Correlates of child marriage at the individual and socio-economic level

Concerning the socio-economic conditions that increase the likelihood of child marriage, several studies have identified parents’ and the girls’ level of education as strong predictors [27–30]. For example, Ali et al. [28] show in their analysis of survey data from Sudan that women who have only acquired primary education are significantly more likely to be married under 18 than those with a higher level of education. It is also known that child marriages particularly often affect illiterate girls [31]. Apart from this, Envulado et al. [32] document a significant positive association between low parental education and increased risk of underage marriage among girls, suggesting that parental educational attainment plays a vital role in shaping marriage-related decisions. At the same time, the authors point out that the economic status of the family is equally important as they observe higher marriage rates among girls whose fathers were farmers, thus relying on lower and more seasonal incomes [32]. In addition to Akter et al. [33] and Hoogeven et al. [34], the study by Kok et al. [26] points to the exceptionally high density of child marriages in regions affected by instability and low income. Corroborating this finding with data from Ethiopia, Aychiluhm et al. [30] show that child marriages take place particularly in rural and impoverished areas of the country. Another study, based in Ethiopia, also explores the links between household wealth and child marriages, revealing that particularly assets owned by a mother are associated with lower rate of girls’ marriage rates, whereas paternal assets do not seem to be a key factor [35]. The strong link between economic hardship and child marriages also has important implications for programmatic efforts. Recent evidence from Bangladesh shows that financial incentives can significantly reduce child marriage, whereas empowerment programmes alone may be insufficient. The findings suggest that early marriage may serve as a strategic signal in the marriage market signalling to prospective grooms and their families that a bride is obedient and adheres to conservative gender norms, particularly in contexts dominated by traditional values. In such settings, altering economic incentives appears more effective than attempting to shift gender norms directly [36].

Apart from this, the effect of poverty is particularly strong in rural areas [32]. Al-Ridhwany et al. [29] confirm an urban-rural divide in child marriage rates in their study from northern Iraq. Corroborating this, several studies suggest that the urban–rural divide in child marriage also reflects differences in prevailing gender norms. Using Afrobarometer data from 34 African countries, Charles [37] shows that support for women’s equal rights is systematically higher among urban residents and among individuals more exposed to extra-local cultural influences. Similarly, evidence from Demographic and Health Surveys across multiple sub-Saharan African countries indicates that, in most cases, rural residence is associated with a higher likelihood of justifying intimate partner violence against women, a commonly used proxy for conservative gender norms, although important cross-country variation exists [38]. Another significant social factor contributing to the prevalence of child marriage rates is parental authority. Aychiluhm et al. [30] argue that child marriages are frequently arranged in contexts of limited autonomy of the prospective bride and substantial pressure exerted by her parents.

Correlates of child marriage at the institutional level

The law around child marriage differs from country to country but often targets the minimum age of marriage and informed consent. To protect girls and boys under the age of 18, countries may adopt various strategies like criminalisation of child marriages, banning marriage below the legally prescribed minimum age, or prescribing a minimum age of marriage without criminalising or banning it [39]. For example, under the Indian Prohibition of Child Marriage Act of 2006, where the minimum age of marriage for girls is 18 and for boys is 21, people involved in performing, conducting, or directing a child marriage can be imprisoned. Further, child marriages can be nullified within two years of the minor reaching adulthood [40]. In Zambia, while the minimum age for marriage is 18 years under the Prohibition of Child Marriage Act, individuals between 16 and 21 were still allowed to get married with parental consent until 2023. In addition, customary laws often took precedence over statutory law, making enforcement of the minimum age difficult [19,41]. Previous studies have shown that legal reforms have a limited impact on child marriage prevalence, and they need to be integrated into broader gender transformative approaches that tackle underlying causes like gender norms, low status of women and girls and cultural values and beliefs [1,42, 43]. Fattah and Camelia [44] used a mixed-methods design in rural Bangladesh and reported high levels of awareness among participants regarding the negative consequences of child marriage. The sample was socio-demographically diverse and included both female and male participants with varying levels of education and self-reported household economic status, ranging from extreme poverty to wealth. The authors further highlight that participants perceived early marriage as a means of protecting family reputation, as it was believed to prevent social stigma in cases where a young girl had experienced sexual assault or harassment.

