Effects of Remittance Flows on the School Attainment of Household Members Left Behind

Using household survey data for Albania, this paper compare decision-making about human capital investment in remittancereceiving households and non-remittance-receiving households. The Cox proportional hazard model is used to capture the effects of remittances. The crucial assumption in the model is that the effect of the covariates is proportional over the entire base line. The vector of covariates includes information such as; children’s demographic characteristics, parental schooling, household income and the presence of remittances. Household incomes are considered separately from remittances in order to identify whether income from remittances have the same effect as other types of household non-labour income in the decision to invest in more years of schooling for household members. The estimation of the survival function indicates that receiving remittances from household members working abroad increases the hazard of leaving school after the end of secondary education. The negative relationship between education and remittances is more evident for males living in rural areas.


Introduction
Remittance flows, funds received from migrants working abroad, have become enormously important as a source of income in many developing countries (Giuliano and Ruiz-Arranz, 2005;Mundaca, 2009).Remittances can increase not only physical investments but also can expand human capital accumulation, such as health and education.This paper examines the contribution of migration and remittances on human capital investment using cross-sectional data for Albania.The question is whether household members who live in households with external migrants complete more grades of school at a given age than do other children.Sending migrants abroad may generate remittances that raise household income and allow children to complete more schooling, but it may also disrupt family life in a manner that hinders children's scholastic progress.
Since the fall of the Berlin Wall in 1989, migration from Eastern Europe including the Balkans has increased sharply.According to World Bank estimates, in 2005 Albania was the fourth-ranked country in the world in terms of share of emigrants in relation to population, with 27.5 percent of Albanians living abroad, mostly in Greece and Italy.In 2006, remittances were 13 percent of Albania's GDP, exceeding by more than three times both the FDI and the total amount of development aid received by the country.There are reasons to believe that this extraordinary volume of migration and remittances is likely to have had extensive consequences for the Albanian economy.In their review of the existing literature, for example Rapoport and Docquier (2006) argue that remittances have short-run economic benefits, and may have long-run implications for households' labor supply decisions, education opportunities for offspring and investment in household businesses.
The specific relationship between remittances and education achivement has been explored by focusing on the impact of remittances on the education of household members.Of particular concern for the process of economic development is how migration affects household investments in human capital.The empirical findings about this impact are ambiguous.The extra income from remittances may allow children to delay entering the workeforce in order to further their studies.However, the departure of wage earners from a household may disrupt family life.Migration may reduce the number of adult role models in the home and may increase the demand on older children to assist in running and supporting the household.These effects may make it more difficult for children to remain in school.Thus, migration may increase or decrease household investment in children's schooling (Hanson & Woodruff, 2003;Hanson, 2007).
By smoothing liquidity constraints, remittances should raise investment in human capital of household members.While schools may be state supported, in developing countries students are expected to pay for their books, transport, and supplies imposing additional costs to the household.Remittances by lifting up the budget constraint of poor households may increase the investment in human capital by increasing the maximum years of schooling the household can afford.
A few studies have examined the potential forward linkage between remittances and education.These studies provide a starting point for analysing the potential growth effects of remittances through human capital formation .Hanson and Woodruff (2003) employ a reduced-form approach to estimate the effect of remittances on children's schooling and health in Mexico.They find a positive relationship between child education and having a family member abroad and argue that remittances are the mechanism that links the two.They control for the potential endogeneity of having a migrant family member by using historical state migration rates and household characteristics.
Cox-Edwards and Ureta (2003) reach similar conclusions about the impact of remittances in El Salvador.They estimate survival functions to show that remittances significantly contribute to reduce the hazard of school leaving in El Salvador using cross-sectional data collected in 1997.Their findings report rural/urban differences in the impact magnitude effect.One problem with this study is that it does not address potential sample selectivity issues and endogeneity of remittances.Thus, the findings could be tested using alternative econometric techniques.
In a subsequent study, Acosta (2006) examine the effects of remittances on age-based demographic sub-groups, using survey data for El Salvador.The evidence in this study suggest that girls and boys between 11 and 14 years of age seem to benefit from remittances in terms of higher enrolment rates, but this positive impact does not apply to boys between 15 and 17 years of age.The study concludes that remittances are used as a substitute for child labor, a practice usually associated with higher school dropout rates (Acosta, 2006;Acosta, et al., 2007).Elbadawy and Roushdy (2010) investigate a sample of Egyptian children who live in remittance-receiving households.They find than remittances have a positive effect on school attendance of boys rather than girls.The effect is particularly evident among those boys who are close to university enrollment age.For girls the positive effect is true only for those between 15 and 17 years of age.Yang (2008) finds that remittances in the Philippines cause only minor improvements in school attendance for children between 10 and 16 years of age.There is a much greater impact in the attendance for boys between 17 and 21 years of age, an increase of remittances amounting to 10 percent of household income leads to a 10 percent increase in these boys' attendance.The literature on the impact of remittances on education suggests that this relationship may vary by context, gender, and age.

