Access the full text.
Sign up today, get DeepDyve free for 14 days.
The aim of terrorism all over the world is to have fear rule over people’s lives. The consequences of terrorist attacks, however, are substantially different across contexts. In this paper we study the association between exposure to Boko Haram’s attacks and households’ fertility choices in Nigeria. We hypothesise that households exposed to terrorism increase their number of children as a way to insure against future unexpected shocks. We test this hypothesis using geolocalised panel data linked to information on terrorist attacks that occurred in the region. Consistent with our hypothesis, terrorism is found to increase fertility (proxied by the number of surviving children per household): a one standard deviation increase in the number of fatalities increases the probability that a household hit by terrorism has a newborn by 1%. This association is robust to the use of difference-in-differences and instrumental variables models—and therefore can be given a causal interpretation. Keywords: terrorism, fertility, Boko Haram, Nigeria JEL Classiﬁcation: J13, I15, D19 1. Introduction The aim of terrorism is to have fear rule over people’s lives. Terrorist groups all over the world share this aim, irrespective of the context in which they operate. The consequences of terrorist attacks, however, can be substantially different depending on the context in which the attacks are perpetrated. In developing countries, terrorism poses a direct threat to achievement of sustainable development goals (SDGs) by undermining ‘peace, justice and strong institutions’ (SDG 16). However, terrorism can also constitute an indirect threat to the achievement of SDGs by generating psychological, social and political effects among the population due to the climate of fear the attacks evoke (Becker & Rubinstein, 2004; Kim & Albert Kim, 2018; Metcalfe et al., 2011; Romanov et al., 2012). In this paper, we study one of these consequences by © The Author(s) 2021. Published by Oxford University Press on behalf of the Centre for the Study of African Economies. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 2 Valentina Rotondi and Michele Rocca focusing on Boko Haram’s terrorist attacks in Nigeria and their effects on household fertility choices. From a theoretical standpoint, terrorism could be positively or negatively associated with fertility choices. Terrorism increases feelings of uncertainty and changes individuals’ risk assessments. This could have a negative effect on fertility choices, as in the case of famine, political events, economic decline (e.g., Lindstrom & Berhanu, 1999) and war (e.g., Agadjanian & Prata, 2002; Woldemicael, 2008). Or, terrorism could have a positive effect on fertility, as has been shown for other shocks, such as war (Easterlin, 1961; Van Bavel & Reher, 2013), bombings (Rodgers et al., 2005) and natural disasters (Nobles et al., 2015). In developing countries, in particular, children may be considered as insurance on future income against unexpected shocks (Lambert & Rossi, 2016; Pörtner, 2001). Parents might therefore react to the increased instability brought about by terrorist attacks by having more children. A large body of research relates reproductive outcomes to conflicts and violence (e.g., Agadjanian & Prata, 2002; Heuveline & Poch, 2007; Woldemicael, 2008), while relatively little work has studied the relationship between terrorism and birth and children’s outcomes (e.g., Bruckner et al., 2019; Camacho, 2008; Di Maio & Nisticò, 2019; Quintana-Domeque & Ródenas-Serrano, 2017). However, the implications of terrorist attacks in terms of fertility choices have been largely neglected in the existing literature, with the few relevant exceptions being Rodgers et al. (2005); Berrebi & Ostwald (2014)and Sanso-Navarro et al. (2018). With this paper, we innovate with respect to the existing literature by studying the effect of Boko Haram in Nigeria on parents’ fertility choices proxied by the number of surviving children per household. We do so by using micro-level geolocalised longitudinal data collected from 2009 to 2017 in Nigeria matched with data on terror events from the Armed Conflict and Location Event Dataset (Raleigh et al., 2010). More specifically, in this paper we test one main hypothesis: households exposed to terrorist attacks increase their quantity of children as a way to insure against future unexpected shocks. Our empirical analysis builds on Bertoni et al. (2018) and adds to this work by using not only the location of events (in a panel data fixed-effects model) but also the timing of events [in a difference-in-differences (DID) design]. We also use instrumental variables estimation techniques to deal with the possibly endogenous nature of terrorist attacks. Our empirical findings confirm our hypothesis by showing that a one standard deviation increase in the number of fatalities increases the probability that a household hit by terrorism has a newborn by 1%. This effect is robust to the use of empirical models aimed at tackling the issue of endogeneity. The remainder of this paper is organised as follows: Section 2 briefly presents the background on which this paper is built; Section 3 presents the data and methods; Section 4 presents results of the empirical analysis; and Section 5 concludes. 2. Conceptual background and hypotheses Does violence affect fertility? A number of studies have so far studied the relationship between violence and fertility showing that violence affects fertility during and after a conflict in a variety of ways (Kraehnert et al., 2019). While most studies (mainly using aggregate measures of fertility) have shown that fertility declines during violent conflicts and rebounds (or even increases) in the early postwar period (Van Bavel & Reher, 2013), other studies have found no significant effects at all (Iqbal et al., 2018). When focusing specifically on low- and Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 3 middle-income countries (LMICs), Urdal & Che (2013) have shown that armed conflicts are associated with higher overall fertility. In LMICs, which are less advanced in the demographic transition, violence and violence-related deprivations may in fact both impede and encourage procreation (Agadjanian & Prata, 2002). While violence can decrease fertility because of demographic changes (i.e., the separation of spouses, reductions in marriage) and biological effects (i.e., reduced fecundity or increased risk of spontaneous abortion, especially when famine coexisted), it can also increase it as a result of reduced access to modern contraceptives and increased sexual violence (Bendavid et al., 2021). Fertility can also increase due to the desire to replace child deaths (Kraehnert et al., 2019), or as a way to insure against future mortality shocks (Hossain et al., 2007). It is therefore not surprising that the existing literature on the subject offers often conflicting results. By focusing on Ethiopia, Lindstrom & Berhanu (1999) provide evidence of short-term declines in the probability of conception while Islam et al. (2016) in Cambodia, Khawaja (2000) in the Occupied Palestinian Territories and Shemyakina (2011) in Tajikistan show that violence increases fertility. On the contrary, some studies have shown that exposure to violence, conflict or political instability resulted in a decline in fertility in Kazakhstan (Agadjanian et al., 2008), Angola (Agadjanian & Prata, 2002), Cambodia (De Walque, 2006; Heuveline & Poch, 2007), Eritrea (Woldemicael, 2008), Ethiopia (Lindstrom & Berhanu, 1999) and Tajikistan (Clifford et al., 2010). Terrorism is a variant of political violence. As such, it is usually defined as the use of extreme violence to obtain political, religious or ideological objectives that is aimed at intimidating a large audience, both those directly touched by the event itself (Enders & Sandler, 2000; Friedland & Merari, 1985) and those affected by general violence and war. To the best of our knowledge, only a few papers to date have studied the relationship between terrorism and fertility. Rodgers et al. (2005) examined the consequences for fertility of the Oklahoma City bombing on 19 April 1995, which caused 168 deaths and injured more than 680 people. Using two different empirical methodologies (a control group interrupted time-series design and a DID design), they show that fertility in Oklahoma County increased after the attack. Berrebi & Ostwald (2014) used a longitudinal macro- level dataset comprising 170 countries from 1970 to 2007 and different empirical models, including instrumental variables regressions, and found that terrorist attacks decrease both total fertility rates and crude birth rates. Sanso-Navarro et al. (2018) used a DID approach to estimate the impact of terrorism on population growth in the municipalities of the Basque Country and Navarre autonomous communities in Spain from 1986 to 2014, and they show that terrorism had a negative and transitory effect on population growth rates in these areas. These three papers focus largely on developed countries. In this paper, we expect fertility to increase as a consequence of an insurance (or ‘hoarding’) mechanism according to which, in the aftermath of a terrorist attack parents increase the quantity of children as a way to insure against future unexpected shocks. This is the main hypothesis this paper wants to test. For a simplified rational-choice model of fertility in the aftermath of a terrorist attack, please refer to the Appendix. 