Livelihood Impact and Risk Management of Drought: The Experiences of Farming Households in the Free State Province, South Africa

26 Drought is one of the main causes of food insecurity, malnutrition and poverty. It is therefore 27 important to understand the perception of farmers on socioeconomic and environmental 28 impacts of drought and the strategies employed to manage it. Using data collected from 301 29 smallholder households in Thaba Nchu, Free State Province, the study contributes to three 30 perspectives: analysing the perceived socio-economic and environmental impact of drought, 31 examining the determinants of the perceived impact, and identifying factors affecting the 32 intensity of drought-risk management practices used by smallholder farming households. 33 Using 11 indicators as a measure of perceived impact, the findings from the principal 34 component analysis (PCA) revealed three main dimensions of perceived drought impact: 35 economic, environmental and social impacts. Different socio-economic and institutional 36 factors have a different influence on the three dimensions. In addition, factors such as age, 37 household size, non-farm work and extension services are significant in determining the 38 intensity of drought-risk management strategies implemented by farmers in the study area. The 39 study therefore recommends that climate risk management be integrated into the provision of 40 extension services, particularly in drought-prone areas such as the Free State Province. 41 Encouraging farmers to engage in non-agricultural economic activities is also crucial, as this 42 can serve as insurance against events such as drought.


53
The evidence of climate change is real, and its impacts is felt all over the world, with 54 agricultural households in both developed and developing economies suffering the most 55 (Wang, 2021). Agricultural production in sub-Saharan Africa (SSA) is highly dependent on 56 weather patterns, particularly high temperatures and rainfall. With the rise in global 57 temperature, inter-annual temperature and precipitation variability will also increase (Dai,58 2013). For those whose livelihoods depends on agricultural activities, these irregular patterns  Extreme variability in rainfall patterns leads to floods and droughts, leading to a reduction 70 in GDP growth in many SSA countries (Kogo et al., 2020). Drought is considered to be the 71 most devastating climate-related events in the world, causing serious damage to agricultural, 72 environmental and socioeconomic activities (Moeletsi and Walker, 2012). Drought occurs 73 when there is a lack of sufficient rainfall over a long period of time (usually months) leading 74 to water shortages and adverse impacts on the lives of people, animals and the loss of vegetation 75 (Msangi, 2004). Apart from a decrease in rainfall giving rise to drought, drought can also result 76 from human activities such as a change in land use (Keesstra, 2007;Mohmmed et al., 2018). 77 Essentially, the occurrence of drought can be divided into four main factors, depending on the 78 impact: agricultural, socioeconomics, meteorological and hydrological (Thilakarathne and 79 Sridhar, 2017). The frequency and severity of droughts have increased significantly in recent 80 years as a result of climate change and agricultural activities in developing economies such as 81 Africa countries (Panthi et al., 2016). 82 South Africa has relatively less rainfall; therefore it is generally described as a semi-arid 83 and water-stressed country (Botai et al., 2016). The average annual rainfall in the country is  Risk Management (DRM) can be defined as strategies put in place to mitigate adverse effects 101 while trying to pursue positive objectives (Hansen et al. 2004). DRM is part of climate risk 102 management techniques that refer to climate mitigation strategies decision-making, where 103 farmers use the information they receive to take action to minimize climate risk and exploit 104 climate opportunities (Hellmuth et al., 2009;Shiferaw et al., 2014). DRM tools can be divided 105 into two: ex-ante DRM (coping mechanisms) and ex-post (adaptation mechanisms), depending 106 on whether the strategy reduce risk exposure or minimize the impact of undesirable outcomes 107 after the shock (Owens et al., 2003). More often than not, farmers dwelling in drought-prone 108 areas modify their production systems such that the likely impacts of the shock could be 109 minimized within acceptable level.  Perception and DRM strategies are the two significant elements in the process of adaptation.

136
Farmers first need to perceived the impact of drought in order to help them take appropriate 137 action to mitigate their vulnerability and build their adaptive capacity (Bryan et al., 2009).

138
Misleading perceptions about the consequences of climate change and variability may lead to 139 no action taken to manage the risk. However, farmers who perceive potential consequences of 140 extreme weather conditions such as drought are more likely to take precautionary measures to 141 minimize its effects and support programs that aims to address it (Niles et al., 2013). Drought    186 This study followed economic theory in order to achieve its objectives. First, PCA was used to  Consider that the following matrix denote the relationship between farmers' perceived 206 impact of drought and a set of explanatory variables:

Analytical frameworks
215 where k P P ... 1 denote principal components that have eigenvalues greater than one and . n k  216 k can be equal to one, where only one factor is retained or greater than one where more than 217 one factors are retained as in the case of this study (see Table 3).  The perceived impact of drought equations with three dependent variables can be estimated 231 through SUR as follows: These equations are described as seemingly unrelated regression model and can expressed as; The SUR model is based on the assumption that the residuals have zero mean or the i X are 236 strictly exogenous, homoscedastic and uncorrelated across observations. However, the 237 equations within the system is correlated.

