Introduction

One of the most noted manifestations of gender discrimination in large parts of the developing world is seen in terms of inequitable inheritance and property rights. In many traditional societies, particularly in Sub-Saharan Africa land inheritance continues to be dictated by customary law, which excludes women from property ownership (Harari, 2019). Women’s legal inability to inherit land leads to undermining their economic security, access to economic opportunities, and further contributes to increased “feminization of poverty” across many of these developing regions of the world (Agarwal, 2003; Kabeer, 2016). It is therefore argued that such forms of discrimination can only be corrected through strong legal reforms, which give women equal rights in the matter of inheritance. According to the World Bank (2012), “Legal reforms aimed at establishing equal inheritance over property are the most promising policy measures to increase women’s voice in society.” Land rights are thus regarded as crucial determinants to bring about more equitable gendered patterns of wealth distribution and household bargaining power. Some empirical studies have shown how assets acquired through inheritance have a positive impact on women’s social mobility (Anderson & Binder, 2015), the agency in intrahousehold decision-making (Anderson & Eswaran, 2009), marriage market outcomes (Kudo, 2015), fertility choices (Jayachandran, 2015), and labour market mobility (Fernando, 2020).

In recent years, there have been reforms in inheritance laws across many developing countries to improve gender equity in inheritance distribution. According to UN Women (2015), 115 countries of the world have now recognised women’s property rights on equal terms with men. Changes in inheritance laws and their impacts on women’s empowerment have been explored in different contexts. These studies present a rather mixed picture. Some studies reveal that gender-positive legislation is positively associated with an improvement in women’s autonomy and well-being indicators. For instance, Heath and Tan (2020) analysed how the passage of the Hindu Succession Act (2005) prompted greater autonomy in household decision-making, increased participation in public life and women’s increased labour supply in high-paying jobs. In another study, Menon et al. (2014) demonstrate that with the passing of the Vietnam Land Law (1993), which tilted land rights more heavily in favour of women, there was an increased allocation of household resources towards health, education and nutrition, away from economic “bads” like tobacco and alcohol. An in-depth case study on Rwanda’s Pilot Land Tenure Regularisation Programme (2004) revealed that this programme improved land access for married women, enhanced tenure security for female-headed households and promoted better recordings of inheritance rights without bias (Ali et al., 2014). In the Peruvian context with the tilting of land rights in favour of women, more women started substituting working from home with working outside in higher-paying jobs (Field, 2007). Harari (2019) found that amendments in inheritance laws in Kenya had a significant impact on health outcomes in terms of reducing the incidence of gender mutation, getting medical assistance during childbirth, and delaying the age of marriage and childbearing.

However, other studies have revealed that gender equitable inheritance laws on their own may not be a sufficient condition for women’s empowerment, and in some cases, the passage of these laws may have adverse impacts on women in actual practice. For instance, Bhalotra et al (2020) showed that with the passage of the HSAA (2005) in India, there was an increased prevalence of sex-selective female abortions in areas with an already high degree of son preference. Other studies reveal how the implementation of HSAA resulted in increased intra-household conflicts, heightened incidents of domestic violence and the prevalence of suicides among both men and women (Anderson & Genicot, 2015). These studies demonstrate that such policy initiatives alone may not be sufficient to positively affect women’s well-being in the presence of rigid societal norms, which oppose women from inheriting parental property. There are also concerns that the positive short-run impacts of these reforms in terms of greater autonomy for women, may dwindle over time (Deininger et al., 2019). Therefore, it is crucial to analyse the second-generation impact of these reforms.

Causal evidence on the second-generation effects of gender-positive inheritance rights is limited. In an empirical study, Deininger et al (2019) studied how in the Indian context, women who were beneficiaries of the HSAA (2005) tended to invest 2.2–2.5 times more in their daughter’s education, compared to non-beneficiaries. Ajefu et al (2022) concluded that there was a lower prevalence of nutritional deficiencies in children whose mothers were beneficiaries of HSAA (2005). However, so far there has been no empirical research linking women’s inheritance with the health outcomes of their children. This question is pertinent because a large body of literature reveals that with increased asset endowments women tend to invest more in their children, leading to enhanced health outcomes in the subsequent generations (Baranov et al., 2020; Duflo, 2003).

