Schooling as a positional good: the Brazilian metropolitan regions in recent decades

Abstract

In this article, I aim to determine whether, in view of the trend of educational expansion in recent decades in Brazil, the role played by schooling in the labor market of Brazilian major metropolitan regions has changed, becoming positional rather than absolute. To this end, I analyze the influence of schooling on people’s income and occupational status. Based on data from the National Household Sample Survey (PNAD-IBGE) for years 1995, 2005 and 2015, I compare ordinary least squares models that use absolute and positional measures of schooling. The results show that the explanatory power is greater for positional measures than for absolute ones, and that this advantage has increased over recent decades. As a result, it is argued that although educational expansion possibly makes the chances of access to a certain absolute schooling level less unequal, it also ends up undermining the opportunity structure related to it.

Keywords:

• educational expansion
• income inequalities
• occupational status
• positional schooling
• social stratification

Acknowledgements

The author would like to thank the National Council for Scientific and Technological Development (CNPq), for providing grants for this research.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 I consider the metropolitan areas as defined by the National Household Sample Survey (PNAD-IBGE).

2 For this reason, even the less optimistic voices about the effects of schooling expansion should recognize its benefits as a consumer good.

3 The conclusions by Rotman, Shavit, and Shalev (2016) for Israel are similar.

4 I removed from the analysis the households located in rural areas, which have a residual proportion within the metropolitan regions.

5 Unfortunately, the available data do not allow us to work with variables over system or type of educational institution.

6 Cases of people with income equal to zero were excluded from the analysis.

7 I converted its scale to a range from 16 (lower status) to 90 (higher status).

8 These two variables (years in current occupation and working time) are not used in the models in which occupational status is the dependent variable.

9 Since absolute and positional schooling present a high correlation, we chose to compare similar models, instead of inserting them within the same model.

10 All models were run in Stata/IC v.15.

Additional information

Funding

This work was supported by the National Council for Scientific and Technological Development (CNPq) [grant number 404641/2016-4]