The Effect of Remittances on Education in Togo

Togo is not exempt from migration phenomenon in the last few years, as the statistics show that it experiences a galloping migratory flow. This trend in migratory flow leads us to question the possible repercussions on the daily lives of households. This article analyses the effect of migrants’ remittances on the education of recipient households with econometric analysis using a binary logit model with instrumental variable. The estimations results show that remittances from migrants as well as income increase the level of education of recipients in countries. The results have also made it possible to note that access to electricity significantly improves the level of education of the household’s members. Suggestions are made to education decision-makers in the perspective of strengthening the education system in Togo and facilitating the procedures of remittances to Togo.


Introduction
Despite the risks that migration presents nowadays, it is for many people in middle-and low-income countries an opportunity especially through remittances from migrants. These remittances are the way for resident households to smooth their spending from basic needs. In recent years, remittances exceed not only foreign direct investment (FDI) but also export earnings and foreign aid. The purpose of this paper is to analyse the effects of remittances on education in Togo.
In 2000 The Millennium Development Goals (MDGs) recognize in education an important way for individuals to realize their capacity and consequently, priority is given to the completion of primary education. Following these, the Theoretical Economics Letters fourth goal of the Sustainable Development Goals reiterates the importance of education and aims to ensure access for all to equitable, free and quality education through all stages of life, including the elimination of gender and income disparities. It also focuses on the acquisition of fundamental and higher-level skills to live in a sustainable society as well as the construction and improvement of educational infrastructure. Aware of its importance in development, education defined according to Emil Durkheim as "the action exercised by the adult generations on those who are not yet ripe for social life" aims to arouse and develop in the child a certain number of physical, intellectual and mental states that require of him and the political society as a whole and the social environment for which he is particularly destined.
71% of the 62.2 million out-of-school children of primary school age and 59% of the 60.7 million out-of-school children of lower secondary school age in the world in 2016 live in developing countries (World Bank, 2020). Children from the poorest households, who live in rural areas and in countries affected by fragility and conflict, are even more likely to be left behind, as are girls in many countries. Like other developing countries, in the WAEMU area, the primary school completion rate is unsatisfactory over the period 2015(World Bank, 2020. With a completion rate of 89%, the Togolese State has made many efforts in recent years to improve schooling conditions. It is followed by Côte d'Ivoire (71.6%), Burkina Faso (65%), Senegal (57%) and Mali (49.6%).
In Togo, enrolment improved at almost all levels of education between 2011 and 2018. Coverage of pre-school education more than doubled over the same period. In 2018, 34% of Togolese children were pre-schooled compared to only 14% in 2011; while 91% of children complete primary school in 2018 compared to 2011 when the rate was 76% (UNESCO, 2020). The priority for education remains high. In 2017, the total state budget allocated to education was 19%.
However, this budgetary priority is mainly driven by current expenditure, which represents 24.6% of the State's operating budget against only 1.9% of the State's investment budget. Despite the low purchasing power of Togolese, in 2015 households spent almost 77,000 francs per student in school. This expenditure is slightly increasing, more precisely by 18%, compared to its level in 2011 and the poorest are those who suffer the most.
However, worldwide remittances received from migrant workers have more than quadrupled between 2000 and 2011, from CFAF 323.1 billion to CFAF 1353 billion. Relative to GDP, they rose from 1.7% in 2000 to 3.6% in 2011. Over this period, they represented on average 3.7% of the Union's GDP. This proportion varies from 0.9% in Côte d'Ivoire to 8.8% in Niger. It is estimated at 2.9% in Benin, 1.3% in Burkina Faso, 3.9% in Guinea-Bissau, 3.4% in Mali and 7.3% in Senegal. Togo is to date, the country receiving more remittances from migrants relative to GDP and stands at 8.8%.
As in other countries of the West African subregion, the evolution of migrants' remittances has also been notable in recent years in Togo. Over the period 2010 The contribution of this work lies in the fact that survey data are mobilized to address the issue at the micro level unlike previous work on Togo which had addressed the subject at the macroeconomic level.
The rest of the paper is organized as follows. The first section presents previous work on the relationship between remittances and education in developing countries, the second section discusses the methodological approach of the study. The third section presents the results of the empirical analysis and the last section (5) concludes.

