Brain Gain or Brain Waste? Horizontal,Vertical,and Full Job-Education Mismatch and Wage Progression among Skilled Immigrant Men in Canada

Abstract

This study examines the incidence and wage effects of vertical, horizontal, and full job-education mismatch for high skilled immigrant and native-born men over a six-year period, using a Canadian longitudinal dataset. Immigrants (particularly racial minorities immigrants) are more likely to be fully mismatched than white native-born Canadians. Full mismatch lowers initial wages, especially for racial minority immigrants. Full mismatch accelerates immigrants' wage growth slightly over time, but this is not enough to narrow the immigrant wage gap over the six-year survey period. The results highlight the importance of disaggregating the different types of job-education mismatch experienced by immigrants.

Introduction

Over the past several decades, many immigrant-receiving countries have focused their immigration policy on attracting highly educated and skilled immigrants. They have done so in response to a perceived shortage of skilled workers and the general sense that highly educated immigrants will better integrate into the host country in the long run than less educated ones. As part of this trend, in the 1990s and 2000s, Canada revised its points-based immigration system to attract more skilled immigrants (Hou and Picot 2016). Despite the focus on educated immigrants who in theory should get good jobs, the initial earnings of new arrivals in Canada declined significantly over the 1990s and remained stable but did not improve between 2000 and 2010 (Hou and Picot 2016).

Researchers seeking to explain the declining outcomes of newer cohorts of immigrants have pointed to the increasing proportion of immigrants arriving from developing countries, macroeconomic conditions disadvantaging all new labor market entrants, and discounting of foreign human capital by employers (Aydemir and Skuterud 2005; Picot and Sweetman 2005; Ferrer and Riddell 2008; Green and Worswick 2012). The latter factor seems especially salient; immigrant workers often have trouble transferring foreign qualifications and skills to the host country’s labor market (Friedberg 2000; Warman, Sweetman, and Goldmann 2015). Many highly skilled immigrants are forced to work in jobs unrelated to their educational background and/or below their skill level and at much lower wages, creating the much discussed job-education mismatch.

The present study adds to the literature by looking at the well-known immigrant wage disadvantage through the lens of job-education mismatch, proposing and testing two models. First, we examine the likelihood of job-education mismatch for racial minority and white highly skilled immigrants relative to white native-born Canadians. Second, we explore the relationship between job-education mismatch and wages over time, testing whether this relationship differs for racial minority and white immigrants relative to white Canadian-born workers. Our study is among the first to distinguish between vertical mismatch (overeducation), horizontal mismatch (relatedness between educational field and job), and full mismatch (both horizontal and vertical mismatch) and examine the effects of each on immigrant workers’ wages using longitudinal data.

The distinction between different types of mismatch is important since each represents different (albeit related) aspects of immigrants’ labor market disadvantage and therefore has unique policy implications. Some highly skilled immigrants, for example, may be able to find work in their field but at a lower level than their education would otherwise predict (e.g., a university-educated engineer working as an engineering technologist). Others may be unable to find work related to their field of study, forcing them to change occupations altogether (e.g., a nurse working as a sales associate). Both possibilities have obvious implications for wages.¹ Our findings provide a better understanding of how the various types of job-education mismatch affect immigrants’ labor market integration.

Vertical and Horizontal Job-Education Mismatch

A large body of research has examined the match between workers’ education and the education required in their job (e.g., Freeman 1976; Duncan and Hoffman 1981; Rumberger 1987; Hartog and Oosterbeek 1988; Verdugo and Verdugo 1989). Most studies measure this concept in terms of the relationship between years of schooling required for a job and years of schooling completed (e.g., Rumberger 1987; McGoldrick and Robst 1996). Workers with a level of education higher than required are considered “overeducated,” while workers with a level of education lower than required are considered “undereducated.” Under- or overeducated workers are considered to be “vertically mismatched” (Mahuteau et al. 2014).

Some studies measure this concept using self-assessment questions that ask workers about their jobs’ educational requirements (e.g., Duncan and Hoffman 1981; Hartog and Oosterbeek 1988; Verhaest and Omey 2006; Galasi 2008). This method, however, is very subjective, and workers may overstate a job’s requirements (Hartog 2000). Another method uses standardized occupational classifications, such as O*Net codes in the United States and National Occupational Classifications (NOC) codes in Canada. These types of measures, however, are not updated every year and may not be accurate given the high costs of thorough and regular updating. In addition, an occupation’s required education is reported on a scale of one to four, not in years or detailed educational levels (Hartog 2000).

The third common method to measure vertical mismatch is the “realized matches” procedure developed by Clogg and Shockey (1984). In this method, years of education completed by the respondent are compared with the mean or mode years of education of all workers in the same occupation (see Verdugo and Verdugo 1989; Kiker, Santos, and de Oliveira 1997; Madamba and De Jong 1997; Quinn and Rubb 2006; Tsai 2010). Workers are defined as being over/undereducated if their completed level of schooling is at least one standard deviation from the mean or mode of others in their occupation. The disadvantage to this method is that it is driven by demand and supply forces and does not necessarily reflect job requirements. It also ignores variations in required schooling across jobs within a given occupation (Leuven and Oosterbeek 2011), producing results with lower rates of mismatch relative to job analysis or subjective (self-assessment) methods.

Around 30 percent of workers in the global labor market are overeducated, according to a meta-analysis by Leuven and Oosterbeek (2011). Overeducated workers earn relatively lower wages than workers with education commensurate with their jobs. A survey of the literature by Hartog (2000) concludes that the returns to overeducation are about half to two-thirds of the returns to the required level of schooling. More recent studies confirm these findings in a variety of contexts (Rubb 2003; Büchel and Mertens 2004; Frenette 2004; Linsley 2005; McGuinness 2006; Dolton and Silles 2008; Chevalier and Lindley 2009).

