Women as decision-makers within households: Does religiosity matter? Evidence from Morocco 1 Imane Chaara2 University of Oxford (ODID) May 2015 (Work in progress) Abstract: The participation of mothers in decisions within their household is recognized as having non-neutral effects and in many instances positive effects, especially in what concerns children health and education. In this paper, we focus on the participation of women in decisions concerning their children’s education and we investigate the relationship that may exist between the religiosity of the mothers and their involvement in education decisions. By analyzing data we collected in Morocco in 2008, we find a positive and significant correlation between the intensity of religious practice of the mothers and their participation in decisions concerning their daughters’ education. This result is essentially true for poorly- or non-educated women, which suggests that religion acts as a compensating factor for the lack of education. Moreover, a positive and significant correlation between the intensity of the religious practice of the mothers and their participation in decisions concerning their sons’ education exists only in mixed families (families composed of both sons and daughters). This suggests the existence of a spillover effect in families composed of both daughters and sons. We argue that, in the Moroccan context, the existence of religious movements that are socially influential may drive our main results. This assertion is based on an analysis of the social action and the discourse of these movements about the role of women in the private sphere and society and about the importance of education. Finally, we do not find a clear significant correlation between the intensity of the religious practice of the fathers and their wives’ participation in decisions concerning their children’s education. Keywords: Household decisions, education, religious movements, Islam. JEL codes: D10, Z1, Z12 1. Introduction 1 2 The author thanks the Agence française de Développement (AfD) for financial support, Jean-Philippe Platteau and participants at Contact information: imane.chaara@qeh.ox.ac.uk 1 Raising school enrolment is poor countries is a universally praised policy. The World Bank has repeatedly recognized in its World Development Report the importance of primary schooling as an input to the social and economic progress in poor countries. The education of girls has received special attention due to the important positive externalities of female education on children and adult health, fertility, and infant mortality. Raising school enrolment may depend on the relative importance of school supply and household demand factors. In the case of Mozambique, for instance, Handa (2002) shows that building more school or raising adult literacy will have a larger impact on primary school enrolment than interventions that raise household income. Huisman and Smits (2009) looked at the effect of household and district level factors on school enrolment in 30 developing countries. Beside the effect of socio-economic factors, demographic characteristics of the family and characteristics of educational facilities at the distric level, they find that empowerment of women is an important determinant of primary school enrolment. Moreover, the strong positive effect of mother education indicates that mothers with more knowledge are in a better position to get their children into school. Huisman and Smits (2009) argue that since most of the mothers in poor countries are illiterate, a major way to make them aware of the benefit of education is through the media. The aim is to publicly disscuss the importance of education and pass information that widens their horizon. Investment in children’s education is a decision taken at the household level. Households are complex structures and a good understanding of the way decisions are taken by household members is crucial in order to understand household outcomes. A natural question is therefore : who takes the decisions ? Given the aforementioned observations, the role played by mothers in the decision-making process is of particular importance. Using data that we collected in three different regions of Morocco in 2008, we observe that when mothers participate in decisions concerning the education of their daughters, the later are less likely to drop out of school after primary education in urban areas and they are more likely to enroll in primary school in rural areas. In this paper, we focus on the determinants of mothers’ participation in decisions concerning their children’s education and discuss the role of a non-conventional factor, which is the religiosity of mothers. The participation of women in household decisions is generally presented as an indicator of their bargaining power within the household (Hashemi and Schuler, 1993; Kabeer, 1999a and 1999b; Malhotra et al, 2002)3. When it comes to determining the bargaining power, the emphasis is often put on the control over tangible resources. However, as pointed out by psychologists (see Foa and Foa, 1980), in addition to tangible resources such as money, education, and occupation, intangible resources such as intelligence, physical attractiveness or love may also have an impact on the bargaining power of family members. In fact, they argue that any trait of behavior can lead to more influence and power if it is valued by members of the family. For instance, Jonhson (2000) explains that among the Fulani tribes 3 Nevertheless, as pointed out by Kabeer (1999a), only a few cultures operate with starkly dichotomized distributions of power, with men making all the decisions and women making none (see also Quisumbing, 2003: chap. 4). Usually, a hierarchy of decision-making responsibilities exists, certain areas being reserved to men, others to women. For instance, in South Asia, it appears that women are mainly in charge of decisions concerning food purchases, children’s health and men are in charge of children’s education, children’s marriage and market transactions in major assets (Sathar and Kazi, 1997; Cleland et al., 1994; Morgan and Niraula, 1995; Hashemi et al., 1996 as reported by Kabeer, 1999a). Furthermore, Quisumbing (2003) does not consider decision-making as a measure of power but she believes that the relative power of spouses is an underlying factor likely to have effects on decision-making. 2 of Western Africa, who primarily are Muslims, family members, especially women, can increase their power in the family by practicing traditional Fulani customs of conjuring the spirits of dead ancestors. Furthermore, as cultural anthropologists emphasize, there are forms of valuation other than economic such as social or religious valuations. Based on the aforementioned literature, we suspect that religious faith may impact on the relative power of each household member. If this is true, women may strategically use religion in order to gain more control, play an important role in their family and access public life. Others have discussed similar assumptions. Platteau (forthcoming: 234) explains that educated and urbanized women may decide to wear the Islamic veil in order to escape traditional norms and the control of their family, and in turn gain access to public life. This understanding is similar though different to Carvalho’s theory about veiling. Carvalho (2013) considers veiling as a commitment mechanism that reduces temptation to deviate from religious norms of behavior. Veiling is then presented as a strategy to seize outside economic opportunities and preserve reputation within the religious community. In the Moroccan context, beyond the strategic use of religion by mothers at the household level, we argue that socially influential religious movements are important actors. In fact, religious movements may represent the underlying factor that influences both the religious practice of people, and the decision process within the household. In Morocco, the two main religious movements are actually influential in a large spectrum of the population and quite popular in the low and lower-middle classes thanks to the supply of social services. Besides their assistance activities, they transmit messages about their vision of education and the allocation of roles within the family. Understanding their action and discourse may therefore be useful to grasp how religion may play a role in the division of tasks between spouses within the household, in particular for matters concerning children’s education. We suggest that religious movements may represent an alternative way to make mothers aware of the benefit of education, on a one hand, and to raise the perceived value of women, describing them as important actors in the private sphere and society, on the other hand. In this paper, by using data we collected in Morocco in 2008, we look at the relationship between the religiosity of the parents, especially the religiosity of the mothers, and the likelihood of mother’s participation in the final decisions concerning their children’s education. We treat separately the decisions concerning daughters and those concerning sons. We use a measure of the intensity of religious practice as a proxy of parents’ faith and religious devotion. We find a positive and significant correlation between the intensity of the religious practice of the mothers and their participation in decisions concerning their daughters’ education. This result is essentially true for poorly or non-educated women, which suggests that religion acts as a factor that compensates for the lack of education. A positive and significant correlation between the intensity of the religious practice of the mothers and their participation in decisions concerning their sons’ education exists only in mixed families (families composed of sons and daughters). It therefore appears that religious faith motivates mothers to participate in decisions concerning their daughters’ education mainly, but there is a spillover effect that makes them more concerned with the education of their sons as well. Finally, we do not find a clear significant correlation between the intensity of the religious practice of the fathers and their wives’ participation in decisions concerning their children’s education. 3 The rest of the paper is organized as follows. The next section provides a brief literature review. Section 3 gives a description of the main religious movements present in Morocco. Section 4 describes the data collection methodology. Section 5 formulates our main hypothesis and presents the methodological approach. Section 6 describes the key variables and presents some descriptive statistics. Section 7 discusses the results and Section 8 concludes. 2. Literature review This section reviews briefly how the household economics literature conceives decision-making processes that take place in households and presents empirical evidence, especially about the effect of women’s involvement. The second part of this section discusses recent contributions in the literature on the link between religion, human capital accumulation and other socio-economic outcomes. How to think about decision-making processes in households? There are two broad categories of intra-household decision models. Unitary models assume that resources are pooled together in the household and the decisions are such that social welfare in the family is maximized. The outcomes depend on the total income, the prices and information available to the household. There are different versions of this model. The unitary model does not impose a particular set of preferences on decision makers, it simply assumes that they act as one and the aggregation mechanism generating this unity ranges from consensus (Samuelson, 1956), weighted summations (Sen, 1966), and existence of an altruistic patriarch that takes into account the preferences of other members without giving them a voice in the decision process (Becker’s Rotten kid theorem, 1981). It may be that parents will invest in children who will offer the highest return to education (Becker and Thomes, 1976), or they may prefer equality, the first born, or a specific gender (Behrman, Pollak, and Taubman, 1982). Collective models (the cooperative and the non-cooperative collective models) assume several decision units4. Household members may have different preferences and they may find a way to challenge and negotiate the patriarch authority. The non-cooperative models assume that there are no binding agreements between household members and decisions are not necessarily Pareto efficient. The cooperative model assume that members acknowledge that cooperation generate benefits but disagree on how to share this benefit. A bargaining process takes place in order to decide how to share resources. Cooperative models assume that a stable decision process exists but they do not specify what the decision process is. The models also assume that all decisions are Pareto efficient. In the cooperative bargaining model, the outcome of the negotiation process depends on the relative bargaining power of each spouse. The respective bargaining power of the wife and husband depends on their respective fallback position. The later depends on economic factors (the relative wage of each spouse, the relative share of spouses in non-earned income, opportunities on the labour market…) and non-economic factors (laws concerning divorce, custody law, property rights, family support, social norms…). Sen adds that the result of a bargaining process may also depend on how each member is perceived to contribute to the household resources. So, perceptions and not only actual facts may impact the negotiation process and the subsequent decisions. 4 For a discussion about these models, see for instance Pollak (1994) and Lundberg and Pollak (1996). 4 The empirical literature on intra-household allocation shows that the unitary model is debatable and that households do not constitute a unified economy (see Bruce, 1989 for a discussion; Quisumbing et al., 1999; Quisumbing et al, 2003). In fact, several studies found that earned and unearned income received by the husband or wife significantly affect demand patterns when total income or expenditure is held constant (see, for instance, Lundberg, Pollak, and Wales, 1997; Duflo, 2003; Quian, 2008). The participation of women in the decision-making process within the household generally leads to nonneutral effects and might have, in many instances, positive impacts (see Duflo, 2012 for a review of the literature). The participation of women is actually not only perceived as a step toward more equity between spouses but also toward more efficiency for several household outcomes. In the literature, correlations have been pointed out between women’s participation in decisions and income management, on the one hand, and different outcomes in terms of children’s education, children’s health and household’s wellbeing, on the other hand. For instance, the World Bank, in its report `Engendering development´ (2001), indicates that several studies conducted in Bangladesh, Brazil and the Ivory Coast show that the management of supplementary incomes by women leads to higher positive effects on household wellbeing than in the case where they are managed only by their husbands (see also Thomas, 1990; Hoddinott and Haddad, 1991). Co-responsibility appears to lead to welfare improvements. Furthermore, some studies point out the fact that heterogeneous preferences exist within households which translate in different priorities depending on the solution of the bargain. Hoddinott and Haddad (1995) show that in the Ivory Coast, men spend relatively more than women on goods such as alcohol and cigarettes and less on goods that benefit children or the household as a whole. Moreover, Thomas (1990), in a study based on survey data on family health and nutrition in Brazil, found some evidence for gender preference: mothers prefer to devote resources to improving the nutritional status of their daughters while fathers give the priority to their sons. Furthermore, Duflo (2003) looked at the effect of the expansion of a social pension program to the black population in South Africa after the end of the apartheid. She found that pensions received by grandmother translate in better nutrition for girls; the effect on boys is lower. Pensions received by grandfather do not translate in better nutritional status for grandchildren whatever the gender. In the Chinese context, Qian (2008) looked at the impact of the exogenous increase in the price of tea after the introduction of Post-Maoist reforms at the beginning of the 80’s. Agricultural incomes are gender-specific and an increase in the price of tea represents an increase in the relative value of female labour. Qian found that an increase in female income, holding male income constant, improves survival rates for girls and education of children. An increase in total income without changing the relative share of each spouse’s income has no effect on survival rates. In specific social contexts, conventional modes of behavior may suggest to the players - the spouses - a focal point equilibrium thus reducing or eliminating the need for negotiations (i.e. equilibrium without explicit bargaining). In the case of marriage, social conventions regarding the rights and responsibilities of husbands and wives may indeed suggest to the spouses a particular equilibrium. England and Kilbourne (1990) argue that women are socialized to be less willing than men to drive hard bargains with their spouses and, hence, that wives get less than they otherwise would. Sen carries the internalization argument a step further, arguing that ‘socialization’ may prevent a woman from recognizing her true interests5. In the same vein, Pollack (1994) argues that economic models of distribution between men and women focus on the subgame of bargaining within a particular marriage and that the real action is elsewhere, in the prior game that determines social norms and gender roles. 