The belief that the unmarried status of young girls tarnishes the family’s reputation following a sexual attack is evident across various regions, including North African countries [21], Ethiopia, Ghana [15,45] and India [7]. Such entrenched value systems appear to be a crucial barrier to the success (or effectiveness) of interventions that seek to enhance the self-determination of young girls [46]. Traditional gender roles thus predetermine the trajectories of girls’ lives. In such contexts, the adolescent desire for intimacy, recognition, and social belonging can often be fulfilled only through marriage. Yet this occurs within a normative framework that defines women’s roles as “passive, abnegated wives and mothers” [47, p. 5]. The prospect of abstaining from child marriage appears scarcely desirable, as besides the pressure exerted by the social milieu surrounding young girls [48], the entire family would face societal sanctions and ostracism [49,50]. Moreover, religious convictions may also impede the expansion of freedoms for young girls. For instance, Ahonsi et al. [45, p. 2] underscore that "major religious traditions in Ghana encourage early marriage". However, it is essential to note that a universal impact of religious tenets cannot be assumed as two studies by Cislaghi et al. [51,52] from Cameroon elucidate that certain traditions permit marriage only upon reaching the age of maturity.

The influence of external shocks on child marriage

A body of literature has examined how external shocks, such as those introduced by unforeseen extreme weather events or natural disasters, may affect girls’ vulnerability to underage marriage [53]. For example, Ahmed [54] conducted a study investigating the repercussions of the 2010 flood disaster in Pakistan, finding that there was a significant decrease in child marriage rates contrary to the authors’ initial expectations. The authors contended that the calamity likely engendered a state of emergency that compelled individuals from forming partnerships as a means of emotional coping. Moreover, Ahmed [54] underscored the possible adherence to legal regulations regarding the minimum age of marriage, notably emphasising the Child Marriage Restraint Act of 1929, which unequivocally prohibits child marriages. However, this observation’s generalisability was challenged by a subsequent study by Ahmed et al. [17], which investigated the impact of a flood disaster in Bangladesh. The researchers highlighted the high poverty rates prevalent in the affected regions and demonstrated an increase in child marriages following the disaster. Drawing on focus group interviews, they argued that parents of married daughters deemed child marriage necessary due to the family’s economic constraints stemming from the disaster. Marriage was perceived as a means to transfer the financial responsibility for the daughter to the groom’s family, thereby alleviating the financial burden the bride’s family faced. Other analyses based on samples from India corroborate this observation [55,56]. Blanco [57] assesses how exposure to different earthquakes within Indonesian provinces between 1994 and 2014 affected girls’ chances of being married under age. Specifically, the author reveals a 44% increase in the annual hazard of girls marrying before the age of 18. In line with previous explanations, the author finds that the daughter’s marriage functioned as a financial coping strategy through the receipt of a marriage payment. Yet, she also reveals evidence for an additional mechanism, whereby the destruction of school buildings caused a drop in school attendance, consequently increasing marriage rates. Significant increases in child marriage rates following different natural disasters are further shown in a study by Das and Dasgupta [58], who analysed the societal consequences of the 2001 Gujarat earthquake in India, as well as in studies by Tsaneva [59] and Carrico et al. [6] who assess the impact of heat waves in South Asia.

Another type of external shock could have been introduced by the COVID-19 pandemic and the linked economic repercussions caused by lockdown measures and considerable employment losses. The pandemic rapidly exposed the disproportionate burden borne by low- and middle-income countries, many of which entered the crisis with limited fiscal space and institutional capacity. In particular, the ability to cushion the economic consequences of the necessary lockdown measures was barely possible in these countries. Before the pandemic began, there were warnings that poverty rates would rise in the short and medium term, as previous work could no longer be carried out due to the pandemic measures [60]. South Asia, where household income is highly dependent on labour migration, was particularly hard hit by travel and movement restrictions [61]. Household income was also severely reduced in African countries due to the pandemic, and poverty rates rose significantly in a short period [62–64].

