Vocational education and training – an engine for economic growth and a vehicle for social inclusion?


  • Anders Nilsson

    Corresponding author
    1. Department of Economic History, School of Economics and Management, Lund University, Sweden
      Anders Nilsson, Professor, Department of Economic History, School of Economics and Management, Lund University, PO Box 7083, SE-22007, Lund, Sweden. Email: Anders.Nilsson@ekh.lu.se
    Search for more papers by this author

  • This paper is an outcome of an initiative by Kristine Bruland, who organized a series of seminars in Paris and Geneva 2007–2009 under the heading ‘Education and economic change: history, theory and governance’. I have benefitted from comments by the participants at these seminars, in particular by Kristine Bruland, Claude Diebolt, Jonas Ljungberg, David Mowery and Bart Verspagen. Comments by the editors and anonymous referees of this Journal have been singularly helpful. The usual disclaimer applies. The research has been supported by the Swedish Council for working life and social research (grant no. FAS 2002-0649), and by the Swedish Research Council (grant no. VR 2006-1931). Generous travel grants from the Stiftelsen Partnerskapet at the Lund University School of Economics and Management are also gratefully acknowledged.

Anders Nilsson, Professor, Department of Economic History, School of Economics and Management, Lund University, PO Box 7083, SE-22007, Lund, Sweden. Email: Anders.Nilsson@ekh.lu.se


Vocational education and training (VET) has in recent years enjoyed a revival for two major reasons. Firstly, it is regarded as a suitable means of promoting economic growth. Secondly, it is seen as a potentially powerful tool for fostering social inclusion. In this review, these assumed effects are critically examined on the basis of the vastly expanding literature in the field. Evidence of the productivity-enhancing effects of VET at company level is quite solid, but evidence of the effect on overall economic growth is far from conclusive. The effects on social inclusion are uncertain because reform of VET systems has not been sufficient and because it has proved difficult to bring about the necessary institutional change. The review identifies policy implications and makes some suggestions for future research.


Vocational education and training (VET) has in recent years enjoyed a revival in academic research as well as in the political arena. In relation to developing countries, bilateral aid agencies, the World Bank and United Nations Educational, Scientific and Cultural Organization (UNESCO) advocate vocational education to reduce poverty, promote economic growth and increase competitiveness (Comyn & Barnaart, 2010). In a European context, it is seen as a major tool in the transformation of the European economy (Bordeaux Communiqué, 2008), and there are numerous examples of presumed effects in countries, regions and specific sectors of the economy (Baum, 2002; Budría and Telhado-Pereira 2009; Mupimpila & Narayana, 2009; Spielhofer & Sims, 2004). There are two major reasons for this. First, VET is regarded as a particularly suitable means of promoting economic growth. The purpose of VET is to provide individuals with skills that are more or less directly applicable in the workplace and it is argued that these are likely to have direct and immediate effects on productivity and consequently upon economic growth. The second reason is the potential for VET to promote social inclusion. Here, the elimination of all barriers to education and training is regarded as a key issue (European Commission, 2007). In this context, VET is of particular interest because there are reasons to believe that it is superior to general education from a socializing point of view, as well as in promoting access to the labour market.

In this review of the existing research, these assumed effects of VET are critically examined. There are, of course, other pertinent issues such as the retraining of workers (Gvaramadze, 2010), training in (and for) an ageing society and lifelong learning (Tikkanen, 2009) and gender-related questions (Butler & Ferrier, 2006; Kraus, 2006; Rommes et al., 2005). However, the assumptions that VET is beneficial to economic growth and to social inclusion are fundamental and ‘classical’. The vocabulary has changed a little but these two assumptions about the benefits of VET have been made repeatedly since the late nineteenth century (that is, when VET in its present form took shape). Empirical evidence to support the assumptions is not that frequent yet the claims are made repeatedly. This justifies a scrutiny of the empirical evidence and an attempt to put them, as well as the assumptions, in their proper context.

The outline of this review is as follows. Several of the problems in presenting strong empirical support for the effects of VET stem from the fact that the concept, although intuitively simple and straightforward, is virtually impossible to define in an unambiguous manner (Stevenson, 2005). The concept is multidimensional and can be organized in different ways. This problem is discussed in the following section. The scrutiny of empirical evidence of the effects of VET begins in the third section with an overview of studies at company level, where focus is on the contribution VET makes to productivity. The few studies that exist on the macroeconomic effects of VET are reviewed in the fourth section. Sections five and six address important aspects of social inclusion. One of the most important means through which education facilitates social inclusion is through socialization – that is, preparing pupils and students for participation in society as adults. Another equally important and associated aspect is that VET could be particularly efficient in the transition from school to work, that is, to counteract youth unemployment. In the concluding section, it is acknowledged that it is difficult to present satisfactory answers to the question posed in the title of this paper. Although research has expanded vastly in recent years, there still exist substantial lacunae in our understanding of VET and some suggestions for further research are put forward. Finally, policy proposals are put forward, two of which are controversial: the regulation of apprenticeship-based trades and professions, and the conscious construction of a two-tier system in order to better fulfil the two main tasks of VET. The aim of the latter would be to promote social inclusion, including among ‘low-achievers’ while providing high-quality education and training to meet today's labour market demand.

