https://orcid.org/0000-0003-2510-1492
![Sample our Education journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5932/TOC-Subject-Sample-2021-Education.jpg"})
ABSTRACT
This paper analyses the prevalence of educational mismatch and its effects on wages in Brazil using a large employer-employee dataset. I find that half of the Brazilian labour market is mismatched, with similar proportions of over- and undereducated. Overeducated (undereducated) workers earn significantly lower (higher) than their co-workers who hold a well-matched job, and the penalty for overeducation is the same as the premium for undereducation. Moreover, the overeducation penalty is about half of the premium for going to university. Further, given the symmetry of the mismatch correcting it yields small effects on aggregate wages.
Acknowledgements
I would like to thank Thomas Gall, Michael Vlassopoulos, Corrado Giulietti, Matthias Parey, Martin Foureaux Koppensteiner, Michela Vecchi, Brendon McConnell, Fernanda Estevan, Hector Calvo Pardo, as well as conference participants at the 2018 Royal Economic Society Annual Conference, and University of Southampton internal seminars for valuable comments and suggestions on earlier drafts of this paper. I also acknowledge the financial support of the Economic and Social Research Council (ESRC). Any errors are my own.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data that support the findings of this study are available from the Brazilian Minister of Labour and Employment – Ministério do Trabalho (MTE). Restrictions apply to the availability of these data, which were used under license for this study.
Notes
1 The authors report that on average, 26% of employees are undereducated and 30% are overeducated.
2 For more details on the expansion and quality of higher education in Brazil see Schwartzman (Citation2004).
3 Administrative records, population censuses and other household surveys use the enumerative classification for statistics purposes. Activities that need information about the content of the job uses the descriptive classification, such as educational activities in firms and unions, immigration services. The latter provides a detailed description of the activities performed at the job, the training requirements and professional experience, and working conditions.
4 For example, the occupational family Economist includes economic analysts, agroindustrial economist, financial economist, industrial economist, public sector economist, environmental economist, regional and urban economist.
5 The authors compared five ways of measure overeducation: job analysis, worker-assessment of the required level to do the job, worker-assessment of the required level to get the job, the mean educational level of realized matches and the modal educational level of realized matches. They concluded that for all variables analysed (wages, job satisfaction, mobility and training) job analysis method was the best among all.
6 The educational system in Brazil is structured as follows: primary education is compulsory, and it is divided into two, basic education I, which goes from 1st to 5th grade and basic education II from 6th to 9th grade. After completing primary education, students start the non-compulsory secondary education or high school. After that, students can enter higher education.
7 As attained education and required education are categorical variables, I use dummy variables to determine whether one is overeducated, well-matched and undereducated in order to avoid measurement error. Translating educational categories into years of schooling is not a straightforward task, for instance, higher education may vary from 4 to 6 years depending on the subject.
8 Identification of the econometric analysis requires sufficient variation of the education level and/or job for the same individual across time.
9 I find that among those who changed their mismatch status, on average 31.74% only improved their education, 52.21% only changed occupation and 16.05% changed both educational attainment and occupation. Similarly, using the National Household Sample Survey (Pesquisa Nacional por Amostra de Domicílios, PNAD) data, Reis (Citation2017) find that 28% of individuals changed their mismatch status and among those, one-third are related to improvements in education and two-thirds did not changed their educational attainment.
10 Another nationwide longitudinal dataset is the Continuous National Household Sample Survey (Pesquisa Nacional por Amostra de Domicílios Contínua, PNAD – Contínua). This dataset was introduced in 2011 and although covers the whole country, is much shorter than RAIS.
11 Table A1 shows the results without interaction terms and the results for each interaction term individually.
12 Note that the category ‘young’ is fixed and consists of workers who are up to 40 years old in 2006. Similarly, ‘elder’ consists of workers who are older than 40 years old at the beginning of the sample.
13 The Hausman specification test rejects the null hypothesis that there is no correlation between individual error and explanatory variables. Therefore, the random effects model is rejected against the fixed effects model for this specification.
14 For more information on the deferred compensation scheme, see Lazear (Citation1979) and Lazear and Moore (Citation1984).
15 The cut off age was picked at 40 years old because it is expected that by this age workers that started their careers as overeducated had time to find a better match job. The fact that younger workers are more likely to be overeducated than their older colleagues is consistent with the search theory, career mobility theory and the view that extra education may compensate for lack of experience.
16 The human capital theory considers that human capital and earnings are proportional to the worker's productivity on the job (Rumberger Citation1987).
17 Using the coarser education categories defined in 5.1.2 the results are similar, changing workers’ education decreases aggregate wages by −0.29% and switching workers’ to a well-matched job increases aggregate wages by 0.27%.
18 Hsieh et al. (Citation2019) using a Roy model of occupational choice for the United States from 1960 to 2008 find that reductions in barriers to occupational choice (for minorities) can explain 15–20% of the increase in aggregate output per worker.