Ethnic background and the transition from vocational education to work: a multi‐level analysis of the differences in labour market outcomes

Abstract

This article focuses on the impact of ethnic background on employment and earnings among people with a vocational education in Norway. I differentiate between three different groups: majority, first‐generation non‐Westerners and second‐generation non‐Westerners. Panel data from several public register databases of the entire population of graduates from Norwegian vocational schools from 1997 to 2001 (N = 54,453) are used. The aim of this article is to uncover ethnic inequality in the labour market among vocational graduates and try to understand these differences. By using multi‐level linear and logistic regression analysis methods it is shown that the majority have significantly higher earnings than the first‐ and second‐generation non‐western ethnic minorities; however, the earning gap is closing with time since graduation. The analyses also show that the ethnic differences in earnings are minor when we compare individuals who are fully employed. Results for employment show that the first‐generation non‐western minorities have a lower probability for full employment as compared to the majority, but the second‐generation minorities experience only a minor disadvantage. Finally, I found that the disparity in annual earnings between the first‐generation non‐western minorities and the majority increases with increasing non‐western ethnic concentration and unemployment rate in local labour markets.

Keywords:

• Ethnic minorities
• Vocational education
• Labour market
• Employment
• Earnings

Acknowledgements

The research reported in this article is part of the project ‘Educational Careers: Attainment, Qualification, and Transition to work’, which is supported by the Norwegian Research Council. The author thanks Arne Mastekaasa and Silje N. Fekjær for their helpful comments, and Øyvind Nicolay Wiborg for assistance in preparing the data for analysis. Errors and omissions remain the responsibility of the author.

Notes

1. The level of education is the same for all persons in the data set. All the graduates have achieved vocational certificates from upper secondary schools.

2. The use of data in this article and the record linkage of individual records are approved by the Data Inspectorate, an independent administrative body under the Norwegian Ministry of Labour and Government Administration.

3. The individuals in this sample have different graduation years, and the annual income is measured in different years for different individuals. The consumer price index is used to make a fair comparison of the annual income in the period 1997–2001, using the following formula: January 1998: 98, 9 (index) January 2001: 107, 6 (index). Based on this formula, the value can be calculated as follows: 107, 6 / 98, 9 = 1,087.

4. I could include calendar years as a variance term, but the number of calendar years included is so low that I decided to take this source of variation into account by means of dummy variables.

5. Bayesian information criterion (BIC) provides a basis for a model selection criterion. The BIC can be written as follows: −2 log L + P log (N), where L = likelihood, P = number of parameters, N = number of observations (Weakliem, 2004).

6. The interaction between ethnic background and time since graduation was positive, but small and mainly insignificant. Additionally the BIC value was higher in the model were this interaction was included (result not shown).