![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
We provide first evidence of the impact of over-education, among natives and immigrants, on firm-level productivity and wages. Our results show that the over-education wage premium is higher for natives than for immigrants. However, since the differential in productivity gains associated with over-education outweighs the corresponding wage premium differential, we conclude that over-educated native workers are in fact underpaid to a greater extent than their over-educated immigrant counterparts. This conclusion is refined by sensitivity analyses, when testing the role of immigrants’ background (e.g. region of birth, immigrant generation).
Acknowledgments
The authors are most grateful to Statistics Belgium for giving them access to the data. Financial support from the Belgian Federal Science Office (BELSPO, IMMILAB project) is also kindly acknowledged.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 According to career mobility theory (Sicherman and Galor Citation1990), over-educated workers should experience greater upward occupational mobility, which in turn should be associated with relatively higher wage growth. Several empirical studies have tested the validity of these predictions and their results are mixed. For the US, Rubb (Citation2006), for instance, find that over-educated workers have higher turnover rates, are more likely to be promoted and experience faster wage growth than their adequately educated work colleagues. In the German context, while Büchel and Mertens (Citation2004) do not support the career mobility hypotheses, Grunau and Pecoraro (Citation2017) conclude that over-educated workers are more likely to be promoted to managerial positions and, that – in terms of wage grows – over-educated workers benefit more from promotions than their well-matched educational peers. The study of Wen and Maani (Citation2019), based on Australian panel data, adds methodologically to most previous studies by adopting a dynamic model and controlling for endogeneity. Their estimates show that the probability of upward occupational mobility and wage growth is lower among over-educated workers (whether or not they are over-qualified). The persistence of mismatch in Australia and the associated wage penalty is also supported Mavromaras et al. (Citation2013), particularly among university graduates. Finally, among other studies, Baert, Cockx, and Verhaest (Citation2013) find that over-education at the beginning of a career in Belgium significantly delays the transition to an adequate job. Therefore, they conclude that over-education is a trap.
2 Firm heterogeneity has notably been assessed by the diversity of the workforce within firms in terms of age, gender, occupations, or employment contracts.
3 More precisely, the authors focus on wage costs.
4 A few other papers have relied on firm-level data to investigate whether over-education is beneficial or harmful to productivity (e.g. Quintini Citation2011; McGowan and Andrews Citation2015; Grunau Citation2016). For the Belgian private sector, the results of Kampelmann and Rycx (Citation2012a) and Mahy, Rycx, and Vermeylen (Citation2015) suggest that the effect is significant and positive.
5 As this paper focuses primarily on the empirical estimation of productivity-wage gaps related to over-education according to workers’ origin, we choose not to reiterate the numerous mechanisms put forward in the theoretical literature. However, following Kampelmann and Rycx (Citation2012b), these mechanisms can be divided into: (a) theories based on efficiency and individual rationality (e.g. when over-educated workers differ from workers with the required level of education in terms of quasi-fixed costs or firm-specific skills, or when efficiency considerations lead firms to compress their wage structure so as to avoid shirking or demotivation (see e.g. McGuinness Citation2006; Cardoso Citation2010)); and (b) institutionalist theories (e.g. when monopsony power, market regulations, wage norms, or collective bargaining are associated with positive or negative rents that differ between over-educated and well-matched workers (see e.g. Quintini Citation2011; Konings and Vanormelingen Citation2015)).
6 Information on workers’ educational attainments, available in 7 categories in our dataset, were reported by firms’ HR departments (based on their registers). We converted that information into years of education, applying the following rule: (i) primary education: 6 years of education; (ii) lower secondary education: 9 years; (iii-iv) general, technical and artistic upper secondary education: 12 years; (v) higher university and non-university education, short type: 15 years; (vi) university and non-university education, long type: 17 years; (vii) postgraduate education: 18 years. Given that information on workers’ levels of education were provided by firms’ HR departments, those levels might be somewhat under-estimated for immigrants. The findings reported in this paper should therefore be considered as a lower bound.
7 Note that, by definition: , i.e. the sum of a worker's years of required and over-education minus her/his years of under-education is equal to her/his years of attained education.
8 The distinction between blue-collar and white-collar workers is based on the International Standard Classification of Occupations (ISCO-08). Workers belonging to groups 1–5 (i.e. managers, professionals, technicians and associate professionals, clerical support workers, and services and sales workers) are classified as white-collars, whereas those in groups 7–9 (i.e. craft and related trades workers, plant and machine operators and assemblers, and elementary occupations) are classified as blue-collars.
9 It covers the following sectors: (i) mining and quarrying, (ii) manufacturing, (iii) electricity, gas, steam and air conditioning supply, (iv) water supply, sewerage, waste management and remediation activities, (v) construction, (vi) wholesale and retail trade, repair of motor vehicles and motorcycles, (vii) transportation and storage, (viii) accommodation and food service activities, (ix) information and communication, (x) financial and insurance activities, (xi) real estate activities, (xii) professional, scientific and technical activities, (xiii) administrative and support service activities.