5 Those last two arguments are reported in Lundberg and Pollak (1997: 26) 5 As a consequence, an external intervention that may modify women perceptions about gender roles and incitivise them to redefine their own interests and act in accordance with those interests may also influence, ceteris paribus, decision processes within household and in turn household outcomes. Religion and Education: a few recent contributions A recent literature has rigorously investigated whether religion has an impact on education, socioeconomic attitudes, and in turn on development. Becker and Woessmann (2009) provide an alternative theory to Max Weber’s Protestant work ethic. By using county-level data from the 19th century Prussia, they find that Protestantism leads indeed to higher economic prosperity but also to better education. They argue that Protestant economies prospered because instruction in reading the Bible generated the human capital crucial to economic prosperity. Botticini and Eckstein (2007) consider that a religious and educational reform in Judaism that took place in the second century AD could explain a massive occupational shift among Jews. From the end of the second century AD, Judaism enforced a religious norm requiring any Jewish father to educate his children. Botticini and Eckstein present evidence that this exogenous change in the religious and social norm had important impact on the Jewish demographic and economic history. First, the cost of educating children in subsistence farming economies led to an occupational shift from agriculture to urban occupations (2nd to 7th centuries). Second, education gave a comparative advantage to Jews people to enter skilled urban occupations during the vast urbanization in the newly developed Muslim Empire (7th and 8th centuries). In the case of Islam and in a more recent period, Myersson (2014) investigates the social effects of Islamic political control by focusing on the case of Turkey. He questions whether Islamic political control may lead to poorer women living standards. The rational behind this assumption is the belief that Islamic parties tend to represent poor and conservative constituencies with correspondingly low support for women rights. By using data on 2600 Turkish municipalities, Myersson looks at the causal effect of Islamic political control on education. He finds that Islamic rule had positive impact on female education, and among others on female secular high school completion. He explains this success by the implementation of pragmatic policies that were successful in overcoming female education barriers (for instance, headscarf ban, mixed classes, and a strongly secular curriculum could have exacerbated existing socio-economic constraints). His results suggest that under specific circumstances, socially conservative politicians can have socially progressive effects. In other terms, he argues that controversial but popular movements may have development-related effects difficult for secular parties to replicate. The Islamic rule also seems to have long-term positive effects: lower female adolescent marriage rate and higher female political participation. Binzel and Carvalho (2013) approach the relation between education and religion from a different perspective. They argue that the typical Islamist activist is an educated youth. According to them, an unexpected decline in social mobility among educated youth and inequalities can lead to a religious revival, led by the educated middle class and not the poor and illiterate. Religion is then presented as a coping mechanism for unfulfilled aspirations. More importantly, they argue that religious participation by educated youth builds organizational capacity that can lead to a popular and long-lasting religious movement that can affect less educated segment of the population as well. 6 3. Religious movements in Morocco According to Clifford Geertz, mass fundamentalism was held in check in Morocco since there are no « mass party, public organization or famous personalities that openly defend a hard line » (Geertz, 1968: 7; the introduction was written in 1991). In the same vein, Willis (2007) points out the fact that the absence of a large Islamist movement has long distinguished Morocco from other Arab states. However, Willis explains that during the last three decades, the socio-political landscape of Morocco has changed. In fact, two main religious movements grew in importance during that period. The first one is a political party, the Justice and Development Party (PJD), associated with the sociallyconservative organization Harakat al-Islah wa at-Tawhid (Movement for Reform and Unity). The second is an organization, the Justice and Charity movement (Jamaa Al Adl wal Ihsan), which is illegal but tolerated by the authorities.6 The purpose of this section is not to provide an exhaustive description of the trajectory followed by these two movements since their creation but to give some details about their main philosophy, their structure, their strategic approach and their social involvement. All these elements will contribute to the understanding of their potential social influence. Both movements are non-violent and their origins go back to the late 60’s and early 70’s. They never formed a coalition because of their differences. On the one hand, they differ in terms of ideology: the PJD has taken a much more conservative line on social issues such as the place of women in society (see Willis, 2007). On the other hand, they follow opposite strategies: the PJD accepted the co-optation system and decided to participate in the official political life7. On the contrary, Al Adl wal Ihsan always stayed out of the political system and decided to act as a critic of the regime by directly challenging the monarchy. Al Adl wal Ihsan Al Adl wal Ihsan was founded by a school inspector, Abdessalam Yassine who was a member of a sufi brotherhood named Boutchichiyya8. Yassine believed that the social, political and economic problems of the kingdom were due to the insufficient attention paid by the state to Islam (Willis, 2007: 151). He asserts that change has to come through social activism. He always criticized the constitution, the concentration of political powers and the ‘sacred’ status of the Monarch. Since the 80’s (the movement exists as an organization since mid-80’s), the main concern of the movement has been social justice. It became more engaged in various social services (schooling, medical care, sanitation…) while at the same time, ensuring religious and spiritual guidance to their 6 Aside from these two main movements, some marginal salafi groups exist since Moroccan independence. According to Pruzan-Jorgensen (2010), salafi movements in Morocco can be divided into three schools of thought mutually interconnected: wataniyya (national) salafism which inspired key figures of the independence movement, Wahhabiya salafism which was officially allowed by King Hassan II as a means to counter the two main religious movements, and jihaddiyya salafim which is a violent movement whose members were accused of acts of terrorism. These groups attract only a few adherents and do not play a dominant social and political role in the Moroccan society. 7 The co-optation consists of an official recognition of the movement by the authorities on the condition that the movement explicitly recognizes the authority and legitimacy of the monarchy. 8 The Boutchichiyya brotherhood grew in importance under the reign of King Mohamed VI, after the Casablanca bombings of 2003 (Bekkaoui and Larémont, 2011). The actual minister of Islamic affairs is a member of the Boutchichiyya brotherhood. The brotherhood re-established the order’s link to scriptural Islam and its followers are mainly found among educated youth (Bekkaoui and Larémont, 2011: 35-36). 7 beneficiaries (Pruzan-Jorgensen, 2010: 15). The organization acts by supplying social services as a substitute to the state and at the same time, it implements its ‘educational’ plan. The movement is constructed around the ideas and writings of a single man, its founder (the Sheikh), and it is based upon Sufi principles9. As a result, a large amount of emphasis is placed on the inner-transformation of the individual (Willis, 2007: 166). The priority must be the inner change of the people in order to change the society. We qualify this approach as a bottom-up approach. The movement stresses the importance of education and the transformation of society along Islamic lines at the individual level prior to the establishment of an ideal political system (Willis, 2007: 168). The foundation of the organization’s work is the dawa: each member makes an effort to become a better Muslim and attempts through example to turn others into better Muslims, so that individual behavior can slowly affect the society as a whole (Cavatorta, 2006). In other terms, dawa implies missionary activities in as many areas as possible of social life. At present, Al Adl wal Ihsan is one of the biggest and well-organized movements in Morocco. They have developed a sophisticated organization with a pyramidal structure and branches throughout the country (Willis, 2007: 164). The main executive bodies of the organization are the Majlis al Shura (Consultative Council) and the Majlis Al-Irchad (Guiding Council) which is the executive body of the organization. The other main bodies are two educational organizations, one for men and one for women, and the `political circle´ which is structured as a political party (it notably includes students’ and women’s sections). However, the democratic nature of the internal organization is disputed. As a matter of fact the Sheikh seemed to make all important decisions without any regard for what other members think (Cavatorta, 2006). Sheikh Yassine died in December 2012. The position of Supreme Guide previously held by Yassine has now been changed to a general secretariat with Fathallah Arsalane and Muhammed Abbadi heading the political and spiritual factions, respectively. The organization benefits from a large support of the population but the number of its members is unknown. Moreover, since the 80’s, it has gained influence among university students and students’ unions. In fact, the association’s recruitment activities are particularly concentrated in the educational sector, targeting teachers and students (Bekkaoui and Larémont, 2011: 45). Furthermore, militants hold discussion groups in private homes and proselytize among members of their social networks (Cavatorta, 2007). Pruzan-Jorgensen (2010) indicates that aside from young teachers, it also attracts and influences doctors, lawyers and engineers. More recently, the association has been gaining adherents among peasants (Bekkaoui and Larémont, 2011: 45). According to Shahin (quoted in Bekkaoui and Larémont, 2011: 44), « the Sufi influence of Yassine and the moral and spiritual program which he devised for his followers make Al Adl wal Ihsane attractive to a wider following, especially middle and lower classes, civil servants, peasants, and workers, who are the usual recruits of Sufi orders ». The Women section of Al Adl wal Ihsan Women were active in the movement since the 80’s but the Women section of Al-Adl wal Ihsan has officially been created in 1998 by the daughter of Sheikh Yassine, Nadia Yassine. The creation of such a section has been initially met with fierce opposition by then second-in-command Mohammed Bashiri (See Gray 2014:99-100). Internal disagreements between Bashiri and the Sheikh about Nadia Yassine’s initiative and other matters led to the departure of Bashiri. This opened the way for Nadia Yassine to finally establish the Women section. The members of this section accept the rejection of polygamy in 9 However, mainstream Sufi figures have argued that Yassine’s knowledge of Sufism is superficial (Willis, 2007: 166). 8 certain cases10, and favor the model of the nuclear family including the belief that woman can have an activity outside the house. They are militants for a feminist interpretation of the religious texts. They support the idea of the emancipation of women through education. Education of women is central and represents the basis of their model. They also support the egalitarian involvement of men and women in family matters. However, they do not defend a model of clear equal rights and responsibilities between men and women. They use the notion of ‘complementarity’ in the roles of men and women instead of full equality though this concept is quite elusive (Ramírez, 2006: 115-116). Gender complementarity emphasizes women’s role as wives and mothers. However, the female leaders of the movement insist on the importance of education for women and their right to pursue a professional career (Gray 2013: 117-18). They also prefer to refer to the concept of gender justice instead of gender equality. But this appears mainly as a terminology issue while the improvement of women’s rights is a common objective. Naciri (1998) explains that some women are receptive to the approach of Muslim movements because it values their social roles within the family. She argues that their behavior may be explained by the fact that Muslim movements have never envisioned gender division of roles in terms of hierarchy, but as noble tasks, which are the responsibility of the Muslim women. Moreover, Meriem Yafout, a former member of the Jamaa who has written a thesis on women in Islamist movements states that those movements offer greater prospect for self-realization and advancement than many secular associations that are characterized by rigid structure of power (Gray, Oxford Islamic Studies online). Party of Justice and Development The PJD exists as a political party since the mid-nineties. Many of its leaders were involved in the organization Shabiba al-Islamiyya (Islamic Youth) which was established in the late 60’s and dissolved in 1975. The organization was led by a school inspector, Abdelkrim Mouti who was disillusioned by the secular drift and ineffectiveness of the political process in Morocco (Willis, 2007: 151). In the 80’s, the former members of Shabiba al-Islamiyya created a number of new organizations. A number of them merged in the 90’s to create the actual Harakat al-Islah wa at-Tawhid (Movement for Reform and Unity). The PJD is, in a sense, the branch of the initial movement which is politicized. The members of the PJD accept the rules of the political game and avoid any criticism about the place of the monarchy in the political system. They focus their action on social and moral issues such as corruption, education, and the place of women in society (Willis, 2007: 159). According to Pruzan-Jorgensen (2010), the PJD is a conservative movement inspired both by the Muslim Brotherhood and by Wahhabi salafism. Yet, it appears flexible and rather pragmatic. On the contrary, the Movement for Reform and Unity is more religiously strict and is characterized by a conservative agenda which includes opposition to the family code reforms (Cavatorta and Dalmasso, 2010: 227). However, political analysts consider that the expressed differences are just obeying a strategy adopted by the two groups (Pruzan-Jorgensen, 2010: 13). Furthermore, Willis (2007) points out the fact that many of the members and supporters of the two main religious movements have views that are substantively more radical than the ones expressed by their leadership. 10 Meriem Yafout, the former leader of the women section, goes one step further considering that monogamy should be the norm and polygamy only the exception (Gray 2013: 89). 9 A women’s association is linked to the PJD party, the Organization for the Renewal of Feminine Consciousness (ORCF). Its members have more conservative views about the ideal family structure: they are favourable to polygamy, the guardianship of the father and marriage for women from the age of 15 (Ramírez, 2006; Dalmasso, 2008). The organization has been also known to support the idea that women should not work outside the house if the husband has the financial means to maintain the family (Gray 2013: 121). ORCF is more attached to a traditional vision of society with a clear division of roles between men and women. The president of the association emphasizes that women should know their duties and rights. However, the rights of women should be defined by considering as a priority the protection of the family instead of its destruction (our own interview). In the same vein, Gray (2013: 121) emphasizes that for ORCF members, women’s rights need to be first and foremost grounded in safeguarding not just the rights of individuals but also those of the children and the family. Unlike the Jamaa Women section, they do not have a well-defined educational project. Adl wal Ihsan appears as the most influential religious movement in Morocco. It is very well organized and very active on the ground notably as a social service provider. The movement opts for a bottom-up approach, starting from the ground in order to change the society. They provide social services, help people and at the same time, ‘educate’ them. Secular education is important - as one of their members asserts, they “have no complex about learning from the Western world” (Willis, 2007:168) - but it has to be complemented by moral education and spiritual teaching. The association attaches importance to the education of women describing them as the basis for the development of future generations and, in turn, for the ideal society they want to construct. As a matter of fact, women make up almost half the membership of the Al Adl wal Ihsan and are extremely active. The PJD choses another strategy. They argue they can participate in the change of society by being involved in the political system. Even if they are also active on the ground, particularly through different associations linked to the party, their approach is more top-down. Since their existence as a party, they won a lot of political support from the voters and achieved very good results since the 2002 legislative elections, becoming by then the third largest party in the parliament. They were thereafter the victims of political maneuverings aimed at keeping their rising influence under control. They accepted the rules of the game and the party’s deputy leader explained in 2000 that « they are frightened of frightening people » (Willis, 2007: 159) in case they grow too rapidly in importance. In November 2011, the PJD emerged as the biggest party in parliamentary elections. Their leader, Abdullah Benkirane, became prime minister and was chosen by the King to compose the new government. The PJD is currently still in power. 