This study contributes to this critical yet under-researched area, by analysing how women’s inheritance has impacted the key anthropometric indicators of their offsprings in the case of Kenya, drawing on subsequent rounds of data from the Kenya Demographic and Health Surveys. Kenya provides a compelling context to analyse these outcomes, as property rights in the country are governed by multiple and overlapping legal domains. Up to 1981 inheritance rights in Kenya were dictated by the customary law of the ethnic group of the deceased individual, and in the case of Muslims by Islamic law. The customary law of all ethnic groups in Kenya denied women any rights over the inheritance of property, whereas Islamic law entailed women to half of the inheritance share that went to their brothers. In 1981 inheritance rights in Kenya were made a matter of statutory law and equal inheritance rights were instituted for men and women, irrespective of their religious and ethnic affiliations. However, in 1990 Muslims were allowed to revert to the Islamic law of succession again. In our empirical investigation, we have considered the timing of the 1981 reform and the subsequent amendment in 1990, as well as the cross-sectional variations in religious/ethnic affiliations to quantify the effects of the reform on a variety of child health and nutritional outcomes, employing a difference-in-difference framework.

Conceptualizing Women’s Inheritance Rights

Parental bequests of wealth are considered the primary mechanism to transfer assets across generations. These affect the patterns of capital accumulation, as well as associated economic gains (Becker & Tomes, 1979). In analysing the nature of inter-generational bequests, traditional models of wealth transfer (Becker, 1974) treat households as a single unit, in which all resources are pooled. It is assumed that the household head is an altruistic planner, who will make economic decisions that maximise both their present consumption, as well as their children’s future income. However, these unitary, gender-neutral models of economic wealth have been criticised by a number of scholars, as these are not deemed to be representative of the heterogeneity in preferences and female bargaining power within the household (Agarwal, 1995; Anderson & Eswaran, 2009). Some empirical studies have demonstrated that the transfers of assets to women have resulted in better household food security, an increase in survival rates, higher investment in schooling, and better educational outcomes for the girl child (Duflo, 2003; Duflo & Udry, 2004; Luke & Munshi, 2011; Qian, 2008).

Harari (2019) has hypothesized that when inheritance laws are made more gender equitable, there are two possible consequences. Following the changes in inheritance laws, an individual household may decide to invest more or less in women’s human capital. This depends on whether physical capital (ancestral property) and human capital (health, education, and nutrition) are seen as substitutes or complements. In cases where human capital and physical capital are considered as complements, more equitable inheritance laws will also translate into higher investment in women’s human capital and a consequent improvement in their socio-health outcomes. This has been demonstrated empirically by scholars such as Roy (2015). Alternatively, if human and physical capital are considered substitutes, as parents are forced to bequeath more physical assets to women, they would tend to disinvest in their human capital formation. Some empirical studies have confirmed this and revealed that in some contexts with the passing of inheritance laws women’s health, education, and labour market outcomes have suffered (Rosenblum, 2015).

Thus, there remains considerable ambiguity on the possible outcomes of these gender-equitable inheritance laws. This paper examines these questions empirically in the context of Kenya in the subsequent sections. We will assess whether amendments in women’s inheritance laws have sustained second generation impacts in terms of improving health and nutritional outcomes of their children, and how have these impacts differed among the varied ethnic and religious groups over time. This inquiry will enable us to generate crucial empirical evidence on this question and provide a useful exemplar to quantify these effects in other contexts as well.

Women’s Inheritance Rights in Kenya

In Kenya, the laws of succession and property rights are governed by multiple and overlapping legal domains. Prior to 1981, there were four separate systems of inheritance-traditional African, civil/Christian, Muslim and Hindu. According to Kameri-Mboti (1995) virtually all ethnic groups in Kenya, prohibited women from inheriting land from their parents and deceased husbands. Islam inheritance on the other hand was governed by the principles enshrined in the Koran. Only one-third of the deceased’s estate can be dealt with by will. The remaining two-thirds is allocated by intestacy rules laid down in the Koran, with fixed shares for specific heirs. A wife receives one-fourth of her husband’s estate, women in polygamous marriages receive one-eighth if they are childless. What is left is divided between sons and daughters in such a way that sons receive twice as much as daughters of their father’s property.

