Discussion paper KDPE 2001 by Amrit Amirapu (University of Kent), M Niaz Asadullah (University of Malaya) and Zaki Wahhaj (University of Kent), March 2020.
Non-technical summary
In developing countries around the world, one in three women marry before reaching the age of 18. The practice of early marriage has wide-ranging adverse consequences. Research has shown early marriage leads to worse educational outcomes, weaker social networks and greater risk of domestic violence for the women who experience them, as well as lower investments in the health and education of their children.
In most countries, there is a minimum age of marriage prescribed by the law. But, in countries with weak institutions, the law is frequently ignored and rarely enforced. By contrast, there is strong evidence, particularly for South Asia, that patriarchal norms, often in the form of pressure to marry following the onset of puberty from the extended family and the wider community, play an important role in the marriage timing of women. Parents, family elders and other members of the extended family play an influential role in marriage decisions and ignoring their wishes can give rise to long-term tensions within the family.
In recent years, a number of countries have adopted legislative reforms to tighten the restrictions on early marriage, or improve the enforcement of the law. Where the state has weak institutional capacity, whether such legislative reforms can affect social attitudes and behaviour relating to marriage is a key question. To address this policy question, we conducted a randomised information intervention in 80 villages following a recent change in child marriage law in the country. We recorded the incidence of female adolescent marriages in the study areas during a 10-month period following the intervention. The information intervention was based on a child marriage bill approved by the Bangladesh parliament in March 2017. The bill had two key elements. First, it introduced harsher punishments for adults who married a minor or facilitated an early marriage. Second, it introduced an `exceptions clause’ that decriminalises early marriages in certain unspecified circumstances, if a court deemed that the marriage was in the best interest of the child. The minimum marriage age remained at 18 for women and 21 for men as it had been under the previous law.
The information intervention took the form of a short drama written and performed exclusively for the study and lasting 2-3 minutes, involving the marriage of a 15-year old girl, that was shown, individually, to study participants. There were three versions of the drama. The first version (henceforth referred to as Treatment 1 video), provided information about the harsher punishments but not the exceptions clause in the new law. A second version (Treatment 2 video), provided information about both the harsher punishments and the exceptions clause. A third version (Control video), made reference only to the previous law on child marriage.
The main participants in the study were respondents in 80 randomly selected rural clusters from an earlier, nationally representative survey of women, conducted in 2014. A number of additional family members were interviewed per primary respondent, with equal probability of 0, 1 or 2 additional interviews. During the baseline interviews, one third of the primary respondents were individually shown one of the treatment/control
videos on a tablet device, with equal probability of being assigned to any one arm. All members of the extended family were interviewed separately and in parallel with the primary respondent and shown the same video. Primary respondents with unmarried adolescent daughters were contacted again 5 months and 10 months after the intervention and asked, for each unmarried daughter, whether the family had taken any steps towards her marriage.
Our information treatments led to a change in participants’ own attitudes and behaviour (including reported attitudes regarding appropriate marriage age and willingness to contribute to a charity that campaigns against child marriage), but did not substantially influence their beliefs about attitudes or practices prevalent in their community. Contrary to expectations, we find that providing information about the harsher punishments for underage marriage specified in the law increased the probability of marriage of adolescent girls by 7.2% points relative to the control group after 5 months (for comparison, the probability of marriage in the control group during the same period was 1.2%). The treatment also increased the probability of receiving a marriage offer by 13.3% points and, conditional on receiving an offer, increased the probability that the offer was accepted by 20.3% points. We obtain similar effects for the 10-month period following the intervention. The effects of the second treatment – in which respondents were informed about both the harsher punishments and the exceptions clause – go in
the same direction but they are smaller in magnitude and statistically insignificant.
We find that the adverse effects of the intervention are concentrated in households where information about the new law was provided to extended family members. Specifically, we find that the effect of either treatment on marriage-related decisions is close to zero when only the mother is informed about the change in child marriage law. But when the same video is shown to her husband or family elders, the adolescent daughters in Treatment 1 are at greater risk of early marriage compared to the control group (the corresponding effects for Treatment 2 are more muted).
The adverse effects of informing parents and extended family members about the harsher punishments for child marriage specified in the new law, although counterintuitive, are consistent with a narrative in which parents hastened to marry off their daughters before strict enforcement of the new law came into effect. Although, we do not have direct evidence of this narrative, two other effects we observe lend support to it. First, the adverse effects are more muted when (i) parents and extended family members are additionally informed about the exceptions clause, which would have made the law seem more aligned with patriarchal norms; (ii) the treatment is restricted to mothers only who are likely to place a greater weight on their children’s welfare.
The absence of an adverse effect in the last case is interesting in its own right, because it implies that mothers do not share the information they are provided with other family members who have a say in the marriage timing of their daughters. This suggests that the objectives of different family members are not aligned: mothers may place greater weight on the welfare of their daughters even while other family members try to abide by patriarchal norms