Defining VET

The difficulties in defining VET come from two sources. The initial problem is to distinguish between general and vocational education. The second and related problem derives from the fact that VET is a multidimensional concept that can be organized in different ways. It is not a new problem, as a quote from as far back as 1895 demonstrates: ‘It passes the wit of man, so far as I know, to give a legal definition of technical education’ (T.H. Huxley, quoted in Moodie, 2002). In the review by Gavin Moodie, four different sets of criteria, used by different researchers, are examined. He points out that they are all problematic and that a proper definition should probably contain elements of them all. He also adds that because the content of VET and its relation to other parts of the education system, and to working life, differs between countries and changes over time, it is not possible to give one satisfactory definition (Moodie, 2002).

To narrow down the concept, it is useful to discuss the various forms through which skills acquisition takes place in society. A large part is through on-the-job training that may or may not result in recognized qualifications (Barber, 2003). In countries such as the United States, where the extended, post-compulsory education is totally dominated by high school/junior colleges which provide the students with a much more general education, job experience in combination with studies has a strong tradition (Goldin & Katz, 2008; Stone, 2002). Under these circumstances, on-the-job training with limited instruction but in combination with frequent job changes and considerable geographical and occupational mobility, constitutes an important but largely unrecorded part of VET (Shackleton et al., 1995; Stern et al., 2004). A recent survey in some of the Organisation for Economic Co-operation and Development (OECD) countries indicates that on average, about 180 hours are spent annually in on-the-job training (OECD 2007). The variation between the countries is large, from about 100 hours in Belgium and Poland to about 300 hours in Ireland and Canada, but the fact that on average about 10 percent of working time is spent on on-the-job training implies that it is a very important activity.

In-company training is a more formalized means of training. It comes in many forms, from extended trainee programmes to short courses, and is mainly provided as continuing VET. In countries such as the United Kingdom, however, in-company training is also a major component of initial VET. The system of National Vocational Qualifications (NVQs), where knowledge and skills acquired by way of work experience are subject to certification processes, implies that in-company training forms the basis of initial as well as continuing VET (Kau, 1998; for an update in 2008, see the country report for the UK at Cedefop's database on national VET systems, ReferNet).

Since the mid-1990s, the NVQ system has been integrated with the Modern Apprenticeship System in England (Brockmann et al., 2010), which underlines that apprenticeship in many ways is similar to in-company training because much of the training takes place in a workplace. The main (but not the only) difference is that an apprenticeship usually is of a much longer duration. After the dissolution of the guilds system in the nineteenth century, the apprenticeship system remained the principal form for vocational training in a number of trades, but in the absence of a regulation the system tended to erode and in many cases apprentices turned into cheap labour. As a consequence, the trade unions became hostile to the apprenticeship system unless they were able, as sometimes was the case, to reach a collective agreement where important aspects such as wages and number of apprentices were settled. However, the training component in the apprenticeship agreements was generally left to the employer's discretion. In a few countries, Denmark and Germany for example, the state instituted an apprenticeship legislation which implied, among other things, recognized training titles and that part of the training should take place in a school.

School-based VET is where all or most of the training takes place in a (vocational) school. In a historical perspective, school-based VET emerged primarily as a supplement to apprenticeship training, but in a few countries where apprenticeship training remained unregulated, France for instance, it emerged as the most important form of vocational education (Greinert, 2005). After World War II, school-based VET became more important, and in many countries the dominant form of VET for young people.

For historical reasons, countries have over time developed systems that are very different from each other. An influential tradition has been established from the study of European systems that emerged in the initial stages of the Second Industrial Revolution, that is, the decades around 1900. There is some consensus that (at least) three ‘ideal models’ emerged and that these models still characterize today's national systems: a market-led system where a labour market characterized by substantial mobility provides much of the vocational training; a school model where most of the VET takes place in schools; and a ‘dual’ model with a strong presence of an apprenticeship system (Greinert, 2005; Thelen, 2004). In other traditions, with a stronger focus on present-day systems and incorporating also Latin America, Asia and Africa, the different roles that could be played by the state are emphasized. David Ashton and his colleagues, for instance, distinguish not only the market-led and the corporatist (or ‘dual’) model but also the developmental state and the neo-market model (Ashton et al., 2000). It appears that researchers from different traditions identify a limited number of models that include a market-led and a corporatist (or ‘dual’) model, but where the role of the state is viewed differently. Often, however, the state is associated with school-based VET.

