heyer, 1996]. Urban migrants are commonly observed to continue to maintain strong rural family connections, even after several generations of urban residence [Lucas and Stark, 1985; Stark and Bloom, 1985; Valentine, 1993; Hoddinott, 1994]. Circular migration, in which family members work for periods in the urban economy then return to their family farms, is noted in several studies [Bigsten, 1988; 1996; Lageman, 1989; Andrae, 1992]. Seasonal migration related to cyclical work opportunities in different locations is also common [Agarwal, 1990; Breman, 1996].

The extended definition of the household can be used to capture diversity of social interactions as well as diversity of location. The household or family social unit has been described as a 'multipurpose unit' [Bryceson, 1996], the members of which have 'multiple social identities' [Bangura, 1994], and the livelihoods of which involve continual processes of renegotiation and redefinition of family, gender, power, and property relations [Berry, 1993; 1997; Hart, 1995].

Reviewing the literature, there is a temptation to draw up typologies of diversification. In particular, some branches of the literature are predominantly concerned with diversification as a matter of survival, emphasising the reasons for diversification born of desperation (poverty, lack of assets, vulnerability, disaster). Other branches of the literature focus, by contrast, on diversification as a matter of choice and opportunity, involving proactive household strategies for improving living standards. Along these lines, for example, 'diversification for survival' has been contrasted with 'diversification for accumulation' [Hart, 1994: 48].

While such typologies are superficially attractive as a means of distinguishing different motives and opportunities to diversify, they also lend themselves to inaccurate and misplaced generality that fails to engage with the complexity of local circumstances. For example, in some societies income-generating options for women are more socially circumscribed in better-off than in poorer households. More generally, diversification obeys a continuum of causes and motivations that vary across families at a particular …