Articles

Working Papers 53 - Nonfarm work and fertilizer use among smallholder farmers in Kenya : A cross-crop comparison

Working Papers 53 - Nonfarm work and fertilizer use among smallholder farmers in Kenya : A cross-crop comparison

Authors : Melinda Smale, Mary K. Mathenge, and Joseph Opiyo

Introduction :

As a consequence of economic and environmental change across rural African communities, nonfarm work contributes a growing share of the household income—especially among smallholder farm families who struggle to survive on diminishing farm sizes with declining land quality. More than a decade ago, Bryceson (2000) reported case study findings of nonfarm earnings that ranged from 55% to 80% of household income. Considering farm surveys conducted in 23 countries during the 1990s and 2000s, Reardon, Stamoulis, and Pingali (2007) reported that nonfarm income represented an average of 34% of rural household income.

 Multiple factors have contributed to this dynamic situation, many of which are context-specific. Barrett, Reardon, and Webb (2001) differentiated them as ‘pull’ and ‘push’ factors. Bezu, Barrett, and Holden (2012) examined the relationship between nonfarm employment and the social mobility of rural households in Ethiopia, concluding that income growth is positively associated with the nonfarm share of income. In the Oromia region of Ethiopia, Van den Berg and Kumbi (2006) found that land-poor households are pushed into nonfarm activities, reducing income inequality. According to Mathenge and Tschirley (forthcoming 2015), smallholder farmers engage in off-farm work as a long-term strategy to deal with anticipated weather risks. In Mozambique, Cunguara, Langyintuo, and Darnhofer (2011) concluded that nonfarm work is a coping strategy for farm households when faced with drought, concurring that poorer households were more likely to engage in less remunerative activities. In Western Kenya, Djurfeldt (2012) finds that while lack of nonfarm earnings aggravates the seasonal variability of income among poorer households, wealthier households utilize these earnings to meet both farm and nonfarm expenditures.

 Recently, comparing longitudinal data among eight African countries, Djurfeldt and Djurfeldt (2014) concluded that “distress-driven” diversification out of agriculture into nonfarm activities appears to have slowed, with households moving in and out of farm and nonfarm work in response to economic incentives. Optimistically, they portray a complementarity among grain productivity, crop diversification on farms, and nonfarm opportunities. By contrast, in a detailed study of land resource use and rural livelihoods in Western Kenya,  Mutoko, Hein, and Shisanya (2014) found that, despite major differences across farm types, land productivity is generally low, intensification lacking, and household reliance on off-farm income rising.

 In this paper, we focus on the decision to intensify crop production through applying inorganic fertilizer. Past research has often demonstrated a negative relationship between off-farm work and investment in agricultural production (Ahituv and Kimhi 2002; Chikwama 2004; Morera and Gladwin 2006; Davis et al. 2009; Davis, Carletto, and Winters 2010); by contrast, Lamb (2003) found off-farm work and input use on crops to be complementary. Soil fertility is a binding constraint to crop productivity in most regions of Sub-Saharan Africa, and there is a general consensus that raising productivity will require at least some inorganic fertilizer in addition to other soil amendments (Bationo 2004). In their study of Western Kenya, Marenya and Barrett (2007) have shown that nonfarm income positively affected the adoption of integrated soil fertility management practices (including mineral fertilizer, stover lines, and manure). Among inputs that enhance soil fertility, cash constraints are thought to be particularly severe for fertilizer, but these depend on credit availability. Recently, Mathenge, Smale, and Tschirley (2014) estimated input demand for fertilizer and hybrid seed in maize production in Kenya. They found that greater earnings from nonfarm sources detracted from use of these inputs, especially in areas with greater productivity potential.

In addition, recognizing that credit sources depend very much on the value chain, we hypothesize a priori that the effects of nonfarm earnings on fertilizer use depend on crop type and nature of nonfarm work. Nonfarm income may provide the means to purchase fertilizer with cash, overcoming the problem of absent credit markets. For example, no formal credit services are provided directly for maize production in Kenya. At the same time, the engagement of household members in nonfarm activities, including informal business and migration to towns for salaried work can divert labor resources from agricultural activities and peak period tasks. In comparison to maize, which is a staple food crop, traditional cash crops such as tea and some export-oriented vegetables have vertically-integrated supply chains in which credit services are bundled with inputs and marketing arrangements (Minot and Ngigi 2010; Maura and Muku 2007).

 Our analysis builds on the work of Mathenge, Smale, and Tschirley (2014) by comparing input demands for fertilizer among crop categories. We compare the role of nonfarm work among three categories of crops: a major food staple (maize), an emerging cash crop (vegetables), and a traditional export crop (tea). We also disaggregate nonfarm income in order to examine differences between the role of informal business as compared to salaried and wage employment.

 We are able to exploit data collected from a panel of 1200 smallholder farm households distributed across the major agricultural zones of Kenya in four waves that span a decade (2000 through 2010). To accommodate the censored structure of the fertilizer application in the case of maize, while controlling for potential endogeneity, we apply an instrumented Control Function Approach (CFA).We employ the Correlated Random Effects (CRE) model to handle unobservable heterogeneity. We use Fixed Effects, two-stage least squares (FE2SLS) in the cases of vegetables and tea, which are continuous variables, and as a robustness check in the case of maize. We define fertilizer application in terms of nitrogen nutrient kilograms per hectare (N kgs per ha). N nutrient kgs is calculated by the percentage content represented by N in the type of fertilizer used. Farmers observe the physical kgs of mineral fertilizer they apply, but most Kenyan farmers today also know the nutrient content of the type they use.

 Next, we summarize the conceptual basis that serves to guide our econometric approach. We then describe the methods, including the data source, econometric model, and operational variables. Results are presented in the fourth section and conclusions drawn in the final section.    

WPS 53/2015  -  Nonfarm work and fertilizer use among smallholder farmers in Kenya: A cross-crop comparison