Only few empirical studies have evaluated the impacts of the COVID-19 pandemic on child marriage rates. For example, one study by Melnikas et al. [65] draws on data from repeated cross-sectional surveys that were conducted as part of the “More Than Brides Alliance” impact evaluation in 609 villages across four Indian states. Observing child marriage rates over the course of five years, the authors are able to show declines in marriage rates between their baseline and midline measurement, which occurred prior to the COVID-19 pandemic, but then found that this trend stagnated in the course of the pandemic and that child marriage rates increased slightly between midline and endline measurement. Conversely, Batheja et al. [66] leverage data from the Indian National Family Health Survey conducted between 2019–2021 (NFHS-5) and reveal a significant increase in the age at marriage when comparing women who got married after the onset of the pandemic to those married before. The authors suggest that the observed decline in child marriage rates may be driven by a negative income shock, which made it impossible for the bride’s parents to afford the wedding and dowry payment for their daughter.

Research gap

To the best of our knowledge, no study to date has provided a systematic cross-context analysis of child marriage during the pandemic using mixed-method approaches and considering cultural practices in conjunction. Understanding the full impact of the COVID-19 pandemic and external or economic shocks more broadly, requires attention to cultural practices, such as marriage-related transfers, which shape households’ strategies for coping with financial strain. Yet, only few studies investigate economic stress by also considering respective local culture context.

For instance, Jibon et al. [67] document pandemic-related economic stress and rising child marriage rates, but without accounting in detail for customs like dowry or bride price. Similarly, studies such as Julianto et al. [68] and a review by Haq et al. [25] highlight economic determinants while neglecting cultural mechanisms. Brandl and Colleran [69] provide a theoretical link between marriage payments and marriage patterns; however, their analysis remains largely conceptual, as it does not provide quantitative evidence to support the proposed framework [69]. Our paper builds on their contributions by offering comparative evidence from India and Zambia. These two contexts are characterised by contrasting marriage institutions, namely dowry and bride price, which allows us to examine how these mechanisms interact. We provide both statistical estimates and in-depth contextual understanding to uncover the micro mechanisms that drive child marriage under pandemic related stress, offering a more comprehensive perspective than previous research. The subsequent section introduces our theoretical framework and outlines the hypotheses guiding our analysis.

COVID-19, marriage customs, and child marriage: Theoretical framework and predictions

To examine how income changes such as those caused by natural disasters or external shocks affect marriage prospects, Corno et al. [7] introduced an analytical model, that will guide the hypotheses we test in this paper. In their model, the researchers make three fundamental assumptions about child marriage. First, it is assumed that girls contribute less to the household welfare than they consume, whereas this effect is not prevalent in boys, who are considered as better able to work. Second, it is assumed that, apart from a financial transaction between the groom’s and bride’s family that is arranged in the context of the marriage, the bride must move-in with the groom’s family to shift the financial situation of both households. Third, the model assumes that there is gender balance in the community, leading to a balance in the marriage market. Based on these three assumptions, the micro-mechanisms of the two effect directions are described below, and hypotheses are formulated. To test the predictions of their model, the authors combined weather data with data from the Demographic and Health Survey (DHS) from Sub-Saharan Africa and India. Their analyses revealed that while periods of drought in African countries significantly increased the probability of child marriages by up to three percentage points on average, a similarly significant but opposite effect was found in the sample from India, that is, a decrease in the probability of child marriages. Corno et al. [7] argued that the heterogeneous findings that they observe between the two regional contexts are driven by a third factor: customary marriage transfers.

