Rainfall shocks and crop productivity in Zambia: Implication for agricultural water risk management

Highlights

• Water related problems significantly undermine agriculture productivity in Zambia.

• The impacts of erratic rainfall on productivity varies depending on the type of crop.

• The differential impacts of rainfall shocks calls for crop specific strategies.

Abstract

This paper investigates the impact of erratic rainfall and related water problems on agricultural productivity. The paper also aims to shed light on the conceptual importance of understanding the incidence and impacts of rainfall shocks for choosing feasible agricultural water risk management strategies both at household and policy levels. To achieve these goals we develop a conceptual framework, use national representative data from Zambia’s crop estimates survey for 2017/2018 farming season, employ fixed effects regression approach, and find that dry spells, excessive floods, incidence of water logging are all detrimental to crop productivity. The crop-based equations also reveal the differential impacts of the rainfall shocks on different crops. Since the effect of water factors including dry spells, floods and water logging on agricultural productivity is dependent on the crop types, it is important for the Zambian government as well as other countries to take this into account when planning and implementing strategies for agricultural water risk management.

Introduction

Agriculture plays a key role in the Zambia’s economy and its share in total GDP is currently around 7%. It is the primary source of livelihoods for many people in the country. In 2012 the sector, which comprises agriculture, forestry and fishing employed more than 50% of the population although the size of the employed population came down to about 26% in 2017 after aligning the definition of the labor force to the 19th International Conference of Labor Statisticians definitions of 2013. Of the 26%, nearly 60% of the employed people in agriculture are in rural areas (Central Statistics Office, 2019). Despite the reduction in the share of employed population, agriculture still remains the highest employer. This is not surprising because the proportion of people dependent on agriculture is even more and stands at more than 85% for the entire sub-Saharan region (The World Bank, 2015). As more than 50% of people that are dependent on agriculture live in rural areas and are very poor, it becomes imperative to improve agricultural incomes, and/or agricultural productivity in order to significantly improve the livelihoods of many people in Zambia and other countries in similar situations in Africa (Binswanger and Townsend, 2000). Evidence based policy planning and implementation aimed at understanding drivers of agricultural performance can improve productivity and lead to livelihoods positive change (Fan, 2009).

The Comprehensive African Agriculture Development Program (CAADP), the Malabo Declaration and indeed the Agenda 2064 of the African Union (to which Zambia is a party) all seek to ensure African economic development as a means for securing better livelihoods for the poorest many in Africa. The CAADP and Malabo Declarations specifically, focus on agricultural transformation for shared prosperity. The key commitments by African Heads of State in those declarations include the Ending of Hunger, eradicating Poverty, and ensuring that agriculture contributes half of the poverty reduction that Africa may achieve, by 2025. The role of agricultural research, investments in agriculture, water management and strengthening resilience are also touted as important in achieving these declarations, perhaps because previous research has also shown that improving agricultural productivity, sustainability as well as profitability is considered a plausible and more practical pathway to poverty reduction for many of the rural households in Africa (Diao et al., 2010; Collier and Dercon, 2014).

Nevertheless, agricultural productivity in general in Africa is low, although there are some positive changes (Loayza and Raddatz, 2010). For instance, Zambia’s cereal yields are low and stand at just around the African Union target of 2 tons per hectare (although there are spatial variations) compared to 10 tons per hectare in other countries (USA, Canada, Chile) outside Africa. The drivers of the widely observed poor agricultural performance in Africa are several and related to the stability and accessibility of the main factors of production, among other things.

Improving agricultural productivity can reduce poverty through several pathways such as through increasing real incomes, employment generation, rural non-farm multiplier effects and food price effects (Benin et al., 2016; Fan, 2009). So, the impacts are both direct and indirect. For example, agricultural productivity can reduce poverty directly by raising farm incomes and indirectly by stimulating demand for inputs and outputs leading to mushrooming of businesses upstream and downstream (Gollin et al., 2014). However, rainfall variability, dry spells, rainfall shocks and other undesired water related factors lead to high costs for smallholder farmers because poor water access can limit future technology uptakes, or poor farmers may invest more in other inputs that may not be productivity enhancing (Amare and Shiferaw, 2017). In any case, erratic rainfall would have a negative effect on agricultural productivity, through any one or all of the pathways in subject, especially in a predominantly rain-fed agricultural system as Zambia, where the proportion of land irrigated to total arable land stands at under 7% (Chilonda et al., 2013).