A form of job-education mismatch receiving less attention in the literature is the relatedness of a worker’s field of study to his or her occupation. “Horizontal mismatch” (Mahuteau et al. 2014) gauges the underuse of education-specific skills (rather than general human capital), something likely to have significant negative consequences for workers, employers, and society as a whole. Horizontal mismatch has been measured using the job analysis method or subjective self-assessment questions. The job analysis method uses standardized occupational classifications such as O*Net and NOC to compare the educational fields common in a particular occupation to the respondent’s field of study (see Wolbers 2003). Drawbacks of this method are that the standardized occupational codes are not updated regularly and information is provided at an aggregate level of the occupation. In the subjective self-assessment method, respondents are asked how closely their educational field is related to the work they do. A potential advantage of this approach is that employees’ field of study is directly compared with the content of their jobs. A potential disadvantage is that employees’ perceptions of horizontal match are by definition subject to self-report bias.

In one of the first studies of horizontal mismatch, Robst (2007) found that about 20 percent of US college graduates reported working in a job not related to their field of study, a situation associated with lower annual income than for those working in a job related to their educational field. According to this study, the income penalty for horizontal mismatch was larger than the penalty for being overeducated. Other studies from the United States, Canada, and Sweden find that employees who work in jobs unrelated to their educational fields not only earn lower wages but also have lower job satisfaction and higher rates of voluntary and involuntary turnover (Wolbers 2003; Bender and Heywood 2009; Nordin, Persson, and Rooth 2010; Boudarbat and Chernoff 2012).

A handful of studies have simultaneously examined both vertical and horizontal job-education mismatch. Verhaest, Sellami, and van der Velden (2017), for instance, look at horizontal and vertical mismatch across 18 countries; they find horizontal mismatch is lower in countries with stronger employment protection and selective educational programs while vertical mismatch is explained largely by labor market imbalances. Using Australian data, Mahuteau et al. (2014) show that for men, horizontal mismatch by itself is not associated with lower wages but that vertical mismatch is. For women, however, horizontal mismatch reduces earnings significantly. The largest wage penalties for both men and women occur when horizontal mismatch occurs jointly with vertical mismatch. In other words, disadvantage is greatest for those working in jobs that both are unrelated to their field of study and require lower levels of education.

Job-Education Mismatch among Immigrant Workers

For immigrants, job-education mismatch is cited as a major source of labor market disadvantage (for a review, see Piracha and Vadean 2012).² Host country employers who are unsure of the true value and applicability of the foreign education and work experience of immigrant applicants may prefer to hire those who have more education than the job requires. However, as immigrants gain local human capital and employers become more familiar with their abilities, immigrants should move up to better jobs where they are less likely to be overeducated (Chiswick and Miller 2009). The devaluation of foreign human capital could also be the result of employers’ racial and cultural biases, in which case, we would not expect overeducation among immigrants to decline over time (Esses, Dietz, and Bhardwaj 2006; Guo 2009).

Previous studies find that immigrants’ rates of mismatch vary with level of education, years since migration, source country, years of foreign work experience, and the type of education, namely, whether vocational or general (Lindley and Lenton 2006; Chiswick and Miller 2009; Dahlstedt 2011). Not only are immigrants more likely to be mismatched than native-born workers, but the wage penalty of job-education mismatch is also greater for immigrants (Lindley and Lenton 2006; Kler 2007; Nielsen 2007; Sanroma, Ramos, and Simon 2008; Wald and Fang 2008; Chiswick and Miller 2009; Sharaf 2013; Joona, Datta Gupta, and Wadensjö 2014).

All studies of immigrant job-education mismatch discussed thus far have focused on vertical mismatch (overeducation), but in practice, most mismatches involve both vertical and horizontal mismatch (see Reitz 2001). For example, new immigrants who are unable to find work related to their educational background are more likely to accept low-level “survival” jobs. In this scenario, the job is both unrelated to the immigrants’ previous education (horizontal mismatch) and below their education and skill level (vertical mismatch). Aydede and Dar (2016) undertook one of the only studies focusing on horizontal mismatch among immigrants in Canada. Using the 2006 census, they found that immigrant workers were significantly more likely to be mismatched relative to their native-born counterparts, but the wage improvement for becoming “properly matched” was not significant. They conclude that it is not occupational mismatch per se that disadvantages immigrant workers but the poor quality of pre-migration human capital.

In one of the few studies to look at both types of mismatch, Nieto, Matano, and Ramos (2013) analyze rates of vertical and horizontal mismatch among immigrants and native-born workers in the European Union (EU) countries, disaggregating immigrants into those from the EU and from non-EU source countries. Immigrants, they found, are more likely to be overeducated than native-born workers, and this effect is higher for immigrants from non-EU countries than those from other EU countries. However, with time in the host country, all immigrants assimilate into the labor market and are more likely to find work matching their years of education. In terms of horizontal mismatch, only immigrants from non-EU countries have a higher likelihood of horizontal mismatch than native-born workers. Moreover, the probability of horizontal mismatch does not decrease with time in the host country. This study does not consider the wage effects of vertical and horizontal mismatch, but its findings suggest that immigrants’ experiences of vertical and horizontal mismatch follow different patterns. Analyzing only vertical mismatch (overeducation) thus overlooks an important aspect of the immigrant labor market experience.

Conceptual Framework and Hypotheses

In this study, we examine rates of vertical, horizontal, and full mismatch among highly skilled racial minority and white immigrant men relative to highly skilled white native-born Canadian men and consider how these mismatches change over time. We also investigate the effect of each type of mismatch (vertical, horizontal, full) on wages over time. We hypothesize the following. First, consistent with previous studies (e.g., Wald and Fang 2008; Chiswick and Miller 2009; Joona, Datta Gupta, and Wadensjö 2014), immigrants will be more likely to experience job-education mismatch than their native-born counterparts. The imperfect transferability of foreign human capital will cause immigrant workers to have greater difficulty finding work commensurate with their level of education and/or their field of study. Second, racial minority immigrants will be more likely to face full (horizontal and vertical) mismatch than white immigrants. Nearly all analyses of immigrants’ labor market assimilation in Canada find immigrants of non-European origins are more disadvantaged than those of European origins.³ Third, years in the Canadian labor market will decrease both white and racial minority immigrants’ rates of vertical and full mismatch but not horizontal mismatch. Over time, overeducated immigrant workers will better understand the Canadian labor market and gain local human and social capital and employers will become more familiar with their abilities and credentials. Therefore, overeducated immigrants should conquer vertical mismatch and move into higher level jobs that match their education level. On the other hand, horizontal mismatch may not diminish with time in Canada; in fact, it may even increase. Immigrants who are unable to find work at their level of education within their field of study may retrain and change career paths.

