Factors Influencing the Double-Up Adoption of Climate Change Adaptation Strategies among Smallholder Maize Farmers in Malawi

Abstract

Maize remains the staple grain in Malawi; hence, the cropping system of most smallholder farmers is dominated by the crop, often mono-cropped for food security. Consequently, Malawi’s agriculture sector is made vulnerable to the adverse impacts of climate change. For instance, crop failure results in food insecurity and the low income of farm households. In response, there are coping mechanisms, which can be adopted by farmers to mitigate these negative climate change effects, namely maize–legume diversification, organic manure use, and the practice of agroforestry. Therefore, the underlying objective of this study was to assess the factors influencing smallholder maize farmers’ decision to engage in the double-up adoption of climate change adaptation strategies in Malawi. Both descriptive statistics and the logistic regression model were employed to statistically analyze these factors, and the results of the analysis revealed that landholding size, inorganic fertilizer use, access to credit, seed access, adherence to extension services, and input coupon access were significant in influencing dual adoption. Furthermore, this study recommends policies, which underscore land access and safeguard the land rights of smallholder maize farmers, and also private sector engagement in complementing government efforts in ensuring increased access to seeds. Additionally, improving farmers’ adherence to agricultural extension services is recommended. Thus, addressing the constraints of small-scale farmers observed in this study will act as an incentive for farmers to consider dual adoption, which is perceived to be a feasible method to combat climate change effects.

1. Introduction

Globally, climate change is one of the major challenges, which has devastating economic impacts on farm households and the whole agricultural sector, with its continued dry spells, unpredicted scanty rainfall, and wind patterns. Thus, climate change exacerbates the countries’ efforts toward achieving the Sustainable Development Goals (SDGs) of zero hunger and no poverty, whereby climate risks threaten the crop production and income of farm households [1].

Sub-Saharan Africa is not exempt; the region is also susceptible to the severe impacts of climate change. Recently, the region has been experiencing fluctuations in temperature and rainfall, leading to prolonged droughts [2], in addition to changes in soil quality and moisture content, growing season variations, flooding, and the low harvest of crop produce translating to the low income of farm households [3]. The resulting shortage of food, poor nutrition, and low income are the push factors, which lead to a poverty trap [4]. Jafino et al. [5] also highlighted that most people in sub-Saharan Africa will be subjected to extreme poverty if the necessary climate actions are not taken by the year 2050.

However, farmers can positively respond to these climate-related shocks through adaptation. Therefore, climate change adaptation simply concerns the adjustments that people can make to lessen the impact of extreme weather events emanating from the change in climatic conditions [6]. Adapting agriculture to climate change is therefore a prerequisite to reduce its impacts on the livelihood of farm households, enabling them to increase production and income [7]. Correspondingly, as a means of adapting to climate change in agriculture, the Food and Agriculture Organization (FAO) [8] emphasized sustainable crop production intensification (SCPI), defined as an eco-friendly approach to farming, with the aim of producing more with the existing landholding size while preserving natural resources; SCPI therefore involves farmers adopting both traditional and modern technologies or practices in their farms to meet their needs of higher yields without having a negative impact on the environment, thus building a resilient crop production system, which is well adapted to climate change and thereby ascertaining both increased productivity and a healthy ecosystem. Some of these practices include maize–legume diversification, the use of organic fertilizer, and the practice of agroforestry.

Maize–legume diversification involves the integration of legumes into a maize-based cropping system, and it is one of the key strategies, which can invigorate farmers’ resilience to extreme weather conditions in Malawi [9]. Grain legumes, such as ground nuts, beans, pigeon peas, and soya beans, are high-value crops, which can also play multiple roles within a maize-based cropping system and hence enhance agricultural productivity [10]. Moreover, these multi-purpose grain legumes are suitable for sustainable intensification, as they adapt to various agro-ecological zones, are biologically able to fix the nitrogen in the soil, and have little to no competitive effect on the first crop, such as maize, in a cropping system [11,12]. Generally, crop diversification leads to increased productivity and production stability; it serves as one of the coping mechanisms for food security, production, and market risks [13], in addition to improving the income of smallholder farmers [14]. Therefore, it can be considered an important step in the transition of small-scale farmers from subsistence to commercial farming [15].

Agroforestry is a systematic agricultural practice, which balances the growing of crops and forests on farmland [16]. Planting leguminous or fruit trees together with maize can yield increasing benefits to farming households by providing manure to the soils, and fruit trees can offer both food and income to farmers. Moreover, the provision of fuel wood and deep-rooted trees can help control soil erosion during flash floods, and trees also act as a windbreak against storms, which can easily damage crops such as maize in a farm. Therefore, agroforestry can also be adopted as a climate change adaptation strategy by smallholder maize farmers [1].

Organic fertilizer use is another climate change adaptation strategy, and it encompasses the application of organic fertilizers to farmland. These organic fertilizers can significantly increase crop yield by enriching the soil in terms of its moisture and nutrient content [17], in addition to adding significant amounts of high-quality vegetative biomass to rejuvenate degraded soils [18]. Apart from being a sustainable option, organic fertilizers, such as compost and green manure, are cheap compared to inorganic fertilizers; hence, they can help farmers efficiently reduce costs in their production process [19].

Overall, this study identified these strategies—namely maize–legume diversification, organic fertilizer use, and the practice of agroforestry—because they simultaneously align within the scope of sustainable intensification and the subsequent adaptation to climate change. Additionally, these strategies can complement each other within a mixed maize-based cropping system; they are also perceived to reduce uncertainties in addition to being cost effective and ecologically viable for improving productivity in farming, thus resulting in increased yield and income while also ensuring environmental sustainability in conserving soils [20].