The Kenyan constitution embraced legal pluralism and recognised the application of customary law and Islamic law in specific instances. Section 82 (4) of the constitution provided that customary law of an individual’s tribe could be applied in cases of adoption, marriage, divorce, burial, and devolution of property (Kameri-Mboti, 1995). However, on July 1, 1981, a Law of Succession Act was passed which sought to implement a uniform code of inheritance in the country. This legislation consolidated all the four systems of inheritance into one uniform statute, applicable to all Kenyans. The Law of Succession Act was rooted in western jurisprudence and established equal inheritance rights for female and male children on their parents’ property, regardless of whether married or unmarried. This legislation provided that if the deceased has one surviving spouse and children, the spouse is entitled to an absolute interest in the deceased’s personal and household property. In case of polygamous unions, the estate is divided among the households according to the number of children in each. According to Mutongi (2007), while the law was non-discriminatory in terms of treatment of the children of the deceased, there remained vestiges of discriminatory practices in terms of treatment of the spouses. For instance, when the surviving spouse is a woman, her interest in the property is invalidated if she re-marries, whereas a surviving husband maintains his interest upon remarriage.

Between 1981 and 1990, there were intense agitations by Muslims who regarded the Law of Succession Act as it repudiated on the constitutional provision of freedom of worship and religion. The government gave in to the pressure mounted by the Muslim community in 1990. The Succession Act was amended by statute to exclude application to Muslims. Since then, inheritance law has remained a contentious issue in Kenya. The concession given to the Muslim population has made it difficult to implement inheritance laws on a uniform basis. This has resulted in less-than-optimal outcomes in the area of equitable property rights for women. There is no official data/quantitative evidence available on the enforcement of the Land Succession law and asset ownership by Kenyan women before and after the reform. However, some qualitative field-based assessments have revealed that this legislation suffers from a number of enforcement problems, as well as strong opposition to women’s asset ownership rooted in local customs and traditions (Cooper, 2011).

Data Source and Empirical Strategy

The paper employs data from subsequent rounds of the Kenyan Demographic and Health Surveys (DHS): 1989 (DHS-I), 1993 (DHS-II), 1998 (DHS-III), 2003 (DHS-IV), 2008/2009 (DHS-V), and 2014 (DHS-VI). As this study is specifically focused on outcomes for women (aged 15–49) and their children (0–59 months), these survey rounds capture relevant information about their socio-demographic profile, human capital indicators, marital relations, reproductive and maternal health, and rich information about the health conditions of their children. For our analysis, our sample includes all women captured in each of the survey waves.Footnote 1

The relevant DHS waves were administered post-reform, with the exception of the 1989 wave, which was only administered shortly before the 1990 amendment. We measure health outcomes for the children whose mothers where beneficiaries of the inheritance reforms and hypothesize that these outcomes are related to women’s beneficiary status from the inheritance reforms. We argue that the health status of the child is rooted in the parental bequests of wealth to their mother, complementary investment in the mother’s human capital formation and marital match affected by women’s premarital endowments. This is explained in further detail in the discussion part.

Specifically, the outcome variables from the DHS survey are four anthropometric measures of child health status: (a) a child is underweight, stunted, and wasted, which is a binary indicator if the standard deviation of the weight-for-age, height-for-age, and weight-for-height of the child (0–59 months) of the woman is less than or equal to -2; (b) a child is underweight is a binary indicator if the standard deviation of the weight-for-age of the child is less than or equal to -2; (c) a child is stunted is a binary indicator if the standard deviation of the height-for-age of the child is less than or equal to -2; (d) a child is wasted is a binary indicator if the weight-for-height of the child is less than or equal to -2.

The DHS survey also includes variables for the age of the woman, which we relied on to compute the squared value of the age. We also control for a series of other unobserved variables, such as those that are peculiar to the ethnicity of the individual, the survey wave, the province of residence, the birth year, and the time trend peculiar to a province and a particular region in Kenya—i.e., the Arid and Semi-Arid Lands (ASAL). These locations comprise the poorest areas in the country (Kalele et al, 2021). The government has identified individuals in these areas as needing specific attention and has also developed targeted poverty alleviation schemes for these areas such as abolition of school fees up to the 4th year of primary school (Harari, 2019).