Fourth, we expect that horizontal job-education mismatch will be associated with greater wage disadvantage than vertical mismatch and that full mismatch will be associated with the greatest wage disadvantage. Vertical mismatch preserves at least some of the specific human capital gained through formal educational qualifications and pre-migration work experience. Thus, vertically mismatched immigrants may be better able to maintain their skills, adapt their expertise to suit the host labor market, and signal their value to employers over time. Workers experiencing horizontal mismatch are unlikely to utilize any of the specific human capital they gained through schooling and previous experience and are essentially “starting over,” which would have a more detrimental effect on wages. The most disadvantaged immigrants would be those who are fully mismatched, that is, working in unrelated and lower skilled jobs.

Fifth, based on the career mobility theory of job-education mismatch, proposed by Sicherman and Galor (1991), we predict that all forms of mismatch will result in faster wage growth over time. From the career mobility perspective, job-education mismatch may be seen as the result of workers deliberately choosing to accept jobs for which they are overqualified so they can receive specific or general on-the-job training to be upwardly mobile in their careers. This may be particularly important for immigrant workers who initially accept jobs for which they are mismatched as a strategy to improve their language skills, gain local human capital, and adjust to the host society in an effort to move up into more appropriate jobs. Finally, we expect that controlling for job-education mismatch should result in a significant reduction in the immigrant/native-born wage gap as the inability to find work commensurate with qualifications is a major contributing factor in immigrants’ labor market disadvantage (Reitz 2011).

Data

We use longitudinal data from the Survey of Labour Income Dynamics (SLID) collected by Statistics Canada. We use three waves of SLID data: Waves 3, 4, and 5 (1999–2004, 2002–2007, 2005–2010). We restrict our analysis to these waves because the question about the match between educational field and occupation was added in Wave 3. SLID participants are drawn from the sampling frame used for the monthly Labour Force Survey (LFS), a national sample of 15,000 households.⁴ Our sample is restricted to males aged 25 to 64 (working age) who have completed a postsecondary degree/diploma and report some form of employment in at least two years of the panel. We include only postsecondary-educated individuals as we are interested in overeducation and field of study–occupation mismatch, and these concepts are not salient for those with less than college education. We exclude all individuals who are unemployed or not in the labor force. Since women experience very different career trajectories relative to men (see Blau, Ferber, and Winkler 2002; Fuller 2008), it is important to separate the analyses by gender. We focus on males here but have conducted the same analyses for females with similar results, albeit with some gender-specific differences. To explore these issues adequately, we will present the results for females in future research.

To increase the sample size of high-skilled racial minority and white immigrant men, we pool data across the three SLID waves. We combine the panels by labeling the first year in each wave (1999 in Wave 3, 2002 in Wave 4, and 2005 in Wave 5) as year zero and the final year of each wave (2004 in Wave 3, 2007 in Wave 4, and 2010 in Wave 5) as year five, with other years similarly labeled. Our final sample size in year zero is 8,703. This includes 7,545 white native-born Canadians, 631 white immigrants, and 527 racial minority immigrants. Our sample has a six-year attrition rate of 19.5 percent. The SLID sample attrition has been found to be nonrandom. Respondents with lower employment attachment and from low-income populations are more likely to drop out of the survey (Boudarbat and Grenon 2013). We apply the SLID longitudinal weight, produced by Statistics Canada, to all our analyses to adjust for attrition and nonresponse.

Likelihood of Job-Education Mismatch

For our first outcome variable, likelihood of job-education mismatch, we examine individual i’s likelihood of vertical, horizontal, and full mismatch in year y. Therefore, we run three separate models: (1) likelihood of horizontal mismatch, (2) likelihood of vertical mismatch, and (3) likelihood of full mismatch. The three categories are mutually exclusive. Horizontal mismatch is measured using a question in the SLID that asks how closely the respondent’s job is related to his educational field. Possible responses are “1” (“closely related”), “2” (“somewhat related”), and “3” (“not at all related”). Those who indicate that their job is not at all related to their education are considered to be horizontally mismatched. We chose to use the subjective self-assessment method for measuring horizontal mismatch because of the numerous disadvantages of the job analysis method (see Somers et al. 2016) and the availability of the self-assessment question in our data set.⁵ No questions in the SLID ask for self-assessments of vertical mismatch or overeducation. Therefore, we construct variables using three other established methods: job analysis, realized matches using the mode, and realized matches using the mean. We test our models using each method of measurement; all produce substantively identical conclusions. In this paper, we present the models based on the conservative realized matches procedure, using the mean, as developed by Clogg and Shockey (1984). The results of the job analysis and realized matches method are available from the authors on request.

We first categorize each respondent’s occupation, using NOC codes to the three-digit level; these reflect internally homogeneous categories based on educational attainment. Next, we calculate the mean years of education of all workers in each occupational code. A respondent is considered to be vertically mismatched if his years of education is greater than the mean education required for a given occupation plus one standard deviation. Respondents who are both horizontally and vertically mismatched are considered fully mismatched. In addition to modeling the likelihood of each type of mismatch at the beginning of the study period, we also examine how the likelihood of each type of job-education mismatch changes over time.

The key explanatory variable in the first analysis is immigrant and racial minority status. This variable is constructed from two separate questions in the SLID. The first asks respondents to identify their country of birth. Those born in Canada are set as native-born Canadians, and those born outside Canada are considered immigrants. The second question asks respondents to self-identify as either white or visible minority, based on the definition provided by the Canadian Employment Equity Act.⁶ Those who identify as visible minority are then asked to indicate their specific minority group. Because of small sample sizes for some specific minority groups, racial minority status is simply dichotomized, coded “1” for “racial minorities” and “0” for “whites.”⁷ From these two questions, we construct a single set of four dummy variables representing: (1) racial minority immigrants, (2) white immigrants, (3) racial minority native-born Canadians, and (4) white native-born Canadians. Although racial minority native-born Canadians are controlled for in the models, they are not discussed in the paper as they are not its focus and their sample size is quite small. White native-born Canadians form the omitted reference category for all our analyses; they represent the majority of the Canadian population and are generally thought of as the “mainstream.”