Specifically, this study focused on the double-up adoption of these strategies because, according to Snapp et al. [21], food insecurity is a problem affecting a large population in sub-Saharan Africa, and this could be worsened by extreme weather events, such as drought and floods, which accelerate soil degradation and reduce its organic matter and nutrient content. Additionally, farmers are also uncertain with regard to the unprecedented weather patterns during agricultural seasons as a result of climate change. Conversely, smallholder farmers have a variety of needs for their livelihoods [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21]. Accordingly, the dual adoption of these strategies in a maize-based cropping system offers farmers a security measure in response to unprecedented weather and helps build and strengthen the climate resilience of farming households by hedging against production, income, and price risks. In essence, a mixed cropping system of at least two sustainable adaptation practices—namely maize–legume diversification, organic fertilizer use, and the practice of agroforestry—in their maize-based cropping system can help in ensuring that their diverse needs are met and sustained. These not only include the immediate needs of food and income but also the long-term need of enriched fertile soils.

However, even though it is of outmost significance to sustainably intensify agricultural cropping systems in an effort to adapt to climate change, White and Crawford [22] indicated that the adoption rates of practices such as crop diversification among farmers in Malawi remain low, citing a lack of farmer education as one of the major reasons. A study by Kankwamba et al. [23] also acknowledged that the agricultural sector in Malawi is highly undiversified. Conversely, Cutforth et al. [24] emphasized that greater diversification is an important decision, which individual farmers can make within their farming systems. Therefore, in order to potentially scale up and harness the adoption of these adaptation practices within maize-based cropping systems, there is a need to assess the embedded factors influencing farmers’ dual adoption choice or decision.

In this regard, some previous empirical studies have also analyzed the determinants of farmers’ decision to adopt the adaptation strategies to climate change, and some of their findings are consistent with those of this study. For instance, a study by Diallo et al. [25] showed that adopters and non-adopters differ in terms of household characteristics, plot levels, and institutional characteristics. Similarly, Zakari et al. [7] indicated that socio-demographic and economic characteristics influence farmers’ decision on the adoption of adaptation strategies. Notably, Webber et al. [26] highlighted that the perception of climate change, labor availability, and farmer socio-economic status determine the adoption decision. Furthermore, Ndamani and Watanabe [27], Di Falco et al. [28], and Ojo and Baiyegunhi [29] indicated that access to both credit and extension services was a key driver regarding adoption. Additionally, Kinuthia et al. [30] recommended increasing education and climate awareness and providing not only financial incentives but also timely and accurate weather forecasts to significantly influence farmers’ choice of response strategies to climate change. Some of these findings are consistent with those of this study, but notably, among the other hypothesized variables included in this study analysis, adherence to agriculture extension services—which takes into account both the access to and use of such services—was found to be one of the important factors regarding the dual adoption of adaptation strategies to climate change.

Imperatively, this study therefore contributes to the existing literature, as it analyzes and identifies the factors influencing smallholder maize farmers’ decision to adjust their level of responsiveness to climate-related shocks through dual adoption. In addition to policy implications, the results of this study can be used by policy makers to design practical interventions, which address the needs of smallholder maize farmers in order for them to respond positively to climate change via the dual adoption of sustainable adaptation strategies.

2. Materials and Methods

2.1. Data Source and Sample Size

This study used secondary data from the national sample survey in Malawi known as the Fourth Integrated Household Survey (IHS4) collected by the National Statistical Office (NSO) in the period between 2016 and 2017. IHS4 is a nationally representative rural household survey administered to 12,447 randomly selected households in Malawi. A stratified two-stage sample design was used for IHS4 [31]. IHS data are primarily collected to provide benchmark poverty and vulnerability indicators for Malawi. Therefore, the sample size for this study was based on the number of households, which had produced maize during that particular agricultural season across regions in Malawi, which totaled 8942 households, as shown in

Table A1. Sample size and representative districts in this study.

Region	Sample Size	Number of Districts
Northern	1512	6
Central	3144	9
Southern	4286	13
Total	8942	28

.

Study Area

This study covered all 28 districts across three regions in Malawi, namely the northern, central, and southern regions, as shown in Figure 1. Briefly, Malawi is an agrarian economy located in the southeast of Africa. According to the World Bank [32], the country has a population of over 20 million people, and approximately 80% of this population is rural and engaged in subsistence agriculture, with maize being their staple crop.

2.2. Data Management and Analysis

IHS4 data were analyzed in STATA 17.0 statistical software, and the data required cleaning before analysis. For instance, there were duplicates of unique household identifiers, which resulted from the way the questions were structured in the questionnaire, rendering data in a long format; hence, to rectify these duplicates, the data were reshaped from long to wide format. Additionally, all the missing values of some variables were recorded as zero. Lastly, each of the reshaped variables from the agriculture and household modules of IHS4 data was merged with existing IHS4 summary data used as master data to form one dataset for the analysis.

2.3. Comparing Characteristics between Double-Up Adopters and Non-Adopters of Climate Change Adaptation Strategies

Descriptive statistics, which constitute frequencies, percentages, means, and standard errors, were used in determining differences in terms of the demographic, socio-economic, and institutional characteristics between dual adopters and non-adopters of climate change adaptation strategies. Therefore, to compare these two groups, the chi-square test and t-test were used to determine the statistical significance of dummy and continuous variables, respectively.

