Submitted by Reginaldo Soares de Freitas (reginaldo.freitas@ufv.br) on 2015-11-10T14:45:17Z No. of bitstreams: 1 texto completo.pdf: 854831 bytes, checksum: abec93f80f725277bee92eb7ce175d15 (MD5) / Made available in DSpace on 2015-11-10T14:45:17Z (GMT). No. of bitstreams: 1 texto completo.pdf: 854831 bytes, checksum: abec93f80f725277bee92eb7ce175d15 (MD5) Previous issue date: 2014-02-17 / Fundação de Amparo à Pesquisa do Estado de Minas Gerais / A educação é considerada como o mecanismo central de desenvolvimento social e crescimento econômico, além de ser o meio mais eficaz de superação às desigualdades sociais e regionais no país. Por essa razão, na última década observou-se no Brasil um esforço crescente para que os investimentos em educação sejam suficientes para que as deficiências seculares desse setor possam ser superadas. Apesar disso, são escassos os estudos que determinem o ganho, mensurado pelos retornos salariais, desse investimento nos anos recentes para o Brasil. Mais que isso, poucos estudos consideraram que, sendo o Brasil um país em desenvolvimento, esse retorno é melhor mensurado a partir de uma análise longitudinal. O presente estudo propôs então a formação de uma base de dados com esse perfil, que capte o efeito da educação sobre a renda salarial ao longo do tempo. A partir dos dados da Pesquisa Nacional por Amostra de Domicílios (PNAD), é construída uma amostra de pseudo-painel, que agrupa os dados de cross-sections repetidas. O período considerado vai de 1999 à 2011, e se divide em intervalos bianuais. Essa base de dados foi desenvolvida seguindo os trabalhos de Deaton (1985), Moffitt (1993), Collado (1997), Mckenzie (2004) e Verbeck e Vella (2005). As coortes, que são a unidade de análise nesse estudo, foram determinadas por gênero, raça, região de residência e período de nascimento. A hipótese feita foi de que o retorno salarial à educação quando medida em uma amostra longitudinal é menor do que aquele medido para dados de cross-section. Partindo de um modelo de Efeitos Fixos para dados em pseudo-painel, e contrapondo seus resultados àqueles obtidos pelo método de MQO, essa hipótese foi verificada. Os retornos para a escolaridade foram definidos em três formatos distintos: escolaridade contínua, por grupo de escolaridade e por diploma. Além disso, esses retornos foram diferenciados por gênero e raça. O retorno médio estimado para a educação, em sua medida contínua, foi de 0,06%. Já os resultados por grupos educacionais, mostrou que o retorno por ano adicional de educação não é homogêneo entre os ciclos educacionais. No grupo com 5 à 8 anos de escolaridade o retorno à renda de um ano adicional de educação é de 0,18%, já no ciclo de 11 anos ou mais de escolaridade esse mesmo aumento de escolaridade eleva a renda em 0,83%. O retorno por diploma mostrou-se persistente em remunerar melhor os diplomas de maior escolaridade relativa. A obtenção do título de graduação representa, na amostra considerada, um ganho salarial de 1,42% em relação ao diploma do ensino fundamental de quatro anos. / Education is regarded as the central mechanism of social development and economic growth, in addition to being the most effective way of overcoming social and regional inequalities in the country. For this reason, in the last decade has been observed in Brazil a growing effort done in investments in education sufficient to overcome secular deficiencies of this sector. Despite this, studies that determine the gain, measured by salary, investment returns in recent years for Brazil are scarce. More than that, few studies considered that, being the Brazil a developing country, this return is better measured from a longitudinal analysis. This study proposed the formation of a database with this profile, which captures the effect of education on wage income over time. From the data of the Pesquisa Nacional por Amostra de Domicílios (PNAD), is constructed a sample of pseudo-panel, which groups data from repeated cross-sections. The period considered going from 1999 to 2011, and is divided into twice-yearly intervals. This database was developed following the work of Deaton (1985), Moffitt (1993), Collado (1997), Mckenzie (2004) and Verbeeck and Vella (2005). The cohorts, that are the unit of analysis in this study, were determined by the gender, race, region of residence and period of birth. The hypothesis made was that the return on salary for education if measured in a longitudinal sample is smaller than that measured for cross-section data. Starting with a Fixed Effects model for data on pseudo-painel, and opposed its results to those obtained by the MQO method, this hypothesis was verified. The returns to schooling were set in three different formats: continuous education, group of years of schooling and diploma. Moreover, these returns have been differentiated by gender and race. The estimated average return for education, in its continuous measure, was 0.06%. Furthermore, the results by educational groups, showed that the return for additional year of education is not homogeneous between educational cycles. In the group with 5 to 8 years of schooling the income return of an additional year of education is 0.18%, additionally, in cycle of 11 years or more of schooling that same increase in schooling raises the rent in 0.83%. The return by diploma was persistent in remunerate better higher school diplomas. The graduation title represents, in the sample considered, a wage gain of 1.42% in relation to the diploma of basic education of four years.

Submitted by James Paiva (jspaiva@uol.com.br) on 2013-03-07T05:40:16Z No. of bitstreams: 1 JAMES_PAIVA.pdf: 2054713 bytes, checksum: 3e9c5bed3b6bee0e396aacde2c0510ce (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2013-03-07T12:37:35Z (GMT) No. of bitstreams: 1 JAMES_PAIVA.pdf: 2054713 bytes, checksum: 3e9c5bed3b6bee0e396aacde2c0510ce (MD5) / Made available in DSpace on 2013-03-07T12:50:35Z (GMT). No. of bitstreams: 1 JAMES_PAIVA.pdf: 2054713 bytes, checksum: 3e9c5bed3b6bee0e396aacde2c0510ce (MD5) Previous issue date: 2013-02-05 / The hypothesis that child labor impacts future income generation negatively, for it harms the formal acquisition of education, is widely accepted by the existing literature on this issue. However, some researchers agree that labor might be beneficial to teenagers once they can develop skills, acquire job experience, or even help them to afford their own education acquisition. Thus, the main goal of this study is to assess if there is an age which the negative impact of the early access to the labor market over income and the conclusion of high school, during the adulthood of Brazilian people, becomes positive. To do so, PNADs (Pesquisas Nacionais de Amostra de Domicílios), a National Census of Household Samples, issues 1992 to 2011, were utilized plus the employment of an econometric technique called pseudo-panel. For this analysis, generations of people born between 1982 and 1991 were observed from the ages 10 to 17 (child labor) and from the ages 20 to 29 (conclusion of high-school & income). The results show that starting at the age of 15, the negative effect of an early access to the labor market over income, between ages 20-29, becomes positive. As per high-school, it was observed that accessing the labor market before the age 15 diminishes the probability for an individual to conclude high school before the age 21. From this age on, labor does not have a negative impact anymore. The second goal of this study is to assess how much of the reduction of the child labor occurrence in Brazil for the past years is due to changes of economic and demographic characteristics of children and families. For these analyses PNADs - National Census of Household Samples, issues 2003 to 2011, were employed plus the methodology of decomposition that divides the variation of child labor into 2 components: (a) changing of the probability of children with the same characteristics (intragroup) to start working – access labor market & (b) changing of the distribution of characteristics (intergroups). The results show that the reduction of the child labor occurrence is due, mainly, to changes on the probabilities. In general terms, the occurrence of child labor took place, more significantly, among individuals of the ages 15 to 17 & household heads with less education. Besides the characteristics mentioned, the reduction between non-white individuals was also significant among individuals from 4-member families. The results show that the reduction of child labor took place, mainly, among children and teenagers from non-white and poor families.