Table 1 presents the summary statistics of the outcome variables for the entire population, and for Muslim and non-Muslim women. Twenty-three percent of the children of the sampled women are underweight, stunted, and wasted (28.5% for Muslim and 22.3% for non-Muslim women). For the individual health indicators of the children, 30.3% are underweight (35.9% for Muslims and 30% for non-Muslims), 31.2% are stunted (33.2% for Muslims and 31% for non-Muslims), and 23.4% are wasted (31.1% for Muslims and 22.5% for non-Muslims). The statistics reveal that children of non-Muslims display better health indicators compared to those of Muslims, and the focus of this study is to examine if the inheritance rights contribute to the disparity in child health outcomes by the religion of the women.

Table 1 Summary statistics of outcome variables
Full size table

Empirical Strategy

This study uses an identification strategy similar to Duflo (2001), Bleakley (2010), La Ferrara and Milazzo (2017), and Harari (2019). In this study, we exploit variation in pre-reform customary inheritance laws across different religious groups using a difference-in-differences framework. Specifically, within the difference-in-differences approach, we consider cohorts exposed and not exposed to the reform across Muslims and non-Muslims religious affiliation. The identification assumption in our analysis entails that in the absence of the change in customary inheritance law, there would have been no difference in child health outcomes between religious group (Muslims and non-Muslims) over time.

The customary inheritance law reform in Kenya occurred in two distinct periods. First, the 1981 Law of Succession grants all women a share of parental inheritance equal to that of their brothers. Second, an amendment was introduced in 1990 which exempts Muslims from the rule. See Table 2 for a summary of the three inheritance regimes in Kenya. Following Harari (2019) we identify that in the pre-1981 regime, non-Muslim women had no inheritance rights (zero share of assets), while Muslim women were entitled to half the share of what goes to their brothers as inheritance. In the periods between 1981 and 1990, the Law of Succession relates to both Muslims and non-Muslims as women are entitled to an equal share in inheritance as their brothers, irrespective of religious affiliation. After 1990, the Law of Succession only applies to non-Muslims. Muslims reverted to the pre-1981 rule which grants women half the share of their brothers’ entitlement.

Table 2 Inheritance regimes in Kenya
Full size table

The empirical specification used in this paper considers a cohort’s exposure to the reform as well as the amendment that follows. We define cohort 1, 2, and 3 equal to 1 for women with birth year between 1959 and 1968, 1969 and 1978, and after 1979 respectively.

To estimate the impact of women’s inheritance rights on child health outcomes, we consider the difference-in-differences specification shown below: v

$$\begin{gathered} Y_{ijrt} = \beta_{0} + \beta_{1} non - Muslim_{i} + \beta_{1} post_{1t} \times non - Muslim_{i} \hfill \\ \;\;\;\;\;\;\;+\,\beta_{12} post_{1,2t} \times non - Muslim_{i} + \beta_{2} post_{2t} \times non - Muslim_{i} + \beta_{3} X_{ijrt} \hfill \\ \;\;\;\;\;\; +\varphi_{e} + \delta_{r} + \theta_{w} + \tau_{t} + \emptyset_{r} \times t + ASAL_{d} \times t + \varepsilon_{ijrt} \hfill \\ \end{gathered}$$
(1)

where \(Y_{ijrt}\) denotes child health outcomes for woman i belonging to ethnicity j, living in province r at time t. Post 1, post 1,2, and post 2 equal 1 for cohorts born between 1959 and 1968, between 1969 and 1978, and after 1979, respectively. \(\beta_{3}\) denotes the coefficients of covariates or control variables, \(\varphi_{e} , \delta_{r} , \theta_{w} , \tau_{t} , \emptyset_{r} , \tau_{t}\) represent ethnicity, DHS wave, province and birth year fixed effects. The regression controls for province x time trend, ASAL x time trend. Standard errors clustered at the religious group x birth year level. Standard errors clustered at the interaction of religious group and birth year.

Results and Discussions

This section analyses the impact of the reform on the health outcomes of children born to beneficiary women. Our empirical analysis examines the impact of the reform on anthropometric indicators of health for children aged 0–59 months, for women exposed to the reform. Since the inheritance rights relates to agricultural lands as well, where factors such as crops grown on land, livestock and cropping area are of particular significance, we have analysed rural and urban households separately in our empirical model.