2.4. Assessing the Factors Influencing Double-Up Adoption of Climate Change Adaptation Strategies

The logistic regression model, or simply the logit model, was used to establish the statistical relationship between the explained variable and explanatory variables, which were hypothesized to influence a farmer’s decision to consider the dual adoption of climate change adaptation strategies. Specifically, double-up adoption was devised in this study as a dummy dependent variable, whereby a farming household was regarded as either a double-up/dual adopter or non-adopter and assigned a value of 0 or 1 in the model, respectively. The demographic, socio-economic, and institutional factors were used as independent variables.

2.5. Conceptual Framework

Conceptually, as shown in Figure 2 below, smallholder farmers make rational decisions according to the environment in which they operate. In line with this, according to the rational choice theory, farmers always want to derive more benefits from their production systems. Hence, in making decisions regarding crop production mixes or systems, they will be motivated by increasing the benefits associated with any technology; otherwise, they do not adopt such technologies in their cropping system [33]. Therefore, the benefits associated with the dual adoption of climate change adaptation strategies—namely maize–legume diversification, organic fertilizer use, and practicing agroforestry—involve sustainably intensifying production, so as to meet both the immediate needs of food and income and the long-term requirements of soil enrichment, fuel wood, livestock fodder, and environmental sustainability, eventually cushioning small-scale maize farmers from the adverse impacts of climate change. However, crop specialization, such as maize monoculture, may lead to yield variations and a highly unstable income, especially in times when farmers are faced with the various risks of climate change, such as drought.

2.6. Theoretical Framework

2.6.1. Theory

The bounded rationality theory stipulates that individuals make choices, which deviate from their utility-maximizing behavior as a result of constraints, such as information asymmetry and uncertainty [35]. Therefore, this theory is adopted in this study, in the sense that smallholder maize farmers may fail to dually adopt climate change adaptation strategies due to being limited by various factors, which can be demographic, socio-economic, and institutional in nature. Correspondingly, McFadden’s [36] random utility model represents the stochastic utility theory, which considers the randomness of preference over discrete choices [37]. This theory is also adopted in this study, considering that random utility is a sub-category of probabilistic choice models used to exemplify individuals maximizing behavior economically [38]. According to Ndhlovu [39], utilities are considered stochastic variables depicting asymmetric information regarding the attributes of substitute choices.

Therefore, this study is based on the overall fundamental assumption of utility maximization, such that a household will dually adopt climate change adaptation strategies if their utility exceeds a certain threshold, which is set at zero; thus, for each household, the utility difference between double-up adoption and non-adoption is a function of observed characteristics, X_i and unobserved characteristics, ε_ί. The utility relationship difference is denoted as Y_i*:

Y_i* = β X_ij + ε_ί

Y_i = 1  if Y_i* > 0

Y_i = 0  if Y_i* ≤ 0

In the above equations, Y* is unobserved, and it is also referred to as a latent variable. Subsequently, it is observed that Y_i = 1 (double-up adopter) if and only if Y_i* > 0, and Y_i = 0 (non-adopter) if Y_i* ≤ 0.

2.6.2. Logit Model

In this study, the empirical approach used to assess the factors driving smallholder maize farmers’ decision to dually adopt adaptation strategies to climate change is based on logit modeling, which uses the maximum likelihood estimation technique. The logit model is used because of the dichotomous nature of the output variable. The estimation of such a dummy-dependent qualitative variable can be carried out using three approaches, namely the linear probability model (LPM), the logit model, and the probit model [40]. Despite being easy to use, LPM has deficiencies of allowing the presence of heteroscedasticity, and the estimated probability has a likelihood of lying outside the 0-1 bound; however, the inferences which can be drawn from applying the logit and probit models on the same data are invariably similar despite the coefficient results of the logit model having a tendency to exceed those of the probit model by a scale factor of 1.6–1.8 [41]. Therefore, this study preferably uses the logit model.

The specification of the logit model stems from the logistic distribution function and is given according to Gujarati [41], as follows:

$$p_i=\frac{1}{1+e^{-({\beta }_1+{\beta }_2{x}_i)}}$$

$$p_i=\frac{1}{1+e^{-Z_i}}=\frac{1}{1+\frac{1}{e^{zi}}}=\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$\frac{e^{Zi}+1}{e^Z}$}\right.=\frac{e^Z}{1+e^Z}$$

(1)

where $Z={\beta }_1+{\beta }_2{X}_i$.

$Pi$ in Equation (1) is the logistic distribution function, which portrays the probability that a household will dually adopt climate change adaptation strategies; then, it follows that (1 − $Pi$) is the probability of non-adoption, and mathematically, this is equal to

$$\frac{1}{1+e^Z}$$

(2)