The practice of bride price

Based on these findings, our first hypothesis is focused on the customary marriage transfer in Zambia: the “bride price”. In this form of marriage transfer, it is assumed that the bride’s family must receive financial compensation as the relocation of their daughter to the household of the groom’s family implies that she can produce economic value for her family-in-law in terms of household or paid labor [7]. Thus, a payment in the form of a bride price is made by the groom’s to the bride’s family. This payment can be particularly high if the bride is considered young and sexually inexperienced at the time of marriage [70]. If a household’s income falls sharply and unexpectedly due to an external shock like a natural disaster, families may have to resort to alternative income generation opportunities, particularly in contexts where poverty rates and economic pressures are already high [71]. Asadullah et al. [55, p. 964] argue that receiving a bride price can serve as a “consumption-smoothing strategy” for families under economic pressure [6,17,57]. Thus, in societies where the practice of bride price is common, child marriage rates can be expected to increase in times of economic shocks. We believe that the same mechanism likely holds true in the context of the COVID-19 pandemic and its linked economic pressures and will therefore test the following hypothesis:

• Hypothesis 1: In the context of Zambia where the “bride price” custom is prevalent and where families are economically burdened by the COVID-19 pandemic, an increase in child marriages can be expected.

The practice of dowry

Our second hypothesis shifts focus to the customary marriage transfer prevalent in India: the “dowry”. A dowry denotes a transfer from the bride’s family to the groom’s family. Dowry payments are often customary in contexts where female employment rates are generally low and where the daughter-in-law is perceived as placing more of an economic burden on the household she lives in [30]. In India, dowry payments may even amount up to several times the annual income of the bride’s family [72,73]. If a household is put under severe economic strain due to an external shock, parents of potential brides are incentivised to suspend or delay marriage arrangements to avoid further financial strain. This prediction is supported by empirical evidence from a panel study conducted in Pakistan: The authors show that in villages where marriage costs are high, early marriage was delayed after households had experienced financial shocks and were therefore more credit constrained [74]. Applying this to the context of the COVID-19 pandemic, we therefore test the following hypothesis:

• Hypothesis 2: In the context of India where the “dowry” custom is prevalent and where families are economically burdened by the COVID-19 pandemic, a decrease in child marriages can be expected.

Study setting

The study focuses on two countries: India and Zambia. Menon [73] estimates that 1.5 million girls in India are married before reaching adulthood, and that almost one in three girls are married before their 18th birthday in Zambia. These high prevalence rates make both countries relevant contexts for studying child marriage. Although the data presented in the following tables may suggest a gradual decline in child marriage rates, the absolute numbers remain alarmingly high [75]. Nevertheless as shown in Table 1, the overall child marriage decreased over the last 25 years, but remains high, meaning that 23.3% of all children were married under age in 2021.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Child marriage rates in India by year.

https://doi.org/10.1371/journal.pone.0346851.t001

For Zambia, the currently available data indicate a downward trend in child marriage rates over the past decades. As shown in Table 2, the rate has decreased from 46.9% in 1993 to 29% in 2018.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Child marriage rates in Zambia by year.

https://doi.org/10.1371/journal.pone.0346851.t002

While this long-term progress in both countries is encouraging, there is currently a lack of comprehensive nationally representative data that would allow for a fine-grained assessment of the impact of the COVID-19 pandemic on child marriage rates, which may have significantly disrupted this positive development. According to Equality Now [78], the pandemic has exacerbated many of the underlying drivers of child marriage across Africa—such as poverty, school closures, and reduced access to protection services, which could drive surges in underage marriages.

In India, data were collected in the second most populous state of the country, Maharashtra. Specifically, we targeted peri-urban and urban settlements around the city of Pune as well as rural villages around the city of Sangli. In Maharashtra, approximately 25% of the population lives in poverty, with higher rates in rural areas [79,80]. In Zambia, we collected data in three districts of Zambia’s Southern Province: Kalomo, Mazabuka, and Monze. This province is characterised by low population density, many families living in rural settlements, and a high poverty rate, with 51% of households having lived in extreme poverty in 2022 [19].