1 A branch of the literature—which is only tangentially related to our work and, therefore, is not reviewed in deep in this paper—shows that fertility surges in the aftermath of a natural disaster (see, e.g., Nobles et al., 2015). Note, too, that a growing literature studies the consequences of in-utero exposure to terrorism and violence (this work is not reviewed here due to space constraints). Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 4 Valentina Rotondi and Michele Rocca Figure 1.: Total Number of Boko Haram Fatalities in Nigeria 2009–2017 (ACLED data). 2.1 Boko Haram Boko Haram, which can be translated into English as ‘Western education is forbidden’, is a terrorist group active in Nigeria since 2002. From April 2011 to June 2017, the group deployed 434 bombers to 247 different targets during 238 suicide-bombing attacks (Warner & Matfess, 2017). They have killed more than 30,000 people in North-East Nigeria and neighbouring countries and contributed to the displacement of 2.1 million people. Figure 1 shows the aggregated ACLED data for Nigeria from 2009 to 2017. The intensity of the red circles is proportional to the total number of fatalities in the area. The crosses (plotted in different shades according to the year) show the exact location of the attacks, including those that did not result in any fatalities. The turning point for Boko Haram occurred in July 2009 when its founder, Mohammed Yusuf, was killed in unclear circumstances during imprisonment by Nigerian security forces. After his death, the leadership of Abubakar Shekau began. The new leadership coincided with a rapid escalation of terrorist attacks during late summer 2010. Terrorist attacks increased even more in summer 2011, when a suicide bomber detonated at a United Nations compound killing at least 21 people. The violence escalated again in 2014 and 2015, forcing the Nigerian government to postpone presidential elections by 6 weeks to guarantee the safety of voters in the region. The three states most affected by Boko Haram’s attacks are situated in the North- East: Borno, Adamawa and Yobe. Of the residents forcibly displaced, about 84% remained within these three states; 8% moved to Northern or Central Nigeria; and 8% moved to Cameroon, Chad or Niger (Bertoni et al., 2018). To the best of our knowledge, only two papers to date have studied the micro-level effects of Boko Haram attacks in Nigeria: Bertoni et al. (2018), focusing on educational outcomes, and Nwokolo (2014), focusing on prenatal exposure to Boko Haram and birth weight. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 5 Bertoni et al. (2018) uses individual panel fixed-effects and DID models to show that Boko Haram fatalities in North-East Nigeria from 2009 to 2016 reduced school enrolment, especially for children who were no longer of mandatory school age. Nwokolo (2014) exploited variation in timing and geography of attacks in a DID design to estimate the effect of in-utero exposure to terrorism and birth weight. The results show that terrorism is negatively related to birth weight for cohorts exposed within 6 months of pregnancy. 2.2 Demography in Nigeria Nigeria’s demographic transition is stalling (Bongaarts, 2006; 2008). According to the latest estimates by the United Nations, with a population of 190 million people in 2017, an average annual population growth rate of 2.7% and a total fertility rate of 5.7 children per woman, Nigeria has the largest population in Africa (United Nations, 2017). This population is young, which is typical of countries with high fertility rates: 46% of Nigeria’s population is under the age of 15 years, whereas only 4% of the population is aged 65 years or older (NPC/Nigeria, N. P. C. and ICF International, 2014). In 2017, life expectancy at birth was 52.6 and 51.2 years, for males and females, respectively, and the infant mortality rate was 64.6 deaths per 1,000 live births. When looking at desired fertility rates, Nigerian women have about one child more than the number they want. This implies that the total fertility rate of 5.7 children per woman is around 15% higher than it would be if all unwanted births were avoided. Among currently married women, 15% use a contraceptive method, only 2 percentage points more than in 2003 (NPC/Nigeria, N. P. C. and ICF International, 2014). According to the UN World Population Prospects 2017 (United Nations, 2017), the population of Nigeria, currently the world’s 7th largest, is projected to become the world’s third largest by 2050. This projected population growth is unlikely to be economically sustainable. In fact, despite Nigeria’s Human Development Index (HDI) value increasing from 0.465 to 0.532 between 2005 and 2017, the current HDI value of 0.532 puts the country in the low human development category (United Nations Development Programme, 2018). In terms of ethnic composition, the Hausa–Fulani account for two-thirds of Nigeria’s population. As for religion, Nigeria’s population is nearly equally split between Christians, mainly living in the south, and Muslims, mainly living in the north. Ethnicity and religion play an important role in shaping fertility patterns in the country (Mberu & Reed, 2014; Mobolaji et al., 2017): fertility rates are generally higher among Hausa–Fulani–Kanuri ethnic groups and among Muslim communities. Yoruba and Igbo women tend to marry later than do Hausa–Fulani women (Odimegwu & Somefun, 2017), and thus they have a higher age at first birth. Male privilege, in terms of entitlements and inheritance rules, is prevalent across ethnicity and religions. This is the context in which Boko Haram is operating, which should be kept in mind when reading the remainder of the paper in the following section. 3. Data and methods 3.1 Data Our empirical analysis is based on the combination of four micro data sources. First, we exploit the richness of the three waves (2011, 2013 and 2016) of the Nigeria General Household Survey Panel (GHS-panel), a nationally representative survey of approximately Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 6 Valentina Rotondi and Michele Rocca 5,000 households conducted by the Nigeria National Bureau of Statistics as part of the Living Standards Measurement Study - Integrated Surveys on Agriculture (LSMS-ISA) project (World Bank, 2019). Households included in the GHS-panel are a random subsample of the overall GHS sample of households. The demographic composition of each household can be derived each wave from the household roster. Along with information on households’ assets and durables, the GHS-panel provides information on household members’ education, labour force participation, health and child development. Our estimation sample comprises 12,488 household-year observations resulting from 4,998 households at wave 1. The GHS-panel provides GPS information that allows us to link this data to our second source of data: the PRIO/Uppsala Armed Conflict and Location Event (Raleigh et al., 2010) dataset, which covers conflict events from 1997 to 2018. The third data source is the geo-referencing of ethnic groups (GREG) dataset (Weidmann et al., 2010), which contains geo-referenced information on ethnic groups around the world. The GREG is based on data and maps from the Atlas Narodov Mira, a project by Soviet ethnographers dating back to the 1960s. Similar to Alfano (2017), we allocate households from the GHS- panel to their ethnic Nigerian homelands by matching the GHS-panel latitude and longitude to the centroid latitude and longitude of the ethnicity in the GREG. The last data source is the Nigeria NMIS facility database (NMIS, 2014), from which we get a picture of the exact location of all health clinics (divided by type of service offered) that were active in Nigeria in 2014. We use this information to link these data to our main GHS-panel dataset. 