Introduction It is plausible that working

Introduction It is plausible that working at an early age has a strong negative impact on future individual earnings and, consequently, on national income. This hypothesis seems to be widely accepted by academics and policymakers. However, there is little empirical evidence supporting it. There is extensive literature on the causes of child labor. However, according to Kassouf (2007) and Emerson and Souza (2011), only a few studies have been carried out on its hazards. According to the literature, the main ones are on its negative impacts on health (see Kassouf et al., 2001; Odonnell et al., 2004) and education (see Heady, 2003), both of which are determinants of human capital and have a positive effect on labor earnings in adult life (henceforth earnings). Taking the Yoram (1967) model seriously, both level of schooling and entry age in the labor market are endogenously and simultaneously decided. In order to claim causality from child labor to adult earnings, some treatment must be applied. Unfortunately, relevant and valid instrumental variables in order to control for the potential endogeneity of the level of schooling and age started to work in the earnings equations are seldom available in the databases commonly used in empirical studies of the consequences of child labor. Emerson and Souza (2011) estimated a model using instrumental variables to obtain more robust estimates than those previously obtained by Emerson and Souza (2003) and Ilahi et al. (2001). It should be noted that a set of instrumental variables that do not vary among individuals, only among Brazilian states, were used. Moreover, since information about the age at which the individuals started to work and about their family background is available only for heads of orexin antagonist or spouses, the sample used by the authors excludes all other individuals with a different status in the family. In sum, due to the specificities of the data set used by the authors to build the instruments used in the estimations, amino acids is hardly possible to apply the same kind of econometric modeling to come up with more evidences about the harms of child labor today. In sum, Emerson and Souza (2011) concluded that child labor is associated with lower adult earnings. Specifically, they observed a quadratic relationship between the age at which an individual started working and his or her earnings in adult life. The impact of entering the labor market is negative for young children (in the sample used) and that negative effects turn positive between 12 and 14 years old. Although a plausible explanation was given by the authors for the maximum point of earnings and the age at which one began to work, in our opinion the estimated maximum point is significantly low even taking into account that child labor occurred several decades before. The authors suggest that the results found may be relevant to other countries with economies structurally similar to the Brazilian one in the 1950s, 1960s, and 1970s. They also claim that in Brazil today it might be better to delay the age at which one starts to work, considering that the environment changed quite significantly. In this context, an important social and economic question emerges: what is the best age to enter the Brazilian labor market today? We found evidence suggesting that not controlling for the number of hours worked might yield a downward bias in the estimate of the maximum point for the age at which one started to work. We found that the effect of the age at which one begins to work on labor income remains negative much beyond the age of 14 found by Emerson and Souza (2011). This issue and others will be discussed in greater detail in Section 4. This paper is structured as follows. Section 2 presents a brief description of the data and samples utilized. Section 3 describes the methodological procedures – empirical models and estimator. Results are discussed in Section 4. Section 5 concludes the study.
Data and samples