Wage Effects of Job-Education Mismatch

Our second analysis examines the wage effects of horizontal, vertical, and full mismatch on highly skilled white and racial minority immigrant men relative to highly skilled white native-born Canadian men. This outcome variable is represented by a series of repeated measures of individual i’s hourly wage in year y. We take the natural logarithm of each respondent’s hourly wage in constant 2010 dollars⁸ to account for inflation.

There are two main sets of explanatory variables: (1) immigrant and racial minority status and (2) horizontal, vertical, and full job-education mismatch. The omitted reference category for immigrant and racial minority status is white native-born Canadian. The omitted reference category for job education mismatch is fully matched. We also examine the interaction between immigrant and racial minority status and job-education mismatch to test whether the wage effect of each type of mismatch has differential effects for white and racial minority immigrants relative to Canadian-born whites. We model the relationship between job-education mismatch and wage in year zero and then examine how mismatch is related to changes in wages over time. Thus, we are able to test whether horizontal, vertical, and/or full mismatch leads to accelerated wage progression for our groups of interest. For both outcome measures, the control variables are age,⁹ education level,¹⁰ marital status,¹¹ presence of preschool-aged children,¹² province of residence,¹³ wave¹⁴ years since migration (set to “0” for “white native-born”),¹⁵ and annualized unemployment rate for the province of residence. All control variables within a given panel are time varying (i.e., their values are allowed to change over the years).¹⁶

Analytical Strategy

To exploit the SLID’s longitudinal panel design, we employ growth curve modeling (GCM). Growth curve models are multilevel models also known as hierarchical linear models (Raudenbush and Bryk 2002), random coefficient models (Longford 1993), and mixed-effects models (Littell, et al. 2006). GCMs are increasingly used to analyze longitudinal panel data (Kwok et al. 2008). Traditionally, longitudinal data have been analyzed using repeated measures ANOVAs or fixed-effects regressions, but GCMs account for the correlation in error terms in longitudinal data. They do not require a balanced panel and allow the inclusion of individuals not assessed at every timepoint. They also provide more precise estimates of individual growth over time and have greater power to detect predictors of individual differences in growth. Lastly, they can include both time-varying and time-invariant covariates (Raudenbush and Bryk 2002; Singer and Willett 2003).¹⁷ Using GCM, we first model the probability of job-education mismatch for high-skilled immigrants and Canadian-born men over time and then model the wage effects of job-education mismatch over time. The growth curve models in this study use PROC MIXED in SAS 9.4. All descriptive statistics and analyses are weighted using SLID longitudinal weights supplied by Statistics Canada. However, the overall sample Ns are unweighted.

Results

The descriptive statistics in Table 1 indicate that racial minority immigrant men are much more likely to have a university degree than white immigrant men or white native-born Canadian men. White immigrant men have, on average, been in Canada for much longer than their racial minority comparators. When we look at the likelihood of job-education mismatch, we find that about 17 percent of native-born white men experience horizontal mismatch without vertical mismatch; that is, they work in a job not related to their educational field yet commensurate with their years of education. For both racial minority and white immigrant men, this proportion is around 12 percent.

Table 1.

Descriptive Statistics of Selected Variables for White Native-Born, White Immigrant, and Racial Minority Immigrant Men (Year Zero).

	White Native-Born	White Immigrant	Racial Minority Immigrant
N	7,545	631	527
Age (years)	40	44	41
	(9)	(9)	(8)
Married	0.714	0.783	0.792
	(0.452)	(0.413)	(0.406)
Preschool-aged children	0.208	0.186	0.257
	(0.406)	(0.389)	(0.438)
College education	0.624	0.588	0.408
	(0.484)	(0.493)	(0.492)
University education	0.376	0.412	0.592
	(0.484)	(0.493)	(0.492)
Years since migration	—	24	14
		(14)	(10)
Horizontal mismatch	0.173	0.123	0.125
	(0.379)	(0.329)	(0.331)
Vertical mismatch	0.191	0.207	0.213
	(0.393)	(0.406)	(0.410)
Full mismatch	0.069	0.137	0.174
	(0.253)	(0.344)	(0.379)

Note. Means reported are for the combined sample of year zero data from each of the three Survey of Labour Income Dynamics (SLID) panels. Standard deviations are shown in parentheses.

Vertical mismatch on its own is more prevalent than horizontal mismatch across all three groups. About 19 percent of white native-born Canadian men are overeducated, or vertically mismatched, without being horizontally mismatched. For both white and racial minority immigrant men, this proportion is around 21 percent. The immigrant/native-born disparity in job-education mismatch becomes more apparent when we examine full mismatch. Only about 7 percent of white native-born Canadian men are both horizontally and vertically mismatched. In contrast, nearly 14 percent of white immigrant men and 17 percent of racial minority immigrant men report being fully mismatched. In other words, immigrants, especially racial minority immigrants, are more likely to work in lower-level jobs unrelated to their field of expertise.

Likelihood of Job-Education Mismatch

We model the likelihood of horizontal, vertical, and full job-education mismatch for our groups of interest using three separate GCMs. The results are presented in Table 2.

Table 2.

Linear Probability Growth Curve Model of Job-Education Mismatch among Men, Years Zero through Five.