In the conceptual framework outlined previously, we noted that the human capital investment in women in terms of their education can be seen either as a substitute or a complement to inheriting physical bequests. We consider the educational outcomes of women as an important channel that reflects the physical bequests by parents. We also consider other outcomes related to the bargaining power and autonomy of women as the potential channels through which the reforms’ affect is seen on the child’s health outcomes. We argue that both these outcomes could be affected by improvements in women’s education as a result of the reform, and which could then translate into significant positive improvements in terms of child health outcomes. These supplementary outcomes reflect decision-making processes of different agents within the household and may be regarded as an indicator of women empowerment. It is plausible that as a result of educational empowerment, women’s household bargaining power and female autonomy would be positively affected through investment in the daughter’s education. These would then translate into improved health outcomes for children of the beneficiary women.

Child Health Outcomes

We consider four child health outcomes in Table 3, (i) the likelihood of the child being underweight, stunted, and wasted (columns 1), (ii) the likelihood of the child being underweight (columns 2), (iii) the likelihood of the child being stunted (columns 3), and (iv) the likelihood of the child being wasted (columns 4). The benchmark specification employs the binary treatment indicator, corresponding to Eq. (1) is presented in Table 3. Column (a) refer to the estimates of the outcome variable for the entire sample, while columns (b) and (c) report the estimates for rural and urban households separately.

Table 3 Child health outcomes
Full size table

The coefficient of the interaction between the dummy for women with birth years 1959–1968 and the non-Muslim dummy (Cohort 1 ×  Non-Muslim) in column 1a is not significant as expected, and there is no heterogeneity in this effect by the location of the household (columns 1b and 1c). The coefficient for women with birth years 1969–1978 and the non-Muslim dummy (Cohort 2 × Non-Muslim) is negative and not significant at the traditional levels. This negative effect is only significant for women whose households are located in urban areas, suggesting a 6.8 percentage point decline in the likelihood of adverse health outcomes for an exposed non-Muslim woman. This implies a long-term child health payoff for non-Muslim women residing in urban areas who were exposed to the second implementation of the reform relative to Muslim women. Although there is no heterogeneous effect in the child’s health outcome for exposed non-Muslim women to the third implementation of the reform by the location of the household, the estimates for women with birth years 1979 (and beyond) and the non-Muslim dummy (i.e., Cohort 3 × Non-Muslim) suggests a 2.9 percentage point significant decline in the likelihood of the child of a non-Muslim woman being underweight, stunted, and wasted compared to the child of a Muslim woman (see column 1a).

This significant reduction in the likelihood of having a child with adverse health outcomes (i.e., underweight, stunted, and wasted) as displayed in column 1a suggests the relevance of full inheritance rights in affecting long-term outcomes of the offspring of female beneficiaries. Our results suggest that non-Muslim women with full inheritance rights had better child health outcomes compared to Muslim women with only half inheritance rights (see the coefficient of Cohort 3 ×  Non-Muslim in column 1a).

The estimates in column 2 of Table 3 consider the outcome variable—the likelihood of the child being underweight, and the results again shows no significant effect for the child of a non-Muslim woman in the first regime (Cohort 1 ×  Non-Muslim) since only Muslims benefited from half of the inheritance rights. However, the estimates for the second and third implementation in column 2a (Cohort 2 ×  Non-Muslim and Cohort 3 ×  Non-Muslim) suggest a 4 percentage point and 3.6 percentage point decline in the likelihood of the child of the non-Muslim woman to be underweight relative to the child of a Muslim woman. We only find a significant effect of the decline in the likelihood of the child of a non-Muslim woman being underweight (column 2c) and stunted (column 3c) for those whose households are in urban locations, especially those who were exposed to the second reform implementation.

Qualitatively, the patterns of the estimates in columns (4a–c) are similar to those found in columns 1 and 2, but estimates are slightly higher in magnitude. For example, a non-Muslim woman that is exposed to the second and third implementation of the reform is 5.6 percentage point and 5.1-percentage point less likely to have a child that is wasted, respectively. These effects are in comparison to the child of a Muslim woman. Again, this effect is only significant for the child of such a woman whose household is located in urban areas.

