Prepare for the unanticipated: Portfolios of coping strategies of rural households facing diverse shocks

Rural households have established various informal strategies to cope with unanticipated shocks. These existing coping strategies are receiving renewed interest, particularly in the context of climate change and in terms of the role they do, and can play, in enhancing households' adaptive capacity. An improved understanding of these strategies, and the factors that influence their application, may support the design of locally relevant adaptation strategies. We explored the nature and prevalence of unanticipated shocks, including natural hazards, experienced by households in two villages in Venda, South Africa, with the villages selected, in part, because of notable differences in precipitation. We considered the influence of shock type, and householdand location-specific characteristics, on the use of various household-level coping strategies. We report on semistructured interviews, administered to 170 randomly selected households, and a participatory rural appraisal. Almost 90% of households reported the experience of at least one unanticipated shock over a prescribed 5-year period, with natural hazards reported by 42%. The type of shock experienced and various household-level characteristics, such as households' access to human and financial capital, influences households' coping response. Households' access to natural and social capital allowed for the protection of ex-ante coping options. Overall, our findings indicate that when possible, households actively manage their coping strategy portfolio, both in response to the shock experienced and in anticipation of future shocks. Generally, households' informal, ex-post coping options appear to be insufficiently robust for the covariate nature of natural hazards, suggesting the need for interventions that support households’ existing coping portfolio.

Digital reproduction on this site is provided to CIFOR staff and other researchers who visit this site for research consultation and scholarly purposes. Further distribution and/or any further use of the works from this site is strictly forbidden without the permission of the Journal of Rural Studies.
You may download, copy and distribute this manuscript for non-commercial purposes. Your license is limited by the following restrictions: 1.
The integrity of the work and identification of the author, copyright owner and publisher must be preserved in any copy.

2.
You must attribute this manuscript in the following format: This is an accepted version of an article by Paumgarten, F., Locatelli, B., Witkowski, E.T. and Vogel, C., 2020. Prepare for the unanticipated: Portfolios of coping strategies of rural households facing diverse shocks. Journal of Rural Studies. https://doi.org/10.1016/j.jrurstud.2020.05.013