3.2 Methods Our analysis includes one baseline empirical model corroborated by a number of sensitivity and robustness checks. 3.2.1 Fixed-effects panel data model First, we exploit the longitudinal nature of our dataset to implement a fixed-effects panel data regression, as shown in equation (1). Y = α + β Conflict intensity + γ ∗ X + μ + θ + γ + (1) j,l,t l,t−1 j t st j,l,t j,l,t−1 We test our hypothesis using as a dependent variable (Y ) a dummy variable with value j,l,t 1 if the household has at least one newborn (NewBorn ), that is, a baby younger than j,l,t 2 years. This variable measures the probability that a household has at least one surviving child; it is conceptually different from fertility. In the remainder of this paper, we will also present some models in which we explicitly control for child mortality. The matrix X includes a set of (time-varying) household-level controls, as depicted in h,t Table 1. μ and θ are household and time fixed effects, respectively; γ are state-specific t st linear time trends so as to account for the fact that Boko Haram’s attacks were mainly 2 We implemented these processes using the programs shp2dta and geonear in Stata. 3 The codes needed to replicate our results are available at the following link: https://osf.io/ev5z8/ 4 The Hausman test suggests that under the current speciﬁcation, individual-level effects are not adequately modelled by a random-effects model. We therefore do not show these results, but they are available upon request. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 7 Table 1: Summary Statistics Variables Mean Std. Dev. Min. Max. N Dependent variables Newborn(d) 0.351 0.477 0 1 12487 Explanatory variable Conflict intensity 0.166 2.278 0 187 12488 Covariates HH size 3.898 2.23 1 30 12487 No. of HH members able to read and write 0.499 0.318 0 1 12487 Urban 0.312 0.463 0 1 12487 Share of Islamic in the HH 0.178 0.375 0 1 12487 HH wealth index 0.019 0.642 -0.914 13.319 12458 No. of HH members occupied 0.436 0.283 0 1 12357 concentrated in the states of Borno, Adamawa and Yobe; and is the error term. Consistent j,l,t with Bertoni et al. (2018), Conflict Intensity is the household-specific evolution of h,t−1 conflict intensity over time and space. First, we define a buffer distance measure around each household (5 km). Second, we compute the total number of fatalities that occurred within each buffer zone in the period from 12 months prior the date of the interview to the latest wave, to account for pregnancy and delivery. This means that if the information on the dependent variable (NewBorn ) was collected in March 2011, the information on the j,l,t independent variable (Conflict Intensity ) is the number of fatalities that occurred in a h,t−1 5-km radius from the household’s location before March 2010: t−2 Conflict intensity = Fatalities(5Km) . (2) l,t−1 l t−1 As a robustness check, we re-run our baseline regression using as a dependent variable a dummy variable with value 1 if there has been at least one fatality within the 5-km radius and changing the buffer zone from 5 to 10 km. Results of the latter exercise are available upon request. All regressions are estimated using robust standard errors clustered at the household level. Ideally, we would use individual-level fixed-effects models to trace fertility choices of the same women over time. However, during data cleaning, it appeared too complex to follow an individual ID along the waves. We were therefore forced to conduct our analysis at the household level. Summary statistics for the variables used in the analysis are presented in Table 1. Attrition. Our sample includes 4,998 households at baseline, of which 67% were interviewed three times. Attrition might invalidate the analysis if it is systematically related to exposure to the treatment, in this case terrorism. To check how attrition bias might affect our results, we build a dummy that takes value 1 if the household is not interviewed in wave three and 5 Please notice that, from a theoretical standpoint, the extensive and the intensive margin of the attack might underlie different tactics. Given data limitation, we are unable to disentangle them properly. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 8 Valentina Rotondi and Michele Rocca zero otherwise, and we regress it on the level of conflict intensity. Results of this exercise are depicted in Table A1 in the Appendix. Reassuringly for our analysis, we find no significant correlation between exposure to terrorism and the probability of leaving the sample at wave 3, not even when conditioning on the covariates used in the empirical analysis. These results suggest it is unlikely that selection bias affected our results. However, results of the model reported in column 2 of Table A1 show that households lost due to attrition tend to be more numerous, have a higher number of employed members and have a higher share of Islamic members, relative to households that remained in the panel. The fact that households lost due to attrition seem to be wealthier than households that remained in the panel could lead to an overestimation of the true causal effect, because our measures of fertility choices are likely to be negatively correlated with wealth. 3.2.2 Difference-in-differences Our second empirical analysis is a DID model in which we exploit the timing of the change in Boko Haram’s leadership. In 2009, the Nigerian army attacked a Boko Haram base and Mohammed Yusuf, the leader who founded the movement at the turn of the century, was captured and killed (Thomson, 2012). At that time, Abubakar Shekau, the founder’s right arm, was presumed dead. Less than 1 year later, however, he appeared online and proclaimed himself the new leader of Boko Haram. Under Shekau’s leadership, Boko Haram changed not only its targets, but also its tactics and geographic scope, moving from sporadic skirmishes with the army to unexpected terrorist attacks targeting civilians, schools and churches. This happened around the summer of 2010. This change in leadership might act as an exogenous variation within our dataset. Recall t−2 that our main explanatory variable (Conflict intensity = Fatalities(5Km) ) is built l,t−1 l t−1 as the total number of fatalities that occurred within a 5-km radius from a household’s location in the period from t-12 months to the previous wave (t-2). Considering that interviews for the first wave of the GHS were held in March 2011, they were conducted in a pre-terrorism environment (see Figure 2). Therefore, we can exploit the variation in pre-post terrorism together with variation in the location of events, in a DID framework as depicted in equation (3): Y = α + β T + β P + β T ∗ P + γ X + μ + θ + , (3) j,l,t 1 j,l,t−1 2 j,l,t 3 j,l,t j,l,t j t j,l,t j,l,t−1 where T is a dummy variable that takes value one if a household (j)located in (l) is affected j,l,t by a terrorist attack and P is a dummy variable taking value 1 after the second wave. β j,l,t 3 is therefore our coefficient of interest. Parallel trend assumption. The main identifying assumption for the DID approach (the parallel trend assumption) is that, absent the treatment, treated and control households would have followed the same demographic trends. By definition, this assumption cannot be verified. One way to test the validity of this assumption would be to test whether treated and control localities were following differential trends in fertility before the change in the Boko Haram’s leadership in 2009. Unfortunately, due to data limitation, we are not able The Raleigh et al. (2010) dataset also includes information on general violence and riots. In this paper, we only consider events classiﬁed as being perpetrated by Boko Haram. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 9 March 2010 1 8 9 10 11 12 Figure 2.: Total Number of Boko Haram Fatalities in 2011 by Month (Raleigh et al., 2010). to provide a direct test of this assumption. However, when comparing total fertility rates (TFRs) of states that experienced more attacks (Borno, Adamawa and Yobe) versus states that have been largely peaceful, we find that they followed similar trends (although states that witnessed more terrorism generally started with higher fertility levels). Endogeneity. Although our specification includes state fixed effects and state-specific time trends, the DID analysis might still be biassed due to endogeneity from time-varying state- level characteristics that are not accounted for in the estimated model. We can partially test for this source of endogeneity by analysing the correlation between pre-terrorism fertility rates and conflict intensity at the state level. To this aim, we regress pre-conflict state-specific TFR obtained through the Demographic and Health Surveys (NDHS, 2009) program on the number of fatalities that occurred at the state level. Results of this exercise (shown in Table A2 in the Appendix) show no statistically significant correlations across specifications even when controlling for child mortality, the presence of on-shore petroleuem and the intensity of light sources detected during nighttime as a proxy for local development. 3.2.3 Instrumental variables Under the assumption that there are no omitted household-level time-varying variables correlated with terrorism (i.e., under the assumption that terrorism is exogeneous), the estimated β in equation 1 gives the true causal effect of conflict exposure on fertility choices. 7 We use total fertility rates obtained through the Nigeria Demographic and Health survey (NDHS, 2009). This measure refers to the 5 years preceding the survey and it is computed as the age-period fertility rate for a synthetic cohort of women. It therefore measures the average number of births a group of women would have by the time they reach the age of 50 years if they were to give birth at the current age-speciﬁc fertility rate. 8 See Figure A1 in the Appendix. Data for these variables were retrieved through the AidData’s GeoQuery tool (Goodman et al., 2019). Boko Haram fatalities (2010) 0 10 20 30 40 50 Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 10 Valentina Rotondi and Michele Rocca However, if terrorism is predetermined (terrorism responds to past terrorism shocks) or endogenous, the fixed-effects estimator is inconsistent. According to Warner & Matfess (2017), and aligning with state-level aggregated data, Boko Haram seems to have had no discernible pattern regarding date, target or nature of their bombings. This makes it unlikely that these terrorist attacks are endogenous. However, although the question of causal ordering is teased out by panel data models, these models can still be biased by omitted variables. Accordingly, the estimated β in equation (3) is a measure of the causal effect when, absent the treatment, the difference between the treatment and control groups is constant over time (the parallel trend assumption). Violation of the parallel trend assumption leads to biased estimation of the causal effect. One way to overcome these problems is to implement a model with instrumental variables. It is not easy to find an appropriate instrument, that is, a variable associated with the probability of experiencing terrorism (i.e., relevant) but not related to demographic outcomes (i.e., exogenous) except through its influence on terrorism. For this instrument, prior studies have used lagged domestic terrorism incidents in neighbouring countries (as in Berrebi & Ostwald (2014; 2016)and Enders et al. (2011)) or distance from the border (as in Rehman &Vanin (2017)). Indeed, Boko Haram’s initial suicide-bombing attacks were concentrated in Nigeria’s more remote border areas where the group appears to be holding large proportions of its resources (Foyou et al., 2018) (see Figure 1). Our first candidate instrument for terrorism is therefore distance from the border (distance from border). According to Pieri & Zenn (2016), Boko Haram’s rhetoric references two precolonial empires: the ethnic Fulani and Hausa-led caliphate of Usman Dan Fodio (1804–1903) and the ethnic Kanuri-led Kanem–Borno Empire (700–1900). Although Boko Haram seems to seek legitimacy in the former Fulani caliphate, its leaders and members are predominantly Kanuri operating in areas of the latter empire (the states of Borno and Yobe, see Figure 1). Obviously, it is easier to commit an attack in areas where you can count on more support. However, ethnic belonging per se does not reach the requirement for a good instrument. In particular, it cannot be considered exogeneous with respect to fertility choices. In fact, differences between ethnic groups might play a role in determining preference for sons (Alfano, 2017; Fayehun et al., 2011). The majority of ethnic groups in Nigeria are patrilineal and patrilocal and only male children are allowed to inherit a large part of a father’s property (Milazzo, 2014), but some minority ethnic groups (e.g., the Ijaw) have marriage and other practices associated with matrilineal systems. Furthermore, different ethnic groups are often characterised by different socio-cultural values that affect children’s health metrics (Adedini et al., 2015). Mberu & White (2011) show, for instance, that ethnicity is a key determinant of premarital sexual initiation in Nigeria, and Becker (2018) finds that the extent to which a woman’s ancestral ethnic group depended on pastoralism is positively associated with the strength of the preference for sons. It is also plausible that persistent horizontal inequality (Archibong, 2018) by ethnic group might affect households’ fertility choices. To address this concern, as in Michalopoulos & Papaioannou (2013b); Michalopoulos & Papaioannou (2013a); Cogneau & Moradi (2014)and Alfano (2017), we exploit the exogeneity coming from European powers’ arbitrary partitioning of African ethnicities into states during the ‘scramble for Africa’ at the end of the 19th century. This arbitrary division of territories into administrative borders disregarded traditional ethnic homelands (Englebert et al., 2002) and led to the partitioning of several ethnicities across newly created states. Since they cannot find support in a group strongly rooted in the territory, we believe terrorist Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 11 groups will have more difficulty finding support in partitioned territories than in more homogeneous areas. Thus, our second candidate instrument for terrorism is living in a territory where a traditional ethnic homeland has been partitioned across two or more states (partitioned ethnicity). We include distance from the market as a further instrument in our empirical analysis. Like other terrorist movements, the declared goal of Boko Haram is to shock and disrupt. To this aim, attacks are often directed towards areas with a greater concentration of residents. Our third candidate instrument for terrorism is therefore a household’s distance from the market (distance from market). Because the three instruments do not vary between one wave and another, we interact them with terrorism. 4. Results This section presents results of the empirical analysis. Table 2 depicts the household panel fixed-effects estimates for the association between Boko Haram attacks and households’ fertility choices in Nigeria using as a main measure for terrorism the total number of fatalities that occurred (conflict intensity) or using as a dependent variable a dummy variable with a value of 1 if there was at least one fatality within each buffer zone (columns 1 and 2, respectively). The positive coefficients reported in the first and third lines of Table 2 suggest that an increase in exposure to terrorism is positively related to fertility. Although the magnitude of the coefficients appears small, when computing standardised coefficients for the model depicted in column 1, we find that a one standard deviation increase in the number of fatalities (+2.4 fatalities) that occurred in the 5-km radius from a household in the previous year resulted, on average, in around a 1% increased probability of having a newborn for each household, ceteris paribus. In a context characterised by high birth rates and high incidence of poverty and malnutrition, this result is not as small as it might appear at a first glance. As an example, in relative terms, the coefficient for conflict intensity is comparable to half the size of increasing by one standard deviation the share of occupied in the HH (β =0.01 and β =-.02 c o for conflict intensity and number of household members occupied, respectively). These results (not reported due to space and time constraints) hold when accounting for ethnic-specific time trends. Results from the control variables suggest, as expected, that fertility choices are negatively related to education but positively related to being Islamic. Our measure of fertility choices is a measure of the probability that a household has at least one surviving newborn; it may be thought of as an indicator of recent fertility net of child mortality. This is an imperfect measure that tends to understate actual fertility levels. In the context of this study, this aspect could be particularly problematic if we consider the fact that women and children—especially school-age children—are often the target of Boko Haram’s attacks (UN Security Council, 2017). Our dataset does not include any information about 10 Notice, however, that since economic growth might be lower and grievances higher in partitioned areas, terror groups might ﬁnd more support instead of less support. 