In the last 10–15 years, there has been an increasing interest in apprenticeship systems in response to problems that have become apparent in many of the state-led systems. Attempts have been made to revitalize the apprenticeship system and to develop it, recognizing that in many countries it has played only a marginal role for a long time (Comyn & Barnaart, 2010; Palmer, 2009; Ryan, 2000, 2001). To some extent, this resembles the situation a century ago, when attempts were made in many European countries to establish modernized versions of the apprenticeship that had been an integral part of the pre-industrial guilds and crafts system that was dismantled in the Liberal era of the nineteenth century. Recent research has cast some doubt on the general character of the old system, however, and demonstrated its diversity, not least with respect to the standing of apprentices in relation to the masters (De Munck et al., 2007). Nevertheless, universality (or ganze Haus in German) was perceived as a characteristic of the system that was re-established in North and Central Europe towards the end of the nineteenth century. The new apprenticeship system, later to be known as ‘the dual system’, was initially regarded with resistance by the trade unions (Thelen, 2004). From their perspective, there was a distinct risk that young workers (apprentices) would become dependent on their employers and foremen (masters). For complex (and not altogether clear) reasons, the resistance of the trade unions was a factor in apprenticeship being re-established in only a limited way in some countries including Sweden (Olofsson, 2005); in others such as the Netherlands, the reluctance of larger enterprises seem to have been more important (Westerhuis, 2007). Regardless of the forces of resistance, the dividing line between full and partial re-establishment was the legal framework surrounding the apprenticeship system. Where this system was legally enforced and extended beyond the traditional handicraft sector, some sort of ‘dual system’ emerged. Where it was not, apprenticeship became a marginal part of the system of VET. Present-day attempts to re-establish apprenticeship systems are dealt with later.

In line with Moodie's recommendation, this paper adopts a pragmatic approach and accepts that the concept of VET is somewhat undetermined and constantly changing. At a general level, the definition given by the European Centre for the Development of Vocational Training (Cedefop) is useful: ‘All structured activities that aim to provide people with knowledge, skills and competencies necessary to perform a job or set of jobs, whether or not they lead to a formal qualification’ (Cedefop, 2009b, p. 8). At a more operational level, VET tends to become a residual when the various parts of a country's education system are defined. In the present review, which is based entirely on previous research and official publications, no attempt is made to change whatever concept that was used in the different studies. To keep the review to manageable proportions, an emphasis is placed on initial VET. Research concerning courses in compulsory school, tertiary level courses and programmes, and labour market programmes is not covered. The discussion of the effects of training at company level is mainly based on studies of on-the-job and in-company training, whereas more comprehensive school-based and apprenticeship programmes are used when the discussion focuses on effects at the macro level.

Productivity effects at company level

Studies of the possible productivity effects of training are important not least because it is a field which is seriously under-documented. The OECD survey in 2007 cited earlier is an exception – in most cases, on-the-job training and probably also much in-company training is not systematically recorded and, as a consequence, its extent is far from known. It is likely to be substantial, however. Data from the Labour Force and National Adult Learning Survey indicate that 40 percent of employee learning is non-certified in the UK (Green & McIntosh, 2006). In addition, on-the-job training is often not reported at all (Kitching, 2008). The conclusion to be drawn from under-reported on-the-job training is that the overall contribution of VET to economic growth will be systematically underestimated. This problem could be ignored if it could be assumed that on-the-job and in-company training constitutes a constant proportion of VET, but such is not the case. This fact alone makes international comparisons hazardous. Over a longer time perspective, the problem is probably accentuated because the proportion of undocumented VET is likely to have changed over time, even if Stern and his colleagues report that surveys of US employees do not reveal any overall trend between 1970 and 2000 in the total amount of on-the-job plus in-company training (Stern et al., 2004). Schøne (2006), on the other hand, reports that training provided by the employer increased between 1996 and 2002 in Norway but that this result was driven by an overall higher level of education among the employees that created an increased need for training.

A number of studies have been made to investigate the effects of training on productivity, with results pointing in different directions. An important reason for the diverging results is the complexity of training in firms. Much of the literature is based, implicitly or explicitly, on the Beckerian distinction between general and specific training (for examples, see Smits, 2008). But as Wendy Smits has pointed out, that distinction is extremely difficult to uphold in empirical analyses. Few skills are completely firm-specific or absolutely general, but the degree of specificity varies and so does the market evaluation of the skills (Smits, 2008).

To keep the survey manageable yet putting recent research in some perspective, two meta-studies are used to cover some of the most reliable studies that were undertaken in the 1990s and early 2000s. These two studies have a similar focus in that they try to establish not only productivity effects but also the distribution of accruing benefits between firm and employees. Furthermore, they endeavour to include only studies that were designed to reveal causal relationships, not just correlation, between training and productivity. This is in fact a major problem in many of the studies undertaken. Very briefly, problems could be caused by lack of control group, possible reversed causality (profitable firms invest more in training, perhaps without too much regard to the effects), the existence of possible time lags between training and discernible effects, and the sheer size of the firms as larger firms generally invest more in training (Abdel-Wahab et al., 2008).

The first meta-study summarizes 19 studies covering 10 countries (OECD, 1998). One main issue in the study was to investigate who benefited from training. Two ‘stylized facts’ were emphasized; firstly, that training generated productivity gains in companies that provided training and that those gains were (roughly) equally divided between workers (wage increases) and companies. The second ‘stylized fact’ was that the impact was greatest in connection with structural change, that is, with changes in work organization, job structure or technological innovation. Most of the studies relate to the early 1990s, but in a few cases comparisons were made between two time-periods, one of them going back to the 1960s.

The second meta-study was published by Cedefop 7 years later (Descy & Tessaring, 2005). It summarizes 13 studies from eight countries which in the main covered research conducted in the late 1990s and early 2000s. The study revealed findings similar to those found by the OECD in terms of ‘stylized facts’. In particular, most of the studies confirmed that both workers and companies benefited from training. The two meta-studies also report that although most studies reported measurable effects from training, not all of them did. Conversely, productivity increases were observed also in the absence of explicit training or capital investments (the ‘Horndal effect’1). However, in companies where substantial changes had taken place, training invariably had productivity effects.