Sampling strategy

Data collection

We collected quantitative data via standardised questionnaires administered on tablets with the assistance of female enumerators. Enumerators in India were recruited via two partner NGOs (the SNEH Foundation in Pune and the Astitva Foundation in Sangli) and were fluent in Marathi and Hindi. Data collection was conducted from 1–30 March 2022. In Zambia, enumerators were recruited by the survey firm IPSOS and were fluent in English and Tonga. Data collection took place between 9 May and 9 December 2022. In both countries, enumerators were extensively trained in interview techniques for collecting sensitive data and research ethics, including informed consent procedures, confidentiality, and referral of at-risk participants. Each interview lasted between 60 and 90 minutes and was conducted at participants’ homes, on school grounds after lessons (for girls still attending school) or in a convenient community location (such as a community hall), ensuring that confidentiality could be guaranteed in the setting. Similar to previous studies conducted with adolescents in low-literacy populations [81], the surveys were designed as audio- and computer-assisted self-interviews (ACASI) with the intention of maximising confidentiality as well as reducing possible social desirability bias and under-reporting. In both countries, we sought written (or fingerprinted, in the case of limited literacy) informed consent from girls who were 18 years or above. For girls under the age of 18 years (the majority of our sample), we adopted a two-stage consent procedure by first collecting written informed consent from a parent or legal guardian and subsequently collecting verbal assent from each girl.

In addition to quantitative data, we collected qualitative data. In India, we collected extensive qualitative data from a total of eight focus group discussions (FGDs) and nine individual in-depth interviews. The focus group discussions were carried out prior to the quantitative data collection with groups of mothers (2), fathers (2), adolescent girls (2), and social workers (2). In-depth interviews were conducted after completion of the quantitative survey, and participants were sampled purposely based on having disclosed any of the following adverse outcomes: (i) having experienced family or intimate partner violence in the previous year, (ii) having been married underage, and (iii) reporting a current or past pregnancy. Interview guides were aimed at eliciting the harmful impacts of the COVID-19 pandemic and at discussing their underlying causes. In Zambia, due to budget constraints, we could only conduct a total of three FGDs. These were carried out prior to the quantitative data collection with one group of mothers and two groups of adolescent girls. Similar to the FGDs in India, discussion guides were focused on the harmful impacts of the COVID-19 pandemic. All discussions and interviews were conducted in the respective local language (i.e., Marathi, Hindi or Tonga) and recorded verbatim and then transcribed and translated into English.

Measures

Analysis

Associations between explanatory variables on the individual, socio-economic, institutional, and external shock levels with the likelihood of child marriage were estimated based on multi-variable linear probability regression analyses that were run separately for the respective samples from India and Zambia. We chose linear probability regression over logistic regression to facilitate the interpretation of coefficients, which indicate percentage-point changes in the likelihood of being exposed to child marriage for each included explanatory variable. For ease of interpretation, we also estimated a logistic regression model to calculate the predicted probability of child marriage for a low- and high-intensity economic shock scenario, with the latter being defined as a household having experienced at least two of the pandemic’s economic repercussions listed in S1 Table. The sampled age ranges are narrow (13–18 in India; 15–19 in Zambia), limiting the potential for meaningful non-linear effects; within these age bands, the likelihood of child marriage tends to increase linearly with age, and including an age-squared term would add unnecessary complexity. Analyses were carried out in R, Version 4.3.3. All data and code is available via https://osf.io/2xgpb/.

Ethics

In India, the study was approved by the ethics committee of the Indian Institute of Technology Gandhinagar (IITG) in November 2021. In Zambia, the study was approved by the National Health Research Authority in July 2022. Both study components were additionally approved by the ethics committee of the medical faculty at the Technical University of Munich (TUM) on 29 December 2021. Additional information regarding the ethical, cultural, and scientific considerations specific to inclusivity in global research is included in the Supporting Information (S1 Checklist in S1 File).

Socioeconomic characteristics, child marriage, and impact of the pandemic

Descriptive information on girls’ socio-demographic status, exposure to child marriage, and COVID-19 impacts are summarised in Table 3, separately for the sample from India and Zambia.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Sociodemographic and COVID-19 impact data from India and Zambia.

https://doi.org/10.1371/journal.pone.0346851.t003

Correlates of child marriage

We explored possible factors associated with child marriage on the (i) individual, (ii) socio-economic, (iii) institutional, and (iv) economic shock level based on separate linear probability regression models (see Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Correlates of child marriage on different levels.