11 The United Nations (UN Security Council, 2017) documented 3,909 children (1,428 boys, 1,021 girls and 1,460 unknown sex) killed and 7,333 children (2,101 boys, 1,459 girls and 3,773 unknown sex) maimed during 474 conﬂict-related incidents between 2013 and 2015. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 12 Valentina Rotondi and Michele Rocca Table 2: Boko Haram and Quantity of Children: Panel Data Fixed Effect (1) (2) Newborn(d) b/se b/se Conflict intensity 0.0044*** (0.001) At least 1 death 0.0983** (0.045) HH size -0.0137** -0.0136** (0.005) (0.005) No. of HH members able to read and write -0.3337*** -0.3337*** (0.023) (0.023) Urban 0.0298 0.0325 (0.063) (0.063) Share of Islamic in the HH 0.0618** 0.0625** (0.025) (0.025) HH wealth index 0.0048 0.0040 (0.008) (0.007) No. of HH members occupied -0.0570*** -0.0571*** (0.020) (0.020) Constant 0.8839** 0.8846** (0.419) (0.419) N. 12327 Mean of dep. var. 0.3526 S.D. of dep. var 0.4778 Note: Fixed-effects panel data model. Covariates as described in Table 1. Cluster-robust standard errors reported in parentheses. *p<0.10, **p<0.05, ***p<0.01 whether a household was a victim of an attack in which a child died. However, we have information about whether the family experienced the death of a child. If child mortality is indeed a mechanism by which terrorism operates, we expect the estimated β in equation (1) to no longer be significant when controlling for child mortality in our estimated model. The coefficients depicted in columns 1 and 2 of Table 3 show this is not the case. 4.1 Heterogeneous effects and mechanisms The results depicted in Table 2 might be interpreted as evidence that Boko Haram is pushing high fertility through terrorism in Nigeria. Indeed, Boko Haram’s rhetoric deems contraception forbidden and pushes women to avoid school, marry early and have numerous children. If our results are only driven by an attempt to dictate religious preferences for larger families, we would expect to find a greater effect of terrorism in the 12 states that introduced Sharia law in 2000. 12 When estimating a model using child mortality as a dependent variable, the coefﬁcient for conflictintensity is not signiﬁcant. Results of this exercise are available upon request. Bauchi, Borno, Gombe, Jigawa, Kaduna, Kano, Katsina, Kebbi, Niger, Sokoto, Yobeand Zamfara. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 13 Table 3: Boko Haram, Quantity of Children and Child Mortality (1) (2) Newborn(d) b/se b/se Conflict intensity 0.0044*** 0.0044*** (0.001) (0.001) No. of children death -0.0320 -0.0312 (0.025) (0.025) Conflict intensity × no. of children death -0.0037 (0.014) Constant 0.8829** 0.8821** (0.419) (0.419) N. 12327 12327 Note: Fixed-effects panel data model. Covariates as described in Table 1. Cluster-robust standard errors reported in parentheses. *p<0.10, **p<0.05, ***p<0.01 Sharia regulates a wide spectrum of relationships between parents and children that might affect parents’ decisions about both the quantity of children to have and their investment in their offspring. First, Sharia law changes the expected costs and economic returns of children. While parents are obliged to maintain their children (boys and girls alike) until adulthood, children have the duty to maintain their parents in old age. This duty slightly differs between boys and girls, but only after marriage: married girls are expected to move with their husband and they cannot transfer money to their family without their spouse’s permission, whereas boys act as the main caretakers of their parents. Indeed, Alfano (2017) shows that the introduction of Sharia law in Nigeria increased fertility by 38% and the duration of breastfeeding by 19%, thereby increasing the infant survival rate. The first panel of Table 4 depicts the results of estimating equation (1) by adding an interaction term between conflict intensity and a dummy variable with a value of 1 h,t−1 for states that introduced Sharia law in 2000. The insignificant interaction term depicted in column 1 suggests it is unlikely that our results are driven only by Boko Haram attempting to dictate higher fertility through terrorism. Panels 2 and 3 of Table 4 dig deeper into the interpretation of our results by adding to the model depicted in equation (1) an interaction term between conflict intensity and h,t−1 the number of household members occupied and the number of children already living in the household, respectively. This is in line with our hypothesis and interpretation of the results as an insurance or hoarding effect. However, the negative and significant interaction term depicted in column 1 of the third panel suggests the positive association between terrorism and the quantity of children is smaller for households that already have more children. This result does not allow us to fully exclude (at least from an empirical standpoint) the hypothesis, however remote, that Boko Haram is specifically targeting smaller families with the aim of increasing fertility through terrorism. What instead we do know with reasonable certainty is that Boko Haram fighters have usually targeted women and girls with rape and other sexual violence, amounting to war crimes. Indeed, according to Amnesty International, more than 2,000 women were abducted by Boko Haram between the beginning of 2014 and spring Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 14 Valentina Rotondi and Michele Rocca Table 4: Heterogeneous Effects Newborn(d) b/se 1. Conflict intensity × Sharia states 0.0072 (0.004) 2. Conflict intensity × no. of HH members occupied 0.0037 (0.008) 3. Conflict intensity × no. of children already in HH -0.0018* (0.001) 4. Conflict intensity × female headed HH -0.0059* (0.003) Note: Fixed-effects panel data model. Covariates as described in Table 1. Cluster-robust standard errors reported in parentheses. *p<0.10, **p<0.05, ***p<0.01 2015 (Amnesty International, 2018). Furthermore, according to a book recently published by Matfess (2017), many women in Nigeria decided to voluntarily join Boko Haram in order to escape poverty. This unconditional use of sexual violence is potentially associated with our fertility measure. In the data, we have no precise information on whether women actually experienced any form of sexual violence. However, we would expect female-headed households to be potentially more exposed when compared to dual-headed households to sexual violence. In Panel 4 of Table 4 we therefore include to the model depicted in equation (1) an interaction term between conflict intensity and a dummy for female-headed h,t−1 household. The significant (but negative) interaction found in Table 4 lower panel suggests that this possible interpretation of our results does not hold. A competing explanation for the results depicted in Table 2 is that Boko Haram attacks change accessibility to health care clinics that provide antenatal and postnatal care, thereby changing the supply side of family planning programs. Unfortunately, we lack panel data on the presence of health clinics at a precise point in time. We cannot, therefore, replicate our main model by using as a dependent variable the number of health clinics active at a particular time point in a geographic unit. However, using data from the Nigeria NMIS facility database (NMIS, 2014), we can correlate the number of attacks that took place after 