Therefore, dividing Equation (2) by Equation (1) results in

$$\frac{p_i}{(1-p_i)}$$

$$\frac{p_i}{(1-p_i)}=\frac{1+e^Z}{1+e^{-Z_i}}=\raisebox{1ex}{$\frac{e^z}{1+e^z}$}\!\left/ \!\raisebox{-1ex}{$\frac{1}{1+e^z}$}\right.=e^{Z_i}$$

$$\frac{p_i}{(1-p_i)}=e^{Z_i}$$

Basically, $\frac{p_i}{(1-p_i)}$ is the odds ratio in favor of being a dual adopter. Put simply, it is the ratio of the probability that a household will dually adopt climate change adaptation strategies to the probability that it will not adopt them. Gujarati [41] gave an example showing that, if P_i = 0.8, then this means that the odds are four to one in favor of the household being an adopter; thus, the probability of adoption is four times higher than that of non-adoption. Therefore, taking the natural log of the odds ratio, the results obtained are

$$\frac{p_i}{(1-p_i)}=e^{Z_i}$$

$$ln\left(\frac{P_i}{1-P_i }\right)=ln\left(e^{z_i}\right)$$

$$ln\left(\frac{P_i}{1-P_i }\right)=Z_i\ln \left(e\right) \mathrm{Note}: ln\left(e\right)=1$$

$$L_i=ln\left(\frac{P_i}{1-P_i}\right)=Z_i$$

(3)

$$Z_i={\beta }_0+{\beta }_{1}X$$

L is the log of the odds ratio, which is linear in both X and Y parameters, and it is referred to as logit [41].

2.6.3. Marginal Effects

The marginal effect is defined as the effect on the conditional mean of the response variable as the causal variable changes. Generally, these marginal effects measure the change in the probability of Y = 1 (double-up adoption) as a result of unit change in a particular causal variable. Therefore, the marginal effects of the logit model are developed by taking a derivative. This is given as $\frac{\delta E(y)}{\delta x}=\frac{e^{({\beta }_1+{\beta }_{2}x)}}{{(1+e^{({\beta }_1+{\beta }_{2}x)})}^2}{\beta }_2$.

The derivative can also be expressed as

$$\frac{\delta E(y)}{\delta x}=p(1-p){\beta }_2$$

where $p=\frac{e^{({\beta }_1+{\beta }_{2}x)}}{{(1+e^{({\beta }_1+{\beta }_{2}x)})}^2}$

2.7. Specification of the Empirical Model

The empirical model was based on the probability of smallholder maize farmers dually adopting adaptation strategies to climate change in their crop system, which is a function of the vector of the input variables, unknown parameters, and error term. The logit model specifies the functional form of Z, where ε_i is an error term, which is independent and normally distributed with a zero mean and constant variance.

Y_i* = β X_ij + ε_ί  Y = Y_i* if Y_i* > 0 (double-up adopter)

Y_i* = β X_ij + ε_ί  Y = 0   if Y_i* ≤ 0 (non-adopter)

where Y_i* = dependent variable;

• X = the vector of factors influencing the double-up adoption of climate change adaptation strategies;
• β = the vector of unknown parameters;
• ε_i = the stochastic error term.

Then, the model is specified as follows:

$${\mathrm{Y}}_{\mathrm{i}}=\mathrm{Log} ({\mathrm{P}}_{\mathrm{i}}/{\mathrm{P}}_{\mathrm{i}}-1)=\beta o+\sum \beta i {\mathrm{X}}_{\mathrm{i}j}+{\epsilon }_i$$

$$\begin{array}{ll}{Z}_i& ={\beta }_0+{\beta }_{1}Landholding\_size+{\beta }_{2}Househod\_size+{\beta }_{3}Inorganic\_fertilizer+{\beta }_{4}Total\_labour\\ & +{\beta }_{5}Seed\_access+{\beta }_{6}Adherence\_to\_extension+{\beta }_{7}Input\_coupon\_access+{\beta }_{8}Access\_to\_credit\\ & +{\beta }_{9}Head\_gender+{\beta }_{10}Head\_age+{\epsilon }_i\end{array}$$

2.8. Definition of Variables

The choice of the explanatory variables for this study was based on the literature, availability, and completeness of such variables within the dataset. These input variables include the age of the household head, the gender of the household head, household size, landholding size, inorganic fertilizer use, seed access, input coupon access, adherence to agriculture extension services, and access to credit. The hypothesized signs of these variables are shown in

Table A2. Hypothesized signs of explanatory variables.

Variable	Unit of Measure	Expected Sign
Household head age	Number of years	+/−
Household head gender	1 = male, 0 = otherwise	+/−
Landholding size	Area in acres	+/−
Inorganic fertilizer use	1 = yes, 0 = otherwise	+/−
Seed access	1 = yes, 0 = otherwise	+
Input coupon access	1 = yes, 0 = otherwise	+
Adherence to extension services	1 = yes, 0 = otherwise	+
Access to credit	1 = yes, 0 = otherwise	+
Household size	Number of household members	+/−
Total labor	Man-days	+

. Furthermore, the study used a qualitative variable defined as a double-up adopter if a smallholder farmer had incorporated within their maize-based cropping system a mix of at least two climate change adaptation strategies, namely maize–legume diversification, organic fertilizer use, and the practice of agroforestry; otherwise, it used a qualitative variable defined as a non-adopter.

2.8.1. Age of Household Head

The age of the household head is a continuous variable presented as the number of years of a particular household member. According to Marenya and Barret [42], age is always equated with experience; the probability of a farming household being an adopter of new agricultural practices tends to increase with age. However, unlike young farmers, older and more experienced ones may be reluctant to switch from traditional practices to new practices. Therefore, the coefficient of this variable is not assigned a prior sign.

2.8.2. Gender of Household Head

The gender of the household head is a dummy variable for which a value of 1 denotes that the household head is male, and a value of 0 denotes that the household head is female. According to Kamanet et al. [43], a male head of household is expected to have a large influence on the adoption of advanced and sustainable farming practices due to their increased access and control over production resources compared to their female counterpart. However, a smallholder farmer can choose to dually adopt climate change adaptation strategies regardless of their gender and merely based on their choice or preference.