Remittances in Albania
There is an ongoing debate on the role of migration in the development of countries with high rates of migration such as Albania.The literature focuses on how remittances are spent by remittance-receiving households and their implications in terms of costs and benefits for the local economy.Researchers disagree over the extent to which remittances-receiving households use these financial resources productively.Some findings suggest the use of remittances mainly for shortterm consumption needs rather than for long-term investments.The extent to which remittances contribute to local development depends upon the household context, circumstances and the way decisions are made.Studies of households on Albania have focused mainly on the decision to work and do not consider how remittances impact human capital investment.Little is so far known about the extent that remittances effect socioeconomic outcomes such as school attainment.
There have been few empirical studies of the impact of remittances on the labor market issues in Albania.Utilizing the Albanian Living Standards Measurement Survey (ALSMS) for 1996, Konica and Filer ( 2009) suggest that remittances have a negative effect on female labor market participation due to higher incomes from household members working abroad.This finding is consistent with studies conducted in other countries (Rodriguez & Tiongson, 2001;Amuendo-Dorantes & Pozo, 2006).In the Albanian case, however, Konica and Filer (2009) find that neither the existence of emigrants in the household nor the amount of remittances received has an effect on male labor force participation.
Using data from the 2005 ALSMS, Kilic, et al., (2009) measured the impact of the past migration experience of Albanian households on non-farm business ownership using instrumental variables regression techniques.These results indicate that households' past migration experiences exert a positive impact on the probability of owning a non-farm business.Using the same dataset, Dermendzhieva (2009) investigated the effect of migration and remittances on labor market participation, obtaining large and negative coefficients for young female and older male household members.
Remittances are not only invested in physical capital, but also productively invested in human capital accumulation, such as education.The Becker (1991) model of investment in education states that families take into consideration their education rate of return and its cost in order to choose the optimal education level for their children; in this model a range

529
of factors may influence the educational attainment.If families have financial constraints the level of schooling for their children will be lower than optimal.By relaxing the household's liquidity constraints, remittances from abroad may facilitate investments in education.

Data
Multipurpose household surveys provide data that can be used to investigate patterns of household investment in human capital or labor market participation.The ALSMS was carried out annually by the Albanian National Institute of Statistics with the technical assistance of the World Bank, from 2002 to 2005.In order to investigate the impact of remittances on the educational attainment, this study draws on data from the 2005 ALSMS.
Remittances are defined as money received by households in the past 12 months prior to the survey in cash or inkind from someone who does not live in the household e.g.child, head of the household or other relatives abroad.The cash remittances are reported in Lek and one issue is related to the monetary measurement error.In-kind remittances are not used in the empirical analysis, because they are mainly consumption items.
As part of the education module, the survey collected data on the highest grade for all household members and current enrolment status for each member aged 6 to 24.My study is restricted to a sample of 4,511 household members aged 13 to 26.Household members participate in the survey for a time and, thereafter are no longer observed (right censoring).The analysis time in the dataset is the highest grade completed by the subjects and the failure time is recorder by the enrolment status; if enrolled the individual is considered "right censored", if not is considered "failed".