Introduction
Rural households inhabit a state of often persistent and complex uncertainty, which undermines their efforts to secure a sustainable livelihood (DFID, 1999). This includes the experience of more unanticipated shocks, including climate-related shocks such as natural hazards. In general, more marginalized groups, including poorer, female-headed, and natural resource dependent households, tend to be more vulnerable to, and less able to cope with, these various shocks (Shackleton and Shackleton, 2004;Börner et al., 2015;Djoudi et al., 2016;Flatø et al., 2017). In developing countries, where public safety-nets are limited, and where poor, rural households have limited access to formal, private insurance, households have established various informal strategies to cope with shocks. These coping strategies consist of short-term responses to immediate threats, through which households, within the ambit of their available resources, aim to minimize the impact of the stress (Blaikie, 1994;Eriksen and Silva, 2009;Coulibaly et al., 2015;Mehar et al., 2016). Common strategies include the liquidation of assets (e.g. livestock); the reallocation of labour; changes in household spending and consumption; the use of formal and informal savings and loans and, the reliance on family-and community-support networks (kinship) and on external organizations (Corbett, 1988;Dekker, 2004;Heemskerk et al., 2004;Cooper et al., 2008;Eriksen and Silva, 2009;Adimassu and Kessler, 2016). More recently, attention has been drawn to the safety-net function of natural resources (including non-timber forest products (NTFPs)), including their role in smoothing consumption and income variations (McSweeney, 2005;Paumgarten and Shackleton, 2011;Mugido and Shackleton, 2017).
Households' coping strategies are often distinguished as either ex-post or ex-ante strategies, with the former described as reactive strategies and the latter, as more anticipatory, precautionary and active strategies (Adams et al., 1998;Cooper et al., 2008;McCarthy and Martello, 2010). Households' reliance on ex-post strategies is often framed within the context of limited control over their coping options and limited access to alternatives, although some suggest that even these expost strategies require some degree of anticipation (Cooper et al., 2008;Börner et al., 2015). For example, the liquidation of assets such as livestock requires their initial, anticipatory accumulation. Households may also purposefully turn to these ex-post options to manage their future insurance options, with Corbett (1988) arguing that households actively sequence and prioritize strategies.
Despite the common bundle of responses, households' ultimate selection and application of coping strategies (whether ex-post or ex-ante), and their overall capacity to cope, has been argued to depend on multiple factors, including the nature and duration of the shock, individual household characteristics (e.g. the gender of the household head), as well as contextual factors such as access to public and financial services, natural resources, and markets (Osbahr et al., 2008;Tran, 2015;Gautier et al., 2016;Van Aelst and Holvoet, 2016). Although these informal coping strategies may prove effective in the face of small to medium idiosyncratic shocks, concerns exist over their ability to insure against large, covariate shocks (e.g. natural hazards), which affect several households simultaneously (Skoufias, 2003;McSweeney, 2005;Günther and Harttgen, 2009;Wunder et al., 2014). Increasingly frequent and intense covariate, climate-related shocks, a possible outcome of climate change, may overburden households' existing coping portfolio and push households and communities beyond their capacity to cope (Smit and Wandel, 2006;Börner et al., 2015;Coulibaly et al., 2015).
Despite these concerns, households' existing coping strategies are receiving renewed interest, particularly within the context of climate change and in terms of the role they do, and can play, in enhancing households' adaptive capacity (Brockhaus et al., 2013;IPCC, 2014;Börner et al., 2015;Hänke and Barkmann, 2017). Eriksen and Silva (2009) argue, that in order to reduce households' overall vulnerability, there is the need to support both their ability to engage short-term coping strategies and to adapt to more long-term risks. Although distinctions between coping and adapting are generally framed on the timescale of response (Smit and Wandel, 2006;Osbahr et al., 2008), others suggest that this dichotomy may be misleading, with coping strategies either morphing into adaptation strategies, or with households using their coping strategies to adapt (Berkes and Jolly, 2001;Fabricius et al., 2007). Strengthening households' existing coping portfolio has therefore been described as a possible means of facilitating early forms of adaptation (Cooper et al., 2008;Castells-Quintana et al., 2018). Within this context, it has been argued that an improved understanding of households' past responses to shocks (including natural hazards), their existing coping options, the opportunities and constraints associated with these, as well as how households combine and protect their coping strategies, may provide useful insights into enhancing coping and facilitating adaptation (Smit and Wandel, 2006;Cooper et al., 2008;Börner et al., 2015;Mehar et al., 2016;Ofoegbu et al., 2016). This study sought to contribute towards this improved understanding. To do so we consider the nature and prevalence of multiple, unanticipated shocks, including natural hazards, experienced by rural households in Venda, South Africa. We explore the range and prevalence of coping strategies used in response to these shocks, and whether certain strategies are more often used to respond to particular shocks. In addition to exploring whether shock type influences the strategy used, we consider whether a range of household socio-demographic and -economic characteristics and location have an influence. Finally, we consider the respective coping strategies' limitations in buffering households against covariate natural hazards.
The remainder of this paper is organized as follows: section two describes the study area, section three describes the data collection and analysis processes, section four describes and discusses the main findings, including i.) households' exposure to multiple shocks, ii.) households' use of, and preference for, specific coping strategies, iii.) the associations between the different shocks and coping strategies and, iv.) the determinants of the strategies used. Section five presents the main conclusions and recommendations.

Study area
The research was conducted in the villages of Bennde Mutale (31°1'48''E; 22°24'14''S) and Vondo (30°21'60''E; 22°56'5''S), in the Vhembe District Municipality of South Africa's Limpopo Province, one of the country's poorest provinces (Statistics South Africa (SSA), 2014; Paumgarten et al., 2018). Prior to South Africa's 1994 democratic transition, both villages fell within the Venda homeland. "Homelands" were territories set aside, under South Africa's apartheid policies, for the resettlement of black South Africans (Thornton, 2002). These areas remain characterized by limited development, poor service provision, low education levels, high dependency ratios, and high unemployment rates, with state welfare grants and migrant remittances making an important contribution to households' income (SSA, 2014). To secure their livelihoods, households engage in diverse, land-based and off-farm livelihood strategies (Venter and Witkowski, 2013;Ofoegbu et al., 2016). Limited access to productive assets, a high incidence of HIV/Aids and food insecurity, and variable (and often extreme) weather conditions, including natural hazards such as floods and droughts, exacerbate household vulnerability (Neves and Du Toit, 2013;Venter and Witkowski, 2013;Shisana et al., 2014). Climate change, including increasing temperatures and decreasing rainfall, is likely to intensify households' overall vulnerability context (Turpie and Visser, 2013 (Mucina and Rutherford, 2011). In both sites, households continue to rely on natural resources, including NTFPs, however as Vondo falls within a more densely populated local municipality, households have less access to unconverted land. They do, however, have better access to the nearest, sizeable commercial centre (i.e. Thohoyandou). Although both sites receive summer rainfall, the difference in mean annual precipitation was one of the primary criteria for the site selection, with Bennde Mutale and Vondo receiving, respectively, 402 ± 141 mm and 1108 ± 371 mm for the period 1981-2014 (Funk et al., 2015). Bennde Mutale is warmer, with a mean annual temperature of 24.7 °C, versus 19.0 °C for Vondo. As such, Bennde Mutale is henceforth referred to as the "dry site", and Vondo as the "wet site". Natural hazards, including floods and droughts, are experienced in both sites.

