An extensive social science literature has examined the effects of climate change on human migration. Prior studies have focused largely on the out-migration of working age adults or entire households, with less attention to migration and other forms of geographic mobility among other age groups, including youth. In this study, we focus on the implications of climate variability for the movement of children by examining the association between climate exposures and the in- and out-fostering of children in sub-Saharan Africa. We link high-resolution temperature and precipitation records to data from the Demographic and Health Surveys for 23 sub-Saharan African countries. We fit a series of regression models to measure the overall associations between climate exposures and each outcome and then evaluate whether these associations are moderated by socioeconomic status, the number of children in the household, and the prevalence of fostering in each country. Precipitation is positively associated with in-fostering overall, and these effects are especially strong among households that already have at least one child and in countries where child fostering is common. We find no overall relationship between either temperature or precipitation exposures and out-fostering, but we do detect significant effects among households with many children and those with more educated heads. In sum, our findings suggest that climate variability can influence child mobility, albeit in complex and in some cases context-specific ways. Given the socioeconomic and health implications of fostering, these results underline another pathway through which climate exposures can affect children’s well-being. More broadly, this study shows that new attention to the links between climate variability, child fostering, and other understudied forms of spatial mobility is needed to fully understand the effects of climate change on human populations.
Similar content being viewed by others
Changing patterns of internal and international migration have been among the most discussed consequences of global climate change (Borderon et al., 2019; Hunter et al., 2015; Rigaud et al., 2018). While headline-grabbing projections of large-scale climate-induced migrations are not supported by current empirical evidence, the existing climate migration literature does indeed document robust associations (both positive and negative) between recent climate exposures and migration (Borderon et al., 2019; Call & Gray, 2020; Grace et al., 2018; Mounirou & Yebou, 2022; Thalheimer et al., 2023; Weinreb et al., 2020). Climate variability and downstream socioeconomic impacts alter patterns of spatial mobility among many exposed populations. Importantly, however, the climate migration literature has to date focused largely on adults, under the (plausible) assumption that ex situ adaptations to environmental impacts are largely undertaken by working-age individuals and to diversify income (i.e., labor migration) (Mueller et al., 2020a, b). Far fewer studies have focused on other types of migration and mobility, including moves among children (Madhavan et al., 2012). We address this gap by focusing on the movement of children between households (i.e., fostering).Footnote 1
This issue merits attention given the plausible conceptual reasons to expect the impacts of climatic variability to affect child fostering as well as the substantive importance of this outcome for children’s welfare. The social and economic impacts of climate shocks can affect the quantity and quality of household resources available for children, the demand for labor, and the likelihood of intra-household conflict and marital dissolution. Each of these changes is likely to influence child fostering dynamics, which have a range of potential consequences. For example, fostering can be used to improve educational outcomes among children from resource-constrained households, to provide household labor, and to reallocate demand for household resources (e.g., food) (Akresh, 2009; Bledsoe et al., 1988; Hampshire et al., 2015; Hedges et al., 2019; Zimmerman, 2003). In these and similar ways, fostering can have important implications for the health and well-being of affected children and the households in which they are embedded.
In the current study, we examine the relationship between exposures to temperature and precipitation anomalies (i.e., z-scores) and child fostering across 23 countries in sub-Saharan Africa.Footnote 2 Analyzing historical climate records and population data from the Demographic and Health Surveys (DHS), we find an overall positive association between precipitation and in-fostering but non-significant temperature effects. We do not detect any overall climate effects on out-fostering. However, tests for heterogeneity across sub-populations reveal that the strength and direction of climate effects on both in- and out-fostering vary. For example, we find that precipitation effects on in-fostering are strongest among households with at least one child and in countries with a higher prevalence of fostering, and we detect significant climate effects on out-fostering among large households and those with more educated heads. Evidence that child fostering patterns are sensitive to climatic variability and its downstream effects underscores the need to consider the effects of environmental change on a broad set of mobility outcomes among both children and adults.
Climate change, population mobility, and child fostering
Climate variability has been shown to change human migration patterns in many regions of the world, including parts of sub-Saharan Africa (Borderon et al., 2019; Call & Gray, 2020; Grace et al., 2018; Gray & Mueller, 2012; Gray & Wise, 2016; Mueller et al., 2020a, b; Mounirou & Yebou, 2022; Nawrotzki & DeWaard, 2018; Thalheimer et al., 2023; Weinreb et al., 2020). In broad terms, climate-related disruptions may influence migration in two ways. First, the social and economic impacts of climate variability may increase migration as affected households send members to work in less-affected places and then remit or return with the income they generate (i.e., geographic diversification of livelihoods) (Ellis, 1998). For example, Gray and Mueller (2012) find that drought increases labor-related migration among men in Ethiopia, and Mounirou and Yebou (2022) find that farmers in northern Benin consider out-migration to be a part of possible strategies for adapting to the economic effects of adverse environmental conditions. Likewise, Call and Gray (2020) find that heat spells can increase both short- and long-term migration in Uganda via effects on livelihoods. In a second scenario, climate-induced disruptions to livelihoods may reduce the resources that are available to fund migration, thereby decreasing mobility (i.e., producing “trapped populations”) (Black et al., 2011; DeWaard et al., 2022). For example, Nawrotzki and DeWaard (2018) show that climate shocks reduce out-migration in poor districts in Zambia, where households may lack the resources to adapt via out-migration. Similarly, Gray and Wise (2016) find that temperature anomalies reduce migration in Kenya and Burkina Faso, and Mueller et al., (2020a) find that heat spells reduce out-migration in Botswana.Footnote 3 Although the direction of climate effects on migration varies across contexts in sub-Saharan Africa, decisions about spatial mobility are clearly influenced by environment-induced changes in household resources (Borderon et al., 2019).
The conceptual framework underpinning most of this prior research largely assumes that migration is used as an economic livelihood strategy (Thiede et al., 2022), undertaken to diversify income and mitigate climate-induced income shocks. However, migration is not only a means by which workers secure income. It may also be used to change resource demands within the household by altering the unit’s size and structure. While this point is central to the demographic theory of the multi-phasic response (Bilsborrow, 1987; Davis, 1963; Kalipeni, 1996), it has received less attention within the climate migration literature. The main exceptions (to our knowledge) have been studies of climate-related disruptions to marriage, which recognize how changes in household structure may be used for climate adaptation and, implicitly or explicitly, account for marriage-related mobility (Carrico et al., 2020; Corno et al., 2020; Gray & Mueller, 2012; Tsaneva, 2020).Footnote 4 For example, Gray and Mueller (2012) find that marriage-related moves among Ethiopian women decline in the aftermath of drought. In contrast, Carrico and colleagues (2020) find that heat shocks increase marriage among women and girls in Bangladesh. These and related examples underline the potential importance of changing household structure within adaptation strategies and demonstrate that the impacts of climate variability on population mobility are not limited to labor migration among working-age adults. Attention to other members of the household and other forms of spatial mobility across the life course is therefore merited.