This study aims to generate empirical evidence on the incidence and impacts of rainfall shocks on agricultural productivity in Zambia to guide agricultural water risk management strategies. Specifically, the paper analyses the effects of dry spells (erratic rainfall) on agricultural productivity in order to understand the role of investments in water management on food production in Zambia. It also seeks to highlight the differential productivity impacts of erratic rainfall in that 2017/2018 agricultural year, depending on the crop in focus. To address these objectives, fixed effects panel regression models are applied using nationally representative household survey data at farm level and 2017/2018 crop estimates for Zambia. The study specifically focused on six key agricultural value chains namely Maize, Groundnut, Sweet potato, Sunflower, Soybean, and Mixed beans, all of which are of importance for rural livelihoods.

The contribution of this paper stems from two stylized facts. First, despite several studies that have investigated the differential impacts of rainfalls, irrigations and rainfall shocks among agricultural sectors, they fail to show the practical importance of the empirical tests and hence unable to convince and guide policy actions. In this paper a conceptual framework is developed that shows the practical links between the heterogeneous impacts of rainfall shocks and strategic choice of water risk management options. Thus, the empirical tests and findings are made more consistent and insightful as a basis for policy formulation and actions. Second, by using a large national dataset that is representative this paper avoids sample bias problems. Furthermore, the use of dry spells information minimizes endogeneity issues because dry spells are vis major and hence outside the choice of the individual farmer. Therefore, this paper fills knowledge gaps that exist in Africa’s agricultural productivity literature, on the impact of erratic rainfall and the differential (heterogeneous) impacts they have on productivity depending on crop type and the district levels/geographical/political administrative regions in which the crops are grown, and its implication on water management.

The rest of this paper is organized as follows: Section 1.1 and Section 1.2 present highlights on the agricultural water management in Zambia and a conceptual framework respectively. Section 2 presents the study methods while the empirical results and a discussion thereof are presented in Section 4 and Section 5, before a conclusion is presented in the final section, highlighting key findings and policy implications.

In Zambia, 67% of farmers are smallholders, cultivating less than two hectares of land (IAPRI Indaba Agricultural Policy ResearchInstitute, 2016). Erratic rainfall is an additional burden to challenges such as fragile soils and poor access to agricultural inputs, markets and improved agricultural practices. Rains in some Southern African countries incuding, Zambia, Malawi and Central-SouthernTanzania, are largely unimodal starting in October and ending in April but exhibit high spatial-temporal variability (Muthoni et al., 2019). Variability in rainfall is mainly determined by north–south movement of the Intertropical Convergence Zone (ITCZ) (Diem et al., 2014) and changes in sea surface temperatures, especially in the tropical pacific (Maidment et al., 2015). Zambia experiences a spate of erratic rainfall in some parts of the country from time and again, although generally, its annual rainfall reveals an increasing trajectory. There are spatial variation in rainfall reception over time and recently, the Northern Province has seen a significant decrease in annual rainfall of about −0.1 to−4 mm year⁻¹(Muthoni et al., 2019). On the other hand, in recent years, other parts of Zambia including almost the entire Western, Southern, Central, Lusaka, and Copperbelt Provinces and extended to limited sections of all the other Provinces have seen an increase in annual rainfall of around 0–16 mm year⁻¹ (Muthoni et al., 2019). In the 2017/2018 period for which the data apply, Zambia’s annual rainfall stood at 893 mm per annum compared to the normal expected rainfall amounts of 946.5 mm per annum.

If the water resources that Zambia has were developed fully for irrigation and drainage, the effects of erratic rainfall would perhaps be modest because irrigated agriculture would compensate for lost output from rain-fed agriculture at the very least. However, while there are many agricultural water management initiatives in some of the districts of Zambia, many parts of Zambia have underdeveloped water resources (Ngoma et al., 2019). Ngoma et al. (2019) show that there are some irrigation efforts in Zambia but such efforts have not yet led to a significant share of irrigation of arable land and more effort is needed. Thus, the impact of any small dry spell in Zambia is likely to be felt heavily by farmers in the affected areas, especially given that internal trade of commodities is not as frictionless as may be desirable. Water management is important for productivity as it is shown in many studies that crop yields are higher in irrigated areas than in rain-fed areas because of the erratic nature of rainfall (Rosegrant and Perez, 1997; Lipton et al., 2003).