Data collection
Quantitative and qualitative data were collected through a participatory rural appraisal (PRA), a semi-structured household survey, key informant interviews and personal observations.
Following a livelihoods approach, data were collected on households' characteristics, their livelihood strategies, their access to different assets (i.e. natural, social, financial, human and physical capital, as described in the Sustainable Livelihoods Framework (DFID, 1999)), their vulnerability context (including their experience of various unanticipated shocks), and the coping strategies employed in response to unanticipated shocks.
The PRA, conducted during voluntary community workshops, consisted of exercises such as the construction of seasonal calendars; historical profiles; time trends; and impact chains, which all allowed for discussions on the community's overall vulnerability context (DFID, 1999;Dazé et al., 2009;Turnbull and Turvil, 2012). Respondents discussed the experience of common climate-related shocks, specifically floods and droughts, their impacts, as well as the measures households, and the community, have historically used to cope with these.
The household survey was administered to 170 randomly selected households, 85 per village, and included questions on households' experience (over the preceding 5 years) of twelve common shocks identified during the PRA ( Table 2). The focus was on unanticipated shocks, for example, unexpected crop losses rather than predictable, seasonal crop shortfalls. Similarly, the costs associated with unpredictable social events (e.g. funerals) were considered, not those associated with more predictable, annual social events such as religious holidays. For each shock, households were asked to identify the coping strategies used, selecting from a list of twelve common strategies, also identified during the PRA.

Data analysis
We grouped the pre-selected shocks into four overarching groups, namely production, financial and human shocks as well as natural hazards ( Table 2). The selected responses were grouped into 8 categories and designated as either internal (e.g. adjustments to household savings or consumption) or external (e.g. assistance from outside the household) strategies.
The first analysis examined whether the type of shock experienced influences the coping strategies used. We represented the frequency of reported shocks and coping strategies in mosaic plots and used simple tests (independence test using Pearson residuals) to identify the strategies that were significantly more or less frequently associated with the different shocks ( Fig. 1). For the second analysis, we used a multivariate analysis to explore whether the adoption of a coping strategy was influenced by explanatory variables related to site and/or household characteristics (Table Annex 1). These variables fell into 5 categories, namely: 1.) basic household descriptors related to location and the demography of the household, 2.) households' available savings and assets (capital stocks), 3.) households' access to kinship and community-support networks, 4.) households' existing livelihood strategies and income sources, and 5.) households' existing human capital. Some skewed explanatory variables were log-transformed before the analysis. We used a logit linear model to explain the use of a strategy. In the version 3.3.2 of the R software (R Core Team, 2016), the "step" function was used to select variables for the model. The function started from a full model and successively dropped or added variables to minimize the AIC score. We reported the significant variables (p < 0.05) of the final model and described whether they had a positive or negative effect on the strategy used.   (Meyer et al., 2016). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Households' experience of multiple shocks
Production shocks were the most commonly reported (Table 2). Multiple factors caused unanticipated crop and livestock losses (workshops). For example, in the wet site, respondents noted crop losses from pest outbreaks while in the dry site cross-border cattle rustling occurs, driven by neighbouring Zimbabwe's economic crisis (Thornycroft, 2017) and livestock is occasionally killed by wildlife escaping the neighbouring conservation areas. Crop and livestock losses were also associated with extreme temperatures and unpredictable rainfall. Production shocks included the loss of, or damage to, household infrastructure and assets, including, for example, termite damage to food/seed storage facilities.
Human shocks were the second most commonly reported. Although primarily associated with an increase in dependents, these shocks included the death of household members and serious illnesses/injuries which resulted in either reduced labour availability (of the patient or the primary caretaker) and/or expensive medical treatment. For example, respondents in both sites reported an increase in HIV/Aids and diabetes, with one elder explaining that historically, it was unusual for young adults to die, with this no longer being the case. In South Africa, an increase in dependents is a complex shock, resulting in increased pressure on the household's existing resources, while simultaneously (potentially) increasing the household's access to government social grants.
The experience of impacts associated with natural hazards, was reported by 42% of households. Both sites experienced floods in 2013, the year preceding the fieldwork, with the dry site worst affected, receiving more rainfall in one week, than its annual average (Du Preez and Seale, 2013; Pringle et al., 2013). These floods, caused by a tropical low-pressure system, were associated with multiple, simultaneous impacts, including production, human and financial shocks (workshops), which may have, therefore, been more commonly reported than usual. For example, crops, livestock, irrigation pumps, agricultural and grazing land, and NTFPs were washed away; community infrastructure was damaged; and houses collapsed, damaging assets and food/seed stores. In the dry site, temporary job losses occurred when a nearby tourist camp was closed due to flood damage. Incidences of malaria and cholera increased. Other natural hazards, and their associated impacts, were also reported, including hail and strong winds. A tornado was experienced in the dry site. In the wet site, landslides destroyed several orchards.
Financial shocks were the least commonly reported, and included the loss of income from wages, remittances, government grants, and money earned through self-employment. For example, in the dry site, the closure of a nearby mine resulted in job losses. Respondents also noted the failure of locally owned businesses, with one household having to close their tuck-shop when a bigger store opened in the village, while another claimed that their business, raising and selling chickens, failed because households in the village preferred, and had access to, bushmeat. Financial shocks were also associated with expensive social events such as weddings and funerals.
Over the selected 5-year period, only 11% of households reported no experience of the selected shocks, while 71% reported the experience of two or more shocks.