We work to address this evidence gap by examining the relationship between climate exposures and child fostering, or the exchange of children beyond the household of their biological parents. In- and out-fostering has historically been and is currently a common practice within many parts of sub-Saharan Africa (Alber, 2018; Cotton, 2021; Goody, 1982), with a diverse set of motivations and important implications for children’s well-being. Fostering prevalence varies across the region (Beck et al., 2015; McDaniel & Zulu, 1996), with rates as high as 30% in Malawi and as low as 3% in Sudan (McDaniel & Zulu, 1996). Fostering prevalence also varies according to a child’s demographic characteristics such as age and sex. For example, older children are more likely to be out-fostered (Cotton et al., 2022; Eloundou-Enyegue & Shapiro, 2004), and girls are more likely to be fostered than boys (Beck et al., 2015; McDaniel & Zulu, 1996). Children are most commonly in-fostered by relatives, especially maternal kin (Cotton, 2021; Cotton et al., 2022; Hedges et al., 2019), for anywhere between a few months and several years (Cotton et al., 2022; Beck et al., 2015; Kielland & Gaye, 2010). Existing evidence suggests children are often fostered for a few years, especially if a child is also receiving an education (Beck et al., 2015; Bledsoe, 1990; Kielland & Gaye, 2010). Children may be out-fostered to rural areas while a parent, often a single mother, migrates to the city to work (Cotton, 2021; Cotton et al., 2022). Conversely, children may be out-fostered to urban areas to get better educational opportunities or to live with better-off relatives (Alber, 2018).
In stylized terms, there are three primary motivations for child fostering. First, children may be fostered to adapt to a change in family structure, such as parental divorce or remarriage (Eloundou-Enyegue & Shapiro, 2004; Grant & Yeatman, 2014). In particular, maternal remarriage increases the likelihood of a child being out-fostered (Cotton et al., 2022; Grant & Yeatman, 2014). This pattern is often driven by the unwillingness of a new husband to care for non-biological children, prompting out-fostering (Cotton, 2021; Cotton et al., 2022; Grant & Yeatman, 2014). Relatedly, children born to unmarried parents are more likely to be out-fostered, often to maternal kin (e.g., the maternal grandmother) (Cotton et al., 2022; Hedges et al., 2019). Second, children may be out-fostered to obtain better opportunities for their future than would be available if they remained within their original household. Likewise, better-resourced households may engage in in-fostering to help the children of family and friends obtain a better education, work, or other opportunities (Archambault & de Laat, 2010; Hampshire et al., 2015; Hoddinott & Mekasha, 2020; Serra, 2009; Zimmerman, 2003). Third, children may be fostered to adapt to changing labor demands or economic circumstances. For example, households with fewer resources may be more likely to out-foster a child in the wake of a shock such as a drought (Kielland & Kebede, 2020).Footnote 5 A child may also be out-fostered while biological parents, particularly single mothers, migrate for better work opportunities (Archambault et al., 2012; Cotton et al., 2022; Gaydosh, 2015). In contrast, potential receiving households may decide to in-foster a child to help meet household labor demand (e.g., collecting firewood and water) (Beegle et al., 2006; Hampshire et al., 2015; Serra, 2009; Zimmerman, 2003).
These motivations for fostering are not mutually exclusive; thus, fostering can be beneficial (or deleterious) for both fostered children, sending households, and/or receiving households (Bachan, 2014; Serra, 2009; Zimmerman, 2003). For example, a household experiencing economic hardship may out-foster a child to receive a better education and to reduce household resource demands. Similarly, a household may in-foster a child to help them receive a better education and/or to help with household chores (Hedges et al., 2019; Zimmerman, 2003). Fostering is not always mutually beneficial for children and receiving households. Fostered children can be expected to help with household labor at the expense of their education (Hedges et al., 2019), though this phenomenon may not be limited to fostered children and is likely gendered (Beck et al., 2015). In some cases, fostered children may also be neglected or otherwise mistreated by the receiving household (Bledsoe, 1990). In short, fostering is one of the multiple ways that households adjust their size and composition in response to changes in their social and economic circumstances, as well as those of their family and kinship networks. It has important implications for the welfare of children and both sending and receiving households.
Given this prior research, we expect that climatic variability and its downstream social and economic impacts may alter the incentives for in- or out-fostering children. In our view, several mechanisms may explain such effects. First, climate-induced declines in agricultural production and revenue (Knox et al., 2012; Mueller et al., 2014), as well as other sources of income (Carleton & Hsiang, 2016), may impose constraints that push households to reduce resource demands via increased out-fostering and decreased in-fostering. Importantly, such dynamics may reflect parents’ motivation to protect and maintain their children’s well-being (e.g., access to sufficient nutrition and education opportunities) amidst environmental shocks, above and beyond meeting imperatives to reduce household size. Second, climatic changes may lead to shortages of water, firewood, and other resources that require physical collection (Massey et al., 2010)—tasks that are often assigned to children (Beegle et al., 2006; Pickering & Davis, 2012). Increased demand for such work may therefore be associated with reduced out-fostering and increased in-fostering as households seek to retain or augment the number of children available to work. Third, climate-related resource constraints may increase conflict among spouses (Heath et al., 2020; Sekhri & Storeygard, 2014), leading to higher probabilities of divorce. Given prior associations between parental divorce and increased out-fostering (Gaydosh, 2015; Grant & Yeatman, 2014), we expect such processes to increase out-fostering and decrease in-fostering. Importantly, we make the provisional assumption that each of these mechanisms will influence fostering in a symmetrical manner. That is, to the extent that exposures to better-than-average environmental conditions reduce resource constraints, demand for child labor, or intra-household conflict, they will have the opposite effects to those outlined above.
Considering these potential mechanisms, we argue that there are plausible reasons to expect climatic variability to be associated with fostering patterns across sub-Saharan Africa. Given the lack of prior evidence, however, there is little basis to expect one of these mechanisms to have a stronger influence than the others. Additionally, the anticipated effects of climate exposures on in- and out-fostering mirror each other theoretically but may diverge empirically given differences in the populations that are at risk of, or most likely to participate in, these processes.Footnote 6 We therefore do not propose formal directional hypotheses about climate effects on child fostering but instead treat this as an open question to be answered empirically.