4. Data collection Data was collected in 2008 in three of the 16 administrative regions of Morocco: Tangier-Tetouan, Casablanca and Souss-Massa-Draa. 542 randomly chosen individuals were interviewed out of which 283 were married. In this paper, we focus on these married people. Within each region, we conducted a survey in both rural and urban areas. We worked with the Office of National Statistics based in Rabat in order to construct a sample of 180 individuals in each region. Apart from the size of the samples, several criteria have been established: the choice of provinces, cities, rural municipalities and districts (in the city Casablanca), the allocation of 10 the samples throughout the provinces and between urban and rural areas, and some selection criteria used at an individual level. Our framework is the sample of the National Survey on employment, which is conducted quarterly. Annually, 60,000 households are surveyed (40,000 urban and 20,000 rural). In each province, the samples of this survey are representative of the population of the province. The choice of provinces, towns and rural municipalities was made with a view to reducing geographical distances to cover during the survey. The allocation between the provinces in the same region was made according to the demographic weight of each entity within the region, while the allocation between urban and rural areas was based on the demographic weight of each group within the province11. In the rural communities, villages were randomly chosen. In villages and towns, individuals were selected randomly using the marital status as a stratification criterion. Other criteria included gender, level of education, age and the living environment (only for Casablanca). However, these criteria have been used only to ensure that sufficient variance exists for each of them. Concerning marital status, we imposed a strict rule using the information provided by the Office of National Statistics about the proportions of each marital status in each province of interest and in each environment. Finally, within each marital category, we chose people randomly, checking that we have sufficient variance for the other criteria. The details concerning the sample creation in each region are presented in Appendix 1. In each family visited, the questionnaire was administered to a randomly chosen individual on the condition that he (she) is over 18 years old. The first part of the questionnaire concerns the whole nuclear family to which the person interviewed belongs as well as people who are not part of the nuclear family but live under the same roof (possibly members of the extended family). The second part focuses specifically on the interviewed individual. The questionnaire is relatively long and contains parts that need lengthy explanations. The author herself oversaw the entire investigation process. She accompanied the teams carrying out the surveys in every village and participated in a large number of the interviews, either partially or for the whole interview. She was thus involved in all the rural and urban surveys. The interviews were conducted either in Moroccan Arabic or in Berber (for the villages of the Ouarzazate province). Our sample is thus distributed across three different regions: the grand Casablanca for which we use the name CASA, the region of Tangier-Tetouan whose denomination in the paper is TETOUAN and the region of Souss-Massa-Drââ which we name OUARZAZATE, and across two residential environments: urban vs. rural. The distribution of married individuals across regions and residential environments is given in Table 1. It appears that the sample is quite equally distributed across urban and rural areas although this distribution is quite unequal within regions and this is just reflecting the characteristics of each region12. 11 Except for the region of Casablanca where the weight of the rural area is artificially inflated since the region is largely urban. 12 From next section onwards, we work with a few less observations. We eliminate from the database the polygamists families (3 observations), people who married before independence (4 observations) and people separated without official divorce (2 observations). 11 Table 1: Distribution of the sample across regions and areas CASA TETOUAN OUARZAZATE TOTAL URBAN 54 (62.07%) 44 (44.90%) 31 (31.63%) 129 RURAL 33 (37.93%) 54 (55.10%) 67 (69.37%) 154 (54.42%) (45.58%) TOTAL 87 98 98 283 5. Main hypothesis and models’ specifications In this section, we formulate the three main hypotheses we want to test. These are discussed in Section 4.1. Meanwhile, Section 4.2 presents the econometric specifications through which we intend to test these hypotheses. 4.1 Hypothesis In Islam, education is, in principle, a central issue. The fact of seeking knowledge is regarded as a sacred duty; every Muslim, male and female, is required to get education. It is recognized that the first word of the Quran revealed to the prophet Muhammed was ‘Iqra’ which means ‘read’, in the sense ‘learn’, acquire ‘knowledge’. However, in practice, this message may be ignored or misunderstood. Furthermore, aside from the religious recommendation, investment in education is constrained by a series of factors: economic conditions, opportunity cost, social norms, lack of opportunities… Because of these constraints, the priority is in general given to boys at the expense of girls’ education. As a matter of fact, in a patriarchal society, the education of girls is not likely to be valued in the same way as the education of boys. In the light of the evidence of gender biased preferences pointed out in several studies (see literature review), we expect that the involvement of mothers in decisions concerning their children’s education may lead to higher school enrolment for girls and to stronger support of girls’ effort while at school. Interestingly, from our data we observe that, in urban areas, the drop-out rate after primary school is negatively correlated with the mother’s involvement in decisions regarding their children’s education (the correlation is significant at 10% for daughters but not for sons). Yet, in rural areas, we observe a positive correlation between school enrolment in primary school and the input of mothers in schooling decisions but the correlation is not significant. However, it is important to note that we work with very small samples from which it is not easy to derive clear significant correlations. The aforementioned results are presented in Appendix 2. Whether to invest in a child’s education is a decision taken within the household to which mothers may or may not contribute. In our context, it is not clear how the role of mothers in children’s education is perceived: is she allowed to have a say or is the decision the prerogative of the father? From a theoretical viewpoint, at least three reasons for the lack of participation of mothers can be thought of. First, the mother may have very bad fallback options; as a consequence her threat point is so low that she is not allowed to play an active role in the decision process within the household (except for some 12 basic decisions). Second, according to Sen, the woman may undervalue herself so that her bargaining position is weaker and she is likely to accept inferior conditions (see Bruce, 1989: 983). Consequently, the mother may be less involved in decisions because she thinks that her opinion is less valuable and her capacity of judgment is lower than that of her husband. Finally, a lack of awareness of the importance of the education may imply the absence of involvement. Several factors are likely to reduce the obstacles to the role mothers may play in education decisions: the fact that the mother exercises an income-generating activity and contributes to household expenses (which is expected to improve her fallback option), the education path of the mother, her age at marriage, etc. In the empirical analysis, we test for the relationship between these factors and the participation of mothers in education decisions. We also test for the role of the 1993 reform of the Family Code. The changes of the legislation offer women some protection against men’s abuses and one may therefore expect them to somewhat enhance women’s bargaining power. In this paper, we focus on the role of religion. Religion is expected to shape parental values as education does. The tenets and teachings of religious faith influence the perceived benefits and costs of numerous decisions that people make over the life cycle, including choices regarding the pursuit of investments in secular human capital (Lehrer, 2004). However, the nature of the impact of religion on the involvement of women in education decisions is not clear. It depends on perceptions concerning the role that women should play in the private sphere and the importance attached to children’s education, especially for girls. We argue that these perceptions are likely to be shaped by the discourse of religious authorities and religious movements. In Section 2, we described the main religious movements active in the Moroccan context, the most influential one being Al Adl wal Ihsan. This organization – and especially the Women section – attempts to reach ‘spiritual awakening’ through education (defined in a large sense). They emphasize the fact that education of women is a key condition for the development of next generations conferring upon them a social role in the construction of their ideal society. The Women section goes even further describing education as a religious duty. Moreover, the women’s section of the association defends the involvement of women on an egalitarian basis in family matters. Furthermore, although Al Adl wal Ihsan is influential in a large spectrum of the population, it is quite popular in the low and lower-middle classes thanks to its supply of social services. Cavatorta (2006) stresses the fact that the provision of social services is the greatest strength of Al Adl wal Ihsan and as a consequence several secular NGO’s try to compete with them in the field. This shows how effective they are in spreading their message and vision of society. Consequently, if the influence of their discourse is effective, in a reduced form approach, we expect to find a positive correlation between the involvement of women in decisions concerning their children’s education and the intensity of religious practice of the spouses. Conversely, if the patriarchal tradition is dominant, we should find that the involvement of women is quite low. Moreover, in such a society, there is no convincing reason to believe that their participation would be related to the religiosity of the spouses. The second main hypothesis we want to test is whether the influence exercised by religious movements provides a ‘substitute’ for education. The central issue is whether religion, through the social interactions that it implies, can play the same role as education regarding consciousness-raising about the importance of children’s education and regarding self-valuation of women with respect to their capacity to play a key role within their household. We expect to find a positive correlation between the involvement of women in decision-making and the intensity of the religious practice among poorly or 13 non-educated women. As for educated women, it is not evident that we will find an additional effect of religion as they are already benefiting from the positive effect of education. We thus expect to find a ‘substitution effect’ between education and religion. This effect is not a priori intuitive. It is only the thorough knowledge of the Moroccan context that led us to formulate the above hypothesis. The third hypothesis tested in the paper is linked to the structure of the family. The education discourse of the Women section of Al Adl wal Ihsan is quite focused on girls’ education and, in turn, on the social role of women. Consequently, one can expect that religious faith motivates mothers to participate in decisions concerning their daughters’ education mainly. Furthermore, we suspect the existence of a spillover effect that makes them more concerned with the education of their sons as well. If this is true, we expect to find a positive correlation between the intensity of religious practice of mothers and the probability that they participate in decisions concerning their sons’ education, but only in families composed of both sons and daughters. 4.2 Models’ specifications The central question of this paper is whether religiosity plays a role, notably through social interactions in the involvement of mothers in their children’s schooling decisions. To address this question, we first estimate the following model to investigate the relationship between the intensity of religious practice and mothers’ involvement in their children’s schooling decisions: yi =𝛼+𝛽1 reli +𝛽2 wife_to_schooli +𝜑Xi +𝜀i (1a) where yi corresponds to either the dummy variable educ_girl or the dummy educ_boy whose value is 1 if the mother is involved in schooling decisions concerning her daughters and her sons respectively; reli , our main variable of interest, represents the intensity of the religious practice of the mother ; wife_to_schooli corresponds to a dummy variable whose value is 1 when the mother at least achieved a primary school education (i.e. she participated in the education system); Xi is a vector of mother i’s characteristics, characteristics of her husband and her nuclear family; and 𝜀i is the error term. For the sample of men, we have the equivalent model: yi =𝛼+𝛽1 reli +𝛽2 husb_to_schooli +𝜑 Xi +𝜀i (1b) where yi is the same as in model (1a); reli , our main variable of interest, represents the intensity of the religious practice of the father; husb_to_schooli corresponds to a dummy variable whose value is 1 when the father has at least achieved the collège level of education (i.e. the first four years of the secondary school)13; Xi is a vector of father i’s characteristics, characteristics of his wife and his nuclear family; and 𝜀i is the error term. Our estimation strategy is potentially vulnerable to endogeneity biases. In fact, the observed association between the outcome, the participation of mothers in schooling decisions, and the explanatory variable 13 We use a different education threshold level (the collège degree ) in the model specified for the men’s sample because a majority of men have actually completed their primary education. 