Methodology
To examine the impact of remittances upon human capital investment decisions, my question is whether girls and boys living in the remittance-receiving households complete more years of schooling than those children living in nonremittance-receiving households.In order to characterize the remittance-receiving households, it is important to determine their position in the income distribution.There are important costs associated to the act of migrating, so it is possible that migrants do not come from the lowest quintiles of the income distribution and therefore that remittances do not flow toward the poorest..This situation is more likely to occur in rural areas where poverty rate is higher and in general the educational achievements are lower than those in the urban areas.
The proportional hazard model (Cox, 1972;Cox-Edwards & Ureta, 2003) will be used to capture the effect of remittances on human capital investment.The model is popular because of its elegance and computational feasibility: Where: is the baseline hazard of leaving school after grade t, which is left unspecified and is estimated is a vector of covariates such as; child's characteristics, location, parental schooling, household income and presence of remittances.
is the vector of parameters which has been estimated.The crucial assumption in the proportional hazard model is that the effect of the covariates is proportional over the entire base line.The baseline hazard, , is given no particular parameterization and can be left un-estimated.There is no assumption about the shape of the hazard over time; it could be constant, increasing, decreasing or anything else. 1

Regression Results
Table 1 shows the estimates of the determinants of the hazard of leaving school.In contrast to findings from other countries comparable to Albania, mothers' education is not statistically significant in this dataset.The regression results do not provide support for the effect of mother's education in the school attainment of household members; we fail to 1 If wrong, such assumptions could produce misleading results about β (Cleves et.al. 2008).The advantage of the semiparametric Cox model is that we do not need to make assumption about . The cost is loss in efficiency, knowing the functional form of For the case of remittance-receiving household, the hazard ratio is 1.196, which means that the presence of remittances increases the hazard of leaving school if compared to non-remittance receiving household by 19.6 percent.In other words remittances positively affect the probability of dropping out of school.The difference in survival estimates is higher for the rural areas and more evident after the secondary school or the twelve grade of schooling (Figure 1).
There is an important difference in school leaving behavior between males and females.The coefficient exp( )=.892 for the indicator variable Female is statistically significant, and means that being a female decreases the hazard of leaving school by 10.8 percent .One possible explanation for this finding is that male students have better labor market opportunities than do female students, and thus face higher opportunity costs of attending school.When estimating the survival function only for household members living in remittance-receiving households (Figure 2), the hazard of leaving school is higher for males than females.This finding suggests that if there is a positive effect of remittances on education, it holds only for females.Incomes net from remittances have a positive effect in lowering the hazard of dropping school by around 14.1 percent for every unit increase in income level.This result is consistent with the previous literature on education, and provides evidence on the fact that income matters for the choice of schooling levels of household members.For those household members living in urban areas the hazard of leaving school after a given grade is 11.3 percent lower than for those living in rural areas.To shed more light on the initial controversial result of the negative effect of remittances on education I included more control variables in my regression.Mothers' education still remain statistically non significant in this dataset.Often time the effect of father's schooling is different from that of mother's schooling.This is why I chose to test for differences in paternal and maternal schooling levels.The results provide support for the effect of father's education in the school attainment of household members, because according to the p-value 0.04 we can reject the null hypothesis that the father's education has no effect on the hazard of leaving school.
Members living in households where fathers are more educated face a hazard of leaving school 9 percent lower than others.Less educated parents exhibit lower preferences for educational over other alternative expenditures.The difference in paternal and maternal education effect can be related with the patriarchal nature of the Albanian society where fathers may have more influence in the decisions related to household member's education.
For the case of remittance-receiving household, the hazard ratio is 1.213, which means that the presence of remittances increases the hazard of leaving school if compared to non-remittance receiving household by 21.3 percent.In other words remittances positively affect the probability of dropping out of school.
Household characteristics such as the number of children below 5 years of age positively affect the hazard of leaving school, increasing it by 7.3 percent for those members living in households with more children below 5 years of age.The finding is not surprising.The number of younger brothers and sisters increases the hazard of living school.
The income net from remittances has a positive effect in lowering the hazard of leaving school by 16 percent.This suggests that household's budget constraints play a significant role in the choice of the levels of schooling for the household members. 2  Urban/rural differences are no longer statistically significant after the inclusion of other control variables.Area of residence also is not significant in the dataset. 3However the distance to school increases the probability of leaving school after a given grade.The coefficient exp ( ) = 1.29 is statistically significant and means that as the distance from school increases the hazard of leaving school increases by 29.1 percent.
There is a significant difference in the hazard of leaving school for male and female.According to the results being a female lowers the hazard of leaving school for all levels by 10.3 percent.This result may be related to the delayed entering in the workforce for female or maybe men have a higher probability to leave school for later migrating.
The negative effect of remittances on education in the Albanian case is also confirmed after the estimation of a standard Probit, with the enrolment status in school during the time of the survey as the dependent variable.Table 2, shows a small negative impact of remittances on school attendance.The first column considers only the impact of remittances on the school attendance while the second adds the control variables that describe child and household 2 This result is consistent with previous literature.Hill and Duncan (1987) measure household income and report that a 10 percent increase in household income is associated with an increase in school attainment of less than 1 percent.In comparison to remittances we can conclude that income net from remittances can be a better proxy for "permanent income" than remittances and this may explain the difference in estimated effects. 3 The sampling frame for the survey was stratified into four regions: coastal, central, mountain, and the capital Tirana The marginal effect of receiving remittances suggests that members in remittance-receiving households have 8.3 percent less probability of being enrolled in a given grade if compared with members that live in non remittance-receiving households.After controlling for children and household characteristics the Probit result in column 2 shows still a negative impact for remittances on school attendance, but it is smaller 1.1 percent and not statistically significant.Apart from mother's education the findings related to other characteristics are in line with those from the Cox proportional model.