The overall goal of this paper is to analyze the association between exposure to climatic variability and the likelihood of out- and in-fostering in sub-Saharan Africa. We address three specific aims to achieve this goal. First, we measure the overall associations between temperature and precipitation exposures and the respective probabilities that households out- or in-foster one or more children. Second, we test for variation in climate effects according to the number of children in the household and the household head’s educational attainment. We take the household head’s education as a proxy for household socioeconomic background, adult knowledge and skills, and similar factors associated with a household’s ability to navigate its social and environmental context (Andriano & Monden, 2019; Behrman, 2019; Fortson, 2008). We expect the number of children in the household to be correlated with resource constraints that may differentially incentivize fostering, a dynamic that may be amplified in the context of environmental changes (Bougma et al., 2015). Third, and finally, we evaluate whether effects differ according to the prevalence of fostering in each country within our sample. We use the prevalence of fostering as an indicator of the normativity of this behavior in a given context, which we expect may be correlated with its frequency of use as a coping strategy (Zimmerman, 2003).Footnote 7
We analyze DHS data for 23 sub-Saharan African countries (mapped in Fig. 1), which we access using the Integrated Public Use Microdata Series-DHS database (IPUMS-DHS, Boyle et al., 2020a). The DHS program has conducted nationally representative household surveys across more than 90 low- and middle-income countries, collecting information on demographics, health, and nutrition among households in the sample (ICF, 2023). Standardization of survey instruments and methodology allows for pooling and/or comparisons across samples (e.g., Davenport et al., 2017; Headey & Ruel, 2022; Thiede & Strube, 2020), which we leverage in this study. We draw on all samples from sub-Saharan Africa that were available in IPUMS-DHS at the time of analysis and for which high-resolution climate records were available (see below). Our final analytic sample includes data from 44 DHS samples collected between 1992 and 2016. We apply the household sample weights provided by the DHS (via IPUMS-DHS) in all analyses.Footnote 8
We use DHS birth records and household roster data to construct a household-level file with counts of out- and in-fostered children and relevant control variables. The birth records include information about each birth to reproductive-aged women in the sample, including information about their survival, current age, and location that we use to identify out-fosters. The household roster provides information on household members, including data on individuals’ age and biological relationship to other household members that we use to identify in-fosters.Footnote 9 Overall, there are 542,342 households included in the final analytic sample for our models of in-fostering. We exclude households without biological children from the out-fostering models since they are not at risk of this outcome, leaving 314,786 households in that final analytical sample.
We combine DHS data with temperature and precipitation records using the cluster (i.e., community) identifiers in the DHS. We use temperature and precipitation data from the IPUMS-DHS contextual files (Boyle et al., 2020b), which include records from a pair of high-resolution climate datasets. Precipitation is measured using the Climate Hazards Group (CHG) InfraRed Precipitation with Station data (CHIRPS), which is generated by combining weather station and satellite data to produce gridded estimates at 0.05° resolution. Maximum temperature is measured using data from the Terrestrial Hydrology Research Group (Sheffield et al., 2006). These data are produced by integrating the Climatic Research Unit’s Time Series (CRU TS) and the National Centers for Environmental Prediction-National Center for Atmospheric Research’s Reanalysis (NCEP–NCAR) product (for details, see Sheffield et al., (2006)). Precipitation and temperature variables are both calculated on a monthly scale and as spatial means within a 10 km radius of each DHS cluster (Boyle et al., 2020b). To protect respondents’ privacy, the publicly released locations of clusters in the DHS are displaced by up to 10 km for rural communities and up to 5 km for urban communities. The use of a 10 km radius for measuring temperature and precipitation ensures that the true location of a given household is captured.
Measures and methods
The focal outcomes are the presence of one or more out- and in-fostered children in the household, respectively. These outcomes are operationalized as dichotomous variables. We use the DHS birth records to identify out-fostered children, who are defined as individuals under age 15 residing outside of the mother’s household (we therefore cannot identify the status of maternal orphans, as discussed in the concluding section).Footnote 10 We use the DHS household roster to identify in-fostered children, who are defined as under-15 individuals without a biological parent residing in the household.
The main predictor variables are temperature and precipitation exposures, operationalized as anomalies during the year prior to the DHS interview. We therefore measure the short-term, near-contemporaneous effects of climate exposures on fostering. This approach is consistent with our assumption that initial climate-related fostering decisions are made vis-à-vis recent environmental conditions and downstream effects and therefore operate with minimal lags.Footnote 11 These anomalies are constructed by calculating the total precipitation and average maximum temperature during the focal year and then finding the difference between those values and their respective means over all other 1-year periods in each cluster’s climate history (1981–2016).Footnote 12 These differences are then standardized over the cluster-specific historical standard deviation, allowing the anomalies to be interpreted as z-scores. Anomalies are locally meaningful since they are produced with respect to community-specific averages, should be uncorrelated with baseline climate, and as such have been shown to be stronger predictors of demographic outcomes than unstandardized climate values (Gray & Wise, 2016; Thiede et al., 2022).
Our analyses center on a series of logistic regression models of out- and in-fostering. We predict each outcome as a function of recent temperature and precipitation exposures, controlling for a series of social and demographic factors and a linear time trend, and including province fixed effects.Footnote 13 We control for the age, marital status, and completed education of the household head, the number of biological children born to surveyed women in the household (i.e., inclusive of out-fosters and exclusive of in-fosters),Footnote 14 and rural (urban) residence, which is measured at the community level.Footnote 15 Province fixed effects account for all time-invariant characteristics at the province level, and the time trend accounts for common changes in climate and fostering so long as they occur linearly over the study period. In addition to an overall model of each outcome, we fit three additional models that respectively include interactions between the climate exposure terms and (a) the household head’s education, (b) the number of children in the household, and (c) the percentage of children who are out-fostered in the country.Footnote 16 As described above, these interaction models allow us to account for three potentially salient dimensions of inequality in climate vulnerability and fostering behavior. Throughout all analyses, we cluster standard errors at the DHS cluster level since this is the scale at which our climate exposure variables are measured. The analytic samples are described in Table 1.