14 of interest, the intensity of religious practice of mothers (or fathers, for the sample of men), is likely to be misleading in the sense that it might partly reflect omitted factors that are related to both variables. As a matter of fact, mothers may possess some intrinsic qualities that simultaneously motivate them to influence decisions concerning their children’s education and influence their religiosity. If the omitted factors could be measured and held constant in the regression, the omitted variables bias would be mitigated. However, it is difficult to specify all of the variables that should be held constant while estimating the relationship of interest. Our main concern is thus a problem of omitted variables that might influence both mothers’ participation in education decisions and the intensity of their religious practice. We will mitigate this problem by interacting the intensity of religious practice with a dummy variable whose value is one when the mother achieved at least a primary school education. Incidentally, it appears from our sample that the two interacted variables are independent. In fact, the difference in the mean levels of the intensity of religious practice between educated and non-educated individuals is not significant (p-value = 0.23; for the sub-sample of female respondents: p-value = 0.72; and for the sub-sample of male respondents: p-value = 0.22). Education is expected to shape parental values and is likely to enhance the awareness of mothers with respect to the importance of children’s education. We should therefore observe a difference in participation between educated and non-educated mothers. If religion is not playing a role, the difference in behavior between educated and non-educated women is expected to be the same for each level of religious practice. If we find that this difference varies with the level of religious practice, it is likely that religiosity plays a role. It is not easy to find an omitted variable that at the same time influences the participation in education decisions by mothers and the intensity of their religious practice and would have a distinct effect on educated and non-educated women. We argue that, in the Moroccan context, religious movements may precisely have this role: their presence and action is one of the mechanisms we have in mind to explain our findings. Besides this factor, it is hard to identify an omitted variable that could meet the same requirements. Because of data constraints and in spite of the above, we cannot claim to establish a clear causal relationship between our dependent and our explanatory variables of interest. We believe, however, that we are able to show interesting correlations which enable us to better understand the private and social implications of religious practice in the Moroccan context. Along the line that we have just suggested, we estimate model (2) in order to check whether the effect of intensification of religious practice on the participation of mothers in schooling decisions depends on the education of the mother. Besides the discussion about the aforementioned endogeneity problem, this specification will allow testing the hypothesis according to which a ‘substitution effect’ exists between education and religion. yi =𝛼+𝛽1 reli +𝛽2 wife_to_schooci +𝛽3 reli x wife_to_schooli +𝜑X +𝜀i i (2) Since our dependent variable is a dummy, two kinds of models can be estimated: a linear probability model and a non-linear model (Probit or Logit). We will use both types of model to check whether our results are sensitive to the use of a different estimation methodology. In the Probit model, the conditional expected value of the dependent variable is given by the following expression: E y rel,wife_to_school,Z =Φ 𝛽R rel+𝛽E wife_to_school+𝛽RE rel*wife_to_school+Z𝜑 = Φ u (3) 15 Where Φ corresponds to a standardized normal cumulative distribution function; Z is a vector of the characteristics of the spouses and the household to which they belong. We use the methodology proposed by Norton, Wang and Ai (2004) to test for the sign and the significance of the interaction term. The results concerning the interaction term are not straightforward. We need to base the test about statistical significance on the estimated cross-partial derivative of the expected value of the dependent variable. Third, we want to test whether the effect of intensification of religious practice depends on family structure. This is done through the following model: yi =𝛼+𝛽1 reli +𝛽2 wife_to_schooli +𝛽3 fam_structi +𝛽4 reli x fam_structi +𝜑X +𝜀i i (4) The variable fam_struct is a dummy variable whose value is 1 when the family is composed only of sons. In this specification, yi corresponds to the dummy variable educ_boy whose value is 1 when mothers participate in their sons’ education decisions. The hypothesis we want to test here is the following: religious faith motivates mothers to participate in decisions concerning their daughters’ education mainly, but there is a spillover effect that makes them more concerned with the education of their sons as well. If that is true, we should not observe any spillover effect and, therefore, no effect of religiosity, in families that are composed of sons exclusively. 6. Key variables and descriptive statistics This section presents the key variables used in the econometric analysis: the indicator of women’s involvement in decisions, intensity of religious practice, education, and a variable identifying the households potentially affected by the 1993 reform of the Family Code since the Code may have influenced women’s bargaining power. The construction of each of these variables is presented in the following sub-sections. In parallel, we present some basic descriptive statistics concerning both the key variables and some additional variables essentially used as controls in the empirical analysis. 5.1 Indicator of mothers’ involvement in decisions One of the sections of the questionnaire was dedicated to the management of household affairs. The respondents were asked who took the final decision concerning a series of matters including the education of theirs sons and daughters. Based on the responses to these questions, we constructed our dependent variable which is a binary variable. It has a value of 1 in the case where the mother was involved in the final decision: either she took the decision alone or both spouses took the decision jointly. The dependent variable takes the value 0 when the husband was the unique decision-maker. In each household, only one of the spouses was interviewed: either the husband or the wife. Table 2 depicts the participation of women in decisions concerning their daughters’ education as reported by women and men. Table 3 illustrates the same, but for their sons’ education. We observe that only 200 households out of 274 have daughters and 226 have sons. 16 Table 2: Women’s participation in decisions about their daughters’ education GIRLS’ EDUCATION WOMEN RESPONDENTS MEN RESPONDENTS TOTAL NO PARTICIPATION PARTICIPATION TOTAL 47 (44.34%) 67 (71.28%) 114 (57%) 59 (55.66%) 27 (28.72%) 86 (43%) 106 NO PARTICIPATION PARTICIPATION TOTAL 51 (45.13%) 84 (74.34%) 135 (59.73%) 62 (54.87%) 29 (25.66%) 91 (40.27%) 113 94 200 Table 3: Women’s participation in decisions about their sons’ education BOYS’ EDUCATION WOMEN RESPONDENTS MEN RESPONDENTS TOTAL 113 226 We observe that mothers’ participation reported by men is lower than the one reported by women. According to the sub-sample of men, 28.72% of women are involved in decisions about their daughters’ education against 55.66% reported by women. Concerning boys’ education, men reported an involvement of 25.66% of women while mothers reported an involvement of 54.87%. The men seem to have a tendency to under-report the participation of women in household decisions14. This behavior might be due to their willingness to appear as the patriarch who takes all the important decisions, in line with what the unitary model suggests. Moreover, the responses reported by men may under-estimate the informal decision-making agency that women might exercise. The same tendency to under-report is observed for all types of environment (rural and urban) and in all regions. However, it seems to be less the case in the Casablanca region. 5.2 The intensity of religious practice One of the sections of the questionnaire is dedicated to religious practices. Each respondent was asked whether she/he obeys a series of practices. In the affirmative, she/he was asked about the frequency of her/his practice. The combination of the responses to these two kinds of questions allowed us to construct several discrete variables: for each practice, the higher the value of the variable, the higher the intensity of the practice. We then used these variables to construct an aggregate index measuring the intensity of the religious practice by applying the Multiple Correspondence Analysis methodology15. 14 An alternative way of interpreting this numbers is that women over-report participation in decisions. However, from our experience in the field, we get ta sense that men are those who under-report. We have in memory various discussions with enumerators who were convinced that men are not telling the truth concerning the participation of their wives in decisions made within households. 15 The multiple correspondence analysis is a factorial technique which can be seen as a generalization of the principal component analysis when the variables to be analyzed are categorical variables. 17 The questions that were used to construct this index concern the obligatory prayers (five per day), the prayer at dawn (fjar), the prayer during the night (qyam layl), the non-obligatory prayers (nawafil), the supplementary prayers of the Ramadan (tarawih), the fasting beyond the Ramadan, the tithing duty called zakat, whether the respondent owns the Quran, whether the respondent owns the Hadith, whether the respondent owns books about the Islamic thought and whether the respondent shares her/his knowledge of religious principles with her/his circle of kin, friends and neighbours. We decided not to keep questions about readings as they will bias the ‘score’ of the respondents who cannot read. We also decided to remove the question concerning visits to the mosque since this question will bias the ‘scores’ in favour of men. We believe it is essential to have a multidimensional measure of the religious practice because the importance that a respondent attaches to each practice is subjective. Consequently, by arbitrarily choosing one specific practice as a proxy for the respondent’s religiosity, we are likely to obtain different rankings across individuals depending on the choice of the practice. Moreover, the use of a multidimensional index takes into account the accumulation of various practices. As for the Multiple Correspondence Analysis, it allows for the existence of proximities between individuals and ‘correlations’ between variables. We constructed the variable religion, as we define it, is the first dimension in the Multiple Correspondence Analysis. It is therefore continuous: the higher the value of the variable, the more intense the religious practice. Table 4 presents the main descriptive statistics for the variable religion when considering the full sample of married individuals and several subsamples. Table 4: Measure of the intensity of religious practice Nr. Obs. 274 Mean .137 RELIGION Std. Dev. .967 Median .286 Min -2.276 Max 1.916 Rural Urban P-value = 0.356 150 124 .116 .1604 .884 1.062 .244 .369 -2.276 -2.204 1.787 1.916 Women Men P-value = 0.0473 141 133 .231 .036 .852 1.069 .303 .252 -2.276 -2.237 1.916 1.815 CATEGORIES Full sample We do not observe a significant difference in terms of intensity of religious practice between urban and rural population whereas women appear as being significantly more religious than men. Within a household, we only have information about religious practice of the respondent. We do not have this information for both spouses. This represents one of the limitations of our data. 5.3 Education Concerning education, we asked the respondent whether she/he has ever been to school. In the affirmative, we asked her/him about the highest level of education she/he completed. In the case she/he is still at school, we asked for her/his actual level. We did the same for the respondent’s partner. We used the answers to these questions to construct categories comprising individuals with different levels of education. We constructed several binaries, each of them representing one specific level: no_educ for people who have never been to school or people who just followed lectures in a 18 mosque; primary for people who achieved a primary school education; secondary for people who achieved the collège level of education (the first four years of the secondary school) and for people who went to the lycée (the three last years of the secondary school) without obtaining the baccalauréat; high_educ for individuals who gained the baccalauréat or a university degree (or an equivalent one). On the basis of all this information, we created two discrete variables wife_educ and husb_educ whose value varies between 1 and 4: 1 for non-educated people, 2 for people who achieved primary education, 3 for people who achieved the collège level or went to the lycée, and 4 for people who got the baccalauréat or a university degree. We also constructed a dummy variable specific to wives’s education, wife_to_school whose value is one for the wives who achieved at least primary school education. This variable permits to distinguish between women who participated in the education system and achieved its first level and the others. For men, we constructed the variable husb_to_school whose value is one for husbands who achieved at least the collège degree. As primary education is more extended among men, we use the collège degree as an education threshold. Table 5 presents the distribution of husbands and wives across the different levels of education. Table 5: Spouses’ education levels WIVES’ EDUCATION no_educ 115 no_educ 4 primary 4 secondary 0 high_educ 123 (44.89%) TOTAL HUSBAND’ EDUCATION primary secondary high_educ 30 23 4 1 58 (21.17%) 19 7 22 2 50 (18.25%) 7 8 12 16 43 (15.69%) TOTAL 171 (62.41%) 42 (15.33%) 42 (15.33%) 19 (6.93%) 274 Most of the observations are on the diagonal of the matrix which means that in a majority of households there is a perfect matching in terms of education (176 household out of 274, i.e. 64.2% of the households). When there is no matching, it is more likely that the husband is more educated than his wife (83 households out of 98, i.e. 85% of the household with no matching). Note that about 42% of the sample is composed of non-educated spouses (115 households out of 274). 5.4 The 1993 reform of the Family Code In 1993, a campaign led by the associative sector managed to collect one million signatures to support the claims of those working towards a change in the Family law which was not reformed since the independence in 1956. The initial Code was founded on a combination of religious values and other rules rooted in customary norms. It was quite conservative and traditionalist. The Women’s Action Union (UAF) founded in 1987 presented in 1993 an open letter to the House of Representatives including claims based on a triple standard: the objectives of Sharia (under Islamic law), social reality, and universal human rights. This letter was first addressed to politicians but King Hassan II reacted by stating that the issue involved religious values and, as such, falls in the purview of the ‘Commander of the Faithful’, and not in the political world. The king then received a delegation of women. Quoting ijtihad16, he stressed that legal prescriptions cannot prohibit what God has not prohibited, and cannot 16 In Browning et al (2006), Ijtihad is defined as follows : « Ijtihad is used to refer generally to the jurisprudential activity in which scholars engage either to interpret the Quran and Sunnah, where an interpretation is required, or to reach a ruling 19 allow what has not been permitted by God. He promised amendments proposed by a committee composed of ulama (theologians and jurists). In 1993, the Family Code was reformed for the first time since Morocco’s independence. The main changes were about marriage and divorce rules. The steps taken toward establishing a polygamous marriage and obtaining a divorce are more complicated than before. In addition, the woman must provide her consent to marriage and if she lost her father, she is no longer subject to matrimonial guardianship. Even if these changes are limited, they offer women some protection against men’s abuses and one may therefore expect them to somewhat enhance women’s bargaining power and self-confidence. Given the nature of the reform, we believe that it might have influenced the behavior of spouses who married after 1993. For each of the respondent, we know the date of marriage. Consequently, we constructed a dummy variable whose value is 1 if the respondent got married after 1993 and 0 otherwise. This variable will allow us to control the potential effect of the reform. Table 6 indicates the proportion of marriages that took place before and after the reform for both the sample of women and men. Table 6: Distribution of marriages across time FAMILY CODE REFORM MARRIAGE BEFORE 1993 MARRIAGE AFTER 1993 TOTAL WOMEN RESPONDENTS 85 (60.28%) 90 (67.67%) 175 (63.87%) 56 (39.72%) 43 (32.33%) 99 (36.13%) 141 MEN RESPONDENTS TOTAL 133 274 We observe that about one third of the individuals who composed the sample married after the first reform of the Family Code. We carried out chi-squared tests in order to check whether the participation of women in education decisions is independent from the date of marriage. Independence is rejected at 1 % level in the case of girls’ education and at 5% level in the case of boys’ education. 5.5 Some additional descriptive statistics Table 7 presents a summary of statistics about some additional characteristics of the spouses and the household: the age of the wife, wife_age; the age of the husband, husb_age; the distance in age between the two spouses, dist_age; the age at marriage for women, wife_age_mar; the age at marriage for men, husb_age_mar; the number of children, nbr_child; and a dummy which indicates whether the wife contributes to household expenditures, wife_exp. involving no clear Quranic pronouncement or prophetic precedent » (153-154). The Sunnah is the example given by the Prophet through his life. Considered as the ideal Muslim, he offered the best guidance (Al Hibri et al, 2006: 153). 