	Horizontal Mismatch	Vertical Mismatch	Full Mismatch
	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Fixed effects
Intercept (probability of mismatch in year zero)	0.210*** (0.019)	0.194*** (0.020)	0.089*** (0.014)
Covariate effect on probability of mismatch in year zero
Racial minority immigrant	0.034* (0.018)	–0.026 (0.020)	0.119*** (0.014)
White immigrant	–0.004	0.021	0.074***
	(0.023)	(0.026)	(0.018)
Slope (annual change in probability of mismatch)	–0.001 (0.001)	-0.001 (0.001)	0.0002 (0.0009)
Covariate effect on change in probability of mismatch
Racial minority immigrant	–0.001 (0.004)	0.009** (0.004)	–0.012*** (0.003)
White immigrant	0.002 (0.004)	–0.003 (0.004)	–0.001 (0.003)
Random effects
Intercept, υ_0i	0.083*** (0.002)	0.114*** (0.002)	0.052*** (0.001)
Slope, υ_1i	0.003***	0.004***	0.002***
	(0.0001)	(0.0001)	(0.00007)
Level 1 error, ∊_ti	46.95***	40.26***	22.97***
	(0.411)	(0.353)	(0.201)
Number of respondents	10,800	10,800	10,800
Number of observations	41,132	41,132	41,132

Note. Controlling for panel, age, level of education, province, annualized unemployment rate in the province of residence, marital status, number of preschool-aged children, and years since immigration (for immigrants only).

*p < .05. **p < .01. ***p < .001.

Table 2 presents the likelihood of each type of job-education mismatch in the initial year of the survey and models how this likelihood changes over time. As the table shows, the reference group (white native-born men of average age, unmarried, with a college diploma and no preschool-aged children, living in Ontario, and from panel three) is predicted to have a 21 percent probability of experiencing horizontal mismatch in the initial year of the survey (p < .01). The likelihood of horizontal mismatch does not change significantly over time. White immigrants are not significantly different from their native-born counterparts in terms of horizontal mismatch; racial minority immigrants are more likely than either of the other two groups to face this type of mismatch.

The reference group is predicted to have a 19 percent probability of experiencing vertical mismatch or overeducation in year zero (p < .01). This likelihood does not change over time, and there is little difference between the reference group and the two immigrant groups. However, racial minority immigrants are slightly more likely to become vertically mismatched over time; the likelihood increases by 0.9 percentage points each year relative to the reference group (p < .05).

The reference group has about a 9 percent probability of full job-education mismatch in the initial year of the survey. The likelihood of full mismatch is 7.4 percentage points higher for white immigrants (p < .01) and 11.9 percentage points higher for racial minority immigrants (p < .01). The difference in the rate of full mismatch between white and racial minority immigrants is statistically significant (not shown). With time, however, racial minority immigrant men tend to move out of jobs for which they are fully mismatched, with their likelihood of full mismatch decreasing by 1.2 percentage points annually relative to the reference group (p < .01).

The GCMs for horizontal, vertical, and full mismatch for men are presented graphically in Figures 1, 2, and 3, respectively.

Figure 1.

Predicted Probability of Horizontal Mismatch, Years Zero through Five.

Figure 2.

Predicted Probability of Vertical Mismatch, Years Zero through Five.

Figure 3.

Predicted Probability of Full Mismatch, Years Zero through Five.

Figure 1 shows that the probability of horizontal mismatch is higher for racial minority immigrants than white immigrants or native-born Canadians and remains fairly consistent for all groups over the survey period. From Figure 2, we see that vertical mismatch is fairly similar for white native-born Canadians and immigrants, whether white or racial minority. Over time, the likelihood of vertical mismatch increases for racial minority immigrants but remains relatively stable for the other two groups. Figure 3 shows a marked disparity in the probability of full mismatch between white native-born Canadians and the two immigrant groups in the initial year of the survey. Over time, however, racial minority immigrants become less likely to be fully mismatched. In fact, by year five, they are about as likely as white immigrant men to experience full mismatch.

Wage Effects of Job-Education Mismatch

Next, we model the wage effects of each type of job-education mismatch and examine how these effects differ for each group over time (see Table 3).

Table 3.

Growth Curve Model of Logged Hourly Wage for Men, Years Zero through Five.

	Coefficient (SE)
Fixed effects
Intercept (average initial wage in year zero)	2.454*** (0.022)
Covariate effect on initial income
Racial minority immigrant	–0.266*** (0.024)
White immigrant	–0.127*** (0.030)
Horizontal mismatch	–0.033*** (0.007)
Vertical mismatch	–0.008 (0.008)
Full mismatch	–0.053*** (0.011)
Racial minority immigrant × horizontal mismatch	–0.030 (0.025)
White immigrant × horizontal mismatch	0.032 (0.030)
Racial minority immigrant × vertical mismatch	–0.006 (0.022)
White immigrant × vertical mismatch	–0.042 (0.026)
Racial minority immigrant × full mismatch	–0.165*** (0.027)
White immigrant × full mismatch	–0.105*** (0.039)
Slope (average hourly wage growth)	0.055*** (0.001)
Covariate effect on wage growth
Racial minority immigrant	–0.005 (0.005)
White immigrant	–0.016*** (0.005)
Horizontal mismatch	–0.006*** (0.002)
Vertical mismatch	–0.0003 (0.002)
Full Mismatch	–0.010*** (0.004)
Racial minority immigrant × horizontal mismatch	0.003 (0.008)
White immigrant × horizontal mismatch	–0.006 (0.009)
Racial minority immigrant × vertical mismatch	–0.013* (0.007)
White immigrant × vertical mismatch	0.005 (0.008)
Racial minority immigrant × full mismatch	0.020** (0.009)
White immigrant × full mismatch	0.031*** (0.011)
Random effects
Intercept, υ_0i	0.158*** (0.003)
Slope, υ_1i	0.005*** (0.0001)
Level 1 Error, ∊_ti	20.97*** (0.194)
Number of respondents	10,300
Number of observations	41,132

*p < .05. **p < .01. ***p < .001.

Table 3 presents a number of interesting findings. First, after accounting for job-education mismatch and a vector of other wage-determining characteristics, racial minority immigrant men face an initial hourly wage disadvantage of 26.6 percentage points relative to Canadian-born white men (p < .001) while white immigrant men face an initial wage disadvantage of 12.7 percentage points (p < .001). Accounting for job-education mismatch does not drastically reduce the immigrant wage disadvantage. The wage disadvantage for racial minority and white immigrants before accounting for mismatch is 29.9 and 15.0 percentage points, respectively (results available on request). Moreover, immigrant men do not experience accelerated wage growth over time to overcome their initial disadvantage, so the immigrant/native-born wage gap persists over the survey period.