External coping strategies
Kinship was the most commonly reported external, and overall, coping strategy, with households seeking assistance from family and community-support networks (e.g. religious groups and informal savings clubs) ( Table 2). Seeking external assistance from government and non-government organizations (NGOs) was also common and included both seeking support through the existing agricultural extension, food aid, and healthcare programmes and post-disaster assistance related to the 2013 floods. The remaining external strategy was seeking loans, with this being the least commonly used strategy overall. Some households received loans from formal money-lending institutions or their employers, but most sought loans from local loan sharks or informal, community-based savings groups. Some local shops sold goods on credit.

Internal coping strategies
Drawing from household savings was the most common internal coping strategy, with households drawing from investments in formal institutions (e.g. banks) or informal community-based savings groups and burial societies. Labour reallocation and household spending reductions were reported by similar proportions of households. Labour reallocation included migration in search of work and casual, community-based work in exchange for money or goods. For example, some flood-affected farmers employed people to repair their fences and replant. Household spending reductions included reductions in the quantity and/or quality of meals consumed. According to respondents, wealthier households were able to cut spending on "luxury" items while poorer households were forced to cut "essentials".
Forty-one percent of households relied on the safety-net function of natural products (e.g. NTFPs) from non-cultivated ecosystems. For example, in the dry site, one woman started selling mopane worms after her husband was retrenched, while fencing poles were harvested to replace flooddamaged fences. In both sites, people relied on medicinal plants to treat human/livestock diseases. Although not the most common strategy, relying on NTFPs was more common than the sale of assets (internal strategy) and loan seeking (external strategy). Livestock was the most commonly sold asset, sold both within the community and externally. Respondents indicated that selling small stock, particularly goats, was easier. Assets such as televisions and radios were also sold. The limited reliance on asset depletion, suggests households are reluctant to decapitalize (i.e. it is a "last-resort" strategy), have limited disposable assets, or that market barriers exist.