20 Table 7: Additional descriptive statistics Variables Nr. Obs. Mean Std. Dev. Median Min Max 272 272 271 271 272 274 272 40.23 47.68 7.53 20.58 28.16 3.51 0.13 11.21 12.96 6.27 5.71 7.01 2.34 / 40 46 7 19 27 3 / 18 22 -20 11 17 0 0 71 99 41 54 71 11 1 wife_age husb_age dist_age wife_age_mar husb_age_mar nbr_child wife_exp We observe that generally husbands are older than their wives. Women marry at a younger age than men. Only a small proportion of women contribute to household expenses, these are mostly the ones who exercise an income-generating activity. On average, there is no significant difference between the two sub-samples (sample of female respondents and sample of male respondents) for the aforementioned variables. The unique difference which is statistically significant is the one concerning the age at marriage of the husband. Table 8 presents these statistical tests. Table 8: Difference in means Variables wife_age husb_age dist_age wife_age_mar husb_age_mar nbr_child wife_exp Men respondents Nr. Obs. Mean Women respondents 132 133 132 131 133 133 131 41.33 48.29 7.11 20.19 27.23 3.41 0.12 Women respondents Nr. Obs. Mean 140 139 139 140 139 141 141 39.19 47.11 7.92 20.95 29.05 3.62 0.14 P-value 0.12 0.45 0.29 0.28 0.03 0.46 0.63 7. Results In this section, we present the results obtained with the econometric estimation of the models described in Section 4.2. The results regarding mothers’ religiosity are described in Section 6.1 and those regarding fathers’ religiosity in Section 6.2. In both sections, we analyze mothers’ involvement in decisions concerning their daughters’ and their sons’ education successively. 6.1 Religiosity of mothers 6.1.1 Education of daughters Table 9 presents the results for the entire sample and for the sub-sample of female respondents (models without interaction term) while Table 10 presents additional results for the sub-sample of 21 female respondents only (with interaction term)17. The dependent variable is the variable educ_girl which is a dummy variable whose value is one when the mother takes part in decisions concerning her daughters’ education. Our variable of interest is the variable religion measuring the intensity of religious practice. In regression (1), we work with the whole sample. We do not have information about religious practice for both spouses within a household; we just have the information for the respondent who is either the wife or the husband. Consequently, the variable religion corresponds to the religiosity of the wife when she is the respondent and to the religiosity of the husband when he is the respondent. Consequently, by using the variable religion as a regressor in a model estimated for the whole sample, we have to assume that there is homogeneity in terms of religious practice within a couple, which is quite a strong assumption. We can somehow relax this assumption by interacting the variable religion with the variable resp_husb which is a dummy variable whose value is one in case of a male respondent. In that case we make another hypothesis: the effect of the religiosity of the spouses depends on who reports it. The interaction term informs us about the additional effect of the religiosity of the husband when it is reported by himself. A third approach consists of estimating a model where the variable religion only appears interacted with other variables: with resp_husb and with resp_wife which is a dummy variable whose value is one in case of a female respondent. We thus look at the relationship between the participation of mothers in education decisions and the intensity of religious practice of either the wife or the husband when each of them reports it for himself. Actually, the last two approaches are equivalent. The results regarding the last approach are reported in regression (1). The majority of the controls have been described in Section 5. The variable rural is a dummy variable whose value is one when the respondent is living in a village. The variable nbr_rooms gives the number of rooms per person living under the same roof as the respondent18. This variable is used as a proxy for the level of wealth of the respondent’s family. The variable marriage_date gives the date of marriage of the couple. This variable is used as a time trend that we use to assess the effect of the 1993 reform of the Family Code. From regression (1), we observe that a positive correlation exists between the intensity of religious practice of mothers and their participation in the decisions concerning their daughters’ education. A positive correlation also exists between the intensity of religious practice of fathers and the latter variable. Yet it is much smaller and not significant. The results of regression (1) also indicate that mothers belonging to couples married after the 1993 reform of the Family Code are more involved in education decisions. Women who went to school and women who exercise an income generating activity are more likely to take part in education decisions. This is in line with intra-household bargaining models where spouses are more engaged in the decision process when they have a better fallback option. Finally, men seem to underreport the participation of women in decisions concerning their daughters’ education (the coefficient of the variable resp_husb appears with a negative sign). Some other estimated coefficients appear with the expected sign but are not significant. Thus a generational effect exists for women: the older ones have a lower probability of taking part in education decisions. Moreover, a positive correlation exists between the age at marriage of women and their participation in the decision process: the older a woman is when she marries, the more likely is her participation in education decisions. This last result is in accordance with some negative personal 17 Given that the choice of interviewing either the father or the mother is random, the selection bias problem that can arise when working on a sub-sample is limited. 18 We exclude the toilets, the kitchen and the bathroom. 22 consequences endured by women in a patriarchal culture which encourages marriages at an early age. However, one can argue that women who accept to marry very young are also less assertive, more submissive. We observe from our data that in the case where spouses knew each other before marriage and personally chose their partner, women married at an older age. We therefore observe that women who married at a younger age have a higher probability of experiencing an arranged marriage. Nevertheless, when we control for the fact that the marriage is arranged or not, the result concerning the age at marriage is still the same. Women living in a rural area and with a higher number of children are less likely to take part in education decisions. Concerning our main result, the positive and significant correlation between the intensity of religious practice and women’s participation in the decisions about their daughters’ education, at least two stories can be thought of. First, women might use religion strategically to achieve some independence and to escape from traditional pressure. Second, through social interactions women are influenced, directly or indirectly, by the discourse prevailing in religious movements. This influence bears simultaneously on their religiosity and their perceived role within the family as well as their perception of the importance of education. They may feel that their value is enhanced when religion confers upon them a social role through their action within the household. By constraining our sample to female respondents (regressions (2) and (3)), we also find a positive and significant correlation between the intensity of religious practice of the mothers and their involvement in decisions about daughters’ education. This correlation is even stronger. Another important parameter is the participation of mothers in the education system. In fact, mothers who at least achieved a primary school education (wife_to_school = 1) are more likely to take part in education decisions about their daughters and this difference is highly significant. Moreover, we observe that the education level of fathers is negatively related to the participation of mothers in education decisions. This may be due to the fact that a father with a high education level considers that he is more able to judge what is best for his daughters or alternatively, the mother is less likely to intervene because the father already takes favorable decisions for the education of their daughters. However, this last correlation is not significant. 23 Table 9: Mothers’ religiosity and participation in decisions concerning their daughters’ education DEP. VARIABLE: Whole sample (1) EDUC_GIRL (LPM) Women (2) EDUC_GIRL (LPM) Women (3) EDUC_GIRL (Probit) 0.143*** (0.0494) 0.694*** (0.198) 0.151 0.304** (0.121) 0.232* (0.119) -0.0357 (0.0279) -0.119 (0.0999) -0.174*** (0.0637) -0.0303 (0.0409) 0.0422 (0.105) -0.00280 (0.0570) 0.00425 (0.00613) 0.00459 (0.0561) -0.000478 (0.0566) 1.807 (113.7) Yes 105 0.451 1.453** (0.595) 0.770 (0.517) -0.167* (0.0938) -0.276 (0.366) -0.718** (0.363) -0.160 (0.180) 0.135 (0.406) 0.0571 (0.196) 0.0230 (0.0201) -0.0339 (0.197) 0.0781 (0.195) -155.6 (390.8) Yes 105 0.340 RELIGION RESP_HUSB REL_RESP_HUSB REL_RESP_WIFE WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant -0.252*** (0.0675) 0.0226 (0.0417) 0.106** (0.0501) 0.172* (0.0942) 0.283*** (0.101) -0.00720 (0.0185) -0.0989 (0.0776) -0.0716 (0.0542) -0.0199 (0.0342) 0.164* (0.0941) -0.0613 (0.0380) 0.00482 (0.00490) 0.0599 (0.0381) -0.0593 (0.0379) 119.9 (76.20) Yes 197 0.377 Regional dummies Observations R-squared Pseudo R-squared 0.431 Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Average Marg. Effects 0.174 -0.036 -0.062 -0.156 -0.035 0.030 0.012 0.005 -0.007 0.017 105 We also observe that women in wealthier families are less involved in education decisions. At first glance, this last result is quite puzzling. However, it could make sense if one considers that our proxy is essentially measuring the wealth of the husband. In fact, we do not know how wealthy each spouse was by the time of marriage. Traditionally, the wealth of the household is considered as belonging to the 24 husband as he is generally the one who exercises an income-generating activity. Until the Family Code reform of 2004, in the case of a separation, the wealth was kept by the spouse able to provide proofs of payment, generally the husband. Now, the law offers a more equitable alternative. The spouses can choose to establish a contract in which they state how to divide their property in the case of divorce. Nevertheless, this contract is not compulsory and by default, the old law applies. Consequently, the wealthier the husband is, the less inclined the woman is to take the risk of entering into conflict with him. The safest behavior for her would be to let him take all final decisions concerning important matters. An alternative explanation may lie in the conservative character of the wealthiest families. According to Bourqia (2005), in the Moroccan context, traditional values persist along with more progressive viewpoints. She highlights the existence of a hybrid basket of values where individuals choose the more appropriate ones depending on their personal interests. The wealthiest families may be the most conservative concerning the division of roles within the household, each of the spouses being confined to a traditional role in accordance with a patriarchal vision. The father is in charge of the support of the household and, in return, he is the central authority within the family. The women does not need to work (her participation to the labour market may even be interpreted as a sign of lower social status)19. In this kind of environment, women may internalize completely the traditional division of roles and do not necessarily intervene in the formal education of their daughters. Finally, we observe that the higher the number of children the less likely the participation of mothers in decisions concerning their daughters’ education. Table 10 presents the results for the models where we introduce an interaction term: the variable religion is interacted with the dummy variable wife_to_school. The purpose of this approach is two-fold. First, we want to test the hypothesis according to which a ‘substitution effect’ exists between formal education and religion. Second, we want to mitigate the potential endogeneity problem arising from the existence of omitted variables which might influence the dependent variable and the variable of interest simultaneously. From regression (4), for which we use the entire sample of female respondents, we observe that a positive and significant correlation exists between the intensity of religious practice of non-educated mothers and the likelihood of taking part in education decisions. On the contrary, a Wald test reveals that the correlation between the intensity of religious practice of educated mothers and the likelihood of taking part in education decisions is not significant. Moreover, educated women are more likely to participate in education decisions. It therefore appears that a substitution exists between education and religion. In order to get the same effect as education on the likelihood of participating in education decisions, ceteris paribus, an increase of 1.78 in the index measuring the intensity of religious practice is necessary20. As it is an index which aggregates a lot of information, it is hard to give an intuition of what such an increase represents. However, an increase of 1.78 is quite a significant one as it corresponds to about two times the standard deviation. Regression (5), which corresponds to the estimation of a Probit model, confirms the previous results. Moreover, in both regressions (4) and (5), we observe once again that women in wealthier families are less involved in education decisions. We also observe that the number of children is negatively related 19 From the sub-sample of women, we observe that women who exercise an income-generating activity are from less wealthy families: the difference in means in terms of wealth across working and non-working women is significant at 10%. 20 This increase has been calculated from regression (4). From regression (5), we observe that an increase of about 1.9 in the index is needed to get the same effect as education. 25 to mothers’ involvement in decisions. Finally, we observe that the education level of fathers is negatively related to the participation of mothers in education decisions. But this last result is not significant. Table 10: Women’s participation in education decisions, religiosity and education level DEP. VARIABLE: RELIGION WIFE_TO_SCHOOL REL_WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant Women Women (4) (5) EDUC_GIRL EDUC_GIRL (LPM) (Probit) 0.219*** (0.0490) 0.390*** (0.115) -0.257** (0.117) 0.184 (0.123) -0.0413 (0.0273) -0.105 (0.0961) -0.191*** (0.0567) -0.0385 (0.0402) 0.0387 (0.107) -0.0157 (0.0515) 0.00368 (0.00595) 0.0169 (0.0504) -0.0125 (0.0509) 1.039*** (0.242) 1.811*** (0.560) -1.158** (0.547) 0.793 (0.537) -0.199* (0.104) -0.252 (0.405) -0.944** (0.405) -0.236 (0.202) 0.0287 (0.406) 0.0274 (0.201) 0.0238 (0.0208) -0.00149 (0.205) 0.0482 (0.200) 26.05 (102.3) Yes 105 0.484 -95.25 (401.9) Yes 105 Average Marg. Effects 0.209 0.401 -0.265 0.162 -0.040 -0.053 -0.190 -0.047 0.006 0.006 0.005 -0.0003 0.010 Regional dummies Observations 105 R-squared Pseudo R-squared 0.465 Sign I.T. Neg. Sig. I.T. * Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Coefficients of other control variables have the expected sign but they are not significant: the 1993 reform of the Family Code is positively related to mothers’ participation in education decisions; 26 mothers who live in villages are less likely to take part in education decisions; the participation of mothers in household expenditures is associated with a higher participation in education decisions; age of the mother is negatively correlated with the likelihood of participating in decisions which could be interpreted as the generational effect; the higher the age of the mother at marriage, the more likely she is involved in decisions concerning her daughters’ education. But these two last results are not robust to the estimation methodology. Robustness checks As a robustness check, we estimate the same models as in regressions (4) and (5) on a restricted sample. We restrict the sample to families where the youngest daughter was no more than 5 years old in 1990. We have in mind the role that may be played by an association as Al Adl wal Ihsan which becomes very active in different strata of Moroccan society from the mid-80’s. Consequently, its discourse is likely to influence people with young children by that time onwards. The results are presented in Appendix 3 and confirm the results found in regression (4) and (5). Finally, we estimate the same model as in regression (4) and (5) but we change the education threshold for women: the interaction term is composed by the variable religion and a dummy variable whose value is one for women who at least achieved the collège level (the first four years of secondary school). We obtained the same kind of results: on the one hand, we observe that a positive and significant correlation exists between the intensity of religious practice of non-educated or poorly educated mothers and the likelihood of taking part in education decisions. On the other hand, a Wald test reveals that the correlation between the intensity of religious practice of educated mothers and the likelihood of taking part in education decisions is not significant. These results are presented in Appendix 4a. Given that the majority of rural women are not educated (89% of female respondents who have daughters) and a large part of urban women have participated in the formal education system (52% of female respondents who have daughters), one could suspect that the difference that we emphasize between educated and non educated women is just due to a difference between rural and urban women. We therefore estimate a model in which we interact the variables religion and rural. The results of this estimation are presented in Appendix 4b. It appears that a positive correlation exists between the intensification of religious practice and the likelihood of women’s participation in education decision in both rural and urban areas. The correlation is thus not statistically different between these two types of environments. Discussion of the results As argued before, we think that religion is likely to shape parental values as education does. However, the nature of the impact of religion on the involvement of women in education decisions is not clear. It depends on perceptions concerning the role that women have to play in the private sphere and the importance attached to children’s education. We argue that these perceptions are likely to be shaped by the discourse of religious movements. In the Moroccan context, we identify two important religious movements and Al Adl wal Ihsane appears to be the most influential. Moreover, the discourse of the Women section of this organization is focused on the role of women in society and the importance of education. 