Climate exposures and out-fostering
We begin by examining the association between climate variability in the year prior and out-fostering, as measured by the absence of one or more out-fostered children from the household (Table 2). The first model (Model 1) measures the overall association across the entire sample. We do not detect significant associations between either temperature or precipitation anomalies in the year prior and out-fostering at conventional thresholds for statistical differences. However, we note that the negative association between precipitation exposures in the year prior and out-fostering (odds ratio (OR) = 0.98) is marginally significant (p < 0.10) and implies an approximately 1.6% decrease in the odds of out-fostering with each standard deviation increase in precipitation (and a corresponding increase in out-fostering after precipitation deficits). Given the uncertainty around this estimate, however, we caution against over-interpretation.
Since the overall estimates may mask variation among sub-populations in our sample, we then fit a series of models that allow climate effects to vary across social and demographic groups. We first examine the interaction between climate exposures and the household head’s education (Model 2), which as noted above may be correlated with socioeconomic background, adult knowledge and skills, and other factors associated with a household’s ability to navigate a changing social and environmental context. We do not find a statistically significant relationship between precipitation exposures in the year prior and out-fostering at any level of education. However, temperature anomalies in the year prior are positively associated with out-fostering among households headed by individuals with a secondary education (OR = 1.04) or higher levels of education (OR = 1.09). The point estimates suggest that each standard deviation increase in temperatures is associated with a 3.5% and 9.0% increase in out-fostering among these groups, respectively. Perhaps contrary to expectations, this result indicates that higher-status households are more likely to out-foster a child after exposure to periods of heat stress, which are typically associated with adverse agricultural and economic outcomes (Baker & Anttila-Hughes, 2020; Schlenker & Lobell, 2010).
Next, we assess whether climate effects on out-fostering are moderated by the number of children in the household (Model 3), which we expect to differentially incentivize fostering in the context of environmental change. Exposures to precipitation anomalies in the year prior are not statistically associated with out-fostering among households with only one child (the minimum in this analytic sample), and the coefficient on the precipitation-by-number of the children interaction term is not statistically significant. According to additional analyses, the marginal effect of precipitation remains statistically non-significant among households with two and three children but is significantly and negatively associated with out-fostering among households with four or more children. For example, according to point estimates, each standard deviation increase in precipitation during the previous year is associated with an approximately 1.7% decrease (OR = 0.98) in the odds of out-fostering among households with four children and an approximately 2.2% decrease among households with five children (OR = 0.98). We calculate and plot the corresponding associated change in the predicted probability associated with precipitation anomalies across households with up to eight children while holding all other variables at their means (Fig. 2). These estimates imply that exposure to precipitation deficits will increase out-fostering among households with many children. The magnitude of these effects is modest but may translate into large changes in the absolute number of children fostered when applied to our large target population.
In contrast to the observed pattern of precipitation effects, exposures to temperature anomalies are negatively associated with the odds of out-fostering among households with one child (OR = 0.97). Each standard deviation increase in temperatures during the previous year is associated with an approximately 2.9% decrease in the odds of out-fostering among this sub-population. However, this effect is moderated by the number of living children in the household, converging toward and then increasing above zero as household size increases. Indeed, the net effect of temperature exposures on out-fostering is positive among households with at least five children. For example, each standard deviation increase in temperature is associated with a 3.0% increase (OR = 1.03) in the odds of out-fostering for households with five children. Associated changes in predicted probabilities are provided in Fig. 3. Taken together, these findings suggest that large households cope with the impacts of negative environmental conditions (i.e., heat, low precipitation) by out-fostering one or more of their children. In contrast, households with few children respond to heat shocks by retaining them, as they have lower out-fostering rates compared to after years with average temperatures.
Finally, we consider whether climate effects on out-fostering vary according to the normativity of fostering (Model 4), as indicated by the rate of out-fostering in each country. Precipitation anomalies in the year prior are negatively associated with out-fostering among countries with relatively low fostering prevalence (i.e., from the minimum of approximately 12% to 22%). For example, each standard deviation increase in precipitation is associated with a 6.0% decrease (OR = 0.94) in the odds of out-fostering in countries where approximately 12% of children out-fostered. Point estimates of the marginal effect of precipitation on out-fostering approach and then exceed zero among countries with higher rates of fostering, but these estimates remain statistically non-significant in all cases. We provide the corresponding changes in predicted probabilities in Fig. 4. Contrary to expectations, precipitation deficits increase out-fostering mainly in places where fostering is rare. We do not find the relationship between temperature and out-fostering to be significantly moderated by fostering prevalence.
Climate exposures and in-fostering
The next set of models examines the relationship between climate exposures and in-fostering (Table 3), as measured by the presence of one or more in-fostered children in the household. While one may expect these associations to operate in the opposite direction of what is observed in the out-fostering models (e.g., a factor that increases out-fostering will reduce in-fostering), we treat this as an empirical question. Indeed, the populations at risk of each outcome differ, and multiple, possibly contradictory mechanisms may drive these processes.
The first in-fostering model (Model 5) examines the overall effects of climate exposures. We find precipitation anomalies in the year prior to be a significant predictor of household in-fostering. According to point estimates, each standard-deviation increase in precipitation is associated with an approximately 3.0% increase in the likelihood that a household in-fosters a child (OR = 1.03). This result implies a reduction in in-fostering after periods of precipitation deficits, which is consistent with a dynamic in which households reduce in-fostering when faced with environment-induced resource constraints. We do not find a statistically significant overall association between temperature and in-fostering.
The second model of household in-fostering (Model 6) evaluates whether these associations vary by the educational attainment of the household head, and we find evidence of such moderating processes for both precipitation and temperature effects. Precipitation exposures in the year prior are not significantly associated with in-fostering among households headed by individuals with a primary school education or less but are negatively associated with in-fostering among households with heads that have attained the highest level of education (OR = 0.86). However, climate effects at other levels of education are not statistically significant, making it difficult to draw strong conclusions about these associations.
Temperature effects on in-fostering are concentrated among households headed by individuals with at least a primary school education. For households headed by an individual with a primary school education, each standard deviation increase in temperatures during the previous year is associated with a 4.2% decrease (OR = 0.96) in the odds of in-fostering. For households headed by an individual with either a secondary education or a higher level of education, the corresponding decreases are 11.5% (OR = 0.89) and 34.3% (OR = 0.71), respectively. As temperature anomalies increase, households with more educated heads are less likely than normal to in-foster a child, which again suggests a dynamic where resource constraints are associated with decreased in-fostering.