Associations between shocks and strategies
Households' coping response is influenced by the type of shock experienced. Although all the coping strategies were reported for all the shocks, certain strategies were either positively or negatively associated with certain shocks. This suggests that households prioritize certain strategies in response to certain shocks (Table 2; Fig. 1). Other studies have also noted the influence of shock type on households' coping response, suggesting the need for either a few effective strategies or a diverse coping portfolio (Takasaki et al., 2004;Wunder et al., 2014;Börner et al., 2015;Mugido and Shackleton, 2017). Asset depletion was the only strategy neither positively nor negatively associated with any of the selected shocks.
Although widely used, kinship was only significantly, positively associated with financial shocks, with households possibly relying on their social capital to tide them over until other livelihood and coping options come into play. Loan-seeking, also an external strategy, was also positively associated with financial shocks. Labour reallocation was positively associated with human shocks, but negatively associated with production shocks and natural hazards. Given the risks associated with migration, households may resort to labour reallocation to cope with the long-term challenge of an increase in dependents (human shock) but prefer to retain their labour locally to recover from the more shortterm disturbances associated with production shocks and natural hazards (Davies and Hossain, 2000;Osbahr et al., 2008;Gao and Mills, 2018). Ofoegbu et al. (2016) found migration to be an uncommon response to drought in Venda, however, others have found migration to be a prevalent response to climate-related shocks, raising concerns of increasing climate change refugees (Börner et al., 2015). Natural products appear to be more useful for coping with production shocks that temporarily affect food supplies and infrastructure, rather than natural hazards. Other studies indicate a greater contribution of NTFPs towards coping with natural hazards (Takasaki et al., 2004;Liswanti et al., 2011;Wunder et al., 2014).
Relying on external assistance was logically, positively associated with natural hazards (e.g. floods) and negatively associated with more idiosyncratic, financial shocks. Drawing from household savings was the only internal coping strategy that was positively associated with natural hazards. Poor households, with limited savings, are therefore more likely to resort to less optimal alternatives (Kazianga and Udry, 2006;Mugido and Shackleton, 2017). Increasingly frequent and intense natural hazards may undermine households' ability to save and their capacity to cope with other shocks and day-to-day expenses (Adimassu and Kessler, 2016). Access to weather insurance may help reduce this risk (Castells-Quintana et al., 2018). Households used savings despite government's post-disaster assistance, with respondents indicating limitations to this support, including limited capacity and funds, corruption, and poor infrastructure. Similar challenges were noted by Twomlow et al. (2008). Government support services may become increasingly strained in the face of climate change, increasing pressure on households' internal, coping options (Dercon, 2002;Gao and Mills, 2018). Given that household-level strategies may be insufficient in buffering against natural hazards, strengthening external social safety-nets is essential for maintaining household welfare (Gao and Mills, 2018).
Households' informal, ex-post coping options may be insufficiently robust for the covariate nature of natural hazards, with households therefore relying on external assistance and ex-ante options (i.e. drawing from household savings). Internal coping strategies, including the reliance on natural products, labour reallocation, and household spending reductions, were all negatively associated with natural hazards, suggesting limitations to their safety-net function. For example, respondents associated floods and droughts with reduced NTFP availability, increased competition for scarce resources, and the need to travel further to harvest and market, challenges also noted by Boafo et al. (2016) and Ofoegbu et al. (2016). Although households may store NTFPs, recent floods damaged households' storage facilities while these available stores may be insufficient in the face of prolonged droughts. Furthermore, in the immediate aftermath of natural hazards, households may be unable to reallocate labour to the collection and sale of NTFPs. Households possibly choose strategies that provide more immediate relief (e.g. drawing from savings), turning to NTFPs later (e.g. to rebuild houses), with this delayed use not captured in our survey. Labour reallocation, by means of outmigration, is only beneficial if the migrant finds work, undermining the usefulness of this strategy for coping with the immediate impacts of natural hazards (Gao and Mills, 2018). Given the costs of migrating, this strategy may also be unavailable to more vulnerable households (Castells-Quintana et al., 2018). The option to reallocate labour may be undermined by decreased health (and therefore human capital) associated with flood-related incidences of malaria and cholera and drought-related hunger and weakness. Households may also need to retain their labour to recover locally. Respondents identified some local job opportunities related to the recent floods, with households and government employing people to assist with repairs.
Although kinship, loans and asset depletion were neither positively nor negatively associated with natural hazards, these strategies also have limitations in the face of covariate shocks. Kinship networks and informal, community-based savings clubs may become strained if several members require support simultaneously (Ligon et al., 2002;Gao and Mills, 2018). Informal money lenders (operating in the absence of formal credit facilities) may be unable to provide sufficiently large loans to numerous afflicted households. Where loans are relied on, increasingly frequent and intense natural hazards may result in repeat borrowing, leading to greater indebtedness (Castells-Quintana et al., 2018). The option to sell assets may be limited when several households try to offload assets simultaneously (Kazianga and Udry, 2006;Börner et al., 2015). Households may be unable to sell their assets quickly enough, or for a reasonable price, if supply outweighs demand. This coping option is also undermined by asset losses associated with these natural hazards (Ngigi et al., 2015;Mehar et al., 2016;Hänke and Barkmann, 2017). For example, respondents in both sites reported livestock losses resulting from floods and droughts, with one respondent noting that these losses are "a problem because we lose our insurance for other things". Finally, asset depletion undermines households' future production and coping options (Del Ninno et al., 2016). Other studies also highlight the limitations of these local, informal insurance options for coping with covariate shocks, with poor households, with limited alternatives being hardest hit (Andrade Pérez et al., 2010;Börner et al., 2015).