27 In September 2014, we carried out a series of interviews with current and former members of the women section. Those interviews took place in Casablanca, Rabat, Marrakech and Tangier. We interviewed both members with important responsibility positions and members at the bottom of the hierarchy active at the local level through their involvement in associations. These interviews bring a series of relevant insights in support of our hypothesis about the influential role of this religious movement, especially among the women population. By interviewing those women, we tried to better understand they action, strategy and discourse. The members of the women section consider that women are the victims of inequitable treatments in the name of religion. In order to change the status quo they state that it is urgent to correct humanbiased interpretation of religious prescriptions. They consider that gender justice can be derived from a scholarly re-reading of sacred texts. According to them, women should obviously play an active role in the re-interpretation efforts and should therefore be involved in important debates on this question. The actions of the women section are twofold: on a one hand, they initiated a theoretical work in order to challenge the dominant Islamic jurisprudence about women’s rights, and on the other hand, they are involved in social action on the ground. They emphasize the need to study religion in order to distinguish between the message conveyed by original sources and the message diffused by the ulama (religious scholars). They consider that both the existence of an interpretation problem and a selective reading of religious prescriptions led to the diffusion of messages that reinforce patriarchy. Patriarchy is perceived as a major obstacle to the advancement of women’s rights. In order to challenge the dominant patriarchal view, they emphasize the need of highly educated women in several disciplines. They want women to mainly specialize in social sciences and jurisprudence because they need competent women to engage in the reinterpretation of religious prescriptions. Education therefore appears as a central element in their fight against the lack of justice. Women members of the movement are highly encouraged to pursue their education. One important project is the Aalimate project through which they motivate women members to undertake more advanced studies. Education is described as the only way to evolve in society and make the society evolve. Education is defined as a responsibility and as a duty for all believers. According to members of the women section that we interviewed, women who do not make the effort to acquire education and educate their children are considered to fail in their commitment to God. Here, it is worth quoting the own words of Nadia Yassine as reported by Gray (2013:102): “It is by meeting the challenges here on earth with a serene mind that we ultimately effect change. We are striving for women to become experts not just in the domain of religion but also in areas such as psychology, political science, law, medicine, etc. By doing this, we revive the prophetic tradition which elevates the value of human beings. We have to remember the history of Islam and reject the paternalistic readings and interpretations that have marginalized women. It is necessary for women to have their own interpretations.” Their action on the ground is motivated by three main objectives: fight against poverty, illiteracy (in a broad sense: ‘lack of education’), and injustice. They privilege the proximity with people and act at the local level. They also train and encourage the involvement of local people. This later strategy may increase the participation rate in projects initiated by the Jamaa. In fact, people are approached by persons they know - their neighbours – and may therefore feel confident. This allows the organization to benefit from a widespread network and a decentralized structure. Activity reports are established at the local level and then centralized. Discussions about the needs of each Moroccan region take then 28 place at the national level and specific strategies are defined based on a precise knowledge of the reality on the ground. So, the members of the women section work with women in neighborhoods. They teach and, speak to them about justice and rights, the importance of education as well as give them a religious education (Ramírez, 2006: 115-116; own interviews with female members). Part of their projects target poorly educated women and/or women from poor socio-economic background. They argue that those women are generally characterized by poor mental and physical health and a poor spiritual knowledge. They organize meetings in neighbourhood to discuss issues related to women and children education. Women who come to those meeting are indeed generally from low socio-economic background and poorly educated. The more educated are asked to play the role of facilitators or supervisors for other unprivileged women. The objective is to offer a platform where women can discuss their needs and difficulties in life and then engage in a dialogue. Women from the Jamaa try to increase the consciousness of women with respect to a series of family and social issues, they listen to these women and they encourage them to take action in a structured and efficient way. They try to create a willingness to take action in order to change their own person and in turn their environment. Another ambitious project is called ‘Sister for Eternity’. This project is aimed to increase the ‘consciousness’ of women and advice them about the ‘right path’ to follow in life (‘guiding advice’). The objective is to get closer to God and the necessary condition is a return to religious sources. The project consists in organizing all-female gatherings led by women. As argued by Gray (2013: 103), those gatherings send a strong empowering signal. Those were initially only open to members of the movement. As the project grew in importance quite rapidly, the women section decided to develop it further by creating the school of ‘Sister for Eternity’ where ‘sisters’ (female members) with good communication skills where trained to become in charge of sessions of the ‘Sisters for Eternity’ project all over the country. The Sisters in charge of sessions were taught basic principles of the Jamaa and how to organize the gatherings. All women interested in the philosophy of the movement were invited to participate. They talk about ‘women enlightment’, which means that women should reach a certain level of consciousness and agency, and this could only happen through education21. They emphasize that women should learn to respect themselves and be confident. They believe that women have a selfesteem problem. Increasing women’s agency is believed to be a difficult process. They think that women can change more easily if they associate the change in their behaviour to their relation with God. They try to convince other women that the change will be valued by God and will in turn allow them to get closer to God. They use religious references in order to induce social change. However, they believe that the change could only occur if people understand their religion and acknowledge that undertaking change is a way to get closer to God. Education was central in the ‘Sister for Eternity’ discussions given that education is a huge issue in Morocco. They define education in a broad sense. Three dimensions are emphasized: learning, spiritual 21 According to a current member of the Women section, 90% of the activities of the section are educational and professional training activities (the figure comes from an internal report). They aim to increase the autonomy of women. They prefer to use the term ‘autonomy’ instead of ‘independence’ to distinguish themselves from the discourse of secular feminists. 29 teaching and physical and mental health. They consider that even if a mother is not educated, if she is conscious of the importance of education, she can play an important role in encouraging her children. According to one of the interviewees: “The mother even if she is not educated, if she knows the message of Islam, she will focus on the education of her children because she understands that this is her responsibility. The child is ‘present’ in her hands”. Also central in their approach is, once again, their willingness to teach women that the dominant interpretations of religious prescriptions are patriarchal. They question those interpretations. They explain that there is no reason to think that these interpretations should be unique and unchangeable. Moreover, they want to explain what are the historical origins of these interpretations. They want to engage in jurisprudential questions, which according to them should not be dissociated from political motivations. They incentivize women to militate for a reinterpretation of religious prescriptions by returning to the sources. This is what they call tajdid, the renewal. The project ‘Sister for Eternity’ became so popular that a series of oppositions arose from the leadership (mainly composed of men) of the Jamaa. Women from the movement characterized those men as inflexible minds who felt threatened by the ‘feminist’ approach adopted by women of the movement. The leadership criticized the fact that women dared to challenge the current interpretation of religious prescriptions and accused them to adopt attitude similar to those of secular feminists. As a consequence, women with important responsibility positions in the movement but also women in charge at the local level decided to leave their positions in order to avoid any polemical situation. This happened in the sensitive period that follows the death of the leader of the movement, Sheikh Yassine. This created a large vacuum in the women section of the movement. Those women considered that they could not be involved in the movement without the respect of two main principles: the acknowledgment of the specificity of the women condition and independence in their action. They consider that they cannot produce change without independence. At present, any project must be accepted by a mixed committee (not only women), which defines strategies that should be followed. Given that the aforementioned conditions were not respected anymore, most of the women with a responsibility position in the movement preferred to put an end to their involvement. However, they still consider that they are part of the movement given that they are faithful to its philosophy. Unlike the women section, the leadership of the movement has no clear strategic plan in order to improve the position of women in society. One of the former members of the women section that we interviewed considers that the movement did not reach a sufficient level of maturity to take advantage of the actions of the women section. Even if the ‘Sister for Eternity’ project has been disrupted after most of women with responsibility positions resigned, local activists consider that their actions on the ground are still motivated by the same objectives and based on the same philosophy. 6.1.2. Education of sons In Tables 11 and 12, we present the results for models where the dependent variable is educ_boy, a dummy whose value is 1 when the mother takes part in education decisions concerning her sons. In Table 11, regression (6) estimates the model for the entire sample. Regressions (7) and (8) are restricted to the sample of female respondents. In regression (6), it appears that there is no significant correlation between the intensity of religious practice of both fathers and mothers and the likelihood of 30 women’s participation in decisions concerning their sons’ education. When we concentrate the analysis on female respondents (regressions (7) and (8)), the correlation is positive and significant at 10%. However, the correlation is weaker than in the case where we analyze the participation of mothers in decisions concerning their daughters’ education. Table 11: Mothers’ religiosity and participation in decisions concerning their sons’ education DEP. VARIABLE: Whole sample (6) EDUC_BOY (LPM) Women (7) EDUC_BOY (LPM) Women (8) EDUC_BOY (Probit) 0.0941* (0.0544) 0.317* (0.187) 0.080 0.261* (0.135) 0.234* (0.127) -0.00982 (0.0324) -0.154 (0.119) -0.245*** (0.0688) -0.0596 (0.0413) 0.0284 (0.143) 0.0143 (0.0610) 0.00576 (0.00639) -0.00613 (0.0613) 0.0166 (0.0611) -32.62 (122.7) 0.919* (0.476) 0.841* (0.455) -0.0466 (0.0986) -0.451 (0.381) -1.031*** (0.325) -0.192 (0.148) 0.0665 (0.444) 0.134 (0.207) 0.0234 (0.0200) -0.0968 (0.211) 0.145 (0.206) -290.0 (414.6) 0.250 RELIGION RESP_HUSB REL_RESP_HUSB REL_RESP_WIFE WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant -0.239*** (0.0649) -0.000504 (0.0422) 0.0522 (0.0567) 0.197** (0.0890) 0.144 (0.0939) 0.0131 (0.0172) -0.122 (0.0808) -0.0765 (0.0544) -0.0163 (0.0320) 0.239** (0.108) 0.0236 (0.0249) 0.00593 (0.00479) -0.0177 (0.0256) 0.0257 (0.0249) -51.00 (50.07) Regional dummies Yes Yes Yes Observations 221 111 111 R-squared 0.320 0.392 Pseudo R-squared 0.344 Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Average marg. effects 0.214 -0.012 -0.122 -0.259 -0.048 0.017 0.034 0.006 -0.024 0.036 111 31 From Table 11, it is also evident that women who married after the 1993 reform of the Family Code are more likely to take part in decisions concerning their sons’ education. Moreover, mothers in wealthier families are less likely to be involved in education decisions. These last two results are essentially true for the women sub-sample. Finally, the participation of mothers in the education system is also associated with a higher probability of taking part in education decisions. In Table 12, we look at the relationship between the intensity of religious practice of the mothers and the likelihood of participating in decisions concerning their sons’ education by allowing differences across groups with specific characteristics: educated vs. non-educated mothers; families composed of both daughters and sons (mixed families) vs. families comprising sons only (non-mixed families)22. Regressions (9) and (10) show that for non-educated women, the more intense their religious practice the more likely their participation in decisions concerning their son’s education. A Wald test indicates that this kind of relationship does not exist for educated mothers. Moreover, women who participate in the formal education system are more likely to be involved in education decisions. Once again, a ‘substitution effect’ between education and religion seems to be operating. In regressions (11) and (12), we observe that in families with both daughters and sons (‘mixed families’), the more intense the religious practice of mothers the more likely their participation in decisions concerning their son’s education. On the contrary, in families exclusively composed of sons, a significant negative relationship exists between the intensity of religious practice of mothers and their involvement in education decisions. We may, therefore, deduce that a spillover effect exists which leads to the participation of mothers in education decisions concerning their sons when they are already involved in decisions concerning their daughters. In fact, if the positive relationship between the intensity of religious practice and the involvement of women in education decision is partly due to the influence of a religious movement such as Al Adl Wal Ihsan, priority may be given to daughters since its educational discourse is focused on the significant social role of women and the importance of their education. Consequently, women who have sons only may be less concerned by the discourse of the association about the role of education. 22 In the rest of the analysis, we use the variable NON_MIXED whose value is 1 in families comprising only sons. This variable corresponds to the variable fam_struct of model (4) presented in Section 4.2. 