10 Our dataset does not provide direct information of the age of firms, so we have proxied this variable with the length of employment of each firm's most senior employee.
11 This is estimated through the ‘perpetual inventory method’ (or PIM, see e.g. OECD (Citation2009) for more details). The PIM incorporates the idea that capital stock results from investment flows and corrects for capital depreciation and efficiency losses. Following standard practice, we assume a 5 percent annual rate of depreciation.
12 The focus of our paper is on the effects of over-education. Therefore, the results regarding under-education will be neither reported nor discussed in the empirical section of this manuscript. Our estimates for under-education are available on request.
13 Note that: , i.e. the sum of the average years of over-education (among natives and immigrants, respectively), years of required education, and the average years of under-education (among natives and immigrants, respectively) in firm j at year t is equal to the average years of education attained by the workers employed in firm j at year t.
14 More precisely, to determine the region of origin of second-generation immigrants, the order of priority is based on the father's region of birth (see e.g. Corluy et al. Citation2015; FPS Employment and Unia Citation2017; Piton and Rycx Citation2021). Put differently, for second-generation immigrants, the father's region of birth is used, unless the father was born in Belgium and the mother born abroad, in which case the mother's region of birth is retained. This choice stems from the fact that: (i) children born in Belgium before June 1st, 2014, were named after their father (the legislation has become more flexible since then), and (ii) correspondence studies have shown that call-back rates depend upon the origin of job seekers’ names (see e.g. Baert Citation2017).
15 By ‘developing’ countries, we refer to both transition and developing countries, as listed in the UNCTAD (Citation2020) classification.
16 In addition, we also tested the sensitivity of our results to the exclusion of occupational dummies. The point raised by an anonymous referee is that the proportion of blue-collar workers in each firm, included in our benchmark specification, is derived from the ISCO 1-digit classification, which can also be used as an indicator of educational requirements. According to this classification, white-collar occupations correspond broadly to medium- to high-skilled jobs, while blue-collar occupations tend to coincide with low- to medium-skilled jobs. As a result, the proportion of blue-collar workers may capture part of the true variation in educational requirements, leading to biased estimates of the latter variable. To test the extent of this potential bias, we therefore re-estimated our benchmark equations by excluding the proportion of blue-collar workers from our control variables.
17 Expected biases associated with OLS and the relatively poor performance and shortcomings of the FE estimator in the context of firm-level productivity regressions are reviewed in Van Beveren (Citation2012).
18 The assumption of persistent productivity, both at the industry and firm level, finds some support in the literature (see e.g. Bartelsman and Doms Citation2000). Researchers ‘documented, virtually without exception, enormous and persistent measured productivity differences across producers, even within narrowly defined industries’ (Syverson Citation2011, 326). Large parts of these productivity differences are still hard to explain. The persistence of wages is also highlighted in the literature (see e.g. Le Bihan, Montornes, and Heckel Citation2012). Wage stickiness is notably the outcome of labour market institutions, adjustment costs, and efficiency wages’ motives.
19 Bond and Söderbom (Citation2005) provide a review of the literature on the identification of production functions. The authors notably highlight that the adjustment costs of labour and capital can justify the use of lagged values (of the endogenous variables) as instruments.
20 The SES is a stratified sample. The stratification criteria refer to the region (NUTS-groups), the main economic activity (NACE-groups), and the size of the firm. The sample size in each stratum depends on the size of the firm. The sampling percentages of firms are equal to 10, 50, and 100% when the number of workers is between 10 and 50, between 50 and 99, and above 100, respectively. Within a firm, the sampling percentages of employees also depend on the size of the firm. The sampling percentages of employees are equal to 100, 50, 25, 14.3, and 10% when the number of workers is between 10 and 20, between 20 and 50, between 50 and 99, between 100 and 199, and between 200 and 299, respectively. Firms employing at least 300 workers are required to report information for an absolute number of employees, which ranges from 30 (for firms with 300–349 workers) to 200 (for firms with 12,000 workers or more). To guarantee that firms report information on a representative sample of their workers, they are asked to follow a specific procedure. First, they have to rank their employees in alphabetical order. Next, Statistics Belgium gives them a random letter (e.g. the letter O) from which they have to start when reporting information on their employees (following the alphabetical order of the workers’ names in their list). If they reach the letter Z and still have to provide information on some of their employees, then they have to continue from the letter A in their list. Moreover, firms that employ different categories of workers, namely managers, blue- and/or white-collar workers, have to set up a separate list in alphabetical order for each of these categories and to report information on a number of workers in these categories that is proportional to their share in total firm employment. For example, a firm with 300 employees (namely, 60 managers, 180 white-collar workers, and 60 blue-collar workers) will have to report information on 30 workers (namely, 6 managers, 18 white-collar workers, and 6 blue-collar workers). Finally, we should point out that there is no threshold at the upper limit for wages in the SES. In other words, wages are not censored. For an extended discussion, see Demunter (Citation2000).