Kinship and external assistance -informal and formal social safety-nets
Kinship was not used exclusively by households with limited coping alternatives. Households with formal savings and livestock were more likely to rely on kinship to cope, suggesting that households with ex-ante coping options avoid depleting their resources and future coping potential, relying instead on essentially free, ex-post strategies such as kinship (Table 3). Households with labour occupied in NTFP collection, possibly more vulnerable households with fewer livelihood and coping options, were also more likely to rely on kinship. Households with self-employed members were, however, less likely to do so, possibly being more self-sufficient. The nature of households' kinship networks influenced their coping choices. Households with family-support networks avoided spending reductions but still applied for loans and depleted assets (possibly within their networks). Households with only non-family kinship networks were more dependent on their savings. These results suggest the need for nuanced considerations of the nature of kinship networks; i.e. although kinship is an ex-post strategy, it is influenced by the nature of households' existing social capital. Others also suggest that households with better social capital, cope better (Adimassu and Kessler, 2016;Ofoegbu et al., 2016). Overall, kinship was a common strategy, used irrespective of location and demography, and as part of a suite of strategies (Paumgarten and Shackleton, 2011;Boafo et al., 2016;Mugido and Shackleton, 2017). Table 3. Summary of the effect of location and household characteristics on the use of coping strategies. Only variables with a significant positive (+) or negative (−) effect are described (p < 0.05).
With respect to external assistance, from government or NGOs, larger households and those with a high proportion of children were less likely to rely on this option. Assuming these households are more vulnerable, this raises a concern that the available external assistance is not accessible to those most in need. Households with formal savings were more likely to rely on external assistance, supporting the argument that this assistance may not be reaching those most in need. However, households with a high equipment value (i.e. asset rich households) were also less likely to rely on external assistance.

Household savings and loans
Although drawing from household savings was the most common internal coping strategy, households endeavour to protect their financial capital by engaging alternatives. Households with formal savings still relied on ex-post coping strategies, including kinship, external assistance and natural products. It is also possible that households' savings are insufficient, with a portfolio of strategies therefore required. Households who have had the time (i.e. older households) and income (i.e. through formally employed members and livestock sales) to accumulate savings were more likely to draw from their savings to cope. Households' ability to save is, however, undermined by limited sources and amounts of income (Mugido and Shackleton, 2017). Households with only nonfamily kinship networks were also more likely to rely on their savings. Households selling NTFPs were less likely to rely on savings, which may reflect the safety-net function of NTFPs, or that households selling NTFPs have limited savings. As with kinship, there is the need for a more nuanced understanding of the insurance function of household savings, including savings type. For example, households with informal savings were less likely to draw from these for unanticipated shocks. Informal, community-based savings clubs often have conditions to use, making them better suited to more anticipated periods of hardship (Paumgarten and Shackleton, 2011).
Generally, ex-post loan-seeking was limited to households with remitting migrants, a possible indicator of better access to, and awareness of, loan facilities. Ofoegbu et al. (2016) found poor access to credit to be a major challenge for households in Venda. This lack of access to financial products, including credit facilities, hinders households' ability to both cope and adapt (Börner et al., 2015;Castells-Quintana et al., 2018). Households with family kinship networks also relied on loans, possibly receiving these from within their networks. Larger households and those with a high proportion of children were less likely to seek loans. Assuming these are more vulnerable households, they may lack the resources to access and repay loans. However, households with labour engaged in NTFP collection, also possibly more vulnerable households, relied on loans, which may reflect limited alternatives. Households with formal savings avoided loan-seeking.