32 Table 12: Religiosity, education and family structure DEP. VARIABLE RELIGION WIFE_TO_SCHOOL REL_WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Women (9) EDUC_BOY (LPM) Women (10) EDUC_BOY (Probit) 0.203*** (0.0590) 0.364*** (0.122) -0.265** (0.110) 0.219* (0.127) -0.0214 (0.0323) -0.115 (0.110) -0.248*** (0.0638) -0.0687 (0.0420) 0.0160 (0.135) -0.0222 (0.0509) 0.00569 (0.00653) 0.0302 (0.0510) -0.0202 (0.0510) 0.831*** (0.261) 1.488*** (0.539) -1.169*** (0.442) 0.886* (0.482) -0.0814 (0.105) -0.255 (0.436) -1.140*** (0.382) -0.233 (0.168) 0.0581 (0.450) 0.0103 (0.199) 0.0253 (0.0218) 0.0377 (0.205) 0.0201 (0.199) Average Marg. effects 0.193 0.372 0.207 -0.019 -0.062 -0.265 -0.054 0.013 0.002 0.006 0.009 0.005 REL_NON_MIXED 41.36 (102.4) Yes 111 0.425 -39.97 (399.5) Yes 111 Regional dummies Observations 111 R-squared Pseudo R-squared 0.389 Sign I.T. Neg. Sig. I.T. *** Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Women (12) EDUC_BOY (Probit) 0.166*** (0.0513) 0.311** (0.131) 0.627*** (0.214) 1.179** (0.546) 0.144 0.227* (0.134) -0.0116 (0.0328) -0.149 (0.117) -0.236*** (0.0671) -0.0560 (0.0414) 0.0306 (0.134) -0.000746 (0.0601) 0.00578 (0.00662) 0.0121 (0.0606) 0.00126 (0.0604) -0.0376 (0.123) -0.424*** (0.129) -1.956 (121.3) Yes 111 0.438 0.711 (0.493) -0.0543 (0.104) -0.436 (0.400) -0.959*** (0.346) -0.159 (0.158) 0.267 (0.464) 0.112 (0.214) 0.0248 (0.0213) -0.0566 (0.220) 0.126 (0.214) -0.0519 (0.505) -1.942*** (0.577) -253.5 (429.0) Yes 111 0.168 Average Marg. effects 0.294 -0.298 NON_MIXED Constant Women (11) EDUC_BOY (LPM) -0.012 -0.106 -0.220 -0.036 0.061 0.026 0.006 -0.013 0.029 -0.012 -0.372 111 0.401 Neg. *** 33 6.2 Religiosity of fathers In this section, we focus on the sub-sample of male respondents. We estimate the models (1b), (2) and (3) specified in Section 4.2 in order to test for the existence of a significant correlation between fathers’ religiosity and the participation of their wives in decisions concerning the education of their children. As before, we use the intensity of religious practice as a proxy for religiosity. 6.2.1 Models without interaction term Table 13 presents the results for the estimation of model (1b). In regressions (13) and (14), the dependent variable is a dummy whose value is one when the respondent’s wife is said to be involved in decisions concerning their daughters’ education. We observe that there is no significant correlation between fathers’ intensity of religious practice and mothers’ participation in education decisions. It is also evident that the education of mothers enhances their probability to participate in education decisions. Moreover, we observe a generational difference: older women are less likely to be involved in decisions. If a woman marries at an older age, she is more likely to participate in the decisions concerning her daughters’ education. When we control for the fact that spouses knew each other before marriage and personally chose their partner, it appears that the positive correlation between the age at marriage and the probability to participate in education decisions exists for both arranged and free marriages23. Mothers who exercise an income-generating activity are more likely to participate in education decisions. Finally, we observe that in households where the father is educated, mothers are less likely to be involved in education decision but this last result is not significant. In regressions (15) and (16), the dependent variable is a dummy whose value is one when the respondent’s wife is said to be involved in decisions concerning their sons’ education. We observe once again that fathers’ intensity of religious practice is not significantly related to the likelihood of their wives’ involvement in education decisions. Most of the results of regression (13) and (14) also hold for the regression (15) and (16). However, there is no significant generational difference and the age at marriage of women is not significantly related to their participation in decisions. 23 These last results are not directly reported in Table 13. 34 Table 13: Fathers’ religiosity and participation of their wives in decisions concerning their children’s education DEPENDENT VAR. : RELIGION HUSB_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS WIFE_EDUC HUSB_PRIMARY WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant Men (13) EDUC_GIRL (LPM) Men (14) EDUC_GIRL (Probit) 0.00193 (0.0416) -0.147 (0.146) 0.159 (0.161) 0.0123 (0.0295) -0.0216 (0.123) -0.0130 (0.0734) 0.133** (0.0627) -0.0192 (0.110) 0.326* (0.169) -0.0922*** (0.0251) -0.00327 (0.00853) 0.0824*** (0.0286) -0.0874*** (0.0244) 176.1*** (49.00) Yes 92 0.403 -0.0169 (0.154) -0.707 (0.648) 0.578 (0.614) -0.0201 (0.119) -0.0337 (0.408) -0.219 (0.392) 0.527** (0.237) -0.272 (0.476) 1.459** (0.698) -0.371*** (0.124) 0.000208 (0.0386) 0.347*** (0.126) -0.355*** (0.122) 713.5*** (245.0) Yes 92 Average Marg. effects -0.004 -0.137 0.138 -0.004 -0.007 -0.048 0.115 -0.057 0.374 -0.081 0.00005 0.075 -0.077 Regional dummies Observations 92 R-squared Pseudo R-squared 0.367 Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Men (15) EDUC_BOY (LPM) Men (16) EDUC_BOY (Probit) -0.0205 (0.0435) -0.0922 (0.136) 0.0170 (0.140) 0.0241 (0.0213) -0.0275 (0.101) 0.0226 (0.0655) 0.168*** (0.0597) -0.0402 (0.0956) 0.501*** (0.146) 0.0178 (0.0220) 0.00380 (0.00775) -0.0182 (0.0251) 0.0216 (0.0205) -43.11 (41.24) Yes 110 0.319 -0.0731 (0.152) -0.379 (0.531) 0.0185 (0.529) 0.0841 (0.101) -0.107 (0.348) 0.0119 (0.306) 0.589*** (0.204) -0.267 (0.423) 1.828*** (0.580) 0.125 (0.199) 0.0243 (0.0311) -0.123 (0.210) 0.140 (0.197) -282.4 (394.6) Yes 110 Average Marg. effects -0.017 -0.080 0.004 0.019 -0.025 0.003 0.134 -0.059 0.518 0.028 0.005 -0.028 0.032 110 0.293 6.2.2. Models with interaction term Table 14 shows the results based on the estimation of models (2) and (3) where the dependent variable is a dummy whose value is one when the respondent’s wife is said to be involved in the final decisions concerning their daughters’ education (regressions (17) and (18)) and their sons’ education (regressions (19) and (20)). The models are estimated by using the observations for the whole sample of male respondents. In Appendix 5, we estimate the same models by restricting our sample to men who have 35 children who were sufficiently young in late 80’s early 90’s (or not yet born). This period corresponds to the period when religious movements described in Section 2 became very active and started gaining importance. We restrict our sample to families where the youngest child (girl and boy respectively) was no more than 5 years old in 1990. In regressions (17) to (20), no significant relationship emerges between fathers’ intensity of religious practice and the likelihood of their wives’ participation in education decisions, and this is the case whatever the education level of the father (below or above the collège level of education). Actually, from regression (17) and (20), we observe a positive and significant (at 10% level) correlation for educated fathers but the result is not stable, in the sense that it does not hold for both estimation techniques. From the restricted sample, the results presented in appendix 5 reveal a positive and significant correlation (at 5%) between the religiosity of the fathers and the participation of mothers in education decisions concerning their daughters but only for educated fathers. If this result is explained by the influence exercised by religious movements, it seems that they are more able to impact on men through their active presence in the education system24. However, an alternative interpretation could be that educated men read more about their religion; they are better informed and realize that, according to religion, women have an important role to play in the private sphere. Nevertheless, this last result is not confirmed for the decisions concerning the education of sons (the result is sensitive to the estimation technique). Table 14 presents additional results similar to those observed in Table 13: more educated mothers have a higher probability to be involved in education decisions; mothers who exercise an income-generating activity are more likely to participate in education decisions. For the first two regressions concerning the education of daughters, we once again observe that older women are less likely to be involved in decisions; the older a woman is when she marries, the more likely is her participation in education decisions. 24 Al Adl wal Ihsan has a youth section that is active on university campuses. In general, the organization is very active in the education system and counts among its members and sympathizers a lot of teachers. 36 Table 14: Fathers’ religiosity, education and participation of women in education decisions DEPENDENT VAR. : RELIGION HUSB_TO_SCHOOL REL_HUSB_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS WIFE_EDUC HUSB_PRIMARY WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant Men Men (17) (18) EDUC_GIRL EDUC_GIRL (LPM) (Probit) -0.0467 (0.0570) -0.153 (0.148) 0.136* (0.0786) 0.181 (0.156) 0.0125 (0.0299) 0.00526 (0.126) -0.0186 (0.0762) 0.151** (0.0623) -0.0144 (0.108) 0.343** (0.152) -0.0881*** (0.0259) -0.00512 (0.00842) 0.0794*** (0.0291) -0.0848*** (0.0253) 171.0*** (50.90) -0.175 (0.199) -0.823 (0.653) 0.545* (0.306) 0.647 (0.633) -0.0234 (0.123) 0.0339 (0.418) -0.214 (0.400) 0.597** (0.256) -0.306 (0.496) 1.434** (0.634) -0.352*** (0.116) -0.00926 (0.0413) 0.333*** (0.119) -0.343*** (0.114) 689.0*** (229.2) Average Marg. effects -0.037 -0.151 0.092 0.151 -0.005 0.007 -0.045 0.126 -0.062 0.358 -0.074 -0.002 0.070 -0.072 Regional dummies Yes Yes Observations 92 92 92 R-squared 0.423 Pseudo R-squared 0.385 Sign I.T. Pos. Sig. I.T. Not sig. Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Men Men (19) (20) EDUC_BOY EDUC_BOY (LPM) (Probit) -0.0735 (0.0530) -0.123 (0.145) 0.157* (0.0830) 0.0442 (0.135) 0.0275 (0.0217) -0.00638 (0.105) 0.0289 (0.0686) 0.181*** (0.0616) -0.0290 (0.0954) 0.473*** (0.133) 0.0268 (0.0227) 0.00209 (0.00766) -0.0253 (0.0256) 0.0289 (0.0216) -57.90 (43.53) -0.282 (0.197) -0.526 (0.571) 0.669** (0.304) 0.199 (0.527) 0.0987 (0.107) -0.0374 (0.360) 0.0396 (0.312) 0.668*** (0.216) -0.265 (0.447) 1.755*** (0.558) 0.322 (0.267) 0.0115 (0.0330) -0.310 (0.277) 0.328 (0.266) -661.1 (533.2) Yes 110 0.344 Yes 110 Average Marg. effects -0.061 -0.104 0.130 0.045 0.021 -0.008 0.009 0.145 -0.056 0.481 0.070 0.003 -0.068 0.072 110 0.323 Pos. Not sig. 37 7 Conclusion Our results support the idea that the role of religion, in particular through the influence of religious movements, is more nuanced than what widespread views would suggest. On the one hand, religious movements impose a conservative view in some issues, for instance, regarding equality of rights between men and women. In that case, in the Moroccan context, they appear to influence mainly educated people (Chaara, 2014) through the education system. On the other hand, they spread a positive message about the importance of education and, in particular, the education of girls. Moreover, they confer to women an important social role through their action within the household. These views are partly conveyed through their social action. However, it appears that the positive message about the social role of women and the importance of education is mainly diffused by the women section of the organization Al Adl wal Ihsan. The role of the women section could grow in importance thanks to the support of Sheikh Yassine, the founder of the movement. Recently, internal divergence between the current leadership and the former leaders of the women section limits the role of women within the organization and probably in society. How persistent the effects of their past initiatives and actions will be is an open question. This would probably depends on the role of local activists. Those that we interviewed consider that their actions on the ground are still motivated by the same objectives and based on the same philosophy. In this paper, we show that a factor such as religion plays a role in the participation of women in household decisions. We find that a positive and significant correlation exists between the intensity of religious practice of the mothers and their participation in decisions concerning their daughters’ education. We believe that this result can be explained by the influence of a religious movement, Al Adl wal Ihsan. The religious movement may influence simultaneously the religiosity of women and their perceived role within the family. The educational discourse of the movement is focused on the significant social role of women and the importance of their education. Women may therefore feel that their value is enhanced as religion confers upon them a social role through their action within the household. Furthermore, we find that the positive and significant correlation between the intensity of religious practice of the mothers and their participation in decisions concerning their daughters’ education is essentially true for non-educated or poorly educated women. We therefore conclude that a ‘substitution effect’ between education and religion seems to be operating. We thus suspect that religion, through the social interactions that it implies, can play the same role as education regarding consciousness-raising about the importance of children’s education and regarding self-valuation of women with respect to their capacity to play a key role within their household, and in turn in society. From a theoretical point of view, the cooperative bargaining model does not seem to be the most convincing approach in order to understand the link between the religiosity of mothers and their involvement in the decisions concerning the education of their children. In our case, it is not clear that religiosity is linked to an improvement in the fallback position of women. The discourse of the women section of Al Adl wal Ihsan may have a consciousness-raising effect and induce a change in women’s perceptions about their set of possible choices. In a reference-dependent preferences setting, one could argue that their reference point has changed. Those women may have higher aspirations. Women may 38 be attracted by the discourse of the women section because it gives them more value and provides them a justification for playing a more active role in their family and in the public life. In this sense, they may feel empowered. Women may also be attracted by the egalitarian emphasis of Islam as interpreted by the women section (see Binzel & Carvalho (2013) for a similar argument regarding the egalitarian moral system of Islam). The discourse of the women section may also produce a change of identity. According to Akerlof and Kranton (2000), identity is a person’s sense of self that dictates how people think they should behave and how the others should behave. When it comes to women movements, Akerlof and Kranton explain that the movements’ goals include reshaping societal notions of femininity and removing gender associations from tasks both in the home and the workplace. The women section also tries to reshape the perceptions about what a woman can achieve. Finally, we do not find a clear significant correlation between the intensity of the religious practice of the fathers and their wives’ participation in decisions concerning their children’s education. This last result is in accordance with our hypothesis given that the main driver is assumed to be the discourse and actions of the women section of Al Adl wal Ihsan. Moreover, the women section is mainly in a position to influence the women population. 39 Appendix 1: Sample creation for each of the regions where the survey took place. In the region of Casablanca, one third of the sample of 180 people is comprised of rural people and the remaining two thirds of urban people from Casablanca city. Two provinces were chosen: the city of Casablanca for the urban survey and the province of Nouacer for the rural one (Nouacer is the second province of the region in terms of population). There are 16 districts in the city of Casablanca but in order to avoid geographic dispersion, we have decided to keep only 8 of them25. For the socioeconomic background, the type of housing was used as a proxy. In order to establish the list of districts from where to select our sample units, we have produced a table showing the number of primary units (geographical areas that include about 300 households) for each type of housing in each of the districts of Casablanca. The different categories of housing are: deluxe (stratum 1), modern (stratum 2), old medina (stratum 3), economic and social (stratum 4), and slums (stratum 5). The medina is found in only two districts (El Maarif and Sidi Belyout), and we chose to keep both of them to include the residents of this urban environment in the sample. These are typically people who settled in Casablanca for several generations and whose socio-economic level is relatively low. Some districts, Hay Mohammadi and Sidi Moumen, have been selected for the preponderance of middle class people, lower middle and lower. Other districts have been chosen for their diverse population: Mers Sultan, Hay Hassani and Ain Chock. A striking feature of the latter districts is the proximity of villas and slums, which are sometimes adjacent to each other. Finally, the districts of Anfa and El Maarif have been selected because of the concentration of their populations in the first two categories of housing. For the region of Tanger-Tétouan, two provinces have been chosen: the Tetouan province and Chefchaouen. The sample of 180 people was split between the two provinces based on the demographic weight of the target group of people aged more than 18 years. Within each province, the division between urban and rural areas was made according to the importance of each type of residential environment within the province. In each province, three communes with primary rural units26 have been selected following discussions with people from the region in order to take account of the accessibility of the villages and to avoid geographical dispersion. In the province of Chefchaouen, the communes which were selected are: Bab Taza, Derdara and Talambote, and in the Tetouan province: Azla, Ain Lahsen and Zaytoune. For the region of Souss-Massa-Draa, we have opted for a single province: the province of Ouarzazate. The 180 people were divided between urban and rural areas according to the importance of each type of residence within the province. For the urban area, Ouarzazate was the only city considered. For the rural areas, we drew a circle of 80km around the city of Ouarzazate, and decided to restrict our attention to the communes that lie entirely within this circle. The criteria of accessibility and geographical spread again motivated our choice of municipalities. A member of the High Commission for the region was consulted in order to help us. The municipalities selected are: Ait Zineb, Ighrem N'Ougdal, Ouisselsate, Skoura and Toundoute. 