21 Note that the coverage of the SBS is not the same as that of the SES, as the former does not cover the entire financial sector.
22 This threshold of 10 observations was almost always met, so this constraint left our sample virtually unaffected.
23 For instance, we dropped some workers with no information on their level of education and a very small number of firms for which the recorded value added was negative.
24 All variables measured in monetary terms have been deflated to constant prices of 2013 by the consumer price index taken from Statistics Belgium.
25 At first glance, this result may seem somewhat surprising, but it follows from the fact that the level of attained education is lower among immigrants than among natives. Indeed, workers with lower levels of attained education are less likely to be over-educated. Restricting our sample to workers with tertiary education, we find that the proportion of over-educated workers is, as expected, higher among immigrants than among natives.
26 The lower contribution of immigrants to this statistic is of course due to the fact that immigrants represent only 26% of workers in our sample.
27 It should be reminded that native-born people are those born in Belgium with both parents born in Belgium, whereas immigrants include foreign-born people (i.e. first-generation immigrants) and those born in Belgium with at least one foreign-born parent (i.e. second-generation immigrants).
28 As highlighted in footnote 16, we also tested the sensitivity of our results to the exclusion of occupational dummies. Results, available on request, corroborate our main findings.
29 The time dummies have been considered as exogeneous, and we used the first and second lags of other explanatory variables as instruments.
30 As highlighted at the end of Section 2.3, the use of moderating variables, i.e. interaction effects between our educational variables and immigrants’ background, mechanically reduces the number of data points in each category and for each period. In addition, for some categories of immigrants, the average number of years of over-education within firms becomes much more limited. The year-to-year variations in the value of these categories are often also smaller. For this reason, the GMM-SYS estimates associated with our sensitivity tests should be interpreted with more caution than those for our benchmark specification. Accordingly, to be conservative, we will first present our pooled OLS results (bearing in mind that they should not be interpreted in a causal way, but rather as conditional means) before moving to those obtained with the GMM-SYS estimator.
31 Given that the technological changes of the past decades appear to be skill/task biased and that low-educated workers are typically less skilled and often doing more routine tasks (Goos, Manning, and Salomons Citation2014), the hysteresis in social norms discussed by Doeringer and Piore (Citation1985) and Skott (Citation2005) could lead to the overpayment of low-educated workers whose productivity might have been negatively affected by technological change, and the underpayment of highly (over-) educated workers whose productivity might have increased.
32 Following Hamermesh (Citation1975) and Akerlof and Yellen (Citation1988), there is an efficiency argument for firms to pay high-productivity jobs below and low-productivity jobs above their marginal products, with a view to compress the overall wage structure.
33 The collective wage bargaining system in Belgium is characterized by a high degree of centralization across firms and wage co-ordination across sectors. Wage bargaining occurs at three levels: the national (interprofessional) level, the sectoral level, and the firm level. Negotiations generally occur every two years on a pyramidal basis. In principle, the bargaining rounds are inaugurated by a national collective agreement defining the national minimum wage and, since 1996, a maximum margin for wage cost growth that may be bargained at lower levels (the ‘wage norm’). The objective of this ‘wage norm’ is to make sure that all parties in the negotiations take into consideration the need for wage restraint in an open economy. After the national agreement, sector-level agreements are negotiated and concluded within Joint Committees that bring together employer and union representatives. Sector-level agreements set industry-wide standards, including very detailed pay scales in terms of wage levels and progress (e.g. promotion, seniority), for all workers covered by the Joint Committee. Sector-level agreements apply compulsorily to all companies in the sector and to their workers, whether or not they are members of the signatory organizations (employers’ organizations or unions). As a result, practically the entire workforce in Belgium is covered by a sector-level agreement. Finally, firm-level agreements can complement sector-level agreements and fix wages and working time, as well as work organization and other aspects of the working life when a union delegation is present. Due to the so-called ‘favorability principle’, firm-level bargaining can only improve (or confirm) the conditions set in the sectoral agreement. The Belgian bargaining system has remained practically unchanged since 1968 and, despite some debates, continues to enjoy a broad support. However, a concern raised by the OECD (Citation2018) is that the maximum wage norm and the detailed pay scales in Belgium restrain the extent to which high-performing firms can offer higher wages, attract skilled workers, innovate, and grow. At the other end, legally-binding wage indexation (in combination with regulation of job protection) makes it difficult for firms in economic difficulty to reduce the wage bill. Moreover, the extension of sectoral agreements to non-signatory firms, although it reduces differences in working conditions among workers in the same sector and ensures a level-playing field for companies, may also have an adverse effect on employment and firms’ performance (Hijzen and Martins Citation2016). For a more detailed discussion, see e.g. Kampelmann and Rycx (Citation2013), OECD (Citation2018), and Garnero, Rycx, and Terraz (Citation2020).