Second, for white native-born men, horizontal mismatch is associated with a 3.3 percentage point decrease in hourly wage in year zero (p < .001) and a 0.6 percentage point decrease in annual wage growth (p < .001), vertical mismatch does not have a wage effect, and full mismatch is associated with a 5.3 percentage point decrease in hourly wage in year zero (p < .001) and 1 percentage point decrease in annual wage growth (p < .001). Third, horizontal and vertical mismatch do not differentially affect the initial wages of white or racial minority immigrant men, but full mismatch has a negative effect on both groups’ initial wages. For white immigrant men, being fully mismatched results in an additional wage penalty of 10.5 percentage points in year zero (p < .001), and for racial minorities, it is 16.5 percentage points (p < .001). Horizontal mismatch does not differentially affect the trajectories of immigrant men’s wages. Vertical mismatch is associated with a slightly decreased annual wage growth of 1.3 percentage points (p < .10) but only for racial minority immigrants. Full mismatch, however, is associated with slightly faster wage growth for both groups of immigrant men. The wage effects of job-education mismatch appear in graphical form in Figures 4 through 7.

Figure 4.

Hourly Wages for Fully Matched Men, Years Zero through Five.

Figure 5.

Hourly Wages for Horizontally Mismatched Men, Years Zero through Five.

Figure 6.

Hourly Wages for Vertically Mismatched Men, Years Zero through Five.

Figure 7.

Hourly Wages for Fully Mismatched Men, Years Zero through Five.

Figure 4 presents the wage trajectories of white Canadian-born men, white immigrant men, and racial minority immigrant men with full job-education match. Immigrants, especially racial minority immigrants, clearly have a significant wage disadvantage even if they have jobs in their field and commensurate with their years of education, and this wage gap does not diminish over the survey period.

Figure 5 presents the wage trajectories for horizontally mismatched men. Horizontal mismatch lowers wages for all men and does not have a differential effect on the wages of immigrant men relative to those of their native-born counterparts. Figure 6 shows the wage trajectories for vertically mismatched men. Vertical mismatch by itself does not have a significant effect on the wage trajectories of white Canadian-born or white immigrant men, but it does reduce the wage growth of racial minority immigrants slightly. Finally, Figure 7 presents the wages of men who are fully mismatched. Full mismatch has a negative effect on initial wages in general, but it especially lowers immigrant men’s initial wages, creating a wider immigrant/native-born wage gap in year zero. For white native-born men, full mismatch decreases wage growth over time. For both white and racial minority immigrants, however, being fully mismatched is associated with relatively faster wage growth over time. This slightly reduces their relative wage disadvantage by the end of the survey period, but the effect is not large enough to have a discernable impact on the immigrant/native-born wage gap over the survey period.

Discussion and Conclusion

This study examines the incidence and wage effects of horizontal, vertical, and full job-education mismatch for high-skilled racial minority and white immigrant men relative to high-skilled white Canadian-born men. Our analysis of the incidence of job-education mismatch yields four key findings. First, white and racial minority immigrant men are not significantly more likely than white Canadian-born men to be vertically mismatched. Second, racial minority immigrant men are more likely to be horizontally mismatched than either white immigrant or white Canadian-born men. Third, both white and racial minority immigrant men are more likely to be fully mismatched than white Canadian-born men, although racial minority immigrants are even more likely to be fully mismatched than white immigrants. Fourth, for racial minority immigrant men, the likelihood of full mismatch diminishes with time in Canada.

The second part of the study examines the wage effects of job-education mismatch. Here, we have four important findings. First, vertical mismatch by itself does not result in significant wage disadvantage for any group of interest. Second, horizontal mismatch by itself is related to significant wage disadvantage for all groups. Third, full mismatch is related to significant wage disadvantage, an effect greater for both groups of immigrant men than white native-born men. Racial minority immigrant men who are fully mismatched face an especially disproportionate initial wage penalty. Fourth, and finally, full mismatch results in accelerated wage growth for immigrant men, although the magnitude of this effect is not large enough to close the immigrant/native-born wage gap over the six-year survey period.

Interestingly, controlling for job-education mismatch does not explain a large proportion of the immigrant/native-born wage gap. Even after accounting for horizontal, vertical, and full mismatch, the sizeable wage disadvantage of immigrants, especially racial minorities, remains virtually unchanged. This result is consistent with the findings of Aydede and Dar (2016) and implies that it is not job-education mismatch per se that disadvantages immigrant workers. Other factors, such as devaluation of foreign work experience, lack of language fluency, and employer discrimination, play equally significant roles.

Although our results are generally consistent with previous findings on immigrants’ wage disadvantage, the unpacking of horizontal, vertical, and full job-education mismatch provides further insight into the underlying causes of the immigrant wage gap. We now know that vertical and horizontal mismatch by themselves do not differentially affect immigrant workers and therefore are not factors in understanding immigrant labor market disadvantage. Full mismatch, though, does play a role, and its most detrimental effect is on racial minority immigrants. Our data suggest that many high-skilled immigrants accept unrelated jobs for which they are overqualified as a way to enter the host country’s labor market and access better paying, more appropriate jobs. Although full mismatch is associated with slightly faster wage growth for immigrants over time, this strategy does not appear to be paying off for most. Our findings highlight the importance of disaggregating the various types of job-education mismatch experienced by immigrants. Most previous studies of immigrants’ job-education mismatch have focused exclusively on overeducation (e.g., Wald and Fang 2008; Chiswick and Miller 2009; Sharaf 2013), which may result in misleading conclusions and policy recommendations. Since full mismatch has the most significant effect on highly skilled immigrants’ economic integration, more research attention on this issue is warranted.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We gratefully acknowledge the financial support of the Centre for Labour-Management Relations, Ryerson University. This research was supported by funds to the Canadian Research Data Centre Network (CRDCN) from the Social Science and Humanities research Council (SSHRC), the Canadian Institute for Health Research (CIHR), the Canadian Foundation for Innovation (CFI), and Statistics Canada. Although the research and analysis are based on data from Statistics Canada, the opinions expressed do not represent the views of Statistics Canada or the Canadian Research Data Centre Network.