25 We do not think it was necessary to survey 180 people in 16 districts since the important thing was that our choice provided sufficient variation in terms of socio-economic status. 26 A primary unit is a geographical area which is composed of 300 households. 40 Appendix 2: The relationship between the participation of mothers in education decisions and the education outcomes of children. In order to study the relationship between the participation of mothers in education decisions and the education outcomes of children, we focus our analysis on the sub-sample of women respondents. We also make a distinction between rural and urban populations. For the urban population, we essentially look at the drop-out rate after primary school. The justification for this choice is that a majority of urban children in our sample have at least achieved primary school education (about 95% of those who stopped going to school at least achieved primary school education; and 98% of those who are still at school and who are more than 12 years old already achieved primary school education). For the rural population, we look at the participation in the formal education system: does a child at least attend primary school? In rural areas, only a minority of our sample goes on beyond primary school: about 14% of those who stopped their formal education obtained a degree higher than the primary certificate. What we observe from our data is that, in urban areas, the drop-out rate after primary school is negatively correlated with mothers’ involvement in decisions regarding their children’s education, and this correlation is significant for daughters but not for sons. We run a Linear Probability Model where the dependent variable DROP is a dummy whose value is 1 when the child dropped-out after the primary school and 0 if she/he continues the formal education after primary. We create clusters for children from the same family. Table 15: Education drop-out in urban areas DEP. VARIABLE: DROP (GIRLS) EDUC_GIRL -0.230* (0.127) 0.00926** (0.00452) 0.00729 (0.0592) 0.104** (0.0386) -0.112 (0.0825) AGE WIFE_EDUC HUSB_EDUC NBR_PIECE EDUC_BOY Constant -0.0615 (0.149) DROP (BOYS) 0.00409 (0.00494) -0.0969 (0.103) 0.121 (0.102) -0.140 (0.124) -0.124 (0.179) 0.205 (0.311) Observations 73 69 R-squared 0.197 0.077 Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 In rural areas, we observe a positive correlation between school enrolment in primary school and the input of mothers in schooling decisions but the correlation is not significant. We run a Linear Probability Model where the dependent variable ATTEND is a dummy whose value is 1 when the child 41 attends primary school. We create clusters for children from the same family. From Table 16, it appears that the education of the father is a decisive factor. We did not introduce the education of the mother in our regression as nearly all of them are illiterate. Table 16: Primary school attendance in rural areas DEP. VARIABLE ATTEND (GIRLS) EDUC_GIRL 0.120 (0.125) 0.201*** (0.0423) 0.0594 (0.156) HUSB_EDUC NBR_PIECE EDUC_BOY Constant 0.263 (0.182) ATTEND (BOYS) 0.184*** (0.0425) -0.0357 (0.161) 0.0969 (0.138) 0.432** (0.168) Observations 94 89 R-squared 0.166 0.098 Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 42 Appendix 3: Robustness checks on a restricted sample of women Table 17: Robustness checks on a restricted sample of women DEP. VARIABLE: RELIGION WIFE_TO_SCHOOL REL_WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant Women Women resticted restricted (6) (7) EDUC_GIRL EDUC_GIRL (LPM) (Probit) 0.197*** (0.0540) 0.403*** (0.117) -0.222* (0.117) 0.131 (0.141) -0.0451 (0.0298) -0.118 (0.105) -0.200*** (0.0574) -0.0483 (0.0452) 0.0721 (0.101) -0.0295 (0.0539) 0.00104 (0.00598) 0.0292 (0.0525) -0.0232 (0.0528) 47.61 (106.0) Yes 93 0.505 0.999*** (0.267) 2.330*** (0.655) -0.836 (0.620) 0.275 (0.558) -0.235* (0.129) -0.137 (0.493) -1.010** (0.484) -0.362 (0.265) 0.295 (0.468) -0.0370 (0.193) 0.0123 (0.0220) 0.0641 (0.199) 0.0130 (0.192) -23.49 (385.2) Yes 93 Average Marg. Effects 0.187 0.472 -0.156 0.053 -0.044 -0.026 -0.189 -0.068 0.056 -0.007 0.002 0.012 0.002 Regional dummies Observations 93 R-squared Pseudo R-squared 0.504 Sign I.T. Neg./Pos. Sig. I.T. Not sig. Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 43 Appendix 4a: Additional results concerning the religiosity of mothers and the participation in decisions concerning their daughters’ education Table 18: Women’s participation in education decisions, religiosity and education level DEP. VARIABLE: EDUC_GIRL (LPM) EDUC_GIRL (Probit) Average Marg. effects RELIGION 0.196*** (0.0507) 0.230 (0.165) -0.147 (0.161) 0.194 (0.121) -0.106 (0.100) -0.0404 (0.0274) -0.187*** (0.0604) 0.491*** (0.119) -0.0244 (0.0404) 0.0671 (0.111) -0.00225 (0.0565) 0.00437 (0.00609) 0.00351 (0.0556) 0.00151 (0.0560) -2.157 (112.4) Yes 105 0.487 1.131*** (0.277) 0.932 (0.637) -0.963* (0.528) 0.715 (0.570) -0.354 (0.456) -0.240** (0.115) -1.060** (0.460) 0.240 WIFE_TO_COLLEGE REL_WIFE_TO_COLLEGE MAR_POST_1993 RURAL NBR_CHILD NBR_ROOMS WIFE_PRIMARY HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant -0.212 (0.210) 0.173 (0.425) 0.0689 (0.213) 0.0313 (0.0224) -0.0402 (0.218) 0.102 (0.212) -202.9 (426.1) Yes 90 0.211 -0.205 0.155 -0.079 -0.051 -0.225 -0.044 0.037 0.015 0.007 -0.008 0.022 Regional dummies Observations 90 R-squared Pseudo R-squared 0.442 Sign I.T. Pos./Neg. Sig. I.T. Not sig. Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 44 Appendix 4b: The religiosity of mothers, the residential area (rural vs. urban) and the participation in decisions concerning their daughters’ education. Table 19: Women’s participation in education decisions, religiosity and residential areas DEP. VARIABLE: EDUC_GIRL (LPM) EDUC_GIRL (Probit) Average Marg. effects RELIGION 0.148* (0.0773) -0.115 (0.114) -0.00880 (0.108) 0.304** (0.122) 0.233* (0.121) -0.0356 (0.0281) -0.173*** (0.0635) -0.0297 (0.0419) 0.0416 (0.107) -0.00256 (0.0575) 0.00423 (0.00615) 0.00446 (0.0566) -0.000263 (0.0571) 1.368 (114.8) Yes 105 0.451 0.640** (0.319) -0.315 (0.400) 0.0946 (0.390) 1.462** (0.588) 0.776 (0.517) -0.168* (0.0946) -0.741** (0.369) -0.168 (0.187) 0.132 (0.403) 0.0553 (0.195) 0.0236 (0.0204) -0.0328 (0.197) 0.0763 (0.194) -151.9 (389.2) Yes 105 0.139 RURAL REL_RURAL WIFE_TO_SCHOOL MAR_POST_1993 NBR_CHILD NBR_ROOMS HUSB_EDUC WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant -0.072 0.022 0.341 0.175 -0.037 -0.161 -0.037 0.029 0.012 0.005 -0.007 0.017 Regional dummies Observations 105 R-squared Pseudo R-squared 0.431 Sign I.T. Pos./Neg. Sig. I.T. Not sig. Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 45 Appendix 5: Robustness checks on a restricted sample of men Table 20: Robustness checks on a restricted sample of men DEPENDENT VAR. : RELIGION HUSB_TO_SCHOOL REL_HUSB_TO_SCHOOL MAR_POST_1993 NBR_CHILD RURAL NBR_ROOMS WIFE_EDUC HUSB_PRIMARY WIFE_EXP WIFE_AGE DIST_AGE WIFE_AGE_MAR MARRIAGE_DATE Constant Regional dummies Observations R-squared Pseudo R-squared Sign I.T. Sig. I.T. Men Men restricted restricted EDUC_GIRL EDUC_GIRL (LPM) (Probit) -0.00716 (0.0550) -0.124 (0.175) 0.152* (0.0833) 0.278 (0.169) -0.0337 (0.0375) 0.151 (0.146) 0.0200 (0.143) 0.173** (0.0717) -0.0677 (0.115) 0.375* (0.200) -0.0698** (0.0286) -0.00695 (0.00895) 0.0665** (0.0318) -0.0857*** (0.0264) 172.3*** (53.07) Yes 75 0.491 -0.139 (0.288) -0.895 (0.950) 1.253*** (0.462) 0.953 (0.722) -0.395* (0.223) 1.425* (0.765) -0.00608 (0.530) 1.138*** (0.371) -0.636 (0.702) 1.804** (0.857) -0.369** (0.145) -0.0887 (0.0545) 0.380** (0.166) -0.482*** (0.165) 966.2*** (332.0) Yes 75 -0.024 -0.133 0.167 0.189 -0.069 0.208 -0.001 0.197 -0.103 0.366 -0.064 -0.015 0.066 -0.084 75 -0.106* (0.0556) -0.0779 (0.145) 0.196** (0.0820) 0.0578 (0.152) 0.0381 (0.0305) 0.0706 (0.110) 0.0619 (0.113) 0.198*** (0.0618) -0.0318 (0.0988) 0.578*** (0.206) 0.0334 (0.0264) -0.00309 (0.00842) -0.0349 (0.0301) 0.0345 (0.0239) -69.29 (48.13) Yes 95 0.328 -0.468** (0.229) -0.343 (0.612) 0.919*** (0.329) 0.249 (0.553) 0.132 (0.128) 0.315 (0.388) 0.169 (0.371) 0.750*** (0.232) -0.328 (0.487) 2.188*** (0.827) 0.363 (0.276) -0.0147 (0.0356) -0.357 (0.288) 0.360 (0.277) -725.8 (555.8) Yes 95 Average Marg. effects -0.102 -0.071 0.180 0.056 0.029 0.066 0.037 0.163 -0.068 0.575 0.079 -0.003 -0.078 0.078 95 0.4891 0.3151 Pos. Pos. Not sig. Not sig. Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Average Marg. effects Men Men restricted restricted EDUC_BOY EDUC_BOY (LPM) (Probit) 46 References Akerlof, G.A., and R. Kranton, 2000. “Economics and Identity”. The Quarterly Journal of Economics, Vol. 115(3): 715-753. Al-Hibri, A.Y., El Habti, R.M., 2006. Islam. In: Browning, D. S., Green, M.C., Witte Jr., J. (Eds). Sex, Marriage, and Family in World Religions: 150-225. Columbia University Press: New York. Becker, G., and N. Tomes, 1976. Child Endowments, and the Quantity and Quality of Children. The Journal of Political Economy, 84(4) : 143-162. Becker, G., 1991. A Treatise on the Family, Harvard University Press. Becker, S.O., and L. Woesmann, 2009. “WasWeber Wrong? A Human Capital Theory od Protestant Economic History”. The Quarterly Journal of Economics, 124(2): 531-596. Bekkaoui, K. and René Larémont, R., 2011. Moroccan Youth Go Sufi. Journal of the Middle East and Africa, 2: 31-46. Behrman, J.R., R.A. Pollak, and P. Taubman, 1982. Parental Preferences and Provision of Progeny. The Journal of Political Economy, 90(1): 52-73. Binzel, C., and J.P. Carvalho, 2013. “Education, Social Mobility, and Religious Movements: A Theory of the Islamic Revival in Egypt”. Working Paper (revise and resubmit at The Economic Journal). Bruce, J., 1989. Homes Divided. World Development, 17(7): 979-991. Bourqia, R., 2005. Les valeurs : Changements et Perspectives. In : 50 Ans de Développement Humain & Perspectives 2025. Botticini, M., and Z. Eckstein, 2007. « From Farmers to Merchants, Conversions, and Diaspora : Human Capital and Jewish History », Journal of the European Economic Association, Vol. 5 (5) :885-926. Cavatorta, F., 2006. Civil society, Islamism and Democratisation; the Case of Morocco. Journal of Modern African Studies, 44(2): 203-222. Cavatorta, F., 2007. Neither Participation nor Revolution – The strategy of the Moroccan Jamiat al-Adl wal-Ihsan. Mediterranean Politics, 12 (3): 381-397. Cavatorta, F. and Dalmasso, E., 2009. Liberal Outcomes through Undemocratic Means: the Reform of the Code de Statut Personnel in Morocco. Journal of Modern African Studies, 47(4): 487-506. Cavatorta, F. and Dalmasso, E., 2010. Reforming the Family Code in Tunisia and Morocco – the Struggle between Religion, Globalisation and Democracy. Totalitarian Movements and Political Religions, 11(2): 213-228. 47 Carvalho, J.P., 2013. “Veiling”. The Quarterly Journal of Economics, 128(1): 337-370. Chaara, I., 2014. “Pro-Women Legal Reform in Morocco. Is Religion an Obstacle?”. CRED Working Paper (Revise and resubmit at EDCC). Dalmasso E., 2008. Family Code in Morocco. State Feminism or Democracy. Unpublished. Duflo, E., 2003. Grandmothers and Granddaughters: Old-age Pensions and Intra-household Allocation in South Africa. The World Bank Economic Review, 17(1): 1-25. Duflo, E. (2012), “Women Empowerment and Economic Development”, Journal of Economic Literature 50(4): 1051-79. Elharras, M., and Serhane, F. 2006. L’application du Code de la Famille: Acquis et Défis. L’association marocaine de lutte contre la violence à l’égard des femmes, UNIFEM and PNUD. Foa, E., and Foa, U., 1980. Resource Theory: Interpersonal Behavior as Exchange. In K. Gergen, M. Greenberg, and R. Willis (Eds.), Social Exchange: Advances in Theory and Research: 74-97. Plenum Press: New York. Geertz, C., 1968. Observer l’islam: changements religieux au Maroc et en Indonésie. Editions la découverte: Paris. Gray, H.D., 2013. Beyond Feminism and Islamism : Gender and Equality in North Africa. I.B. Tauris : London & New York. Handa, S., 2002. « Raising Primary School Enrolment in Developing Countries. The relative Importance of Supply and Demand ». Journal of Development Economics, 69 : 103-128. Hashemi, S. M. and Schuler, S. R., 1993. Defining and Studying Empowerment of Women: A Research Note from Bangladesh. JSI Working Paper 3, JSI Research and Training Institute, Boston. Hoddinott, J., and Haddad, L., 1991. Household Expenditures, Child Anthropometric Status and the Intrahousehold Division of Income: Evidence from the Côte d’Ivoire. Paper 155. Princeton, Woodrow Wilson School - Development Studies Hoddinott, J., and Haddad, L., 1995. Does Female Income Share Influence Household Expenditures? Evidence from Cote D’Ivoire. Oxford Bulletin of Economics and Statistics, 57(1): 77-95. Huisman, J., and J. Smits, 2009. “Effects of Household- and District-level factors on Primary School Enrolment in 30 Developing Countries”. World Development, 37(1): 179-193. Johnson, M., 2000. The View from the Wuro: A Guide to Child Rearing for Fulani Parents. In J. DeLoache and A. Gottlieb (Eds.), A World of Babies: Imagined Childcare Guides for Seven Societies. Cambridge University Press: Cambridge, UK. Kabeer, N., 1999a. The Conditions and Consequences of Choice: Reflections on the Measurement of Women’s Empowerment. UNRISD Discussion Paper, No. 108. 48 Kabeer, N., 1999b. Resources, Agency, Achievements: Reflections on the Measurement of Women's Empowerment. Development and Change, 30: 435-464. Lehrer, 2004. Religion as a Determinant of Economic and Demographic Behavior in the United States. Population and Development Review, 30(4): 707-726. Lundberg, S. J. and Pollak, R., 1996. Bargaining and Distribution in Marriage. The Journal of Economic Perspectives, 10(4): 139-158. Lundberg, S. J. and Pollak, R., 1997 . Bargaining and Distribution in Marriage. In Economics of the Family and Family Policies: 17-32. Eds. I. Persson and C. Jonung. Routledge: London. Lundberg, S. J., Pollak, R. and Wales, T. J., 1997. Do Husbands and Wives Pool their Resources? Evidence from the United Kingdom Child Benefit. The Journal of Human Resources, 32(3): 463 – 480. Malhotra, A., Schuler, S. R. and Boender, C., 2002. Measuring Women’s Empowerment as a Variable in International Development. Background paper prepared for the World Bank Workshop on Poverty and Gender: New Perspectives. Gender and Development Group, World Bank, Washington, DC. Myersson, E., 2014. “Islamic Rule and the Empowerment of the Poor and Pious”. Econometrica, 82(1): 229-269. Naciri, R., 1998. The Women’s Movement and Political Discourse in Morocco. United Nations Research Institute for Social Development, Occasional Paper 8. Norton, E. C., Wang, H., Ai, C., 2004. Computing interaction effects and standard errors in logit and probit models. The Stata Journal, 4 (2): 154-167. Platteau, J.P., forthcoming. Religion, Politics, and Development. Is Islam a Special Problem? Pollack, R., 1994. For Better or Worse: The Roles of Power in Models of Distribution within Marriage. The American Economic Review, 84(2): 148-152. Pruzan-Jorgensen, J. E., 2010. The Islamist Movement in Morocco. Main Actors and Regime Responses. Danish Institute for International Studies (DIIS) Report. Quian, N., 2008. Missing Women and the Price of Tea in China: The Effect of Sex-Specific Earnings on Sex Imbalance. Quarterly Journal of Economics, 123(3): 1251-1285. Quisumbing, A. R. and Maluccio, J. A., 1999. Intrahousehold Allocation and Gender Relations: New Emperical Evidence. The World Bank: Policy research Paper on Gender and Development, Working Paper Series No. 2. Quisumbing, A. R. (Ed), 2003. Household Decisions, Gender and Development. A Synthesis of Recent Research. Ifpri: Washington D.C. 49 Quisumbing, A. R. and Maluccio, J. A., 2003. Resources at marriage and intra-household allocation: Evidence from Bangladesh, Ethiopia, Indonesia, and South Africa. Oxford Bulletin of Economics and Statistics, 65(3): 283-327. Ramírez, A., 2006. Other feminisms? Muslim associations and women’s participation in Morocco. Etnográfica, 10(1):107-119. Samuelson, P.A., 1956. Social Indifference Curves. The Quarterly Journal of Economics, 70(1): 1-22. Sen, A., 1966. Labour Allocation in a Cooperative Enterprise. The Review of Economic Studies, 33(4): 361371. Thomas, D., 1990. Intrahousehold resource allocation: An inferential approach. Journal of Human Resources, 25 (4): 635-664. Willis, M., 2007. Justice and Development or Justice and Spirituality? The Challenge of Morocco’s Nonviolent Islamist Movements. In Maddy Weitzman, B. and Zisenwine, D. (Eds), The Maghrib in the New Century. Identity, Religion, and Politics. University Press of Florida. The World Bank, 2001. Engendering Development: Through Gender Equality in Rights, Resources and Voice. A World Bank Policy research Report. The World Bank and Oxford University Press. 50
© Copyright 2024