The results presented in Table 3 are robust according to a number of checks: First, there are some districts where land is communally owned, and it will be difficult to attribute land ownership to individual households. Hence, these districts were exempted from the reform.Footnote 2 In our initial analysis, we included all districts in the benchmark specification because it was impossible to identify which individuals are unaffected by the reform due to this exemption status owing to the lack of information about the location of the parental assets or the nature of such assets. However, we subsequently follow the approach adopted by Harari (2019) by re-estimating the benchmark specification for only the exempt districts, and we typically find insignificant treatment effects in Panel A of Table A1 in the appendix. Therefore, this result can be interpreted as further evidence that we are capturing the effects of the reform.

Second, in Panel B, we estimate the specification for only women that are older than age 20 years, in order to avoid censoring issues and picking up effects from early childbirth or early marriage entry.Footnote 3 The estimates in Panel B of Table A1 show that the estimates are robust to the exclusion of women that are younger than 20 years from the sample, but we do not see heterogeneous difference in the effect for the outcome variables—child is underweight, stunted, and wasted, child is underweight, and child is wasted—by the location of the household of the woman. Third, our results are robust to the inclusion of the household-level wealth index in Panel C of Table A1.

Despite the fact that this variable can be seen as an endogenous control (Harari, 2019), it is reassuring that the estimated effect of the reform is not affected by the inclusion of the household wealth variable. In Panel D of Table A1, we perform a falsification test that arbitrarily assigns the Muslim group to a “placebo” religion. This “placebo” religion now becomes the Muslim group and is treated as our comparison group. We find a precise zero difference in the outcome variables for the full sample, which provides further reassurance that the effect seen in Table 3 is driven by the religion of the beneficiary woman.

Having established the robustness of the main results, we further explore a supplementary check that reflect siblings’ competition for household resources to understand some other factors that could be driving the estimated results in Table 3. In Table 4 we employ an additional indicator of the intra-household competition for resources among siblings. It is likely that the estimated effect in Table 3 will depend on the number of siblings of the woman who will compete for the available household resources, and thus limit the extent of parental bequest. Hence, the number of siblings could be a source of treatment heterogeneity, showing that the effects of the reform are less pronounced for a woman with a large number of siblings—and hence a smaller potential inheritance, which could have significant long-term effect on her offspring’s. The indicator—Num. Siblings—is a dummy if the woman has two or more siblings, and the estimates in columns 1–4 shows no significant effect of the number of siblings of the sampled woman in terms of the health outcomes of her offspring. Likewise, the estimate of interest is the variable Post 1968 × Non-Muslim × Num. Siblings—which reflects the effect of exposure to the reform for non-Muslim mothers, conditioned on the number of her siblings. The estimates from Table 4 also show a zero effect, suggesting that the number of siblings of the sampled woman does not drive the effect of the reform on child health outcomes.

Table 4 Child health outcomes conditioned on the number of siblings of the woman
Full size table

Alternative Mechanisms and Limitations

As explained in the conceptual framework, the link between the reform implementation and child health outcomes can be seen through the human capital investment channel in the optimal bequest hypothesis, such that parental investment in their daughter’s education may be seen as a complement or substitute to inheriting physical assets. In this section, we examine this link, and other outcomes that could emanate from changes in educational outcomes of the woman. Although we find evidence that several aspects of the lives of non-Muslim women change from the reform, we argue that it is unlikely these changes single-handedly explain better child health outcomes, highlighting a role for our focus on education, bargaining power, and woman’s autonomy. However, we acknowledge that we cannot rule out these alternate channels entirely. Given the possibility of other mechanisms linking the reform to child’s health outcomes, our results still link a reform that increases women’s inheritance rights to their offspring’s health. Finally, we do not perform a robustness check on these channels and they should only be examined as suggestive evidence. We only report the results of the full sample, as there is not much interest in the separate estimates for rural vs. urban households.

Human Capital Outcomes and Marital Market Outcomes

The reform implementation could potentially affect the human capital development of the sampled women or the human development conditions upon which they enter marriage or begin childbirth, which could in turn affect child health outcomes. For example, on one hand, parents can decide to invest in their daughter’s education as a complement to inheriting physical assets, if say, educated daughters can better manage family property. On the other hand, it is possible that forward-looking parents making human capital investments in their daughters might alter such investments based on their daughter’s legal entitlement to inherit ancestral property. Parents could perceive future inheritance as a lifetime transfer to their daughter (Heath & Tan, 2020), and may reduce other investments in their daughters in order to bring the total investment closer to their perceived pre-reform allocation. The direction of this effect is theoretically ambiguous.