Literature Review
The relationship between migrant remittances and education in countries of origin has been addressed by several theoretical and empirical studies. Previous work has shown that remittances that ease liquidity constraints can have a positive impact on human capital formation (Askarov & Duucouliagos, 2020;Ballo & Alphonsine, 2018;Chea & Wongboonsin, 2020;García Nájera & Cuecuecha Mendoza, 2020;Medina & Cardona, 2010;Pickbourn, 2016).
As part of the work on the reasons that motive migrants to remit, the theory of the implicit co-insurance contract implies that the family invests in the education and subsistence of the migrant in their home country and will eventually be able to finance the costs of his future migration. In return, the migrant must support the family by sending part of their income earned in the host country.
These new resources will enable the remaining family to cope with possible negative shocks (illness, unemployment, poor harvest, etc.) and to undertake new projects. In a second step, the migrant finances the costs of possible migration of other family members (Amuedo-Dorantes et al., 2006;Gubert, 2002;Calero et al., 2009;Agarwal & Horowitz, 2002).
These resources constitute a kind of private insurance that covers the socio-economic, health and environmental risks incurred by the migrant's native  Studies by Mansuri (2006) and Boutin (2011) show that these funds significantly reduce child labour of both genders. Others, on the other hand, prove that these funds have no influence on the work of male children (Acosta, 2006) or, on the contrary, influence only that of boys (Yang, 2008) or girls (Calero et al., 2009).
The impact of transfers on reducing labour supply is relatively large for girls and children living in rural areas.
In order to see the impact of parental education on the demand for education and on the supply of child labour following remittances, Hanson and Woodruff (2003) and Gang et al. (2008) reveal that educated parents tend to encourage their children to continue their education, rather than bringing them into the workforce at an early age in order to smooth out parents' income fluctuations. Rapoport and Docquier (2005) and Adams and Cuecuecha (2010) show that, in particular, the effect of parental absence on children's education is a crucial issue. While remittances received from a migrated parent help meet short-term consumption and the long-term physical and human investment needs of

Methodology for Empirical Analysis of Effects on Education
Several estimation methods have been addressed in studies on the relationship between remittances and human capital. In addition, the adoption of an econometric method for estimating these effects has often encountered difficulties. In addition, these estimates most often involve using micro data on individuals to explain and predict their interactions with several alternatives (Amemiya, 1981;Maddala, 1983;Gertler & Glewwe, 1989;Gouriéroux & Monfort, 1989). Other studies considering the relationships between the determinants affecting schooling through remittances have used correlation analysis. This approach is limited by the fact that simple correlations between a few variables can be highly influenced by other variables so that each contains the false effects of the other variables (Cimmyt, 1993). Econometric analyses using the ordinary least squares method have often been used. The problems of heteroscedasticity generated by this approach have highlighted the invalidity of the tests carried out on the basis of these methods (Gouriéroux & Monfort, 1989). Following these successive methods, our empirical analysis focuses on the probability of going to school given the receipt of remittances. To do this, we will use logit estimation and instrumental variable (IV) logit estimation to also take into account the problem of potential endogeneity of migrant remittances.
In this research, in view of the binary choice facing the individual, we estimate the econometric model through a logit regression with intrusive variables. This approach is fundamentally inspired by Raut and Tanaka (2018) where * Y represents the probability of being educated by individuals in a household; i X represents all the explanatory variables of the schooling of individuals; 0 β and β , a vector of parameters to be estimated; i ε , the term random error.  Rapoport, 2006;Gang et al., 2008, etc.). The use of externally sourced remittances from migrants to explain the level of education attained by individuals could lead to potential endogeneity biases. In other words, these remittances can be correlated with the error term in the functional specification of the model or some variables not taken into account by the model can influence the remittances of Togolese migrants. Among other things, remittances can be used as a pretext to finance children's school fees, and our basic equation explains schooling by, among other things, migrant remittances. We then face a two-way causality leading us to resort to the use of instrumental variables. The literature review highlights the distance between the child's home and school as a reason for remittances. To take into account the potential endogeneity bias in migrant transfers and the distance between the place of residence and schools, we will carry out endogeneity tests using the Rivers-Vuong (1988). Two steps are suggested by the latter in order to carry out the tests.
First step: estimate the endogenous variable "migrant remittances" by regressing it over all independent variables (instruments and explanatory variables); Second step: introduce the residue from the regression in the first step into the parent equation explaining education. The nullity of the residue coefficient is qualified thanks to the student test.
• Choice and relevance of instruments The results from our estimates by not taking into account the potential endogeneity bias can be biased if certain exogenous variables are also endogenous, The first category of instruments takes into account potential shocks impacting household income (Gang et al., 2008). The instruments suggested by McKenzie and Rapoport (2006) fall into the second category of instruments.
One example is the history of migration networks that can continue to influence the migration decisions of household members. In the context of Togo, the migration history has probably not influenced the costs of transfers, which will in turn affect the transfer of migrants.
In this second category of instruments, we also use the instrumental variable hold receiving transfers. We think that in the context of our study, the family environment could influence parents' decision to send their children to school Bouoiyour et al. (2018). This instrument has been tested to control its influence on migrant remittances. In addition to this instrument, exceptional family events (weddings, traditional festivals, funerals, Mecca-hajj, etc.) will allow the family to gather, which requires an increase in the flow of migrants' remittances.  (Table 1).