Reallocation of labour
Households' access to, and ability to reallocate, labour has an important effect on their coping choices. Overall, ex-post labour reallocation was the third most commonly reported coping strategy, however households with human capital constraints were unlikely to consider this option. This included households with a high proportion of children and those with labour committed to NTFP collection. Households who have successfully traded labour in the past, including those with remitting migrants and self-employed members, continue to rely on their human capital. Migration involves "start-up" costs, which may be reduced when there is an existing migrant able to support new arrivals. Therefore, although labour reallocation is an ex-post option, it is affected by households' existing human and social capital. Similarly, Hänke and Barkmann (2017) identified labour reallocation as a common strategy, practiced more by households already engaged in waged labour, although they found households resorted waged farm labour rather than outmigration.

Household spending reductions
Reductions in household spending and consumption were less common in households with high physical, natural and social capital. Although two thirds of households reported this ex-post strategy, older households, those with a high equipment value (physical capital), with livestock (natural capital), and with family-support networks (social capital), were less likely to reduce spending, suggesting access to alternative coping options (Tran, 2015). Similarly, Mehar et al. (2016) found that livestock-owning households were less likely to reduce food intake. Households with remitting migrants reduced spending, suggesting limits to migrants' capacity to provide support, including delayed support, requiring households to make temporary cuts. Larger households, those with a high proportion of children, and those with labour occupied in collecting water, were also more likely to reduce spending and consumption. Although households may choose to reduce consumption to preserve their assets and future options, poorer households are likely to be worse affected, having to cut essential items while their wealthy counterparts cut luxuries (Eriksen and Silva, 2009;Börner et al., 2015;Castells-Quintana et al., 2018). These consumption decreases have been associated with increased NTFP extraction, which may explain why households in the dry site, with better access to, and already engaged in the sale of, NTFPs, were more likely to reduce spending (McGarry and Shackleton, 2009;Mugido and Shackleton, 2017;Paumgarten et al., 2018).

Natural products
The safety-net function of NTFPs was limited for female-headed households, suggesting genderspecific limitations to this coping option. This differs to other South African studies which found households relied on the safety-net function of NTFPs irrespective of the gender of the household head (Paumgarten and Shackleton, 2011;Mugido and Shackleton, 2017). Although NTFPs are argued to be of greater benefit to poor, rural women and female-headed households, others suggest that women may have limited access to, and control over, these resources (Posel, 2001;Takasaki et al., 2004;Mehar et al., 2016). The notable difference in the proportion of households reporting the daily-net (98%) versus safety-net (41%) function of NTFPs emphasizes the possible limitations to this safety-net function. Börner et al. (2015) also noted a lower than expected reliance on natural resources. Mugido and Shackleton (2017) identified NTFPs as a common coping strategy, especially in low-potential agro-ecological zones, although only 35.6% of their sample reported this safety-net function. We found that households already selling NTFPs, and those with livestock, were more likely to rely on NTFPs to cope, suggesting an existing knowledge of these resources, their harvesting areas, and their markets (Boafo et al., 2016;Ofoegbu et al., 2016). Households with formal (rather than informal) savings were also more likely to rely on NTFPs to cope suggesting that households purposely rely on NTFPs to protect their alternative and future coping options.
Households engaged in the sale of NTFPs were less likely to deplete assets or rely on savings to cope. However, they were more likely to reduce spending and turn to NTFPs, suggesting that they may be poorer, with a more limited portfolio of coping options. It is also possible that their existing traditional ecological knowledge allows them to rely on NTFPs as a gap-filler. By using NTFPs for minor crises, these households are able to maintain their cash and assets, and thereby their future welfare (Skoufias, 2003;Boafo et al., 2016). Paumgarten and Shackleton (2011) found households relied on the safety-net function of NTFPs irrespective of household wealth, while Hänke and Barkmann (2017) found almost all households, irrespective of their primary economic activity, relied on the gap-filling function of wild foods in response to crop failure. McSweeney (2005) argues that access to these natural resources should be maintained, particularly if alternatives are limited.