With these general findings as a background, the current literature presents several other interesting results. A perhaps somewhat unexpected result is that in a few cases, general training had stronger effects than specific training. As noted previously, this dichotomy is difficult to uphold but even on a sliding scale, there are results indicating that firm-specific training is becoming less common. As more complex jobs emerge, the generality of training increases (Harrison, 2000; Loewenstein & Spletzer, 1999; Strietska-Ilina, 2008). An additional factor is found in labour markets with a compressed wage structure, where it can be inferred that skilled workers are paid below their productivity, thus enabling the company to extract economic rents2 from their investment in training (Schøne, 2006).

Most, but not all, studies included in the meta-studies were able to establish causal links from training to productivity, and they could also demonstrate that training was profitable to the firm. Dearden and her colleagues constructed a database with information on (among other things) training from the UK Labour Force Survey and on production statistics (capital, wages, labour and output) from the Annual Census of Production, covering the period 1983–1996. They found a statistically and economically significant effect of training on productivity and on wages. Interestingly, about half of the productivity increase was reflected in wages, strengthening the result from the meta-studies that the gains from training were about equally divided between workers and companies (Dearden et al., 2005). Following a contemporary trend, Abdel-Wahab and his colleagues focus on profitability (rather than productivity) as the main outcome of training and assess the efficiency of public grants to firm-based training (Abdel-Wahab et al., 2008). In their study of the British construction industry, they match information from two separate databases on financial data and on training grants as a proxy for hard-to-get information on training. They find that training does increase profitability but only modestly.

Thus, studies at firm level indicate that training on average has positive economic effects. However, these effects are not universal. Training does not automatically increase productivity and increasing productivity does not necessarily translate into increased profitability. This could in part be attributed to the difficulties to actually observe the acquired skills (Pankhurst, 2010). Furthermore, even when increased skills seem to raise company productivity, a recent study indicates that this effect mainly refers to high-level qualifications. Vocational training has, according to this study, only a small impact on productivity (Galindo-Rueda & Haskel, 2005).

With these results in mind, it is hardly surprising that many companies do not systematically calculate or measure the costs and benefits or the training they provide. One common concept to assess effects from training is known as the Kirkpatrick Model where four levels have been developed: reaction, learning, behaviour and results. However, a review of the practice in the field revealed that only about 10 percent of the companies used all four levels and in fact 70 percent restricted their evaluation of training to level 1 (reaction, usually by using end-of-course feedback). The results confirmed and indeed almost mirrored a similar investigation from the late 1990s in the United States (Sutton & Stephenson, 2005). The inherent difficulties in measuring the effects, combined with the fact that training is often seen in a wider context of motivation and commitment, imply that a large number of companies (56 percent), according to surveys conducted by Eurostat, do not evaluate training effects (Gruber et al., 2009). The notion of a wider context presumably also explains the findings in several studies that it is the firms – not the workers – that pay for most training, including training of a more general character (Autor, 2001; Loewenstein & Spletzer, 1999; Schøne, 2006).

The question of net economic benefits from training is crucial also in more extensive programmes, in particular apprenticeship programmes that are based on long-term relationships between the apprentice and the enterprise. The interest in apprenticeship-based VET has increased since the early 1990s (this theme is developed below), but one of the factors that seem to be problematic in most countries is that employers, not least small and medium-sized enterprises, consider the costs too high (Billet & Smith, 2005). In Germany, where the major part of VET is apprenticeship-based and altogether about 50 percent of a youth cohort gets upper secondary education in a vocational programme (Kupfer, 2009), employers' costs have been a matter of concern for some time. In a study exploring firm-level data, Dionisius and her colleagues have compared the costs and benefits of apprenticeship training in Germany and Switzerland. The apprenticeship system is fairly similar in the two countries, but previous studies have shown that training is on average profitable during the training period for Swiss firms whereas German firms have a net cost for the training. These studies have been made at an aggregate level but Dionisius et al. (2009) have been able to match firm level data for a more detailed analysis. Their study shows quite convincingly that the strongest single factor to explain the difference is that Swiss apprentices are engaged in productive work more often than their German counterparts. To some extent, this is explained by the higher wage-level in Swiss firms, but in addition they suggest that the main reason for the differences is labour market regulation. The Swiss labour market is more flexible than the German labour market (and Swiss apprentices leave the training firm to a much higher rate than the German apprentices), which induces Swiss firms to make a productive use of the apprentices (Dionisius et al., 2009).