Note: Coefficients from linear probability regression model. *p < 0.05, **p < 0.01, ***p < 0.001. Level 1 (yellow): individual level, Level 2 (blue): socio-economic level, Level 3 (red): institutional level, Level 4 (green): economic shock level. Coefficients at different levels of the theoretical framework were estimated in separate models to avoid multicollinearity and endogeneity concerns. In each model, we controlled for individual-level factors, namely age and caste in India.

https://doi.org/10.1371/journal.pone.0346851.g002

In the sample from Zambia, only two factors were significantly associated with the outcome variable. Specifically, the strongest predictor of girls’ marital status was their age, with a six percentage point increase in the probability of being married with every additional year of life. In addition, the rate of child marriages was significantly lower in households with higher living standards and wealth. The effect translates into a significant one percentage point decrease in the probability that a girl is married under age with every additional asset or living standard indicator that a household acquires. Notably, we did not observe any significant associations between child marriage and parents’ education status or girls’ internalised gender norms. Importantly, contrary to the predictions made by our Hypothesis 1, we did not observe a significant increase in child marriage rates among households that were more adversely affected by the economic pressures of the COVID-19 pandemic. In fact, the coefficient for the COVID Economic Stress Index even points in the opposite direction, albeit not being significant. Thus, we must reject Hypothesis 1, as we find no indication that the economic shock induced by the COVID-19 pandemic has prompted households to seek alternative income sources through a bride price and marrying off their daughters as a consumption-smoothing strategy.

In the sample from India, girls’ age was also a significant correlate of child marriage, translating into an eight percentage point increase in girls’ likelihood of being married with every year of age. Girls from lower castes were also more at risk of being married under age, whereby their probability of being married was 3.5 percentage points higher as compared to girls who were from higher castes. On the socio-economic level, girls whose fathers and/or mothers had completed secondary schooling were significantly less likely to become child brides, thus suggesting that parents’ higher education is a protective factor with respect to child marriages. Similar to what we observed in Zambia, girls from economically better-off households were also less likely to be married under age, whereby any additional asset or living standard indicator translated into a 0.5 percentage-point decrease in the probability of child marriage. In addition, our analyses revealed a significant link between child marriage and the institutional-level factor of gender norms. Specifically, girls who were more approving of violence against women (reflecting stronger internalised patriarchal gender norms) were significantly more likely to be married before reaching the age of 18, with every additional scenario (out of 7) in which a girl considers it justifiable for a husband to beat his wife translating into a 0.7 percentage point increase in the probability of being married. Finally, our analysis fails to support the predictions made by our Hypothesis 2. While—based on our theoretical framework—we expected a decrease in child marriage rates within a dowry-based system like India, our regression reveals the opposite trend. Specifically, we observe a significant 0.8 percentage-point increase in the likelihood of girls being married underage with each additional economic pressure faced by a household due to the COVID-19 pandemic.

Lastly, for a more nuanced interpretation of the economic shock effect, we calculate the predicted probability of child marriage for a low- and high-intensity economic harm scenario. As shown in Fig 3, in Zambia, we do not observe any significant differences in the predicted probability of child marriage when comparing a hypothetical household that has experienced no substantial economic shock due to the pandemic with a hypothetical household that has suffered significant economic harm, for example through income losses or food shortages. Notably, the predicted probability even drops from 6.30 percent to 4.02 percent with higher economic pressures, thus suggesting that families in Zambia were not resorting to child marriages with the intention to smooth consumption by means of a bride price. Conversely, in India, we report a significant increase in the predicted probability of child marriages: While in the low-intensity economic harm scenario, there is only a 6.50 percent chance that a girl is married under age, this risk almost doubles and increases to a 11.35 percent probability for girls to become child brides in the high-intensity economic harm scenario. The significant surge in the projected likelihood of child marriages in India implies that dowry payments did not function as a hindrance to the establishment of formal unions in face of the economic strains experienced during the COVID-19 pandemic.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Predicted probability of child marriage for different economic harm scenarios.

