Trends and overview of Conservation agriculture in Africa

Get Complete Project Material File(s) Now! »

CHAPTER 4 THE YIELD IMPACT OF CONSERVATION AGRICULTURE: AN ASSESSMENT OF SMALLHOLDER FARMERS IN ZIMBABWE

Introduction

Yields in smallholder farming systems of southern Africa have remained appallingly low despite technological innovations such as fertilisers and improved seeds (Baudron et al., 2012). In many cases farmers cannot guarantee food security from their own production and very few smallholders are able to sell surplus harvest to generate income (Marongwe et al., 2011). Many causes of agricultural stagnation have been suggested, with some observers emphasizing the Malthusian link between rapid population growth, low agricultural productivity, and resource degradation (Knowler & Bradshaw, 2007; Rockstrom et al., 2009; Mazvimavi, 2011); others emphasising market, government and institutional failures (DFID, 2004; Diagana, 2003) or bio-physical factors such as climate and soils (DFID, 2004; FAO, 2011). Despite the debate regarding the causes, there is consensus about the need to devise strategies to improve food production in order to address food insecurity in the twenty-first century in Africa (Conceição et al., 2011). Increases in food production in Africa must come through increased productivity based on the adoption of new technologies. Agricultural intensification is necessary because many regions of SSA are no longer land abundant (Mwangi, 1996). It is also imperative to involve smallholders in the intensification efforts so as to enhance access to food for vulnerable people.
Conservation agriculture strives to achieve acceptable farm profits with high and sustained production levels while concurrently conserving the environment (Steiner & Bwalya, 2003). Adoption of CA by farmers in several African countries has shown potential to improve rural livelihoods through sustainable and intensified production (Silici et al., 2011). Conservation agriculture is being promoted in response to low agricultural productivity, chronic household food insecurity and environmental degradation linked to conventional tillage and nutrient mining. This innovation constitutes a package of agronomic practices characterised by three principles that are linked to one another namely: a) reduced or eliminated mechanical soil disturbance, b) better use of production inputs and therefore greater cover with crop residues; and c) diversification of crop species grown in sequences and /or associations (FAO, 2008).
Worldwide experience of CA over the past four decades has demonstrated how the simultaneous application of a set of practices of minimal mechanical soil disturbance, organic soil cover and diversified cropping can lead to greater and stable yields (Kassim & Friedrich, 2010). Land preparation and cropping methods in CA also enable efficient use of rainwater which considerably reduces the risk of crop failure due to drought and make the soil a better environment for the development and functioning of plant roots (Reicosky, 2008). Twomlow et al., 2006; Nyagumbo, 1999, Fowler & Rockstrom, 2001, explained the advantages of CA compared with traditional cultivation practices as being its ability to diversify production, increase social capital through farmer groups and decrease dependence on food aid.
When practised correctly, CA stabilises crop yields, thereby increasing household food security and economic and social wellbeing (Solís et al., 2009). Grain yield of maize, teff and wheat have been reported to double under CA-based practises compared to conventional farming in Ethiopia, Ghana, Tanzania and Malawi (Ito et al.,2007), Kenya (Rockstrom et al.,2009) and Mozambique (Nkala et al.,2011; Grabowski, 2011). Haggblade and Tembo (2003) reported that early CA adopters increased crop productivity by 30 to 70% in Zambia. These findings were also observed by Mashingaidze and Mudhara (2006), who reported that crop yields for maize increased by up to 3.5 tons per hectare in Zimbabwe for farmers practising CA. Hassane et al. (2000) evaluated the impact of planting basin, and use of fertiliser and manure on millet crops in Niger and found that over a five year period, farmers experienced yield gains of up to 511%. Yield differences ranged between 20 to 120% higher for CA-managed fields compared with conventionally managed fields in Latin America, Asia and Africa (Pretty et al., 2006; Landers, 2007; Erenstein et al., 2012; FAO, 2008; Hengxin et al., 2008; Rockstrom et al., 2009). The ability of CA technology to yield higher productivity than existing practices was the main reason why farmers decided to adopt the new innovative technology (Twomlow et al., 2008; Muchinapaya, 2012). While there is evidence of CA gains in the literature, there are also studies that present a sharply contrasting assessment of the impact of CA. Giller et al. (2009) suggests that empirical evidence is not clear and consistent on CA contributions to yield gains. Their study notes concerns that include decreasing yield in CA. This chapter addresses methodological problems in other studies to provide more definitive measurements of yield impact of CA as practised by Zimbabwe´s smallholders.