In conclusion, recent research basically confirms previous results that training on average increases productivity. There are two main reasons why these recent results are more convincing than the older ones: the accessibility of micro-level data (firms and individuals) has increased and the increasing use of sophisticated econometric techniques permits causality conclusions rather than correlation (for an informative yet accessible introduction, see Schlotter et al., 2010). However, the productivity effects are not universal; they are difficult (or costly) to identify and in many cases depend on institutional arrangements outside the firm. As a consequence, employers could find it rational not to invest in training or to invest only in programmes that are motivating but with uncertain productivity effects. Ever since the days of the early human capital literature in the early 1960s, researchers and politicians have expressed concern for possible underinvestment in training (Becker, 1964; Billet & Smith, 2003, 2005), referring primarily to a situation where the firm is undersupplied with skilled labour for fear of poaching from competitors. However, the observation that the labour market for skills is imperfect contributes to further complications (Smits, 2008). A wide range of factors, including motivation and commitment, operate to prevent trainees using, or using fully, what they have learned, hence preventing or weakening the training/increased productivity or profitability link. These imperfections imply that even skills with a high degree of generality are difficult to transfer. As a consequence, training in a firm that potentially could be very productive in another firm may not be transferred, and as a result we would observe only a small productivity increase from the training. That fear is even more understandable now because there are studies that show positive external effects from VET because it also increases productivity in other firms located in the proximity (Galindo-Rueda & Haskel, 2005). That phenomenon could be generalized to promoting not only productivity and economic growth but also socially desirable activities in general (Middleton et al., 1993). The problem of possible underinvestment is revisited in the section on policy implications.

Private and social returns to VET

As VET typically increases productivity at the company level, it seems plausible that it should be possible to observe a similar effect in macroeconomic development. The relation between education in general and economic growth is ambiguous and it is often difficult to establish a causal relation from education to economic growth because the mechanisms involved are complex and often indirect (Björklund & Lindahl, 2005). Education in literature, philosophy or history, for example, does not increase a person's measurable productivity much (with the possible exception of extended study that enables the person to take a position as university professor, for instance). However, such education may help to form ‘social capital’ that could be useful and productive in the long run (Bourdieu, 1986; Coleman, 1988; Putnam, 1993, 2000). VET, on the other hand, could be expected to affect economic growth also in the short run, because of its direct effects on productivity at the company level.

According to human capital theory, a high rate of return to investments in human capital could indicate that it also constitutes an engine of growth. Investment in education, in this case vocational, is assumed to increase labour productivity and, as a consequence, economic growth. It presupposes, however, functioning markets and in particular a functioning labour market. It is well known that labour markets are characterized by several imperfections, but in this section they are assumed to be almost perfect. As a consequence, a high rate of return on investments in VET indicates the promotion of growth.

There exists an abundance of estimates of rates of return to education, both private and social returns, for the last 40 or 50 years. Most of these studies, however, refer to levels of education (primary, secondary and higher). Comparatively few studies have been made of specific education sectors, including studies on returns to VET. In an OECD study, to take a much-used source, the estimates of rates of return to education, both private and social, refer only to various levels of education (OECD, 2005). VET is not a separate category. One rare exception in the recent literature is a study of Botswana, using a neoclassical standard growth model.3 In this study, enrolment in university and vocational and technical institutions, respectively, is used as variables for labour with different amounts of human capital. Even though the observation period is rather short (1974–2004, that is, 31 observations) the results are interesting. According to this study, vocational and technical education contributes far more to Botswana's fast growth than university education (Mupimpila & Narayana, 2009).

However, the study on Botswana is something of an exception in contemporary research. This is, to some extent, explained by lack of adequate data, but the main reason is conceptual difficulty. General and vocational education can, to a large extent, be substituted for each other, which renders it very difficult to assess the unique contribution of VET. A related problem concerns the screening or sorting effects of the education system. Returns to education are usually measured by wage differentials but parts of them are not caused by education and, as a consequence, they should be disregarded. The residual factors are usually combined into ‘ability’. The conceptual difficulty arises from the fact that individuals with differing ‘ability’ tend to sort themselves when they choose between various strands of education. An alternative interpretation is that the schooling system reproduces social inequalities in such a way that children from working class families are sorted into vocational rather than theoretical strands of education. Either way, the return from, for example, a VET programme, may reflect the original sorting rather than human capital accumulation during (and after) the training period.

In spite of the difficulties involved, some studies of rate of return to VET have been made in the past and the results from a number of them are reported in Psacharopoulos and Patrinos (2004). The number of studies is not very big – 17 altogether – and only one (fairly old) study refers to an industrialized country (France in 1977). The mean value of the private rates of return for technical/vocational education at the secondary level is 10.5 percent. The corresponding rate for academic/general education is 10.6. In other words, this meta-study indicates that the private returns to VET are equal to those of general secondary education.

That result in isolation is difficult to interpret because it refers to a small number of countries that are either ‘Newly Industrializing’ (Taiwan and Indonesia, for example) or decidedly ‘Developing’ (Botswana, Cameroon, Togo and others). The results are corroborated, however, by a study on Germany where the return to schooling was estimated for the period 1976–1987. Using a Mincerian schooling model,4 it was found that an additional year in school increased earnings by about 6 percent throughout the period. In the next step, dummy variables5 were used to identify the effect of apprenticeship training which was on average about 7 percent. The study is not directly comparable with the rates of return reported by Psacharopoulos and Patrinos (2004) – but the main result is that apprenticeship in Germany is economically beneficial to about the same extent as general education (Kau, 1998). In a series of studies on the apprenticeship system in England, McIntosh has found similar results. Two studies, using data for the period 1996–2005 from the Labour Force Survey, showed very similar results. Men with apprenticeship training had a rate of return of about 7–8 percent whereas those of women were substantially lower. These are overall figures; focusing more on the effects of the Modern Apprenticeship System, results were better, in particular for women (McIntosh, 2005, 2007). A third study, where different intermediate vocational qualifications were investigated, has estimated returns that were even higher, ranging from 6 to 16 percent (McIntosh, 2009).