Intensity of drought-risk management strategies -Count data modelling 242
The number of strategies used by individual farmers described the dependent variable in order 243 to identify factors that influence the intensity of the drought-risk management strategies. 244 Cameron and Trivedi (1990) argued that the number of strategies adopted was a measure of is the number of drought-risk management 264 practices employed, and L is a set of predictors.  and  are the parameter estimates. Greene

265
(2008) observed that the number of drought-risk management strategies used,  is given as:    Table 2 shows that the drought 308 management strategy most frequently used in the study area was NGO intervention, which was   346 Descriptive statistics on the socio-economic variables used in the models to achieve the study 347 objective are shown in  Nhemachena, 2008). Dissemination of information is key to managing any kind of risk, 360 including drought. Sources of information, such as farmer groups and extension services, as 361 well as access to climate change information (e.g. information on early warning systems) are 362 therefore key to managing the risk of drought. Table 2 shows that some 45% are members of   in their inability to take food rather than malnutrition. With regard to wildlife migration and 378 frequent wild fires, they believe that drought has an important role to play. The mean severity 379 score for the drying of water resources exceeds 3, suggesting that drought is perceived to have 380 some serious consequences for their water bodies.   The impact of drought on the environment may result in damage to the quality of the air and 424 water, forest and range fires and wildlife habitats and damage to the quality of the landscape.  437 In order to guide effective and efficient policy options for climate change in general, and 438 drought in particular, it is important to understand the factors that influence farmers' perceived 439 impact of drought in three perspectives: economic, environment and social. Table 5   The respondent's gender leads to a positive perception of the economic and environmental   490 Determinants of intensity of drought-risk management practices adopted by agricultural 491 households in the study area are reported in Table 6. Two count data models were estimated: should therefore be the focus model.

504
From Table 6, two variables of household demographic characteristics: age and household 505 size influence the intensity of drought-risk management practices. The age of the respondent 506 responds negatively to the number of drought-risk management practices adopted, suggesting 507 that elderly is less responsive to adopting a number of strategies to combat the negative impact 508 of drought compared to young and energetic respondents. This could be attributed to the fact 509 that the younger generation is more innovative, less risky, always looking for information and 510 ready to improve agricultural productivity. Older farmers may not be aware of modern 511 agricultural innovations and strategies to minimize the risks and adverse effects of drought.

512
This is consistent with the study of Ali and Erenstein (2017). Household size is negatively 513 correlated with the number of drought-risk management practices adopted by farmers. This 514 could be because drought-risk measures such as crop insurance, maintaining flexibility (e.g.,

515
planting drought-tolerance crop varieties) are quite expensive and households with large 516 financial burdens due to many dependents may not afford to purchase these coping strategies.   impact on their crops may purchase crop insurance, which will be compensated by insurance 554 companies for a proportion of their crop losses. They may also register and receive disaster 555 assistance from NGOs and the government. Drought may, however, create winners. This is 556 because farmers who have perceived the higher impact of drought on food prices may take

574
PCA results show that farmers understand how climate change, particularly drought, affects 575 their lives and the environment, and that, on average, farmers feel that the impact of drought is 576 serious. Gender, age, household size and farming experience are demographic factors that tend 577 to influence their perception of the impact of drought. Access to information on climate change 578 was also a significant factor in how farmers perceived the impact of drought. Knowledge is a 579 significant precursor for people to form their perception, which is intended to help them adapt 580 appropriately to situations. In the event of drought, farmers with prior information may be less 581 vulnerable as they may have prepared themselves in advance. It is therefore important to 582 strengthen improved early warning systems in local communities in order to reduce 583 vulnerability. The dissemination of such climate information to provide prospects with lead 584 times of two to six months prior to the start of the event will help farmers plan their adaptation 585 and coping strategies.

586
The results further indicated that obtaining assistance from NGOs and communities and 587 selling or mortgaging their farmland are common drought-risk management strategies used by 588 farmers to cope with the negative outcome of the drought. Factors that have a significant effect 589 on intensity of drought-risk management strategies include age, household size, non-farm work 590 and extension services. The study therefore recommends that demand-driven extension 591 services that address the needs of people at a particular point in time is critical to the lives of 592 farmers. Climate risk management can be integrated into the provision of extension services, 593 particularly in drought-prone areas such as the province of Free State. In this case, farmers are 594 not only trained in agricultural technology, but also in the management of drought-risk. This 595 makes farmers less vulnerable to climate shocks and serves as a strategy to combat food 596 insecurity and long-term poverty. Encouraging farmers to engage in non-agricultural economic 597 activities is also critical, as this can serve as insurance against events such as drought. Availability of data and materials 602 The datasets used to analyse this study are available from the corresponding author on 603 reasonable request. The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.