Educational mismatches and earnings in the New Zealand labour market

ABSTRACT

Mismatch of educational skills in the labour market is an emerging topic in the field of labour economics, partly due to its link to labour productivity. In this paper, we examine the incidence of educational mismatch and its earnings effects in the New Zealand labour market. Using micro-data drawn from the Household Labour Force Survey and the New Zealand Income Supplement (HLFS/NZIS) for the years 2004–2007, we find a noteworthy incidence of both over- and under-education in New Zealand – approximately half of workers in some occupations were well matched to their jobs. We also find that earnings returns to required years of education exceeded the returns with over- and under-education, with a greater earnings penalty associated with under-education. We test hypotheses on three alternative models of educational mismatch. Our results imply that public spending on education is not wasteful. However, better allocation of workers to jobs may be needed to increase overall labour productivity.

KEYWORDS:

• Over-education
• under-education
• educational mismatch
• wage effects

JEL CLASSIFICATIONS:

• J24
• J31

Acknowledgments

The authors wish to thank three anonymous reviewers of this journal for insightful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Surpassed only by Belgium (6.6%), Denmark (8.8%), Iceland (7.6%), and Norway (8.8%).

2. The New Zealand government has taken a proactive approach in addressing productivity in the economy with the establishment of the New Zealand Productivity Commission in 2011 (http://www.productivity.govt.nz/about-us/the-commission). The mandate of this commission is to provide advice to the government on increasing productivity in a way that directly supports the overall well-being of New Zealanders.

3. Because the highly skilled immigration policy leads to a migrant workforce, that is, on average, more educated, the focus of this paper is on the effects of over-education and to a lesser extent, on under-education.

4. See, for example, http://www.stats.govt.nz/browse_for_stats/population/mythbusters/brain-drain.aspx

5. Empirical analysis utilising the dummy variable specification is available from the authors.

6. CURF data employed reports the occupation of employment for each respondent based on the one-digit New Zealand Standard Classification of Occupations 1999 (NZSCO99). Wen and Maani (2014) have examined the effect of one-digit versus two-digit occupational classification, based on Australian data. They find that interestingly the incidence of over-education is not sensitive to this choice of the classification.

7. The results from this study allow us to approximate job skills using years of formal education, as actual skill matrices were not available from the HLFS/NZIS survey.

8. There are some minor differences across localities, but these are not as notable as the above. For example, there are slightly lower rates of over-education, and higher under-education in Wellington and Canterbury (22.7% and 21.8% are, respectively, classified as over-educated in the two localities (compared to the national average of 23.4%); and 39.4% and 37.0% are, respectively, classified as under-educated (compared to the national average of 37.1%).

9. For an example of such information, we look towards New Zealand job search websites with emphasis on professionals: http://www.gradconnection.co.nz/, http://www.seek.co.nz/, http://www.auckland.ac.nz/uoa/auckland-careerhub

10. The mode method adopted, as discussed, is the preferred method across studies. Results based on the mean method are also available from the authors. These results are qualitatively compatible with the results based on the mode method reported, in finding higher wage returns to required education, compared to each year of over- and under-education. The three hypotheses (H1, H2, and H3) are also rejected across the models, providing support for the assignment model.

11. We also examined clustering by locality, which resulted in qualitatively similar results. The standard errors became slightly larger, but coefficient significance levels did not change. We find that adjustment by industry is more appropriate in this setting, by allowing error correlation due to specific industry characteristics.

12. Although skilled workers were underutilised, the prior paragraph notes that workers in New Zealand who had more than the required education were still slightly more productive than their peers who were adequately educated. However, the marginal compensation that they receive from more years of education was not as high as the marginal compensation from a year of adequate education.

13. In this study, we are unable to differentiate between an educational mismatch and a skill mismatch due to the data-set not containing approximations of skills needed for particular jobs. Mavromaras, McGuiness, and King Fok. (2009), for example, focus on skill mismatches.

14. Occupation group one (10–13 years of required education (mode)) consists of the occupations: Legislators, Administrators, and Managers; Professionals, Technicians, and Associate Professionals; Service and Sales Workers; and Trade Workers (See Table 3). Group two (15–18 years of required education (mode)) consists of Clerks; Agriculture and Fisheries Workers; Plant and Machine Operators and Assemblers; and Elementary Occupations.