Do these results imply that VET constitutes an ‘engine of growth’? Are the rates sufficiently high to motivate such a claim? Human capital literature does not provide any benchmarks even if there are examples where those rates have been compared with rates of return on investment in physical capital. This seems to be a rather dubious exercise, however, and in any case the rates for VET are not higher than those for general secondary education. From a human capital perspective, a better assessment can be made if social rates of return are also taken into consideration. Psacharopoulos and Patrinos (2004) provide estimates of the social rates of return for most of their sample that indicate that the social returns to VET are slightly lower than those to general secondary education (mean value 11.7 compared with 15.5). This is hardly a significant difference, in particular, as the study also demonstrates the huge variance in rates between different countries. That is mainly because of the different methodologies used in the various studies, but may also reflect some of the problems pointed out previously, particularly far from perfect conditions in the labour market. These issues are more or less standard reservations but there is one more point to be made.

A large number of people engage in VET but the overall result in terms of human capital production is, surprisingly enough, not that great. Calculations of a nation's stock of human capital where VET constitutes a separate category are very unusual, but Ljungberg and Nilsson (2009) have presented estimates for Sweden for the period 1870–2000. They indicate that VET represents no more than 1–3 percent of the total stock of human capital, with the peak at 3 percent in 1970. This estimate must be considered to be a lower boundary because the definition of VET is restricted to school-based forms at the secondary level. In-company training and apprenticeship training, not to mention on-the-job training, are not included in the estimates, but of these omissions probably only in-company training is important. In Sweden, apprenticeship training has not been an important part of the system of VET since the early twentieth century (Nilsson, 2008; Olofsson, 2005), and on-the-job training is never included in calculations of the stock of human capital because of the lack of reliable material. Thus, in the Swedish context where the bulk of VET is provided in schools at the secondary level, the lower boundary is probably not very far from the correct value. The conclusion is that VET has constituted only a minor part of the stock of human capital in Sweden over a long period of time and that this result probably also holds for several other countries.

In conclusion and with all the problems in mind, it is hardly surprising that it is difficult to find hard empirical evidence that VET has significant effects on economic development. The effects – if any – are probably small and difficult to detect. Thus, conclusions from the rate of return studies cannot be very strong. At most, they indicate that VET may constitute an engine of growth, but if so, that engine is not stronger than general secondary education.

Promotion of social inclusion

Problems with groups at risk in society seem to have increased in recent decades, although to some extent ongoing problems may now simply be better documented. These groups are heterogeneous and consist of, among others, early school leavers, low-skilled workers, the unemployed and immigrants. The reasons for being at risk are different for each group as well as for parts of a group (De la Fuente Anuncibay 2007). Immigrants, for instance, constitute a very heterogeneous group and the risks vary with cultural and linguistic distance between the country of origin and the new country (Dahlstedt & Bevelander, 2010). Thus, it would be superficial to believe that one set of policies could address the various problems (Preston & Green, 2008). The European Commission has identified six priorities as crucial to promote social inclusion. They include, in part, defensive measures such as the promotion of adequate social protection schemes, policies to reduce poverty among children, immigrants and ethnic minorities, and others. On the more proactive side, education in general and VET in particular are key elements in policies to counteract social exclusion. One active policy concerns the labour market but two priorities refer specifically to education and training (European Commission, 2007):

  • 1Improving access to education and lifelong learning by those most at risk of social exclusion
  • 2Making a concerted effort to prevent dropping out of school and promoting a smooth transition from school to the workplace

There are two major and related reasons why education is a particularly important policy issue in this context. The first is that the schooling system is a major venue for transmitting values, norms and codes of behaviour to young people, or in other words, in their socialization. The second reason, and this is where the training part is more relevant, is in the transfer from school to work. These two presumed effects are examined briefly in this and the following section.

Before the Second World War and in many countries well into the 1960s, schooling and the accompanying socialization process lasted for 6–8 years in primary school for the majority of the population. The main exception from this ‘rule’ was the United States, where the high-school system brought about a substantially higher general level of education (Goldin & Katz, 2008). In the present-day situation, however, this socialization is far from completed at the end of the compulsory schooling phase. The main challenge in the ‘elimination of all barriers to education and training’ (European Commission, 2007) is to counteract early school leaving among young people who are tired of school, as well as to induce them to continue their education at the (upper) secondary level – and perhaps further. In a recent report to the Commission, it is acknowledged that VET can ‘play a central role in tackling the problem of poor motivation and alienation’ (McCoshan et al., 2008). That challenge has a strong social component in that young people from homes with a weak academic tradition continue their education and training to a smaller extent than other social groups. Because social recruitment to vocational programmes is less skewed than to theoretical programmes at the secondary level, an expansion of the former groups of programmes should enhance the total recruitment, and hence socialization of young people.