2.8.3. Landholding Size

Landholding size is a continuous variable referring to the total area of land in acres owned by a farmer. Larger landholding size determines the number of crops, and in general, the cropping system, which a farmer can use in their farm, and it makes them less risk averse and thus more likely to engage in sustainable mix cropping. However, according to Yirga and Hassan [44], farmers with small landholding size are forced into intercropping in order to produce more on their limited land. Therefore, the effects of landholding size on farmers’ dual adoption choice may not be assigned a prior expectation in an empirical model.

2.8.4. Total Labor

Total labor is a continuous variable captured in man-days, and it includes both family and hired labor utilized for pre- and post-planting, including harvesting activities. Farming households hire labor to supplement insufficient domestic labor. Therefore, the hypothesized sign of the coefficient for the total labor variable is positive.

2.8.5. Inorganic Fertilizer Use

Inorganic fertilizer use is a dummy variable and refers to whether smallholder maize farmers used fertilizer during the agricultural season. Fertilizer is one of the agriculture inputs, which increases the level of productivity of the farming sector. However, practices such as organic fertilizer use and maize–legume diversification can enrich the soil with the required nutrients; hence, the hypothesized sign of the coefficient of the variable is not certain.

2.8.6. Seed Access

Seed access is a dummy variable, where a value of 1 is assigned to farmers with access to seeds, and a value of 0 is assigned otherwise. Smallholder farmers can only produce crops when they have seeds [45]. Therefore, having access to seeds is hypothesized to have a positive influence on the double-up adoption of climate change adaptation strategies.

2.8.7. Input Coupon Access

Input coupon access is a dummy variable for which a value of 1 denotes households with access to an input coupon, and a value of 0 denotes otherwise. An input coupon is hypothesized to positively influence the double-up adoption of climate change adaptation strategies, in the sense that by allowing farmers access to subsidized agricultural inputs such as fertilizer, maize and legume seeds become affordable at the market.

2.8.8. Adherence to Agricultural Extension Services

Adherence to agricultural extension services is a dummy variable, and it is denoted as 1 if a smallholder farmer had access and was able to follow or use the agricultural extension services rendered to them, and it is denoted as 0 otherwise. The adaptation practices highlighted in this study—such as maize–legume diversification, organic fertilizer use, and agroforestry—are perceived to be knowledge-intensive and require technology awareness as a precondition for their adoption [46]. Therefore, adherence to agricultural extension services is hypothesized to positively influence farmers’ choice on the dual adoption of climate change adaptation practices.

2.8.9. Access to Credit

Access to credit is a dummy variable for which a value of 1 is assigned to households with access to credit, and a value of 0 is assigned otherwise. Access to credit is viewed as leverage over financially constrained smallholder farmers, enabling them to purchase the much needed agricultural inputs for the farm [47]. Therefore, access to credit is hypothesized to positively influence smallholder farmers’ dual adoption choice.

2.8.10. Household Size

Household size is a continuous variable entailing the number of members within the household. Studies by Kankwamba et al. [23] and Sichoongwe et al. [48] used this variable as a proxy for labor availability, such that larger households have more labor to cope with the increased labor intensity associated with a mixed cropping system. However, it is also more likely for these larger households to specialize in staple grains in order to produce more food and supply the surplus labor to other off-farm activities, so as to buy what is not produced [20]. Therefore, the coefficient of household size is not assigned a prior sign in influencing farmers’ decision on dual adoption.

2.9. Diagnostic Test

Diagnostics for binary choice models (BCMs), such as the logit model, requires the results of analyses to be valid, as the model has to meet and satisfy the assumptions of BCM. Therefore, in this study, before running the logit model, some diagnostic tests were performed to establish precise statistical inference and verify whether the model fits the data sufficiently well. Some of these tests included (i) a goodness-of-fit test, carried out to check whether the overall model is statistically significant; (ii) a heteroscedasticity test, intended to cross-check a phenomenon where the variance of the outcome variable is not the same for any independent observation, causing very high standard errors and inconsistent sample estimates; and (iii) a multicollinearity test, as multicollinearity is a problem, which arises when two or more explanatory variables in a regression model are highly correlated, making it impossible to distinguish the accurate independent effect of such parameter estimate on the explained variable.

3. Results and Discussion

3.1. Results of Diagnostic Tests for Econometric Problems

The likelihood ratio test, as provided by the logit model, indicated that the overall model was significant at 1%, implying that the estimated logit model fits the data precisely. As shown in

Table A3. Test for heteroscedasticity.

	Null Hypothesis (Ho)
Test		Test Statistics	Prob > Chi2
Breusch–Pagan test	Constant variance	Chi2(1) = 96.83	0.0000

, heteroscedasticity was tested using the Breusch–Pagan test, and the problem of unequal variance was detected within the data. Therefore, the robust standard error was used as a remedy to deflate the standard errors, thus rectifying the heteroscedasticity problem. Moreover, to detect the presence of multicollinearity in the data, both the variance inflation factor (VIF) and pairwise correlation were employed. All variables had VIF values, which were less than 10, as shown in

Table A4. Test for multicollinearity using variance inflation factor.

Variable	VIF	1/VIF
Inorganic fertilizer use	1.16	0.864992
Input coupon access	1.14	0.876121
Household size	1.10	0.908593
Head gender	1.10	0.911290
Landholding size	1.08	0.922115
Head age	1.08	0.924116
Seed access	1.04	0.963141
Access to credit	1.04	0.964276
Adherence to extension services	1.04	0.965663
Total labor	1.01	0.991580

. Equally, the pairwise correlation values for the dummy variables were all less than 0.7, as shown in

Table A5. Test for multicollinearity for dummy variables using pairwise correlation.

