ABSTRACT
We model schooling as a sequential process and examine why some children are left behind. We focus on the factors that explain selection at early stages of the education system. Our findings for Uruguay suggest that long-term factors, such as parental background or ethnicity matter across all education stages while the effect of short-term factors, such as family income, wear out as individuals progress in the education system, suggesting a severe selection process at early stages.
Acknowledgments
We are grateful to Albrecht Glitz, Cristina López-Mayan, Raúl Ramos and Roxana Gutiérrez-Romero for insightful comments. We appreciate the comments and suggestions received at the 2015 LACEA Meeting, and seminars at the Economic Institute of Universidad de la República, Regional Quantitative Analysis Group (AQR) at Universitat de Barcelona, and at the Departament d’Economia Aplicada, Universitat Autònoma de Barcelona.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Primary school was made compulsory in 1877, universal primary schooling was achieved in the 1950s (Manacorda Citation2012).
2 Previous studies for Uruguay show that lower high school drop-out is strongly associated with parental educational background, household income, adolescents' socio-emotional factors like taste for maths and science in primary school, and repetition (Failache et al. Citation2018; de Melo and Machado Citation2015). Motivation and labour motives are pointed out by adolescents as the main issues inducing dropping-out from the educational system (Cardozo, Citation2010). Ethnicity has also been found to shape educational attainment in Uruguay (Porzecanski Citation2008).
3 Enforcement of compulsory schooling laws is lax, as in many developing countries (Manacorda Citation2012). That is why we observe children dropping out from lower high school.
4 By education level, shares are 78.3% for lower high school, 71.8% for upper high school, 80% for university, and 78% for non-university high education.
5 Since 2007 the UdelaR is making big efforts in terms of territorial decentralization to give greater opportunities to students living outside the capital city. Likewise, some private universities start locating in different regions of the country.
6 87% of the population aged 20–29 lives in cities larger than 5,000 inhabitants, according to the 2011 Census.
7 Afro-descendance is captured in the ECH through the following question: ‘Do you believe you have … (black or afro, Asian, white, native, other) descent?’. The respondent can choose more than one option of racial descent. For this study, individuals reporting having black or afro descent are classified as afro-descendants. Non-afro-descendants are all individuals reporting not having afro-descent (thus, including whites, Asian, native or other). Almost 90% declares only white descent, while less than 5% declares having native or other descent.
8 This technique ensures consistent estimators (Rosenman et al. Citation2010).
9 Countries with a larger proportion of retained students have a lower performance in PISA test (OECD Citation2011; Ikeda and García Citation2014). Also, Hill (Citation2014) employs repetition to define low-achievers.
10 Note that cognitive ability is likely to be influenced by child’s environment, such as parental education, issue that is controlled for in the analysis.
11 Other risky behaviours used in the literature, such as having had intercourse or emancipation before 15 were found not statistically significant for schooling progression.
12 Checchi (Citation2006) provides an overview of the influence of supply of education and education financing on education attainment.
13 13% of the population aged between 12 and 29 declared having a first job (for at least three months) before 15 years old in the ENAJ. Likewise, 24% of secondary school drop outs aged less than 15 declare "because started to work" as the main reason for leaving education.
14 We use the cmp command in Stata. See Roodman (Citation2010).
15 Table A.1 in the Appendix shows that not accounting for selection overestimates the effects of the key variables on education attainment for girls and boys separately.
16 See Gao and Postiglione (Citation2015) for a recent survey.