However, this would not necessarily increase access to all types of vocational education. The report by McCoshan et al. (2008) underlines the diversity of VET systems in Europe and points at weaknesses that are of particular interest in this context. In the Mediterranean countries, employers generally take a limited part in the provision of VET, and in most Central and Eastern European countries, the rebuilding of former socialist systems is still troublesome. The role of VET is dependent on the social and political contexts in which it operates. In particular, involvement from employers (and trade unions) seems to be crucial if VET is to have the desired effects of promoting social inclusion. There is at present a trend across Europe to increase employer participation in VET, but this refers in particular to the opening up of pathways from initial training into post-secondary programmes, as well as into higher education. Promoting social inclusion requires further efforts.

In Europe, the best example of a VET system that actively promotes social inclusion is probably the ‘dual system’ (in Germany and in modified versions also in Austria, Denmark and Switzerland), which is said to teach not only technical and practical job skills but also social and personal behaviours (Deissinger, 2004). Since the 1990s, attempts have been made in several European countries to revitalize the apprenticeship system, not least because of its socializing effects. One of the more successful attempts was made in Ireland in 1993–1995, with a standards-based apprenticeship system in which the social partners were given a role similar to that in the ‘dual’ system. This introduction coincided with the famous boom of the Irish economy, however, which renders it difficult to have strong opinions about causality. One study claims that the new system facilitated the boom by providing central sectors, in particular the construction industry, with precisely the skills that were demanded (O'Connor, 2006). Ryan (2000), too, in a review of apprenticeships systems in smaller European Union (EU) countries, is very positive about the Irish system and holds that it demonstrates that institutions can be changed in a way that facilitates more efficient vocational education. Ryan's assessment is important, because the other ‘small EU countries’ that form his sample – Austria, Denmark and the Netherlands – have an institutional setting that resembles the German case, at least in terms of labour market institutions, whereas Ireland has ‘inherited’ its setting from Great Britain. In fact, Ryan's paper is as much an assessment of England's failure to adapt its institutions to support the apprenticeship system that was reformed in the 1990s.

That system, Modern (the word ‘modern’ was dropped in 2004) Apprenticeships, was launched in 1994 and included two main components: National Vocational Qualifications and Key Skills Certificate (literacy, numeracy and communication skills). Subsequently, a number of changes have taken place, including a division between Foundation and Advanced apprenticeships. The employer receives a subsidy to cover costs for employing a young person under this scheme (Brockmann et al., 2010). The English system has been meticulously investigated and in that process heavily criticized. In an interview-based study of large employers and focusing on the advanced component of the programme, Ryan and his colleagues conclude that the programme only to limited extent contributes to national educational objectives and that there is a large variation in the way the employers take responsibility for the programme (Ryan et al., 2006). However, Guile and Okumoto (2007) argue that the reason for the employer's reluctance to be involved in apprenticeship training is precisely the government's attempts to promote educational goals. They base their arguments on a study of apprentices in the ‘creative and cultural sector’, which is quite unusual in the literature. This sector, however, is dominated by small and medium-sized enterprises that almost certainly have a different perspective from large companies. An even more critical assessment is made by Brockmann and her colleagues in a paper based on official reports. They conclude that the present system is characterized by low status, poor standards, and in general, an undemanding route for low-attaining students. The educational component is low or completely absent and the partnership involvement is weak, in particular because the trade unions and the institutions for further education are excluded (Brockmann et al., 2010).

Some further examples can be found in Italy, Estonia and the Netherlands. In Italy, a major obstacle in the apprenticeship system was removed in 2003 by eliminating ‘dead ends’– at all levels it has become possible to continue studying at a higher level. In Estonia, a pilot project on apprenticeship became operational in 2004. It targets two groups, one of which is dropouts from school without a lower secondary education. That target group is precisely the one apprenticeship programmes are supposed to reach. In the Netherlands, the social partners and the Ministry of Education, Culture and Science signed an agreement that was formalized by law in 2003. The agreement includes learn–work trajectories (or unpaid apprenticeships) that imply a closer cooperation between regional labour market authorities, companies, and schools in order to reduce dropout rates and put more emphasis on working and learning in practice (examples can be found in the country reports in Cedefop, 2009a). These three cases demonstrate that various methods have been tried in different countries, because the experience from and status of apprentices differ. In part, the methods have been tried to reach those young people that enter neither vocational nor theoretical education. That group has been subject to a lot of attention because structural changes, at least in European countries, have more or less eroded their possibilities for entering the labour market without qualifications over and above comprehensive school (Sanders, 2005).

Recent research does not permit the drawing of strong conclusions concerning the effects of new initiatives in VET on social inclusion. This section has shown that (1) most new initiatives are based on a revival or reinforcement of an apprenticeship-based system, and (2) because the institutional settings differ between countries, the outcome of the new initiatives differs, too. The institutional settings are path-dependent (Thelen, 2004), which implies that it is difficult to change them, but not impossible, as the Irish case demonstrates. However, the Irish case includes some unanswered questions because the system was implemented during an unprecedented economic boom. How will it fare during the present economic crisis that has hit the Irish economy hard? In addition, the new system was restricted to a limited number of sectors, completely dominated by men. How would it behave if it was expanded into a more comprehensive system? The latter is in fact a concern of a general nature. In many of the countries that have reformed their systems, VET has traditionally been a marginal part of the education system. Erica Smith has, in a research paper on school-based VET in Australia, identified some important issues that emerge when a marginal system is being transformed into a mainstream one. Increasing costs and difficulties in recruiting competent teachers, as well as problems in integrating graduates into the labour market, are among the most important ones (Smith, 2004). Smith's study underlines that at least one conclusion can be drawn: merely changing (parts of) the VET system is insufficient to bring about significant changes in social inclusion. It is also necessary – but much more difficult – to bring about changes in the institutional setting.