Notes

References

Aydede

Dar

. 2016. “The Cost of Immigrants’ Occupational Mismatch and the Effectiveness of Postarrival Policies in Canada.” IZA Journal of Migration 5(1):9.

Aydemir

Skuterud

. 2005. “Explaining the Deteriorating Entry Earnings of Canada’s Immigrant Cohorts, 1966–2000.” Canadian Journal of Economics 38(2):641–72.

Banerjee

2009. “Income Growth of New Immigrants in Canada: Evidence from the Survey of Labour and Income Dynamics.” Relations Industrielles/Industrial Relations 64(3):466–88.

Battu

Sloane

P. J.

. 2002. “To What Extent Are Ethnic Minorities in Britain Overeducated?” International Journal of Manpower 23(3):192–208.

Bender

K. A.

Heywood

J. S.

. 2009. “Educational Mismatch among Ph.D.s: Determinants and Consequences.” In Science and Engineering Careers in the United States: An Analysis of Markets and Employment, edited by Freeman

R. B.

Goroff

D. L.

, 229–55. Chicago: National Bureau of Economic Research, Inc.

Boudarbat

Chernoff

. 2012. “Education-Job Match among Recent Canadian University Graduates.” Applied Economics Letters 19(18):1923–26.

Boudarbat

Grenon

. 2013. “Attrition and the Estimation of Employment and Wage Equations: The Canadian Survey of Labour and Income Dynamics.” Applied Economics Letters 20(12):1139–44.

Blau

F. D.

Ferber

M. A.

Winkler

A. E.

. 2002. The Economics of Women, Men, and Work. 4th ed. Upper Saddle River, NJ: Prentice Hall.

Büchel

Mertens

. 2004. “Overdeucation, Undereducation, and the Theory of Career Mobility.” Applied Economics 36(8):803–16.

10.

Chevalier

Lindley

. 2009. “Overeducation and the Skills of UK graduates.” Journal of the Royal Statistical Society Series A 172:307–37.

11.

Chiswick

B. R.

Miller

P. W.

. 2009. “Educational Mismatch: Are High-Skilled Immigrants Really Working at High-Skilled Jobs and the Price They Pay If They Aren’t?” IZA Discussion Paper 4280, Institute for the Study of Labour.

12.

Clogg

Shockey

. 1984. “Mismatch between Occupation and Schooling: A Prevalence Measure, Recent Trends and Demographic Analysis.” Demography 21:235–57.

13.

Dahlstedt

2011. “Occupational Match: Over- and Undereducation among Immigrants in the Swedish Labour Market.” Journal of International Migration and Integration 12(3):349–67.

14.

Dolton

P. J.

Silles

M. A.

. 2008. “The Effects of Over-Education on Earnings in the Graduate Labour Market.” Economics of Education Review 27(2):125–39.

15.

Duncan

G. J.

Hoffman

S. D.

. 1981. “The Incidence and Wage Effects of Overeducation.” Economics of Education Review 1(1):75–86.

16.

Esses

V. M.

Dietz

Bhardwaj

. 2006. “The Role of Prejudice in the Discounting of Immigrant Skills.” In The Cultural Psychology of Immigrants, edited by Mahalingam

, 113–30. Mahwah, NJ: Laurence Erlbaum Associates.

17.

Ferrer

Riddell

. 2008. “Education, Credentials, and Immigrant Earnings.” Canadian Journal of Economics 41(1):186–216.

18.

Freeman

1976. The Overeducated American. San Diego, CA: Academic Press.

19.

Frenette

2004. “The Overqualified Canadian Graduate: The Role of the Academic Program in the Incidence, Persistence, and Economic Returns to Overqualification.” Economics of Education Review 23(1):29–45.

20.

Friedberg

R. M.

2000. “You Can’t Take It with You? Immigrant Assimilation and the Portability of Human Capital.” Journal of Labour Economics 18(2):221–51.

21.

Fuller

2008. “Job Mobility and Wage Trajectories for Men and Women in the United States.” American Sociological Review 73(1):158–83.

22.

Galasi

2008. “The Effect of Educational Mismatch on Wages for 25 Countries.” Budapest Working Papers on the Labour Market BWP 8.

23.

Green

Kler

Leeves

. 2007. “Immigrant Overeducation: Evidence from Recent Arrivals to Australia.” Economics of Education Review 26(4):420–32.

24.

Green

Worswick

. 2012. “Immigrant Earnings Profiles in the Presence of Human Capital Investment: Measuring Cohort and Macro Effects.” Labour Economics 19(2):241–59.

25.

Guo

2009. “Difference, Deficiency, and Devaluation: Tracing the Roots of Non-Recognition of Foreign Credentials for Immigrant Professionals in Canada.” Canadian Journal for the Study of Adult Education 22(1):37–52.

26.

Hartog

2000. “Over-Education and Earnings: Where Are We, Where Should We Go?” Economics of Education Review 19(2):131–47.

27.

Hartog

Oosterbeek

. 1988. “Education, Allocation and Earnings in the Netherlands.” Economics of Education Review 7:185–94.

28.

Hou

Picot

. 2016. “Changing Immigrant Characteristics and Entry Earnings.” Analytical Studies Branch Research Paper Series 374:1–35.

29.

Joona

P. A.

Datta Gupta

Wadensjö

. 2014. “Overeducation among Immigrants in Sweden: Incidence, Wage Effects and State Dependence.” IZA Journal of Migration 3(9):1–23.

30.

Kiker

B. F.

Santos

M. C.

Mendes de Oliveira

. 1997. “Overeducation and Undereducation: Evidence for Portugal.” Economics of Education Review 16(2):111–25.

31.

Kler

2007. “A Panel Data Investigation into Over-Education among Tertiary Educated Australian Immigrants.” Journal of Economic Studies 34(3):179–93.

32.

Kwok

O. M.

Underhill

A. T.

Berry

J. W.

Luo

Elliott

T. R.

Yoon

. 2008. “Analyzing Longitudinal Data with Multilevel Models: An Example with Individuals Living with Lower Extremity Intra-Articular Fractures.” Rehabilitation Psychology 53(3):370–86.

33.