Asset depletion
Asset depletion by households engaged in livestock sales, by older and larger households, and by those with a high number of non-remitting migrants, suggests that existing market experience, accumulated assets, greater need, and limited alternatives may all influence households' decision to decapitalize. Households with family-support networks were also more likely to deplete assets, possibly selling assets within their networks. Female-headed households and those with labour constraints (i.e. labour engaged in water collection) were less likely to deplete assets, suggesting gender-and labour-related barriers to asset accumulation and sales (Tran, 2015). Asset depletion may also be a strategy of last resort. Börner et al. (2015) also noted that the female-headed households in their African subsample were less likely to sell assets. Surprisingly, households in the dry site, where livestock ownership is more prevalent, were also less likely to deplete assets. Limited and variable market demand, veterinary controls on the movement of animals, and the multiple functions of livestock, may all discourage sales (Ofoegbu et al., 2016). Hänke and Barkmann (2017) suggest that where cattle fulfil an important socio-cultural role, households may choose to rather sell less culturally important small stock. Other studies have considered asset accumulation as an exante risk management strategy and an ex-post coping strategy, with mixed views on this strategy's robustness (Dercon, 2002;Kazianga and Udry, 2006;Deressa et al., 2009). Households selling NTFPs and crops were also less likely to deplete assets.

Conclusion and recommendations
Rural households in Venda, South Africa, experience a mix of unanticipated production, human, and financial shocks, as well as natural hazards. Although a range of informal coping strategies was identified, our findings suggest that households' response to these shocks is influenced by the type of shock experienced. This suggests that households choose specific coping strategies based either on the effectiveness of the strategy in buffering against the shock, or on households' active intent to manage their broader coping portfolio in anticipation of future shocks.
With respect to natural hazards, households' informal, ex-post, household coping strategies appear to be limited, with households relying on external assistance from government or NGOs, and ex-ante household savings. The reliance on external assistance is expected however, our findings suggest that more vulnerable households are less likely to rely on this option. Further investigation into whether there are barriers to accessing these formal, external social safety-nets is advised. Given the possibility of increasingly intense and frequent natural hazards, associated with climate change, this assistance, and the institutions that provide it, will need to be increasingly robust, appropriately targeted, and accessible to the most vulnerable. The key role of household savings in helping households cope with natural hazards is also concerning if these events occur with increasing frequency and intensity. Without alternative options, households' ability to accumulate savings will be hindered, undermining their efforts to secure a sustainable livelihood, reducing their overall coping capacity, and limiting their ability to make the necessary investments in adaptation. How to increase households' access to alternative financial services, including credit and climate insurance, needs further exploration.
In addition to exploring whether shock type influences households' coping choices, we explored whether a range of household characteristics and location have an influence. Generally, households' access to human and financial capital influenced the coping strategies used. Our findings also suggest that female-headed households have fewer coping options, being less likely to deplete assets or rely on NTFPs. Female-headed households may face barriers to accumulating and/or disposing of assets, and to accessing NTFPs, which need to be explored to address existing genderbased inequalities. This includes understanding why female-headed households were less likely to rely on NTFPs, particularly given that these are often considered an important safety-net for those with limited alternatives. Larger households, and those with a high proportion of children, also appear to have more limited coping options, relying on negative strategies such as spending and consumption reductions, which may negatively affect their human capital. The influence of location on households' willingness to deplete assets to cope provides insights into the insurance role of livestock and suggests that although households may accumulate livestock as an ex-ante insurance strategy, this is not their primary reason for doing so, with livestock fulfilling multiple functions. It also suggests barriers to selling livestock. It would be useful to better understand at what point households are prepared to discount the benefits of retaining their livestock, versus selling it to cope, as well as what support services could be put in place to enhance the insurance value of livestock.
Natural products were not often used to cope with natural hazards, suggesting limitations to their safety-net function for covariate, climate-related shocks. However, access to natural capital, as well as social capital, allows for the protection of households' assets and savings, and the avoidance of spending and consumption reductions, thereby contributing towards households' overall resilience. Households' reliance on natural and social capital is, therefore, not necessarily an indicator of vulnerability, but may be a considered means of managing their future insurance options. By relying on natural products to buffer against other shocks, households may be better able to retain their savings for natural hazards.

Funding
This work was supported by South Africa's National Research Foundation (NRF 92241), the Global Change Institute (through the Carnegie Corporation of New York (B874.R01)), the Centre of Excellence in Tree Health Biotechnology, the University of the Witwatersrand and the Center for International Forestry Research.

Compliance with ethical standards
The Research Office of the University of the Witwatersrand granted the ethical clearance to conduct the research (H120803). All ethical standards were adhered to. The relevant local authorities granted permission to conduct the research and formal, free, prior and informed consent was obtained from all participants. Anonymity was assured.

Declaration of competing interest
None.