The third model of household in-fostering evaluates whether effects vary by the number of children in the household (Model 7). The association between precipitation anomalies and in-fostering is significantly moderated by this dimension of household size, and in a manner that suggests households with the largest number of children are most likely to reduce in-fostering when exposed to adverse environmental conditions (i.e., low precipitation). Exposure to precipitation anomalies during the previous year is significantly and positively associated with in-fostering among households with at least one child. Among households with a single child, each standard deviation increase in precipitation is associated with an approximately 2.0% (OR = 1.02) increase in the likelihood of in-fostering. The magnitude of this association grows as the number of children in the household increases. For example, the marginal effect of precipitation increases to 12% (OR = 1.12) among households with 8 children. We provide the associated changes in predicted probabilities across other household sizes in Fig. 5. The relationship between temperature and in-fostering is not significantly moderated by the number of children in a household.
Finally, we test for differences in climate effects on in-fostering according to the normativity of fostering. Precipitation exposures in the prior year are significantly and positively associated with in-fostering in countries where the percentage of children out-fostered is above approximately 21%. For example, a one standard deviation increase in precipitation anomalies is associated with a 5.0% (OR = 1.05) increase in the likelihood of in-fostering among households located in countries with around 21% of their children out-fostered. We visualize corresponding changes in predicted probabilities at other levels of fostering normativity in Fig. 6. We find temperature exposures in the year prior are marginally significantly (p < 0.10) associated with in-fostering at the tails of the distribution. In places with low levels of fostering (less than approximately 17%), we observe a negative association between temperature exposures and in-fostering, while there is a positive association in places with high levels of fostering (greater than approximately 39%). We visualize the corresponding changes in predicted probabilities in Fig. 7. This result suggests that in areas with stronger fostering norms, households are more likely to take in a child after heat shocks. In contrast, households in areas with weaker fostering norms appear less likely to in-foster children after such adverse conditions.
Discussion and conclusion
In this paper, we have examined the relationship between households’ climate exposures and their odds of both in- and out-fostering children. In doing so, we aim to bridge the gap between the literatures on climate change and migration (which has focused almost exclusively on the geographic mobility of adults) and on child fostering (which has given only limited attention to the effects of environmental shocks) (Akresh, 2009; Fussell et al., 2014; Hoffmann et al., 2020; Jensen, 2000). Our results support three general conclusions, which are largely consistent with the expectation that fostering is a part of many households’ strategies for coping with the effects of climatic variability.
First, exposure to precipitation deficits in the year prior is associated with a reduction in in-fostering overall across the entire sample. Our estimates suggest that households cope with lower precipitation (which implies negative agricultural and economic outcomes) by avoiding the addition of new dependents. Importantly, we find that precipitation deficit-associated reductions in in-fostering are concentrated among households headed by better-educated individuals and in countries where fostering is relatively common (heat shocks increase in-fostering in places where fostering is very common). The former finding may reflect the fact that better-resourced households are more likely to in-foster children at baseline and are thus more sensitive to climate shocks. The latter finding is consistent with other results that show fostering is most sensitive to resource constraints in places where fostering is rare, perhaps because it is mainly under conditions of duress that fostering occurs in such locations.
Second, the relationship between climate exposures in the previous year and out-fostering is not statistically significant overall, but we detect statistically and substantively meaningful associations among select sub-populations. Households headed by individuals with relatively high education increased out-fostering in the year after periods of high temperatures, and households with large numbers of children also increased out-fostering in the year after precipitation deficits. The latter finding is consistent with the assumption that larger households face greater baseline resource constraints, which translate into disproportionate stresses in the aftermath of environmental shocks. On the other hand, this dynamic could also be consistent with a scenario in which households with enough children to meet labor needs out-foster their children to get better opportunities or to help another household with labor (Beck et al., 2015; Hoddinott & Mekasha, 2020; Serra, 2009). The increase in out-fostering among households with better-educated heads is somewhat unexpected and contrary to the patterns observed across household sizes. We raise two potential, speculative interpretations. First, less-educated households are more likely to be economically marginalized, and as such, they may respond to shocks by retaining children to provide labor as a coping mechanism. This dynamic would be consistent with the so-called vicious circle hypothesis of population dynamics and environmental change (Biddlecom et al., 2005). Second, better-educated households may have more social networks and/or economic resources to support out-fostering and other adaptation strategies in the context of environmental stress. In line with findings about trapped populations from the climate migration literature, the least-resourced households may experience reductions in mobility after environmental shocks (Black et al., 2011; Nawrotzki & DeWaard, 2018). Consistent with the results from the in-fostering model, we also find that exposures to precipitation deficits are associated with increased out-fostering in contexts where fostering is relatively uncommon. Again, this result implies that fostering in such places appears to be driven more by environmental shocks and associated socioeconomic stressors than other factors.
Third, the models of in- and out-fostering are not entirely symmetrical. The general implication is that fostering motives vary not only by social group and context but also according to households’ positions as senders or receivers of fostered children. However, it is also important to note that several of the results are consistent across the out- and in-fostering models. Specifically, we find that exposures to adverse environmental conditions in the prior year (e.g., high temperatures and low precipitation) increase out-fostering among households headed by better-educated individuals and in contexts where fostering is uncommon (relative to the other countries in our study) and have corresponding negative effects on in-fostering among these populations. For these households, at least, climate shocks incentivize out-fostering and disincentivize the in-fostering of children from other households.
While these results provide important baseline evidence about the relationship between climate exposures and fostering, the study is nonetheless characterized by several limitations that should be addressed in future studies. First, the cross-sectional nature of the DHS imposes at least two important constraints on our analysis. For one, we cannot observe households that were exposed to the climatic conditions in question but out-migrated from the DHS communities before the survey. Although such climate-induced whole-household migration is believed to be rare (Bohra-Mishra et al., 2014), it is likely to be selective on factors that are correlated with the fostering behaviors of interest. The cross-sectional nature of the DHS also restricts the availability of appropriate control variables. Many variables measured at the time of the survey are sensitive to the focal climate exposures and therefore not appropriate controls. A second key limitation is that we cannot measure the exact timing and duration of fostering arrangements. We cannot determine exactly when the fostering event occurred vis-à-vis the exposure period, requiring strong assumptions about baseline fostering patterns. We likewise cannot distinguish between short- and long-term fostering arrangements, which may be differentially affected by the impacts of climate variability and which may have distinct implications for children’s welfare.Footnote 17 These limitations highlight the value of collecting and/or using panel data with detailed fostering histories in future studies.