	Seed access	Inorganic fertilizer use	Adherence to extension services	Access to credit	Input coupon access	Head gender
Seed access	1.000
Inorganic fertilizer use	0.2829	1.000
Adherence to extension services	0.2231	0.3281	1.0000
Access to credit	0.0492	0.0461	0.0494	1.0000
Input coupon access	0.0435	0.4066	0.1496	0.0050	1.0000
Head gender	−0.0389	−0.0247	−0.0214	−0.0462	0.0236	1.0000

. Both of these results imply that there was no multicollinearity problem in the data.

3.2. Descriptive Analysis

Table 1. Adoption of climate change adaptation strategies across regions.

Region							Total
Maize–Legume Diversification	North		Central		South
	Frequency	Percent	Frequency	Percent	Frequency	Percent	Frequency	Percent
Non-diversifier	859	22.49	1543	40.39	1418	37.12	3820	42.72
Diversifier	653	12.75	1601	31.26	2868	55.99	5122	57.28
Total	1512	16.91	3144	35.16	4286	47.93	8942	100
Maize: Organic fertilizer use
Non-adopter	1194	18.18	2306	35.12	3066	46.70	6566	73.43
Adopter	318	13.38	838	35.27	1220	51.35	2376	26.57
Total	1512	16.91	3144	35.16	4286	47.93	8942	100
Maize: Agroforestry
Non-adopter	667	14.10	1347	28.48	2715	57.41	4729	52.89
Adopter	845	20.06	1797	42.65	1571	37.29	4213	47.11
Total	1512	16.91	3144	35.16	4268	47.93	8942	100

below specifically presents an insight in terms of the adoption rate of each climate change adaptation strategy across the three regions in Malawi. Out of the three climate change adaptation strategies considered in this study, maize–legume diversification was the most highly adopted option among the total sampled population of 8942 smallholder maize farmers, with the majority of adopters found in the southern region of Malawi. Categorically, 5122 farm households diversified their maize with grain legumes, representing 57.28%, while 3820 households did not diversify, representing 42.72% of non-diversifiers. The second most highly adopted strategy was the practice of agroforestry within maize-based farms, where, out of a total of 8942 sampled smallholder maize farmers, 4213 households adopted the planting of trees in their maize farms, representing 47.11%, while 4729 households did not plant trees, representing 52.89% of non-adopters. Lastly, organic fertilizer use in the maize-based cropping system appeared to be the least adopted strategy, as only 2376 households used organic fertilizer, representing 26.57%, while 6566 households did not use organic fertilizer, representing 73.43% of non-adopters.

Table 2. Double-up adoption of climate change adaptation strategies.

Region							Total
Double-Up Adoption of Climate Change Adaptation Strategies	North		Central		South
	Frequency	Percent	Frequency	Percent	Frequency	Percent	Frequency	Percent
Non-adopter	967	18.04	1797	33.53	2595	48.42	5359	59.93
Adopter	545	15.21	1347	37.59	1691	47.20	3583	40.07
Total	1512	16.91	3144	35.16	4286	47.93	8942	100

below shows the overall dual adoption rates of adaptation strategies to climate change. Out of 8942 households, which produced maize, 3583 households had dually adopted climate change adaptation strategies, representing 40.07%, while 5359 households did not dually adopt climate change adaptation strategies, representing 59.93% of non-adopters. Additionally, dual adoption was higher in the southern region, with 47.20% of households, followed by 37.59% and 15.21% of households in the central and northern regions, respectively. Furthermore, this study’s results are also in line with findings in the IHS4 report, which indicated that the southern region registered the highest proportion of plots, which were intercropped, followed by the central and northern regions [31].

Sichoongwe et al. [48] indicated that the adoption of climate change adaptation strategies such as crop diversification is influenced by a variety of characteristics. Some of these factors include socio-economic, demographic, and institutional characteristics. The results in

Table 3. Descriptive statistics for dummy variables using chi-square test.

Characteristics	Double-Up Adopters		Non-Adopters
	(n = 3583)		(n = 5359)
Demographic and Socio-Economic Factors	Frequency	Percent	Frequency	Percent	p-Value
Head gender (male = 1)	2562	71.50	3660	68.30	0.120
Inorganic fertilizer use (yes = 1)	2682	74.85	3613	67.42	0.000
Seed access	1978	55.21	2533	47.27	0.000
Institutional factors
Input coupon access (yes = 1)	977	27.27	1180	22.02	0.000
Access to credit	1006	28.08	1135	21.18	0.000
Adherence to extension services	2590	72.29	3329	62.12	0.000

below, obtained using the chi-square test, revealed that inorganic fertilizer use, access to seed, access to input coupon, access to credit, and adherence to extension services were all statistically significant at 1%. This implies the existence of associations between these dummy causal variables and the dual adoption of climate change adaptation strategies in which the majority of dual adopters had access to seed, input coupons, and credit; used inorganic fertilizer; and adhered to extension services more than non-adopters.

The results in

Table 4. Descriptive statistics for continuous variables using t-test.