Leuven

Oosterbeek

. 2011. “Overeducation and Mismatch in the Labour Market.” IZA Discussion Paper Series 5523:1–56.

34.

P. S.

2001. “Market Worth of Immigrants’ Credentials Educational Credentials.” Canadian Public Policy 27(1):23–38.

35.

Lindley

Lenton

. 2006. The Overeducation of UK Immigrants: Evidence from the Labour Force Survey. Sheffield, UK: University of Sheffield.

36.

Linsley

2005. “Causes of Overeducation in the Australian Labour Market.” Australian Journal of Labour Economics 8:121–43.

37.

Littell

R. C.

Milliken

G. A.

Stroup

W. W.

Wolfinger

R. D.

Schabenberger

O. O.

. 2006. SAS for Mixed Models, 2nd ed. Cary, NC: SAS Institute Inc.

38.

Longford

N. T.

n.d. “Regression Analysis of Multilevel Data with Measurement Error.” British Journal of Mathematical and Statistical Psychology 46(2):301–11.

39.

Madamba

A. B.

Jong

G. F. De

. 1997. “Job Mismatch among Asians in the United States: Ethnic Group Comparisons.” Social Science Quarterly 78(2):524–42.

40.

Mahuteau

Mavromaras

Sloane

Wei

. 2014. “Horizontal and Vertical Educational Mismatch and Wages.” Presented at 25th Australian Labour Market Research, Melbourne.

41.

McGoldrick

Robst

. 1996. “Gender Differences in Overeducation: A Test of the Theory of Differential Overqualification.” American Economic Review 86:280–84.

42.

McGuinness

2006. “Overeducation in the Labour Market.” Journal of Economic Surveys 20(3):387–418.

43.

Nielsen

C. P.

2007. “Immigrant Overeducation: Evidence from Denmark.” World Bank Policy Research Working Paper 4234. World Bank, Washington, DC.

44.

Nieto

Matano

Ramos

. 2013. “Skill Mismatches in the EU: Immigrants vs. Natives.” IZA Discussion Paper 7701:1–26.

45.

Nordin

Persson

Olof Rooth

. 2010. “Education-Occupation Mismatch: Is There an Income Penalty?” Economics of Education Review 29(6):1047–59.

46.

Oreopoulos

2011. “American Economic Association Why Do Skilled Immigrants Struggle in the Labour Market? A Field Experiment with Thirteen Thousand Resumes.” American Economic Journal: Economic Policy 3(4):148–71.

47.

Picot

Sweetman

. 2005. The Deteriorating Economic Welfare of Immigrants and Possible Causes, Update 2005. Ontario: Statistics Canada.

48.

Piracha

Vadean

. 2012. “Migrant Educational Mismatch and the Labour Market.” International Handbook of the Economics of Migration 6414:1–27.

49.

Quinn

Rubb

. 2006. “Mexico’s Labour Market: The Importance of Education-Occupation Matching on Wages and Productivity in Developing Countries.” Economics of Education Review 25:147–56.

50.

Raudenbush

S. W.

Bryk

A. S.

. 2002. Hierarchical linear models: Applications and data analysis methods. 2nd ed. Thousand Oaks, CA: Sage Publications.

51.

Reitz

J. G.

2001. “Immigrant Skill Utilization in the Canadian Labour Market: Implications of Human Capital Research.” Journal of International Migration and Integration 2(3):347–78.

52.

Reitz

J. G.

. 2011. “Taxi Driver Syndrome: Behind the Scenes Immigration Changes are Creating New Problems on Top of Old Ones.” Literary Review of Canada 19(2):20–22.

53.

Robst

2007. “Education and Job Match: The Relatedness of College Major and Work.” Economics of Education Review 26(4):397–407.

54.

Rubb

2003. “Post-College Schooling, Overeducation, and Hourly Earnings in the United States.” Education Economics 11(1):53–72.

55.

Rumberger

R. W.

1987. “The Impact of Surplus Schooling on Productivity and Earnings.” Journal of Human Resources 22(1):24–50.

56.

Sanroma

Ramos

Simon

. 2008. “The Portability of Human Capital and Immigration Assimilation: Evidence from Spain.” IZA Discussion Paper No 3649, Institute of Labor Economics.

57.

Sharaf

M. F.

2013. “Job-Education Mismatch and Its Impact on the Earnings of Immigrants: Evidence from Recent Arrivals to Canada.” ISRN Economics 2013(2013):452358.

58.

Sicherman

Galor

. 1990. “A Theory of Career Mobility.” Journal of Political Economy 98(1):169–92.

59.

Singer

J. D.

Willett

J.B.

. 2003. Applied Longitudinal Data Analysis. Oxford, UK: Oxford University Press.

60.

Somers

Cabus

S. J.

Groot

Maassen Van den Brink

. 2016. “Horizontal Mismatch between Employment and the Field of Education: Evidence from a Systematic Literature Review.” TIER Working Paper Series 16/02, Maastricht University.

61.

Tsai

2010. “Returns to Overeducation: A Longitudinal Analysis of the U.S. Labour Market.” Economics of Education Review 29(4):606–17.

62.

Verdugo

. 1989. “The Impact of Surplus Schooling on Earnings: Some Additional Findings.” Journal of Human Resources 24(4):629–43.

63.

Verhaest

Omey

. 2006. “The Impact of Overeducation and its Measurement.” Social Indicators Research 77(3):419–48.

64.

Verhaest

Dieter

Sellami

Sana

Velden

Rolf van der

. 2017. “Differences in Horizontal and Vertical Mismatches across Countries and Fields of Study.” International Labour Review 156(1):1–23.

65.

Wald

Fang

. 2008. “Overeducated Immigrants in the Canadian Labour Market: Evidence from the Workplace and Employee Survey.” Canadian Public Policy 34(4):457–79.

66.

Warman

Sweetman

Goldmann

. 2015. “The Portability of New Immigrants’ Human Capital: Language, Education, and Occupational Skills.” Canadian Public Policy 41(Supplement 1):S64–79.

67.

Wolbers

M. H. J.

2003. “Job Mismatches and Their Labour-Market Effects among School-Leavers in Europe.” European Sociological Review 19(3):249–66.