Third, we only measure out-fostering among children born to reproductive-aged women in the household given the data collected in the DHS. We are therefore not able to measure the out-fostering of children among older women (e.g., ages 46+), which may be quite common. While, to our knowledge, there are no strong reasons to expect patterns of fostering among such children to respond differently to climate exposures than those in our sample, any such systematic differences would bias our estimates. Fourth, we are similarly unable to measure the out-fostering of women who have died (i.e., we do not study maternal orphans), which could introduce biases if women’s mortality is correlated with climatic shocks and such women are systematically different than those who survive. In each case, additional analyses—and perhaps new data collection efforts—are needed to better understand and measure such selection processes.
Despite these limitations, new attention to the links between climate variability and child fostering is merited in our view. Additional research on this topic is needed to improve our conceptual understanding of the links between environmental change and geographic mobility, pushing the now-extensive demographic literature on climate change and migration in new directions. This additional research is also substantively important given the significant health and socioeconomic consequences of fostering for children and their families.
While it is possible that fostering may occur within very small geographic areas (e.g., within a village or neighborhood), we assume that a change in guardianship involves a move across a non-trivial distance in most cases.
Temperature and precipitation anomalies capture short-term climate variability (i.e., short-term deviations from long-term climate averages), which is distinct from climate change (i.e., long-term shifts in the temperature and precipitation distributions). Despite these differences, it is currently standard practice among population-environment scholars to measure the demographic impacts of climate variability and (implicitly or explicitly) use those estimates to inform expectations about the likely effects of future climate change (e.g., Bohra-Mishra et al., 2014; Davenport et al., 2017; Thiede & Strube, 2020).
Importantly, however, Mueller et al. (2020a) also find evidence of displacing effects in their study. For example, stresses associated with precipitation deficits increase out-migration in Botswana and Kenya.
One other exception is a recent simulation of the links between rainfall and return migration in Thailand by Entwisle and colleagues (2020). This analysis highlighted the economic demands that returning migrants may place on receiving households.
Relatedly, in some situations, children may be fostered in anticipation of future shocks (Kielland, 2016).
For example, while all households are at risk of in-fostering, out-fostering can only occur among households with children.
The motivations for this sequence of interaction models were determined before the analysis, drawing on prior demographic research and considering what DHS data could be used appropriately. As described below, we do not include moderators (or controls) that could have been plausibly affected by the focal climate exposure terms.
The household sample weight is variable hv005 in the original DHS files, and is used as recommended by the DHS program. According to DHS documentation, “The household weight…for a particular household is the inverse of its household selection probability multiplied by the inverse of the household response rate in the stratum” (Croft et al., 2018: 1.31).
We exclude 18,102 households with a large number (9+) of children to avoid outliers that we expect may reflect measurement error in many cases.
Information on who the child resides with (e.g., other parent, relatives, and others) is only available for some DHS surveys and is therefore not used in the analysis.
This assumption is important to our analysis and merits at least two points of elaboration. First, we are making assumptions about the initial decision to in- or out-foster a child, which is distinct from questions about the duration of fostering. The latter is important but cannot be analyzed precisely using our data—a limitation we discuss more in the concluding section. Second, previous research does not provide clear guidance about the presence and length of potentially lagged effects. We therefore tested the robustness of our findings in a supplementary model that included additional controls for temperature and precipitation anomalies in each of the five years prior to the survey (Models 9–12 and 17–20). Our main conclusions about the effects of temperature and precipitation in year t−1 did not change after controlling for conditions in these earlier years.
This measurement strategy does not capture intra-annual variation in temperature and precipitation, differences in which may be masked by annual averages. This limitation could be substantively meaningful in some situations. For example, a location that experienced alternating periods of much-above and much-below average temperatures or precipitation could have the same annual anomaly as a location that experienced average conditions consistently.
We use the term province to describe all first-level sub-national administrative units.
We henceforth refer to the number of children in the household for brevity. We do not measure the number of children born to non-surveyed women in the household (i.e., older women without birth histories) because of challenges measuring non-resident children.
We do not include controls (e.g., household wealth) or moderators that are measured at the time of the survey and potentially influenced by the focal climate exposures and fostering decisions. Such variables constitute poor control variables (Angrist & Pischke, 2009). We also note that the DHS collects relatively limited information on household social and economic dynamics (e.g., excluding detailed income or consumption records), which is appropriate given its focus on demographic and health outcomes but imposes some limitations on our analysis. We ran an additional robustness check controlling for at least one woman in the household working in agriculture (Models 13-16 and 21-24). Results from these analyses were consistent with our main findings.
We take the average across all years of data for countries with multiple surveys in the analytic sample.
We are likewise unable to detect climate effects on very short-term fostering arrangements that occurred between the exposure period and the DHS interview.
Akresh, R. (2009). Flexibility of household structure child fostering decisions in Burkina Faso. Journal of Human Resources, 44(4), 976–997.
Alber, E. (2018). Transfer of belonging. Brill.
Andriano, L., & Monden, C. W. (2019). The causal effect of maternal education on child mortality: Evidence from a quasi-experiment in Malawi and Uganda. Demography, 56(5), 1765–1790.
Angrist, J. D., & Pischke, J. S. (2009). Mostly harmless econometrics: An empiricist’s companion. Princeton University Press.
Archambault, C., & de Laat, J. (2010). Social mobility in children’s mobility? An investigation into child circulation among the Maasai of Kenya. Children and migration: At the crossroads of resiliency and vulnerability, 187–206.
Archambault, C. S., de Laat, J., & Zulu, E. M. (2012). Urban services and child migration to the slums of Nairobi. World Development, 40(9), 1854–1869.
Bachan, L. (2014). Anticipatory child fostering and household economic security in Malawi. Demographic Research, 30, 1157–1188. https://doi.org/10.4054/DemRes.2014.30.40
Baker, R. E., & Anttila-Hughes, J. (2020). Characterizing the contribution of high temperatures to child undernourishment in Sub-Saharan Africa. Scientific Reports, 10(1), 1–10.
Beck, S., Vreyer, P. D., Lambert, S., Marazyan, K., & Safir, A. (2015). Child fostering in Senegal. Journal of Comparative Family Studies, 46(1), 57–73. https://doi.org/10.3138/jcfs.46.1.57
Beegle, K., Dehejia, R. H., & Gatti, R. (2006). Child labor and agricultural shocks. Journal of Development Economics, 81(1), 80–96. https://doi.org/10.1016/j.jdeveco.2005.05.003
Behrman, J. A. (2019). Polygynous unions and intimate partner violence in Nigeria: An examination of the role of selection. Journal of Marriage and Family, 81(4), 905–919.