Characteristics	Double-Up Adopters		Non-Adopters
	(n = 3583)		(n = 5359)
	Mean	Standard Error	Mean	Standard Error	p-Value
Age of household head	44.90	0.2681076	44.47	0.2262526	0.2168
Landholding size (area in acres)	1.83	0.276157	1.34	0.0195725	0.0000
Household size	4.59	0.0327898	4.39	0.026778	0.1200
Total labor	1910.15	189.4065	1195.45	142.8828	0.2020

below, obtained using the t-test, revealed a statistically significant difference in terms of landholding size between double-up adopters and non-adopters, such that the mean landholding size for dual adopters was 1.83 acres, which was higher than the mean size of 1.34 acres for non-adopters, and this result was significant at 1%.

3.3. Factors Influencing Farmers’ Decision to Dually Adopt Climate Change Adaptation Strategies

3.3.1. Results of Logit Model

The results of the logit model in

Table 5. Results of the logit model and its marginal effects on factors influencing dual adoption of climate change adaptation strategies.

	Dependent Variable: Double-Up Adoption of Climate Change Adaptation Strategies
Explanatory Variable	Coefficient	Marginal Effects Coefficient	Standard Errors	p-Value
Landholding size (area in acres)	0.2151 ***	0.0515 ***	0.0066 ***	0.000
Household size	0.0045	0.0011	0.0029	0.706
Inorganic fertilizer use (1 = yes, 0 = no)	0.1398 ***	0.0333 ***	0.0124 ***	0.007
Total labor (man-days)	4.87 × 10−6	1.17 × 10−6	0.000	0.144
Seed access (1 = yes, 0 = no)	0.2920 ***	0.0698 ***	0.0107 ***	0.000
Adherence to extension services (1 = yes, 0 = no)	0.3328 ***	0.0787 ***	0.0112 ***	0.000
Input coupon access (1 = yes, 0 = no)	0.2206 ***	0.0534 ***	0.0133 ***	0.000
Access to credit (1 = yes, 0 = no)	0.2910 ***	0.0706 ***	0.0128 ***	0.000
Head gender (1 = male, 0 = otherwise)	0.0157	0.0038	0.0121	0.757
Head age (years)	0.0008	0.0002	0.0003	0.592
Constant	−1.4144		0.1236

Log likelihood = −5807.7248, LR Chi² (10) = 291.86, Prob > Chi² = 0.0000, Pseudo-R² = 0.0354. Note: Significance level: *** (1%).

show that the overall model was significant at 1%, implying that all input variables included in the model were jointly capable of explaining the variations in the explained variable. Therefore, the dual adoption of adaptation strategies to climate change is dependent on landholding size, inorganic fertilizer use, seed access, adherence to agricultural extension services, input coupon access, and access to credit. These variables were all statistically significant at 1%. However, household size, total labor, household head gender, and age were found not to be significant.

3.3.2. Landholding Size

This study shows that the landholding size was significant at 1%, and the sign of its coefficient was positive, as expected, demonstrating the existence of a direct relationship between landholding size and the dual adoption of adaptation strategies to climate change. Therefore, increasing the landholding size of a smallholder maize farmer by one acre will lead to a corresponding increase in the farmer’s probability to dually adopt by 5.2% considering the household size, inorganic fertilizer use, total labor, seed access, adherence to extension services, input coupon access, access to credit, head gender, and head age constant. Land is considered one of the limited resources owned by small-scale farmers; thus, increasing the landholding size will always motivate farmers to diversify by producing a variety of crops [48]. Furthermore, this study result also concurs with the findings of Maggio et al. [49], who cited landholding size as a vital factor regarding the adoption of multiple cropping systems as compared to the crop specialization of maize in sub-Saharan African countries. Therefore, to ensure that smallholder maize farmers are able to dually adopt climate change adaptation strategies, policies safeguarding land rights and promoting increased access to land need to be prioritized.

3.3.3. Inorganic Fertilizer Use

The variable of inorganic fertilizer use was significant at 1% and had a positive coefficient, implying a direct relationship between the use of inorganic fertilizer and double-up adoption of adaptation strategies to climate change; farm households who used inorganic fertilizer were 3.3% more likely to dually adopt than those who did not use inorganic fertilizer, considering landholding size, household size, total labor, seed access, adherence to extension services, input coupon access, access to credit, head gender, and head age constant. This study finding agrees with that of Sichoongwe et al. [48], who found that the availability and use of fertilizer give farmers an incentive to adopt certain farming technologies, such as crop diversification, since most smallholder farmers lack fertilizer. Additionally, a study by Kanyamuka [46] stressed the existence of a complementary relationship between the use of inorganic fertilizer and the intercropping of legumes in a maize-based cropping system, unlike in a rotation pattern. Therefore, in Malawi, most smallholder farmers practice intercropping more than rotation due to small landholdings, hence the need for inorganic fertilizer.

3.3.4. Seed Access

The variable of seed access was significant at 1% and had a positive coefficient, indicating a direct relationship between seed access and the double-up adoption of adaptation strategies to climate change. The marginal effects of this variable indicate that smallholder maize farmers with access to seeds were 6.9% more likely to dually adopt than those who did not have access to seeds, considering variables such as landholding size, household size, inorganic fertilizer use, total labor, adherence to extension services, input coupon access, access to credit, head gender, and the age of the household head constant. This finding is also consistent with that of Kanyamuka [46], who indicated that the adoption of farming technologies such as intercropping and the rotation of legumes in a maize-based cropping system largely relies on seeds, hence the need to increase the access to and availability of both legume and maize seeds. This can be achieved by the government and private sector increasing investment in the research and development of improved seeds for crops, which will be both tolerant to extreme weather conditions and made available to farmers at relatively affordable prices.