Background

The type of CA under study is called the planting basin method, which was initiated in Zimbabwe by Brian Odrieve in the 1990s (Muchinapaya, 2012). This method involves planting crops directly into the land which is protected by mulch using minimum or no-tillage techniques and this is aimed at conserving soil and water. Crop rotation is an essential component which calls for farmers to alternate legumes with their maize crops in order to improve soil fertility, but they are often averse to giving up field space where they normally grow their major crops. Mazvimavi et al. (2008) and Marongwe et al. (2011), reported that basin CA was introduced in Zimbabwe on a large scale in the 2003/04 season. This was implemented primarily through programmes aimed at improving the livelihood and food security status of smallholder farmers in Zimbabwe. A comprehensive package of CA has been promoted by NGOs and national agricultural research and extension departments throughout Zimbabwe. It consists of several key practices, namely dry-season land preparation using minimum tillage systems (for example basin planting), crop residue retention, nitrogen fixing crop rotations and precise fertiliser application. CA has been promoted by different partners and involved in the supply of input packages (fertiliser and seed) to farmers who were willing to set up CA demonstration plots. Yield gains from demonstration trials were attributable to multiple factors such as, timely planting of CA fields, availability and precision placement of fertilisers and better moisture conservation (Nyagumbo et al., 2009; Marongwe et al., 2011).
Yield benefits from CA-managed trials encouraged diffusion of CA to other farmers. However, farmers tended to practise CA on relatively smaller portions of their land holdings because of the extra labour required for weeding, and the challenge of retaining crop residues on fields because of communal grazing pressure (ICRISAT, 2009). The Food and Agriculture Organisation estimated that area under CA in Zimbabwe was 139 300 ha, constituting about 9% of area under cereals in 2012 (FAO, 2012). Empirical studies that have been carried out to assess the impact of CA in Zimbabwe use various methods and analytical approaches, ranging from on-station and on-farm agronomic experiments to broader household surveys (Nyagumbo, 1999; Siziba, 2008; Mupangwa, 2009, Musara et al., 2012; Nyamangara et al., 2013; Ndlovu et al., 2013). Most of the impact studies tend to attribute all yield and welfare differences to CA. However, this can be faulty in the absence of robust quantitative approaches capable of isolating effects of other exogenous factors. Ascribing causality of change in yield to CA without first establishing a counterfactual situation could be oversimplification of a complex process. This poses a serious challenge especially when a study makes use of cross- sectional data and does not have a longitudinal (time) dimension. In such analyses the measured impact of CA on yield can be biased by unmeasured or unobservable variation in household conditions. Studies that use longitudinal data focus on agronomic impacts such as yield and soil properties, but generally fail to control for household-level covariates that may have important interactions in the production process. This evaluation however, takes advantage of the longitudinal nature of the data set to control for unobservable household-level factors.
By observing the same farmers in successive seasons of CA practice in a non-experimental setting, it is possible to compare CA with alternative conventional farming practices within the same households (i.e households practising both technologies). The primary interest is on the impact of CA on maize production, since it is a staple crop grown by more than 80% of the sample farmers. The purpose of the study is to estimate the yield impact of applying CA practices. This is achieved by using a unique data set that captures, at the plot level, maize production under CA and alternative conventional farming practices across different agro-ecological regions.
The main hypothesis of the study is that applying CA has a positive and significant impact on yield across agro-ecologies. By testing this hypothesis, the study seeks to justify the use of CA by smallholder farmers. Econometric approaches that capture OLS and household fixed effects were employed in order to determine the impact of CA adoption on maize yield.