To address the possibility that human capital investments are driving the relationship between the reform and child’ health outcomes that we estimated earlier, we begin by assessing whether the reform indeed appears to have affected the human capital of the women in our sample. Specifically, in Panel A of Table 5, we examine women’s education, height, age at marriage and childbirth using the identification strategy previously detailed. We find from column (1) that the likelihood of not being educated declined by 19.1 percentage points for non-Muslim women following the implementation of the second reform and 15.2 percentage point decline following the third reform implementation. Likewise, evidence from column (2) only shows significant decline in the likelihood of not being able to read and write by 25.6 percentage points for non-Muslim women exposed to the second reform and 25.8 percentage points for those exposed to the third reform. While column (3) shows a similar pattern for the outcome variable—dummy if the woman has only primary education. This result is similar to the findings in Harari (2019) that the implementation of the inheritance rights in Kenya led to a sizeable reduction of the gender education gap (increase in female education), which reflects a compatibility with the optimal bequest hypothesis. Nonetheless, we do not find an effect on the health of the exposed non-Muslim women (height indicator) but find significant increase of the age at first birth by 1.240 for non-Muslim women exposed to the third reform, and 1.423 years in the age at marriage entry. Similar significant positive effect is seen for non-Muslim women that are exposed to the second reform.

Table 5 Mechanisms
Full size table

Another possible alternative model that generates our results is if the reform affected the characteristics of the husbands of exposed women, and this change in marital matching led to improved health outcomes of their children. Note that because we find an improvement in the human capital outcomes of exposed women (their education, delay birth and marriage entry), then the changes in average husband characteristics are relatively likely. Essentially, we expect an improvement in the marital match, which has been shown to affect the health outcome of children (Chou, et al., 2010; Hahn et al., 2018).

We test whether the reform led to any changes in husband characteristics, namely, (i) his education, (ii) age, and (iii) the age difference between him and his wife. Panel B of Table 5 shows that there is indeed evidence that reform has an effect on the marital matches of exposed non-Muslim women. For example, we find higher estimates for the education of husbands, younger age or closer in age to their wives for women exposed to the second and third implementation of the reform. Thus, non-Muslim women who benefited from the reform seem to marry more educated spouses, younger husbands, and keeping the spousal age gap closer to theirs. In essence, the reform may have led to improvement in the extent of assortative mating, which is likely if say, more educated mates desire to match with women, who are better educated because of the parent’s bequest. Overall, we find sufficient evidence that the reform may have changed non-Muslim women’s human capital or choice of partner outcomes, as reflected in the measures available in the Kenyan DHS.

Female Bargaining Power, Autonomy and Labour Market Participation

Our conceptual model posits that the link between a woman’s autonomy or bargaining power and the health outcomes of her children is her ability to make decisions that could impact her children and the control of her earnings or economic resources within the household. As with the wife’s human capital outcomes and marital match discussed previously, we begin by estimating the effects of the reform implementation on key intra-household decision making outcomes in our sample, using our primary identification strategy detailed in Sect. 3. Panel C, D, and E of Table 5 test the effects of the reform implementation on the woman’s decision-making participation, female autonomy, and labour market participation.

We measure the decision-making participation using two indicators that reflects the woman’s health and usage of husband’s earnings, while we measure female autonomy as whether the husband insists on knowing her where about and does not permit her to meet her female friends, and labour market participation is measured using the woman is currently working.Footnote 4 We find no effect of the reform enforcement on female autonomyFootnote 5 (see Panel D), but we find improvements in women’s bargaining power over their health (see column 10) and how to spend the earnings of their husband/partner (see column 11). Specifically, we find 10.1 percentage points and 3.5 percentage points increase in the likelihood of the non-Muslim woman being solely responsible for decisions regarding her own health with exposure to the second and third reform implementation, respectively. Likewise, being exposed to the second and third implementation of the reform led to 13.1 percentage points and 8.3 percentage points increase in the likelihood of a non-Muslim woman being involved in decisions regarding the usage and spending of husband’s earnings. These results fit with the context of women’s increased bargaining power, which could affect the health outcome of their offspring. Finally, the labour market participation effect as shown in Panel E reflects a cooperative household model in which a woman’s level of earned income increases her bargaining power over her income (which we call bargaining power effect) and thereby her influence over how to spend such earnings on the health of her offspring. If the bargaining power effect leads her to work, then the increased earnings from such work translates to better health outcome of her offspring. Indeed, we find evidence of a positive labor market supply for non-Muslim women that are exposed to the reform (see Panel E).