This section provides the conceptual foundation of the link between agricultural water risk management and the nature and extent of the risks. The aim is to develop empirically testable hypotheses that can guide and show the practical use of the empirical analyses and findings in agricultural water risk management. It is assumed that there exists three broadly classified agricultural water risk management strategies. Firstly, there is the preventive agricultural water management strategy which includes constructing irrigation structures, water drainage, flood control and other similar strategies. Secondly, there are crop production or food security strategies such as crop diversification or selection/specialization. Thirdly, there are coping strategies such as the use of agricultural insurance either in the form of crop index insurance or weather index insurance.

However, the choice of these agricultural water risk management strategies should depend on two factors. These factors measure the size and consequences of the risks occurring at a given location. The first one isthe incidence pattern of the risks/shocks in affecting households at given space and time. This measures the correlation of the risks’ incidence at province level and is measured by the distribution of the rainfall shocks in affecting households in a province at a time. This relates to whether the risks are jointly and equally affecting households at any given space and time. The second one the differential idiosyncratic impacts of the shocks on agricultural sub-sectors (commodities) and is measured by the correlation of the risks’ impacts across crops. This relates to whether the risk is idiosyncratic or covariate in affecting different crops. These two indicators show the nature and extent of agricultural risks caused by rainfall shocks. Using these indicators, one can identify the type of agricultural water risk management strategy that would be feasible at province level.

Fig. 1 describes how the correlations of the incidences and impacts of rainfall shocks affect the choice of agricultural water risk management strategies. The horizontal line measures the correlation of the risks’ incidence. The correlations become negative if only one of the risks occur at a given time and space. That means households in a given village are being affected by one rainfall shock at a given time. The correlation becomes positive if households are hit by more than one shocks at a time. The vertical line represents the correlation of the productivity impacts of the risks across crops. The correlation of the impacts is considered as positive if the impacts of the shocks are uniform across crops and shocks. It is considered as negative if the impacts of the shocks are heterogeneous across crops. Depending on the direction of correlations of the two line, we get four panels or quarters, which define the priority agricultural water management strategy.

The North-East (NE) panel of Fig. 1 connects the positive correlations of both the incidence and impacts of the shocks. This means that more than one shock affect households at a time and the shocks affect all the crops uniformly. This is a widespread risk for the household to manage either through crop diversification or agricultural insurance. Possibly, the insurance premium will be excessively very large. Thus, the most feasible option is to invest in water management practices such irrigation, drainage and flood controlling structures both by the household and the public support system. The South-East (SE) panel connects the positive correlation of incidence and the negative correlation of the impacts of the shocks. Even if the shocks affect households simultaneously at a time, their impact on crops are heterogeneous. Hence, households can simply manage the risk through crop diversification. The South-West (SW) panel connects the negative effects of the shocks both on households and crops. In this panel, households are affected by only one shock at a time and that shock affects crops heterogeneously. This is the lowest risk of all cases. Thus, a crop-index insurance could be sufficient to manage the rainfall risks. The North-West (NW) panel connects the uniform (positive) impacts of the chocks on crops and the negative correlation of the incidence of the shocks at given location (province in our case). The shocks affect the households alternatively, but that shock affects the crops uniformly. In this case, the strategy depends on the number of households being affected by the shock at a time. If only few households are affected by the dominant shock in a given location, which is likely for the case of flood or waterlogging, weather-index insurance will be a feasible option. If most households are affected by the shock, which is likely for dry spell, insurance would be more costly than investment on agricultural water management practices. Thus, the most feasible strategy could be investment in agricultural water managing practices.

Therefore, agricultural water risk management strategy at a given location must depend on the empirical answers to the following research questions (1) How correlated are the incidence of the shocks at household and regional level? (2) How heterogeneous are the impacts of the shocks across crops? The subsequent sections are designed to provide empirical evidences to address these questions and draw strategic options that best fit the Zambian agricultural water risk management.