Analytical framework

The yield impact of practicing CA is measured through Cobb-Douglas production function estimation. Cobb-Douglas production function, estimates the quantitative effect of two or more inputs on output (maize yield in this case). The estimation uses various specifications in both ordinary least squares (OLS) and household fixed effects frameworks.
Where Ypt is yield of maize on plot p, in year t. Xpt represents the production inputs such as seed fertiliser on plot p, in year t. The amount of seed, basal fertiliser and top dressing fertiliser are the logs of the positive mounts of these inputs applied to the plot in question. The natural log of the yield of maize is regressed as a function of the natural logs of positive quantities of inputs such as seed, area, basal and top- dressing fertiliser. Total cropped area is log of the sum of all plots which have been cultivated by a household measured in square meters (m²). Z represents various household, plot, and regional factors affecting yield. T is the year indicator representing the year and also round of the survey and ε is the error term which is normally distributed.
CA is an indicator variable for a plot (or household) on which CA is practised in a particular round of the survey. For a plot to be coded as 1, the farmer identified the plot to be a CA plot. CA plot are thus defined in terms of the land preparation in which case it will be digging basins. Digging planting basins is what distinguishes CA plots from non-CA plots. In terms of the practices reported by the farmer, these vary in accordance with the intensity of adoption. Due to a number of farmer specific constraints, the range of practices is from one to eight however, digging of basins is a requirement for a plot to be coded as 1.The dummy variable for CA =1 means that at least one part of the CAzim package ( basin land preparation) has been applied to a plot. When the land preparation does not involve digging of planting basins then the plot is coded as 0.
Five model specifications of the yield function were estimated using OLS with indicator variables for round and natural region (NR)

Specification A

Xpt in equation 1 refers to seed only. Specification A is the general function whereby yield is expressed as a function of seed and CA technology is used as a dummy. This specification captures the impact of basic CA. Basic CA is merely digging basins however, it may or not include all the three principles of CA on some of the plots. Many farmers who use this CA method will use other technologies and practices as well. As these practices and technologies like top dressing fertiliser, basal fertiliser, manure and multiple weeding are not included in the regression, some of their effect will be reflected in the CA coefficient. Thus this specification gives an estimate of the upper end of the CA impact.

Specification B

Xpt in equation 1 refers to seed, fertiliser (both basal and top dressing), manure and weeding (Lower end of CA effects). The full CA package includes basic CA i.e the three principle of CA plus additional components which are included in the CA package promoted in Zimbabwe. The yield function is similar to that specified in Specification A but the difference is in that, weed frequency and fertility management are included in the specification. Though CA is used as an indicator variable, practices that enhance the effect of CA such as manure application and weeding frequency, are also specified. Specifically including these practices will reduce the estimated coefficient on CA in Specification B compared to Specification A, because any positive effects of fertilizer, manure and weeding will be captured separately, rather than being pooled with the CA indicator. Specification B gives a yield function with positive quantities of inputs such as seed, basal and top-dressing fertilisers, weeding frequency gives the lower end in terms of CA impact on yield. The impact of CAzim would be the sum of the effects captured in the coefficients on CA, weeding, and fertilizers.

Specification C

Specification C is similar to Specification B but for the CA technology variable: a detailed variable that captures the depth of the technology use (number of techniques applied) is used instead of an indicator variable. CApt in equation 1 refers to the number of CA techniques applied to a plot in a particular year. The CA variable is not captured as an indicator variable (0; 1) but represented by a count (1;8) since they are 8 distinctive techniques defining CAzim. Zero implies that the plot is non-CA 8 means all techniques were applied.