Predicted probabilities with robust standard errors based on logistic regression model, controlling for age and caste (in India). COVID-19 Economic Harm was coded as 1 (“yes”) if a household had suffered two or more economic repercussions (see S1 Table), and as 0 (“no”) otherwise.

https://doi.org/10.1371/journal.pone.0346851.g003

Qualitative findings on the associations between COVID-19 economic pressures, marriage customs, and child marriage

Findings and implications for the cultural context of India

In India, 9% of adolescent girls included in the study were married underage, with significant age gaps between spouses. Marriages were predominantly arranged by families, dowry was paid in 24% of the marriages and economic pressures during the pandemic increased the likelihood of child marriage.

Our findings on the individual, socio-economic, and institutional correlates of child marriages integrated into our conceptual framework align with existing literature. Similar to previous studies, we also find that individual-level factors such as higher age and lower caste were associated with an increased risk of child marriage [84,85]. In many LMICs, menarche translates into marriage eligibility, which is deeply rooted in patriarchal norms like the maintenance of purity before marriage and upholding the family reputation [86–88]. As a result, societal and familial pressure to marry off girls may intensify as they grow older. A previous study conducted with the same sample of 3049 adolescent girls in India presented that social expectations around marrying early were one of the main drivers of adolescent marriage and pregnancy [53]. Significant age differences between adolescent girls and their partners, as seen in this sample, can also lead to power imbalances in the relationship, leading to loss of sexual and reproductive health-related decision-making power and even intimate partner violence [89].

The higher incidence of child marriage among lower caste families reveals the compounded disadvantages and social marginalisation faced by these groups. Other studies also corroborate these findings [90,91]. At the socio-economic level, parent’s education level and household wealth were negatively associated with the probability of child marriage. Studies from other LMICs have also shown that higher parental education and wealth protect against child marriages [35,46,92]. Additionally, the institutional-level factor of gender norms, here captured by a measure of justifying violence against women, emerged as a critical factor influencing the incidence of child marriage. Previous studies from India and Pakistan have also presented that women married as children were more likely to justify spousal violence [93,94]. Others have highlighted that women who were married as children or married as adolescent girls face increased violence and controlling behaviour from their partners [81,95,96].

Finally, with regards to the impact of economic shocks, our analysis did not support the hypothesis derived from our conceptual framework that child marriage rates would likely decrease in consequence of the economic shock caused by the COVID-19 pandemic. In contrast, we found significantly higher rates of child marriages among families that experienced substantial economic hardships due to the pandemic. Nearly one-fifth of the child brides in our sample reported that the timing of their marriage was influenced by the pandemic, causing their marriages to occur earlier than they would have under normal circumstances. Qualitative interviews with married adolescents highlighted that the lack of dowry demand and low wedding costs made marrying during lockdown lucrative. While most previous studies observe a negative relationship between economic shocks and child marriage rates in dowry-based societies, a study from Bangladesh also presented that in the aftermath of a flood, marriages were less expensive as most households suffered economically and social expectations around dowry and wedding expenses were lower, leading to families marrying off their daughters [17]. Likewise, another study in the same setting from 2020 showed that households coped with economic hardship by marrying off their daughters to reduce household consumption [6].

Findings and implications for the cultural context of Zambia

In Zambia, the rate of child marriage was slightly lower at 7%, with a higher proportion of marriages driven by romantic relationships rather than family arrangements. Higher household wealth was strongly associated with lower rates of child marriage in both countries. But while the COVID-19-related economic shock played a crucial role in the family’s financial situation, it did not lead to an increase in child marriages as predicted by our conceptual framework.

Like in the Indian context, in Zambia as well, inequitable gender norms that prioritise women’s roles as wives and household caretakers and men’s role as the breadwinner often result in low investment in girls’ education and future economic prospects. The drivers of child marriage in this sample were highlighted during interviews with girls. They mentioned that school closure due to the pandemic, economic pressures leading to non-payment of school fees, and prolonged periods of tedium due to the lockdown made families and girls lose interest in future education. Similar channels were also identified in a qualitative study in Indonesia after the pandemic [13]. A combination of inequitable norms and a lack of avenues to pursue education pushes girls to form financially beneficial relationships in return for sex, leading to school dropout, adolescent pregnancies and subsequent child marriages [97–99].