Discussion and Conclusions
This paper examines the relationship between remittances and school attainment with cross-sectional data using the Cox proportional hazard model.The estimation of the survival function indicates that receiving remittances from household members working abroad increases the hazard of leaving school after the end of secondary education.The negative relationship between education and remittance receiving status is robust to using a Probit regression and after adding other control variables.
The negative relationship between education and remittances is more evident for males living in rural areas.One explanation of this result is that remittances fuel further migration rather than further education at home.It is possible that remittance-receiving household members will later migrate not valuing the local education enrolment.A second explanation is that the absence of a household head may lead to less parental control in the household, thus negatively affecting children's school enrolment.This absence may result in the need for children to undertake household work in substitution of members living abroad.Thirdly, higher drop rates can be related to the way remittances are used.Households with migrants may invest remittances in higher return activities that provide alternative avenues for skill formation and higher returns than staying in school.In sum, although parents' migration may helps children to benefit economically in the short-run, the lack of parental care may cause problems that affect long term children's welfare.The absence of a parent may change the decision-making process within the household, and children may end up dropping out of the school.
Females have a higher probability of staying longer in school than do males, both in urban and rural areas.This result holds even when the survival function is estimated for remittance-receiving households only.Another explanation is that, in patriarchal contexts, men are generally the frontrunners of international migration (Stecklov, et al., 2010).Surprisingly mother's education does not affect the hazard of leaving school.However after adding the father's education in the regression, the results provide support for the effect of father's education in the school attainment of household members, because we can reject the null hypothesis that the father's education has no effect on the hazard of leaving school.This difference in paternal and maternal effect might be characteristic of the cultural context of Albania, when men may have more influence in the decisions related to household member's education. 4onsistent with similar findings in other countries is the differential result in rural and urban areas.Schools are more available in urban areas, living conditions and the organization of the economy in the way it affects the cost of attending school is more favorable in urban rather than in rural.The answer to the relationship between remittances and schooling has fundamental policy implications.If remittances have an impact on schooling, we can conclude that there is support for Government funds to be transferred to poor households in order to enhance children's education.Another policy implication derived from a positive effect of remittances on education levels is the reduction in the remittances transaction costs.
It is important to further investigate what children are doing instead of going to school and find if living in migrant households increases the probability for boys migrating themselves at school ages.Migration networks and information may increase the probability of migration in the future, making migration and education substitutes of one another.The immediate substitution effect by which the opportunity cost of staying longer at school increases due to the higher earning probability abroad induces children from migrant households to leave school earlier.Because the reduction in years of schooling might be a conscious choice of remittance-receiving household members in the face of better opportunities abroad it should be less of a policy concern than a restriction on schooling due to budget constraints.One possible policy solution can be related with the improvement of post compulsory school system especially in the rural areas where the drop out of school rate is higher.

Figure 1 :
Figure 1: Estimated survival functions in urban and rural areas, with and without remittances

Figure 2 :
Figure 2: Estimated survival functions, by gender, in urban and rural areas

Table 1 :
hypothesis (p-value 0.21) that the mother's education has no effect on the hazard of leaving school.Estimates of the determinants of the hazard of leaving school

Table 2 :
Estimates of the determinants of children's school attendance