3.3.5. Adherence to Extension Services

Even though access to agricultural extension services is a prerequisite in any technology adoption, both the access to and use of such extension services are crucial for smallholder maize farmers to dually adopt adaptation strategies to climate change. The variable of adherence to extension services was statistically significant at 1%, and its positive coefficient depicted a direct relationship between adherence to extension services and the dual adoption of adaptation strategies to climate change. Smallholder maize farmers who had access to and used the extension services provided were 7.9% more likely to dually adopt than farmers who did not adhere to the extension services provided. Strategies such as maize–legume diversification, organic fertilizer use, and the practice of agroforestry are considered knowledge-intensive technologies; hence, their dual adoption will definitely require a better and deeper understanding by farmers, specifically in ensuring the proper and successful management of crops and the whole cropping system in general. This result concurs with that of Maggio et al. [49], where access to extension services was identified as an important pull factor driving the adoption of legume incorporation into cropping systems. Therefore, the policy advice is that extension service providers should use participatory approaches to ensure not only the access of farmers to extension services but also their active participation to allow them to develop an in-depth understanding, so as to apply the acquired knowledge in their farms.

3.3.6. Input Coupon Access

This study shows a direct relationship between input coupon access and the dual adoption of adaptation strategies to climate change, as the variable had a positive coefficient and was significant at 1%. Therefore, smallholder maize farmers who had access to input coupons were 5.3% more likely to dually adopt than those with no input coupon access, considering landholding size, household size, inorganic fertilizer use, total labor, seed access, adherence to extension services, access to credit, the gender of the household head, and the age of the household head constant. This finding is in line with the notion that input coupons administered through the Farm Input Subsidy Program (FISP) by the Malawian government have facilitated the availability and affordability of agricultural inputs such as maize and legume seeds, as well as fertilizers, among smallholder maize farmers, hence encouraging the adoption of strategies such as maize–legume intercropping [50].

3.3.7. Access to Credit

Access to credit had a positive coefficient, implying the presence of a direct relationship between access to credit and the double-up adoption of adaptation strategies to climate change, and the variable was significant at 1%. Smallholder maize farmers with access to credit were 7.1% more likely to dually adopt than those who did not have access to credit, considering landholding size, household size, inorganic fertilizer use, total labor, seed access, input coupon access, adherence to extension services, head gender, and head age constant. Basically, most smallholder farmers in Malawi are characterized as being resource-poor; hence, they lack the financial capital necessary to purchase the required agricultural inputs for production. Therefore, access to credit from either public or private lending institutions lessens the financial burden on smallholder farmers and enhances the availability of agricultural inputs, which trigger farmers’ decision to consider dual adoption. This finding coincides with that of Lambrecht et al. [51], who highlighted that access to credit is one of the institutional factors, which expedites resource-constrained farmers’ access to relatively unaffordable and externally purchased agricultural inputs, such as inorganic fertilizer and improved seeds. Therefore, the policy implication is that, rather than operating as an individual, farmers should be encouraged and motivated by the government and stakeholders in the private sector to form and join farmer cooperatives or associations, where they can either obtain the required credit from fellow members or use their membership as a loan security from financial institutions.

3.3.8. Limitation of the Study

The limitation of this research is that the study relied on a secondary dataset; hence, some variables of interest, which would have been included in the model for analysis, were omitted on the basis of completeness, as they had a large number of missing values. Therefore, this study only focused on and analyzed the demographic, socio-economic, and institutional characteristics of smallholder maize farmers as determining factors toward their dual adoption decision of climate change adaptation strategies.

4. Conclusions and Recommendations

To summarize, this study focused on smallholder maize farmers in Malawi, who were categorized into dual adopters and non-adopters of climate change adaptation strategies, namely maize–legume diversification, the use of organic fertilizer, and the practice of agroforestry. Using IHS4 data, descriptive statistics were computed to compare the demographic, socio-economic, and institutional characteristics of dual adopters and non-adopters. Specifically, the chi-square test for dummy variables revealed that the majority of dual adopters had access to seeds, input coupons, and credit; used inorganic fertilizer; and adhered to extension services more than non-adopters. Furthermore, the t-test results portrayed a significant difference in terms of the mean landholding size, with dual adopters having 1.83 acres, which was higher than the mean size of 1.34 acres for non-adopters. Further, to determine the factors influencing the dual adoption decision of smallholder maize farmers, the logit model was used. Based on results of the analysis, this study found that landholding size, access to credit, adherence to agricultural extension services, input coupon access, inorganic fertilizer use, and seed access were all statistically significant in terms of influencing the dual adoption of climate change adaptation strategies. Importantly, attending to the needs of smallholder maize farmers, as highlighted in this study, can potentially scale up or intensify the level of farmers’ responsiveness to climate change through the dual adoption of these sustainable adaptation strategies capable of meeting and sustaining not only farmers’ immediate needs of food and income but also their long-term need of enriched fertile soils.

Therefore, the policy implications and advice based on this study’s findings are as follows: a vast investment in research is needed to develop new good-quality varieties of both maize and grain legume seeds, which are well adapted to different ecological zones; adherence to agricultural extension services needs to be enhanced through the use of participatory approaches to improve farmers’ participation in and understanding of adaptation practices; increased land access and policies defending land rights need to be institutionalized to ensure land ownership, which can trigger long-term decisions by farmers regarding their farms, such as practicing agroforestry. Furthermore, farmers need to be encouraged to form and join farmers’ cooperatives or associations, which can ease and improve their access to credit from financial institutions.