Specification D

Specification D attempts to capture the impact of labour by excluding weeding frequency in the specification. The impact of labour on yield is captured by disaggregating the household size into various age groups. The yield function includes the dummy variables for use of top-dressing and basal fertiliser, as well as year dummies to capture shift in weather and policy.
Zpt in equation 1 refers to the disaggregated household size into various age groups and exclude weeding frequency Xpt in equation 1 captures quantity of inputs used except for basal and top dressing fertiliser where an indicator variable is used instead.
Lag variable of CA is defined as dCAp (t-1) and captures the history of CA application in a given plot. The lag of CA is an indicator variable implying CA was applied to the plot in the previous season. The greater the number of years CA was practiced in the plot the more the benefits, if the lag is t-1 it means CA was implemented on the plot the previous year whereas t-2 implies CA was implemented on the plots two seasons prior to the current.

DEDICATION 
DECLARATION 
ACKNOWLEDGMENT
TABLE OF CONTENTS 
LIST OF TABLES 
LIST OF FIGURES 
ABBREVIATION AND ACRONYMS
ABSTRACT
CHAPTER 1 INTRODUCTION
1.1 Background
1.2 Problem statement
1.3 Focus of the study
1.4 Objective of the study
1.5 Research hypothesis
1.6 Justification and relevance
1.7 Summary of research methods
1.8 Organisation of the thesis
CHAPTER 2  LITERATURE REVIEW 
2.1 Introduction
2.2 Basic concepts
2.2.1 Conservation agriculture
2.2.2 Technology adoption
2.2.3 Adoption intensity
2.2.4 Partial adoption
2.2.5 Dis-adoption
2.3 Trends and overview of Conservation agriculture in Africa
2.3.1 Justification of Conservation agriculture for smallholder farmers in SSA
2.3.2 Constraints to adoption of Conservation agriculture in SSA
2.4 Historical developments of Conservation agriculture in Zimbabwe
2.4.1 Promotion and uptake of Conservation agriculture in Zimbabwe
2.4.2 Current state of Conservation agriculture research in Zimbabwe
2.5 Review of empirical studies on technology adoption
2.5.1 Studies on technology dis-adoption
2.5.2 Methodological issues in technology adoption
2.6 Theoretical models of adoption and dis-adoption
2.7 Theoretical framework for the study
2.8: Summary
CHAPTER 3 RESEARCH DESIGN AND METHODS 
3.1 Introduction
3.2 Research approach
3.2.1 Panel survey approach
3.2.2 ICRISAT panel survey
3.2.3 Sampling strategy for ICRISAT panel survey
3.2.4 Determination of the ICRISAT sample size
3.3 Study area
3.4 Data collection
3.4.1 Attrition
3.4.2 Study sample size
3.4.3 Crop production data
3.4.4 Summary of household data
3.5 Description of what constitute CA practices used in the study
CHAPTER 4  THE YIELD IMPACT OF CONSERVATION AGRICULTURE: AN ASSESSMENT OF SMALLHOLDER FARMERS IN ZIMBABWE
4.1 Introduction
4.1.1 Background
4.2 Analytical framework
4.2.1 Data
4.2.2 Description of yield function variables
4.2.3 Expected impact of explanatory variables on yield
4.3 Results and discussion
4.3.1 Comparative analysis: Area cultivated
4.3.2 Comparative analysis: Maize yield
4.3.3 Comparative analysis: Planting dates
4.3.5 Ordinary least squares (OLS) Results
4.3.6 Household fixed effects model results
4. 4 Conclusion
CHAPTER 5  FACTORS AFFECTING ADOPTION INTENSITY OF CONSERVATION AGRICULTURE aMONG SMALLHOLDER FARMERS IN ZIMBABWE 
5.1 Introduction
5.2 Analytical framework
5.3 Results and discussion
5.4 Conclusion
CHAPTER 6 ABANDONMENT OF CONSERVATION AGRICULTURE BY SMALLHOLDER FARMERS IN ZIMBABWE
6.1 Introduction
6.2 Analytical framework
6.3 Results and Discussion
6.4 Conclusion
CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS 
7.1 Introduction
7.2 Summary findings
7.3 Conclusions
7.4 Key implications for policy
7.4 Direction for future research
REFEFRENCES
GET THE COMPLETE PROJECT