Results and Interpretation of the Analysis of the Effect of Remittances on Education
• Descriptive analysis. Our results showed that the probability of studying increases by 0.070 when (1) Education 1.000 (2) Transfers −0.203* 1.000 ( with the results of the work of Yang (2008). This result is explained by the fact that in the African context in general and Togolese in particular, through remittances from migrants, young boys are led to take over from the family as leaders and therefore should make greater use of the transferred funds for their education.
In contrast, the probability of studying decreases in convexly with the age of household members receiving transfers. As for the distance between the house and the nearest primary school as well as the nearest secondary school, the estimates are not significant. As for the household wealth level index, the estimates are only significant for the poor, the rich and the richer. These increase the probability of studying by 0.18 respectively; 0.316; 0.326. These results can be answered by the high cost of education except for the free primary school introduced since 2008 in Togo.
According to the regression results, marital status is a key factor in the education of individuals. Our estimates show that being married decreases with a significance threshold of 1% the probability of dating recipient members of 0.28.
This result may be due to the fact that in developing countries such as Togo, being married further complicates access to education since once married, men and women must perform tasks that give them various trajectories other than increasing their level of education. The wealth of a household is correlated with the level of education of the individual. Indeed, if the wealth quintile remains positive, the probability that the individual will increase his or her level of education also increases. In particular, this probability increases by 0.019 with the level of wealth of the household.
The results of the estimates also show us a correlation between the possession of durable goods and education. Considering electricity, the probability of education increases by 0.022 with electricity possession (Raut & Tanaka, 2018). This finding remains consistent with the findings of de Haas (2006) findings that international migrant households invest more than others in housing, agriculture and other family businesses.
In addition, in order to specify the effect of migrants' remittances on education taking into account gender, the results show that remittances significantly attenuate male education. With regard to women's education, remittances have a negative but not significant effect. These results are counterintuitive and do not corroborate the results of the work of some authors (Calero et al., 2009;Hanson & Woodruff, 2003;Mansuri, 2006). This could be explained by the fact that in low-income countries such as Togo, the collection of remittances is much more directed towards micro-entrepreneurial investment expenditures and therefore will not be able to contribute to the accumulation of human capital (Woodruff & Zenteno, 2001) (Table 3 & Table 4).

Robustness of estimates of the effect of remittances on education
Finally to ensure the robustness of the results obtained through logit regression, in this section, we have used estimation by instrumental variables and the   The validity test of the instruments allowed us to select two of them, including the number of children of all adult members and the expenses for family events.
The investment in real estate being rejected as an instrument must either be re- have less concern for housing and therefore allows him to focus his efforts on education.
Through probit regression, funds received from migrants, gender of the head of household, wealth quintile, homeownership and electricity possession increase the probability of increasing the level of education significantly except for homeownership which remains insignificant.
With regard to household size, the results of the estimates show that it increases the probability of lengthening the level of education. Results remain counterintuitive given that in Africa and particularly in Togo the size of the household is a limiting factor in schooling given the high number of people in households and the difficulties in taking care of the education expenses of the latter.
As we have several instruments, it is likely to face a problem of overidentification. We then have to go through an overidentification test. We must verify that the residue of the reestimated equation is not correlated with our instruments.
This is equivalent to testing the overall nullity of the regression coefficients by including the residuals. The null hypothesis (H0) will be accepted if the calculated statistic (N*R 2 ) has a value lower than the theoretical value (Table 5).

Conclusion
This article analyzed the effect of migrant remittances on the education of recipient  this study was that the remittances that households receive from migrants contribute to a significant improvement in their level of education. Our results also showed that these transfers do not impact the education of a girl and a boy in the same way. However, a comparative analysis of the effect of these transfers shows that remittances seem to encourage boys to step back from education, unlike girls. Similarly, our results suggest that living a long distance from primary school reduces a child's incentive to attend in order to increase their human capital. The results of this research provide a new perspective on migrant remittances with more recent data to complement the qualitative aspect of recipients not taken into account by the work already done.
However, without questioning the results of our research, we recognize cer-