Conclusion, Policy Implications and Limitations of the Study

We studied the Kenyan inheritance right, which was implemented in two regimes, such that in 1981 the reform formally established equal rights for men and women—regardless of religious affiliation, and in 1990 an exemption was established for Muslims to revert to the religious succession laws. We exploit the timing of the 1981 reform and subsequent 1990 amendment, and the cross-sectional variation in the religious affiliation of the women (Muslim vs. non-Muslim), to estimate the causal impact of the reform on child’s health outcomes, in a difference-in-differences framework.

We find that the inheritance reform improved the health outcomes of the children of beneficiary non-Muslim women. Non-Muslim women that are exposed to the second and third implementation of the reform are respectively 2.9 and 1.8 percentage points less likely to have a child that is underweight, stunted, and wasted. We find some evidence that this estimated effect is stronger for urban dwellers that benefited from the second implementation of the reform. We also find similar evidence when we consider the probability of the child of the exposed woman being, (i) underweight, or (ii) wasted. Our conceptual model explains that through a parent’s optimal bequest in investing in the human capital of daughters as a complement or a substitute to inheriting physical assets, as with other improvements in the woman’s outcomes, the child’s health outcomes will subsequently be impacted. By highlighting the relationship between human capital outcomes that leads to better marital match, intra-household bargaining, and labour force participation of women, our results suggest that the reform affects child’s health outcome not just through education but also through the choice of her spouse, ability to bargain over the utilization of household resources, and ability to earn such resources. Therefore, targeted policies that affects the empowerment of women—not just the inheritance rights reform that we study, but also other reforms to education, improvement of marital choices, bargaining power, and labor participation—can have a generational health effect.

Limitations of the Study

As with all empirical research, we must be careful when transferring the generalizability of our findings to other developing country contexts. Kenya represents an “extreme case” where religious identities have an overwhelming influence on women’s access to inheritance rights. This may not hold true in other developing country contexts, where other socio-economic variables such as caste, class and occupational divisions may also influence women’s inheritance rights. A second limitation of the work is that while we have focussed on indicators of child well-being such as stunting, wasting and malnourishment, it would be instructive to see how women’s access to property inheritance affects the psychological well-being of their offsprings. This question can be explored through subsequent qualitative and field-based work in this emerging area. Thirdly, our sample was limited to children in the age group of 0–59 months, there can be a subsequent study on the effect of women’s inheritance rights on health and educational indicators of children in older age groups as well.

Policy Implications

While improved health child’s health outcomes are important indicators of sustainable development, our results provide a new channel linking the improvement of women’s inheritance rights to the attainment of the Sustainable Development Goal 3—on good health and well-being. Previous literature has focused on a battery of issues, including the link between women’s empowerment through access to economic resources—e.g., (un-) conditional cash transfer—and human capital investments as a mechanism for this result (Barber & Gertler, 2010; Barham, 2011; Chari et al., 2017). Our results suggest that there can be important multiplier effects from policies that improve women’s access to ancestral inheritance. Since there is evidence that such policy can increase a parent’s desire to invest resources for the human development of their daughters (Heath & Tan, 2020), policies that seek to transfer or make access to equal access to inheritance between sons and daughters can prompt parents to invest in their daughter’s education, which has long-term effect on other outcomes in her later life. The flip side of this compounding is that parents making human capital investments in their daughters might alter such investments based on their daughter’s legal entitlement to inherit ancestral property. Our results suggest that reforms to women’s access to inheritance, such as the Kenyan Inheritance law, can propel parents into considerably human capital investment in their daughters, which is similar to the findings in Harari (2019).