2014 with the number of clinics (divided by type of service offered) active in 2014 in the local geographic unit of reference. Results of this exercise are shown in Table A3 in the Appendix. The insignificant correlations found across specifications between terrorism and the presence of health clinics seem to suggest that a supply-side mechanism is not present in our data. 4.2 Addressing endogeneity In this section, we present results of the empirical models aimed at tackling the issue of endogeneity. We first report results of the DID model and then those of the instrumental variables model. Table 5 depicts results of the DID model described in Section 3.2.2. The upper panel of Table 5 reports results of a simple DID, and the lower panel reports results of a kernel propensity score matching DID whereby the weights derived from the kernel density function are used to implement a propensity score matching on the covariates by imposing common support. When combining DID with matching techniques, the aim is to difference out the Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 15 Table 5: Robustness Newborn (d) DID 0.3614*** (0.040) N. 12327 Kernel PSM DID 0.3182*** (0.039) N. 12321 IV 0.0033** (0.002) F-stat. (first stage) 40.16*** Hansen J stat. χ (p-value) 0.4675 N. 11491 Note: DID. Covariates as described in Table 1. Cluster-robust standard errors reported in parentheses. *p<0.10, **p<0.05, ***p<0.01 permanent confounders of the true causal effect (as in the DID) while capturing transitory shocks (as in matching techniques). The results of the average treatment effect on the treated are in line with the main findings in Table 2. The results of our instrumental variables approach are depicted in Table 5. Good instru- ments should be relevant and valid: that is, they should be correlated with the endogenous regressors and orthogonal to the errors. We cannot test whether our instruments are truly exogenous. The discussion in Section 3.2.3 is intended to justify our claims of exogeneity by reasoning. What we can test, however, is the strength of our selected set of instruments. The bottom panel of Table 5 reports the F-statistics for the first-stage regression and the Hansen J overidentification test. The F-statistics reported in Table 5 let us reasonably maintain that the selected instruments are not weak, and the fact that the Hansen J statistic is far from the rejection of its null (i.e., the instruments are valid, they are uncorrelated with the error term) gives us confidence that our set of selected instruments is appropriate. The results of Table 5 confirm our main finding with respect to fertility choices. 5. Conclusions In this paper, we have shown that terrorism might have a positive impact on the quantity of children that parents decide to have. More specifically, our empirical analysis of the impact of Boko Haram attacks in Nigeria shows that, ceteris paribus, a one standard deviation increase in the number of fatalities (+2.4 fatalities) that occurred in a 5-km radius from a household resulted, on average, in around a 1% increased probability of having a newborn. This result can be given causal interpretation. We interpret this result as an insurance or hoarding effect, whereby parents decide to have more children to insure against future shocks when hit by terrorism. The policy implications of our work are substantial. In a context of already high fertility and extreme poverty, the instability brought about by terrorism might exacerbate an already Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 16 Valentina Rotondi and Michele Rocca dramatic self-perpetuating poverty trap. When this young population enters working age, the ratio of the non-working-age population to the working-age population will decline. If the labour market can absorb the increased number of working-age individuals, other things being equal, per capita income will increase. Nigeria could exploit the potentials of this youth bulge to steer this young population away from the ongoing violence. However, if these young adults are unemployed, the demographic bomb created by high fertility, coupled with increased life expectancy and no jobs, will lead to further social and political instability. Note that Boko Haram recruits children in Nigeria (UN Security Council, 2017). Many of these children are abducted, but others are given up by their parents to obtain security guarantees or for economic gain. This fact might offer a competing interpretation (with respect to our explanation of children as an insurance through labour market participation) of the positive association between terrorism and fertility, but it also shows that these dramatic choices are rooted in poverty and will ultimately undermine the sustainable development goal of achieving peace, justice and strong institutions by 2030. The demographic literature distinguishes between the quantum and tempo of fertility, but in this paper, we only considered a (limited) measure of the quantum of fertility. Bongaarts & Feeney (1998) refer to the quantum as the average number of children born to women in a cohort, and the tempo as the timing of births by mothers’ age within the cohort. Terrorism could affect not only the quantum but also the tempo of fertility. Due to data limitations, this aspect was left out of this work; it offers an excellent starting point for future research. Acknowledgments This project has received funding from the European Research Council under the European Union’s Horizon 2020 Research and Innovation Programme (grant agreement no. 694262), project DisCont - Discontinuities in Household and Family Formation. The authors declare that they have no conflict of interest. Conﬂict of interest statement The authors declare that they have no conflict of interest. Supplementary material Supplementary material is available at Journal of African Economies online. References ACLED. ‘ ACLED-Armed Conflict Location and Event Data’, http://www.acleddata.com/about- acled. Raleigh, C., Linke, A., Hegre, H. and Karlsen, J. (2010) ‘Introducing ACLED-Armed Conflict Location and Event Data’, Journal of Peace Research, 47(5) 651–660. Adedini S. A., Odimegwu C., Imasiku E. N., Ononokpono D. N. (2015) ‘Ethnic differentials in under-five mortality in Nigeria’, Ethnicity & health, 20(2): 145–62. Agadjanian V., Dommaraju P., Glick J. E. (2008) ‘Reproduction in Upheaval: Ethnic-Specific Fertility Responses to Societal Turbulence in Kazakhstan’, Population Studies, 62(2): 211–33. Agadjanian V., Prata N. (2002) ‘War, Peace, and Fertility in Angola’, Demography, 39(2): 215–31. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 17 Alfano M. (2017) ‘Islamic Law and Investments in Children: Evidence from the Sharia Introduction in Nigeria’, technical report, Centre for Research and Analysis of Migration (CReAM). Amnesty International (2018) Amnesty International Report 2017/2018: The State of the World’s Human Rights. Amnesty International UK. Archibong B. (2018) ‘Historical Origins of Persistent Inequality in Nigeria’, Oxford Development Studies, 46(3): 325–47. Becker A. (2018) ‘On the Origins of Son Preference and Female Genital Cutting’, Mimeo. Becker G. S. (1960) ‘An economic analysis of fertility’, in Demographic and Economic Change in Developed Countries. New York: Columbia University Press, pp. 209–40. Becker G. S., Lewis H. G. (1973) ‘On the Interaction Between the Quantity and Quality of Children’, Journal of political Economy, 81(2, Part 2): S279–88. Becker G. S., Rubinstein Y. et al. (2004) Fear and the Response to Terrorism: An Economic Analysis. University of Chicago Mimeo, p. 93. Bendavid E., Boerma T., Akseer N., Langer A., Malembaka E. B., Okiro E. A., Wise P. H., Heft-Neal S., Black R. E., Bhutta Z. A., Black R., Blanchet K., Boerma T., Gaffey M., Langer A., Spiegel P., Waldman R., Wise P. (2021) ‘The Effects of Armed Conflict on the Health of Women and Children’, The Lancet, 397(10273): 522–532. Berrebi C., Ostwald J. (2014) ‘Terrorism and Fertility: Evidence for a Causal Influence of Terrorism on Fertility’, Oxford Economic Papers, 67(1): 63–82. Berrebi C., Ostwald J. (2016) ‘Terrorism and the Labor Force’, Journal of Conflict Resolution, 60(1): 32–60. Bertoni E., Di Maio M., Molini V., Nisticò R. (2018) ‘Education is Forbidden: The Effect of the Boko Haram Conflict on Education in North-East Nigeria’, Journal of Development Economics, 141: 102249. Bongaarts J. (2006) ‘The Causes of Stalling Fertility Transitions’, Studies in Family Planning, 37(1): 1–16. Bongaarts J. (2008) ‘Fertility Transitions in Developing Countries: Progress or Stagnation?’ Studies in Family Planning, 39(2): 105–10. Bongaarts J., Feeney G. (1998) ‘On the Quantum and Tempo of Fertility’, Population and Development Review, 271–91. Bruckner T. A., Lebreton É., Perrone N., Mortensen L. H., Blondel B. (2019) ‘Preterm Birth and Selection In Utero Among Males Following the November 2015 Paris Attacks’, International Journal of Epidemiology, 48(5): 1614–22. Camacho A. (2008) ‘Stress and Birth Weight: Evidence from Terrorist Attacks’, American Economic Review, 98(2): 511–5. Cigno A. (1993) ‘Intergenerational Transfers Without Altruism’, European Journal of Political Economy, 9(4): 505–18. Clifford D., Falkingham J., Hinde A. (2010) ‘Through Civil War, Food Crisis and Drought: Trends in Fertility and Nuptiality in Post-Soviet Tajikistan’, European Journal of Population/Revue Européenne de Démographie, 26(3): 325–50. Cogneau D., Moradi A. (2014) ‘Borders That Divide: Education and Religion in Ghana and Togo Since Colonial Times’, The Journal of Economic History, 74(3): 694–729. De Walque D. (2006) ‘The Socio-Demographic Legacy of the Khmer Rouge Period in Cambodia’, Population Studies, 60(2): 223–31. Del Carpio X. V., Loayza N. V., Wada T. (2016) ‘The Impact of Conditional Cash Transfers on the Amount and Type of Child Labor’, World Development, 80: 33–47. Di Maio M., Nisticò R. (2019) ‘The Effect of Parental Job Loss on Child School Dropout: Evidence from the Occupied Palestinian Territories’, Journal of Development Economics, 141: 102375. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 18 Valentina Rotondi and Michele Rocca Easterlin R. A. (1961) ‘The American Baby Boom in Historical Perspective’, The American Economic Review, 51(5): 869–911. Enders W., Sandler T. (2000) ‘Is Transnational Terrorism Becoming More Threatening?’ Journal of Conflict Resolution, 44(3): 307–32. Enders W., Sandler T., Gaibulloev K. (2011) ‘Domestic Versus Transnational Terrorism: Data, Decomposition, and Dynamics’, Journal of Peace Research, 48(3): 319–37. Englebert P., Tarango S., Carter M. (2002) ‘Dismemberment and Suffocation’, Comparative Political Studies, 35(10): 1093–118. Fayehun O., Omololu O., Isiugo-Abanihe U. (2011) ‘Sex of Preceding Child and Birth Spacing Among Nigerian Ethnic Groups’, African Journal of Reproductive Health, 15(2): 79–89. Foyou V. E., Ngwafu P., Santoyo M., Ortiz A. (2018) ‘The Boko Haram Insurgency and Its Impact on Border Security, Trade and Economic Collaboration Between Nigeria and Cameroon: An Exploratory Study’, African Social Science Review, 9(1): 7. Friedland N., Merari A. (1985) ‘The Psychological Impact of Terrorism: A Double-Edged Sword’, Political Psychology, 6(4): 591–604. Goodman S., BenYishay A., Lv Z., Runfola D. (2019) ‘Geoquery: Integrating HPC Systems and Public Web-Based Geospatial Data Tools’, Computers & Geosciences, 122: 103–12. Heuveline P., Poch B. (2007) ‘The Phoenix Population: Demographic Crisis and Rebound in Cambodia’, Demography, 44(2): 405–26. Hossain M. B., Phillips J. F., LeGrand T. K. (2007) ‘The Impact of Childhood Mortality on Fertility in Six Rural Thanas of Bangladesh’, Demography, 44(4): 771–84. Iqbal N., Gkiouleka A., Milner A., Montag D., Gallo V. (2018) ‘Girls Hidden Penalty: Analysis of Gender Inequality in Child Mortality with Data from 195 Countries’, BMJ Global Health, 3(5): e001028. Islam A., Ouch C., Smyth R., Wang L. C. (2016) ‘The Long-Term Effects of Civil Conflicts on Education, Earnings, and Fertility: Evidence from Cambodia’, Journal of Comparative Economics, 44(3): 800–20. Khawaja M. (2000) ‘The Recent Rise in Palestinian Fertility: Permanent or Transient?’ Population Studies, 54(3): 331–46. Kim D., Albert Kim Y.-I. (2018) ‘Mental Health Cost of Terrorism: Study of the Charlie Hebdo Attack in Paris’, Health Economics, 27(1): e1–e14. Kraehnert K., Brück T., Di Maio M., Nisticò R. (2019) ‘The Effects of Conflict on Fertility: Evidence from the Genocide in Rwanda’, Demography, 56(3): 935–68. Lambert S., Rossi P. (2016) ‘Sons as Widowhood Insurance: Evidence from Senegal’, Journal of Development Economics, 120: 113–27. Lindstrom D. P., Berhanu B. (1999) ‘The Impact of War, Famine, and Economic Decline on Marital Fertility in Ethiopia’, Demography, 36(2): 247–61. Matfess H. (2017) Women and the War on Boko Haram: Wives, Weapons, Witnesses. London: Zed Books. Mberu B. U., Reed H. E. (2014) ‘Understanding Subgroup Fertility Differentials in Nigeria’, Population Review, 53(2): 23. Mberu B. U., White M. J. (2011) ‘Internal Migration and Health: Premarital Sexual Initiation in Nigeria’, Social Science & Medicine, 72(8): 1284–93. Metcalfe R., Powdthavee N., Dolan P. (2011) ‘Destruction and Distress: Using a Quasi-Experiment to Show the Effects of the September 11 Attacks on Mental Well-Being in the United Kingdom’, The Economic Journal, 121(550): F81–F103. Michalopoulos S., Papaioannou E. (2013a) ‘National Institutions and Subnational Development in Africa’, The Quarterly Journal of Economics, 129(1): 151–213. Michalopoulos S., Papaioannou E. (2013b) ‘Pre-colonial Ethnic Institutions and Contemporary African Development’, Econometrica, 81(1): 113–52. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 19 Milazzo A.et al. (2014) ‘Son Preference, Fertility and Family Structure: Evidence from Reproduc- tive Behavior Among Nigerian Women’, technical report, The World Bank. Mobolaji J. W., Bisiriyu L., Bamiwuye S. O. (2017) ‘Contraceptive Discontinuation Among Nigerian Women: Exploring the Ethnic Variations’, IFE Research Publications in Geography, 14(1): 47–58. NDHS. (2009) ‘DHS data, https://dhsprogram.com/Publications/Publication-Search.cfm?ctry_id= 30&country=Nigeria, National Population Commission - NPC/Nigeria and ICF Macro. Nige- ria Demographic and Health Survey 2008. Abuja, Nigeria: NPC/Nigeria and ICF Macro. Neher P. A. (1971) ‘Peasants, Procreation, and Pensions’, The American Economic Review, 61(3): 380–9. NMIS (2014) ‘NMIS data’, https://energydata.info/dataset/nigeria-nmis-health-facility-data- 2014. Nigeria NMIS facility database. The Office of the Senior Special Assistant to the President on the Millennium Development Goals (OSSAP-MDGs) & Columbia University. Nobles J., Frankenberg E., Thomas D. (2015) ‘The Effects of Mortality on Fertility: Population Dynamics After a Natural Disaster’, Demography, 52(1): 15–38. NPC/Nigeria, N. P. C. and ICF International (2014) Nigeria Demographic and Health Survey, Nwokolo, A. (2014) ‘Terror and Birth Weight: Evidence from Boko Haram Attacks’ (Doctoral dissertation, PhD thesis, Berkeley). Odimegwu C., Somefun O. D. (2017) ‘Ethnicity, Gender and Risky Sexual Behaviour Among Nigerian Youth: An Alternative Explanation’, Reproductive Health, 14(1): 16. Pieri Z. P., Zenn J. (2016) ‘The Boko Haram Paradox: Ethnicity, Religion, and Historical Memory in Pursuit of a Caliphate’, African Security, 9(1): 66–88. Pörtner C. C. (2001) ‘Children as Insurance’, Journal of Population Economics, 14(1): 119–36. Quintana-Domeque C., Ródenas-Serrano P. (2017) ‘The Hidden Costs of Terrorism: The Effects on Health at Birth’, Journal of Health Economics, 56: 47–60. Rehman F. U., Vanin P. (2017) ‘Terrorism Risk and Democratic Preferences in Pakistan’, Journal of Development Economics, 124: 95–106. Rodgers J. L., John C. A. S., Coleman R. (2005) ‘Did Fertility Go Up After the Oklahoma City Bombing? An Analysis of Births in Metropolitan Counties in Oklahoma, 1990–1999’, Demography, 42(4): 675–92. Romanov D., Zussman A., Zussman N. (2012) ‘Does Terrorism Demoralize? Evidence from Israel’ Economica, 79(313): 183–98. Sanso-Navarro M., Sanz-Gracia F., Vera-Cabello M. (2018) ‘The Demographic Impact of Terror- ism: Evidence from Municipalities in the Basque Country and Navarre’, Regional Studies, 53(6): 1–11. Shemyakina O. (2011) ‘The Effect of Armed Conflict on Accumulation of Schooling: Results from Tajikistan’, Journal of Development Economics, 95(2): 186–200. Thomson V. (2012) ‘Boko Haram and Islamic Fundamentalism in Nigeria’, Global Security Studies, 3(3): 46–60. UN Security Council (2017) ‘Report of the Secretary-General on Children and Armed Conflict in Nigeria (s/2017/304)’, http://undocs.org/S/2017/304. United Nations (2017) ‘World Population Prospects: The 2017 Revision, Key Findings and Advance Tables’, Department of Economic and Social Affairs, Population Division, Working Paper No. ESA/P/WP/248. United Nations Development Programme (2018) ‘Human Development Indices and Indicators’, in 2018 Statistical Update. United Nations Development Programme. Urdal H., Che C. P. (2013) ‘War and Gender Inequalities in Health: The Impact of Armed Conflict on Fertility and Maternal Mortality’, International Interactions, 39(4): 489–510. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 20 Valentina Rotondi and Michele Rocca Van Bavel J., Reher D. S. (2013) ‘The Baby Boom and Its Causes: What We Know and What We Need to Know’, Population and Development Review, 39(2): 257–88. Warner J., Matfess H. (2017) ‘Exploding Stereotypes: The Unexpected Operational and Demo- graphic Characteristics of Boko Haram’s Suicide Bombers’, technical report, Combating Terrorism Center at West Point: United States Military Academy. Weidmann N. B., Rød J. K., Cederman L.-E. (2010) ‘Representing Ethnic Groups in Space: A New Dataset’, Journal of Peace Research, 47(4): 491–9. Werding M. (2014) ‘Children Are Costly, But Raising Them May Pay’, Demographic Research, 30: 253. Woldemicael G. (2008) ‘Recent Fertility Decline in Eritrea’, Demographic Research, 18(2): 27–58. World Bank (2019) ‘Nigeria National Bureau of Statistics. General Household Sur- vey Panel (GHS-panel) 2015–2016, http://microdata.worldbank.org/index.php/catalog/2734/ study-description#page=export-metadata&tab=study-desc. Appendix A model Children in LMIC can be seen as an investment good (which indirectly increases parents’ utility by increasing their life-time income or life-time consumption [see, e.g., Neher (1971); Cigno (1993); Pörtner (2001)]) rather than a consumption good (which directly increases parents’ utility [see Becker (1960)]). In fact, when institutions such as social security are missing and markets (credit, insurance) are imperfect, parents have an expected return from their children – in the form of child labor or the provision of financial support in older age – to such an extent that having an additional child can be seen as securing a minimum level of income (Agadjanian & Prata, 2002, Lambert & Rossi, 2016). When parents decide their optimal number of children, they weigh benefits and costs. In this model, parents live in two periods: present, p, and future, f. For simplicity, let us assume that the benefits of raising children include the expected income from child labor at time future (W), and the costs include the opportunity cost of the mother’s time and the costs associated with raising each child (i) at time present. Parents are subject to a budget constraint that depends on the present output (Y ), family endowments acquired at time present (H), the number of children (N), and the cost (investment) per child (i). Assuming For a non-technical introduction to analyses of fertility based on a rational-choice paradigm, see Werding (2014). 15 See the Appendix for a simple model. 16 We assume that parents can (wish to) control the number of births, which is not always the case in developing countries, where traditions and cultural heritage often play a decisive role in adoption of contraceptive methods. We also assume that the cost of controlling fertility is negligible. For a description of contraceptive prevalence and desired fertility among Nigerian women, see section 2.2. For simplicity, we exclude any emotional component of raising children and any psychological cost from sending children to work. We are aware that this is a simpliﬁcation of the true value of children. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 21 that there can be no inter-period transfer of income except through children, this means: Y = C + iN + H (A.1) p p Where C is the part of the current output (Y ) that is consumed at time present (p). p p Consumption at time future (C )isgiven by: C = WN + Y (A.2) f f Where Y is the household’s future output. As in Becker & Lewis (1973), from (A.1) we know that, for a given income constraint and a given level of present consumption (C ), the costs associated with raising children (i) imply a trade-off between having fewer high-quality, high-cost children (characterized by higher earning potential in the future) and having more low-quality, low-cost children (characterized by lower earning potential in the future). Households have two possible strategies: 1. Increase the number of children (N) and reduce the investment in each child (i) 2. Keep constant both the number of children (N) and the investment in each child (i) Which strategy will prevail is an empirical question, and it crucially depends on the relationship between the expected income from child labor and parental investment, i.e., W and i, respectively. We can assume that, similar to Del Carpio et al. (2016), a child’s expected wage in labor market W is a (positive) function of (1) family endowments (H), such as productive assets in the household (e.g., tools, land, capital), and (2) parental investment in each child at time present (i), such that W(i, H) = Hw ,where w is the salary the child gains in the labor market and it is assumed to be exogenously given. Solving (A.1)for N and substituting into (A.2) we obtain: Y − C + H p p C = Hw + Y (A.3) f f Parents wish to pick i and N so as to maximize their chosen combination of present and future consumption. Therefore: ∂C Hw Y − H − C ln (w) i − 1 f p p = = 0 (A.4) ∂i i ∗ 1 By solving equation (A.4)for i we obtain: i = . Assuming w is exogenous, substi- ln w ( ) ∗ ∗ tuting i into equation (A.1) and solving for N we obtain: N = ln (w) +Y − C − H . p p ∂N Therefore, a reduction in H due to a terrorist attack increases N ( < 0). ∂H We rule out the possibility of reducing the number of children as ethically implausible. Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 22 Valentina Rotondi and Michele Rocca Table A1: Attrition (1) (2) Household interviewed in Household interviewed in waves 1 and 2 waves 1 and 2 b/se b/se Conflict Intensity 0.0002 0.0007 (0.001) (0.001) HH size -0.0023* (0.001) # of HH members who know to read -0.0006 and write (0.010) Urban 0.0059 (0.007) Share of Islamic in the HH 0.0258*** (0.006) HH wealth index -0.0018 (0.003) # of HH members occupied 0.0351*** (0.010) Observations 12488 12327 Note: Ols. Covariates as described in Table 1. Standard errors robust to heteroskedasticity reported in parentheses. * p<0.10, ** p<0.05, *** p<0.01 Table A2: Pre-terrorism characteristics and conﬂict intensity at the state level (1) (2) (3) (4) (5) Total Boko Total Boko Total Boko Total Boko Total Boko Haram Haram Haram Haram Haram Fatalities Fatalities Fatalities Fatalities Fatalities b/se b/se b/se b/se b/se TFR (2008) 0.036 0.217 (0.073) (0.164) Child mortality (2000) 0.185 0.218 (0.160) (0.164) On shore petroleum -0.095 -0.129 (0.074) (0.091) Nightlights (2012) -0.011 -0.106 (0.023) (0.082) Constant 0.000 0.000 0.000 0.000 0.000 (0.166) (0.161) (0.161) (0.163) (0.164) R-squared 0.062 0.047 0.047 0.017 0.011 N 38.000 38.000 38.000 38.000 38.000 F 0.490 1.749 1.755 2.016 1.676 OLS. Robust standard errors in parentheses. Standardized coefficients. * p<0.10, ** p<0.05, *** p<0.01 Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 Terrorism and Fertility Choices in Nigeria 23 Table A3: Access to health care clinics and conﬂict intensity (1) (2) (3) (4) Total Boko Haram Fatalities in 2014 b/se b/se b/se b/se # of clinics providing: Maternal delivery 9.750 (11.372) # of clinics providing: Child vaccination -5.258 (10.356) # of clinics providing: Family planning -0.590 (9.371) # of clinics providing: Antenatal care -1.252 (11.470) R-squared 0.747 0.747 0.747 0.747 N 34068 34068 34068 34068 OLS. Robust standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01 Downloaded from https://academic.oup.com/jae/advance-article/doi/10.1093/jae/ejab030/6479813 by DeepDyve user on 04 January 2022 24 Valentina Rotondi and Michele Rocca Abia Abuja Adamawa Akwa Ibom Anambra Bauchi Bayelsa Benue Borno Cross River Delta Ebonyi Edo Ekiti Enugu Gombe Imo Jigawa Kaduna Kano Katsina Kebbi Kogi Kwara Lagos Nasarawa Niger Ogun Ondo Osun Oyo Plateau Rivers Sokoto Taraba 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 Yobe Zamfara 2008 2010 2012 2014 2008 2010 2012 2014 Figure A1: Total Fertility Rates from Nigeria’s States (2008-2013) Note: Total fertility rates are computed from the DHS for the 5 years preceding the survey. TFR
Journal of African Economies – Oxford University Press
Published: Dec 27, 2021
Access the full text.
Sign up today, get DeepDyve free for 14 days.