The significant association between household wealth and lower child marriage rates aligns with previous research from LMICs, which emphasises the protective effect of economic stability against early marriage [100–103]. In our sample, despite significant economic hardships that households faced during the COVID-19 pandemic, like loss of employment or income, there was no corresponding increase in child marriage rates, suggesting that Zambian families employed different coping mechanisms than their Indian counterparts. Families in Zambia did not marry their daughters in lieu of receiving the bride price as a consumption smoothing strategy. This finding is in contrast to previous studies that indicated an elevated risk of child marriages in societies with bride price traditions following natural disasters or income shocks [6,18,26,55,56,58,59,83]. One possible explanation for our findings is that financial hardship during the pandemic affected households broadly, including families of prospective grooms. This may have reduced their ability to afford high and attractive bride price payments, thereby lowering the incentives for parents of potential brides to marry off their daughters early during periods of widespread economic distress. While our findings clearly differ from those reported in Corno et al. [7] and from the predictions made by our conceptual framework, it is important to note that the COVID-19 pandemic represents an unprecedented external shock that is not fully comparable to natural disasters. In particular, the prolonged economic consequences of lockdown policies likely imposed a distinct form of financial strain on households, differing from the strain typically observed during environmental or humanitarian disasters, which are often accompanied by substantial inflows of humanitarian aid and relief programmes. Moreover, our results indicated that social distancing policies were deliberately leveraged to reduce expenditures on daughters’ weddings, a mechanism specific to the pandemic context and unlikely to generalise to other types of shocks. Against this background, our conceptual framework cannot be validated in the specific context of the COVID-19 pandemic.

Limitations and avenues for future research

This study has several limitations. The cross-sectional nature of the data does not allow us to draw causal conclusions on how the risk factors are linked to child marriage. To mitigate this, we present a conceptual framework based on previous studies and qualitative interviews with adolescent girls to unpack potential channels linking economic, social, cultural, and disruptive factors to child marriage outcomes. Apart from this, we did not collect any data prior to the pandemic based on which we could estimate the increase in child marriage rates caused by the pandemic. However, we apply an extensive list of measures that represent nuanced variations in the extent to which households were negatively affected by the pandemic. Further, data collection was limited to two rural and semi-urban locations in India and three rural locations in Zambia, which may affect the generalisability of our findings. Lastly, the relatively small number of focus group discussions conducted in Zambia limits the depth of contextual insights that can be drawn for this setting.

Future studies should aim to collect longitudinal data and leverage quasi-experimental designs to establish causal relationships between the highlighted risk factors and child marriage. Additionally, incorporating qualitative interviews with not only adolescent girls but also their parents and relatives could provide deeper insights into the coping mechanisms and cultural contexts that drive child marriage, helping to unpack the nuanced ways in which economic, social, and cultural factors interact to influence marriage decisions.

Implications

The findings from this study have significant implications for intervention development and policy making aimed at reducing child marriage in LMICs both during public health emergencies (PHEs) and beyond. Restrictive gender norms are a significant barrier for girls in achieving their aspirations and avoiding child marriage. Eroding this would require interventions to promote communication and support from their parents and community. Such programmes could help adolescents and parents to develop ways of coping with conflict and setting shared boundaries. One such intervention is the Families Matter! Programme in Tanzania that aims to empower caregivers to discuss sexuality and practice positive parenting skills with their children [104]. Building on this, future interventions could be adapted to include discussions and activities specifically focused on sexual reproductive health and child marriage.

A recent systematic review of the association between PHEs and various gender outcomes highlighted that PHE-related response strategies like home confinement and closure of schools, along with associated economic losses at both national and household levels are key channels behind the increase in child marriage in LMICs [53]. While these measures are essential to curb new infections, governments and policymakers should focus on complementary measures to safeguard adolescents’ health and life outcomes. These include providing safe spaces for them to socialise and receive education, organising mobile youth-friendly health and counselling services, subsidising or removing school fees to allow girls to return to school, and integrating child marriage prevention messages into the broader policy landscape [105–109]. Lastly, addressing structural barriers like poverty, low education, and limited employment opportunities requires continuous development of policies and programmes at a national level to increase the economic resilience of families, especially during crises. This could include social protection programmes, job creation initiatives, and support for small businesses.