Biddlecom, A. E., Axinn, W. G., & Barber, J. S. (2005). Environmental effects on family size preferences and subsequent reproductive behavior in Nepal. Population and Environment, 26(3), 183–206.
Bilsborrow, R. E. (1987). Population pressures and agricultural development in developing countries: A conceptual framework and recent evidence. World Development, 15(2), 183–203.
Black, R., Bennett, S. R., Thomas, S. M., & Beddington, J. R. (2011). Migration as adaptation. Nature, 478(7370), 447–449.
Bledsoe, C. (1990). “No success without struggle”: Social mobility and hardship for foster children in Sierra Leone. Man, 25(1), 70–88. https://doi.org/10.2307/2804110
Bledsoe, C. H., Ewbank, D. C., & Isiugo-Abanihe, U. C. (1988). The effect of child fostering on feeding practices and access to health services in rural Sierra Leone. Social Science & Medicine, 27(6), 627–636.
Bohra-Mishra, P., Oppenheimer, M., & Hsiang, S. M. (2014). Nonlinear permanent migration response to climatic variations but minimal response to disasters. Proceedings of the National Academy of Sciences, 111(27), 9780–9785.
Borderon, M., Sakdapolrak, P., Muttarak, R., Kebede, E., Pagogna, R., & Sporer, E. (2019). Migration influenced by environmental change in Africa. Demographic Research, 41, 491–544.
Bougma, M., LeGrand, T. K., & Kobiané, J. F. (2015). Fertility decline and child schooling in urban settings of Burkina Faso. Demography, 52(1), 281–313.
Boyle, E.H., King, M., & Sobek, M. (2020a) IPUMS demographic and health surveys. Version 8 . Minneapolis, MN: IPUMS and ICF. https://doi.org/10.18128/D080.V8
Boyle, E. H., King, M. L., Garcia, S., Culver, C., & Bourdeaux, J. (2020b). Contextual data in IPUMS DHS: Physical and social environment variables linked to the Demographic and Health Surveys. Population and Environment, 1–21.
Call, M., & Gray, C. (2020). Climate anomalies, land degradation, and rural out-migration in Uganda. Population and Environment, 41, 507–528.
Carleton, T. A., & Hsiang, S. M. (2016). Social and economic impacts of climate. Science, 353(6304),aad9837.
Carrico, A. R., Donato, K. M., Best, K. B., & Gilligan, J. (2020). Extreme weather and marriage among girls and women in Bangladesh. Global Environmental Change, 65, 102160.
Corno, L., Hildebrandt, N., & Voena, A. (2020). Age of marriage, weather shocks, and the direction of marriage payments. Econometrica, 88(3), 879–915.
Cotton, C. (2021). An enduring institution? Child fostering in sub‐Saharan Africa. Population and Development Review, padr.12439. https://doi.org/10.1111/padr.12439
Cotton, C., Clark, S., & Madhavan, S. (2022). One hand does not bring up a child. Demographic Research, 46, 865–904.
Croft, T. N., Marshall, A. M., Allen, C. K., Arnold, F., Assaf, S., & Balian, S. (2018). Guide to DHS statistics. Rockville: ICF, 645.
Davenport, F., Grace, K., Funk, C., & Shukla, S. (2017). Child health outcomes in sub-Saharan Africa: A comparison of changes in climate and socio-economic factors. Global Environmental Change, 46, 72–87.
Davis, K. (1963). The theory of change and response in modern demographic history. Population Index, 29(4), 345–366.
DeWaard, J., Hunter, L. M., Mathews, M. C., Quiñones, E. J., Riosmena, F., & Simon, D. H. (2022). Operationalizing and empirically identifying populations trapped in place by climate and environmental stressors in Mexico. Regional Environmental Change, 22(1), 29.
Ellis, F. (1998). Household strategies and rural livelihood diversification. The Journal of Development Studies, 35(1), 1–38.
Eloundou-Enyegue, P. M., & Shapiro, D. (2004). Buffering inequalities: The safety net of extended families. In Cameroon. SAGA Working Paper.
Entwisle, B., Verdery, A., & Williams, N. (2020). Climate change and migration: New insights from a dynamic model of out-migration and return migration. American Journal of Sociology, 125(6), 1469–1512.
Fortson, J. G. (2008). The gradient in sub-Saharan Africa: Socioeconomic status and HIV/AIDS. Demography, 45(2), 303–322.
Fussell, E., Hunter, L. M., & Gray, C. L. (2014). Measuring the environmental dimensions of human migration: The demographer’s toolkit. Global Environmental Change, 28, 182–191.
Gaydosh, L. (2015). Childhood risk of parental absence in Tanzania. Demography, 52(4), 1121–1146. https://doi.org/10.1007/s13524-015-0411-4
Goody, E. (1982). Fostering and occupational roles in West Africa. Cambridge University Press.
Grace, K., Hertrich, V., Singare, D., & Husak, G. (2018). Examining rural Sahelian out-migration in the context of climate change: An analysis of the linkages between rainfall and out-migration in two Malian villages from 1981 to 2009. World Development, 109, 187–196.
Grant, M. J., & Yeatman, S. (2014). The impact of family transitions on child fostering in rural Malawi. Demography, 51(1), 205–228.
Gray, C., & Mueller, V. (2012). Drought and population mobility in rural Ethiopia. World Development, 40(1), 134–145.
Gray, C., & Wise, E. (2016). Country-specific effects of climate variability on human migration. Climatic Change, 135(3–4), 555–568.
Hampshire, K., Porter, G., Agblorti, S., Robson, E., Munthali, A., & Abane, A. (2015). Context matters: Fostering, orphanhood and schooling in sub-Saharan Africa. Journal of Biosocial Science, 47(02), 141–164.
Headey, D. D., & Ruel, M. T. (2022). Economic shocks predict increases in child wasting prevalence. Nature Communications, 13(1), 2157.
Heath, R., Hidrobo, M., & Roy, S. (2020). Cash transfers, polygamy, and intimate partner violence: Experimental evidence from Mali. Journal of Development Economics, 143, 102410.
Hedges, S., Sear, R., Todd, J., Urassa, M., & Lawson, D. W. (2019). Earning their keep? Fostering, children’s education, and work in north-western Tanzania. Demographic Research, 41, 263–292. https://doi.org/10.4054/DemRes.2019.41.10
Hoddinott, J., & Mekasha, T. J. (2020). Social protection, household size, and its determinants: Evidence from Ethiopia. The Journal of Development Studies, 1–20. https://doi.org/10.1080/00220388.2020.1736283
Hoffmann, R., Dimitrova, A., Muttarak, R., Crespo Cuaresma, J., & Peisker, J. (2020). A meta-analysis of country-level studies on environmental change and migration. Nature Climate Change, 10(10), 904–912. https://doi.org/10.1038/s41558-020-0898-6
Hunter, L. M., Luna, J. K., & Norton, R. M. (2015). Environmental dimensions of migration. Annual Review of Sociology, 41(1), 377–397. https://doi.org/10.1146/annurev-soc-073014-112223
ICF. (2023). “Methodology.” The DHS Program. http://www.dhsprogram.com. [June, 22, 2023].
Jensen, R. (2000). Agricultural volatility and investments in children. American Economic Review, 90(2), 399–404.
Kalipeni, E. (1996). Demographic response to environmental pressure in Malawi. Population and Environment, 17(4), 285–308.
Kielland, A. (2016). The role of risk perception in child mobility decisions in West Africa, empirical evidence from Benin. World Development, 83, 312–324. https://doi.org/10.1016/j.worlddev.2016.01.008
Kielland & Gaye. (2010). Climate change and the role of children in household risk management strategies in rural Senegal (Report No. 69811). World Bank. https://openknowledge.worldbank.org/server/api/core/bitstreams/edd0cd61-a9dd-5c5b-a228-97e9c2868c8a/content
Kielland, A., & Kebede, T. A. (2020). Drought vulnerability and child mobility in rural Senegal. Forum for Development Studies, 47(3), 427–445. https://doi.org/10.1080/08039410.2020.1739122
Knox, J., Hess, T., Daccache, A., & Wheeler, T. (2012). Climate change impacts on crop productivity in Africa and South Asia. Environmental Research Letters, 7(3), 034032.
Madhavan, S., Schatz, E., Clark, S., & Collinson, M. (2012). Child mobility, maternal status, and household composition in rural South Africa. Demography, 49(2), 699–718.
Massey, D. S., Axinn, W. G., & Ghimire, D. J. (2010). Environmental change and out-migration: Evidence from Nepal. Population and Environment, 32, 109–136.
McDaniel, A., & Zulu, E. (1996). Mothers, fathers, and children: Regional patterns in child-parent residence in sub-Saharan Africa. African Population Studies, 11(1), 1–28.
Mounirou, I., & Yebou, J. (2022). Perceptions of climate risks and migration of agricultural producers in northern Benin. Society & Natural Resources, 36(2), 190–208.
Mueller, V., Gray, C., & Hopping, D. (2020a). Climate-induced migration and unemployment in middle-income Africa. Global Environmental Change, 65, 102183.
Mueller, V., Gray, C., & Kosec, K. (2014). Heat stress increases long-term human migration in rural Pakistan. Nature Climate Change, 4(3), 182–185.
Mueller, V., Sheriff, G., Dou, X., & Gray, C. (2020b). Temporary migration and climate variation in eastern Africa. World Development, 126, 104704.
Nawrotzki, R. J., & DeWaard, J. (2018). Putting trapped populations into place: Climate change and inter-district migration flows in Zambia. Regional Environmental Change, 18(2), 533–546.
Pickering, A. J., & Davis, J. (2012). Freshwater availability and water fetching distance affect child health in sub-Saharan Africa. Environmental Science & Technology, 46(4), 2391–2397.
Rigaud, K. K., de Sherbinin, A., Jones, B., Bergmann, J., Clement, V., Ober, K., & Midgley, A. (2018). Groundswell: Preparing for internal climate migration. Washington,, DC: World Bank.
Schlenker, W., & Lobell, D. B. (2010). Robust negative impacts of climate change on African agriculture. Environmental Research Letters, 5(1), 014010.
Sekhri, S., & Storeygard, A. (2014). Dowry deaths: Response to weather variability in India. Journal of Development Economics, 111, 212–223.
Serra, R. (2009). Child fostering in Africa: When labor and schooling motives may coexist. Journal of Development Economics, 88(1), 157–170.
Sheffield, J., Goteti, G., & Wood, E. F. (2006). Development of a 50-year high-resolution global dataset of meteorological forcings for land surface modeling. Journal of Climate, 19(13), 3088–3111.
Thalheimer, L., Schwarz, M. P., & Pretis, F. (2023). Large weather and conflict effects on internal displacement in Somalia with little evidence of feedback onto conflict. Global Environmental Change, 79, 102641.
Thiede, B. C., Randell, H., & Gray, C. (2022). The childhood origins of climate-induced mobility and immobility. Population and Development Review, 48(3), 767–793.
Thiede, B. C., & Strube, J. (2020). Climate variability and child nutrition: Findings from sub-Saharan Africa. Global Environmental Change, 65, 102192.
Tsaneva, M. (2020). The effect of weather variability on child marriage in Bangladesh. Journal of International Development, 32(8), 1346–1359.
Weinreb, A., Stecklov, G., & Arslan, A. (2020). Effects of changes in rainfall and temperature on age- and sex-specific patterns of rural-urban migration in sub-Saharan Africa. Population and Environment, 42(2), 219–254. https://doi.org/10.1007/s11111-020-00359-1
Zimmerman, F. J. (2003). Cinderella goes to school the effects of child fostering on school enrollment in South Africa. Journal of Human Resources, 38(3), 557.
Ronnkvist acknowledges support from the University of Wisconsin-Madison’s University Fellowship. Support for this fellowship is provided by the Graduate School, part of the Office of Vice Chancellor for Research and Graduate Education at the University of Wisconsin-Madison, with funding from the Wisconsin Alumni Research Foundation and the UW-Madison. Ronnkvist also acknowledges the support of the Center for Demography and Ecology (P2C HD047873 & T32 HD007014). Thiede acknowledges the assistance provided by the Population Research Institute at Penn State University, which is supported by an infrastructure grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (P2CHD041025). Thiede’s work was also supported by the USDA National Institute of Food and Agriculture and Multistate Research Project #PEN04623 (Accession #1013257). Barber acknowledges support from The Pennsylvania State University’s College of Agricultural Sciences undergraduate research award program.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Below is the link to the electronic supplementary material.
About this article
Cite this article
Ronnkvist, S.R., Thiede, B.C. & Barber, E. Child fostering in a changing climate: evidence from sub-Saharan Africa. Popul Environ 45, 29 (2023). https://doi.org/10.1007/s11111-023-00435-2