A remedy for youth unemployment?

The concept ‘social inclusion’ is multifaceted and while the previous section mainly dealt with ways to stimulate enrolment in education, this section concerns the possibilities for young people to get access to the labour market. That is perhaps the most important prerequisite for social inclusion. Youth unemployment has become a major social and economic problem in the industrialized countries, with consequences such as underutilization of human capital, possible skill shortages in various sectors of the economy, low income among young people, and potentially increasing drug abuse and criminality (Björklund & Lindahl, 2005; Dettmann & Günther, 2009; Meager, 2009; Preston & Green, 2008; Strietska-Ilina, 2008). In many European countries, youth unemployment was below 5 percent in the early 1970s. It soared to unprecedented heights from the mid-1970s in countries such as France, the Netherlands and the United Kingdom. After some recovery, youth unemployment increased dramatically again from the early and mid-1990s even in countries that for a long time had exhibited low figures in this respect, such as Denmark and Sweden. The unemployment figures for young people have remained high (if fluctuating) ever since (Cedefop, 2009b, ch. 2)

VET has often been put forward as a particularly suitable remedy for youth unemployment. It is hardly an exaggeration to say that vocational training programmes, at least in the European countries, have been major recipients of funds designed for reduction in youth unemployment. The results are mixed, which is hardly surprising because the causes behind increasing youth unemployment and the subsequent difficulties to reduce it are complex and intertwined. In this context, the main interest lies in the effects of school-to-work mechanisms, but in a non-experimental situation, it is impossible to isolate this completely from the other major causes. For that reason, it is unsatisfactory to look at youth unemployment per se; instead, it is necessary to look at relative youth unemployment (that is, unemployment in the under 25 year age group compared with unemployment in the 25–74 year age group) as that largely cancels out differences in the macroeconomic development between countries.

In principle, VET enhances the possibilities of getting access to the labour market in several, though not mutually exclusive, ways. VET increases skills and competencies in young people, which make them more competitive in the labour market. However, as numerous studies have demonstrated that may not be the main reason why VET facilitates labour market entry. A separate and possibly additional effect is that at least some forms of VET makes young people ‘insiders’ rather than ‘outsiders’ in the labour market. That effect is usually associated with an apprenticeship system. Apprenticeship is the dominant form of VET only in a few European countries, particularly Austria, Denmark and Germany. These are characterized by a ‘dual system’ in which apprenticeship has a pivotal role. To see the extent to which these countries stand out in comparisons with other countries, relative youth unemployment has been calculated for a number of European countries in Figure 1. As Figure 1 shows, relative youth unemployment has generally exhibited a better development in these countries than the average of the 15 countries studied, but the pattern is not clear-cut. Relative youth unemployment has also been below the average in Ireland and in the Netherlands, two countries where the dual system is not a prominent feature. However, in these two countries, other forms of apprenticeship play an important role in the VET system (Ryan, 2000).

Figure 1.

Relative youth unemployment in ‘EU-15’, Austria, Denmark, Germany, Ireland and the Netherlands, 1988–2007.
Source: Eurostat, 2010.

This is an important reason for the present tendency in Europe to promote apprenticeship systems in countries where it has been playing only a minor role. In some cases, France for instance, apprenticeship has for a long time been a viable alternative to school-based vocational education, even if it is quantitatively much smaller. In other cases, such as Spain, apprenticeships are restricted to the handicraft sector and/or generally regarded as a second-best alternative by students and employers alike. At the same time, it should be pointed out that young people are not necessarily attracted to apprenticeships. France reports that the number of apprenticeship contracts has declined in the recent years, and in the Netherlands and Finland, the number has stagnated. Similar developments are also reported from countries with an established dual system. In Denmark and Germany, the number of apprenticeship contracts has diminished and the Federal Germany Ministry of Education has even expressed fears about the future of the system (this paragraph is based on Nilsson, 2007).

A crucial problem, however, is to ascertain causality between VET, irrespective of what form it takes, and improved chances to get access to the labour market. Steven McIntosh reports a positive association between vocational qualifications and being in employment in England, but also notes that the direction of causality is not clear (McIntosh, 2009). In a review article published almost 10 years ago, Ryan (2001) identified macroeconomic development, the ‘double skill bias’, labour market institutions and the school-to-work mechanisms as the major causes. He notes that these mechanisms seem to have had different effects in various countries, which implies that national institutions (not only in the labour market) are important. Later studies have confirmed this assessment. Budría and Telhado-Pereira, (2009), using a logit model6 on individual data for Madeira in Portugal, found that vocational training increases the probability of employment, in particular in the long run (2 years after the completion of the training programme). However, they also found that the effect is stronger with higher previous education, and in particular, that the effects are very strong for a person with a tertiary education prior to the training programme. In Sweden, with a VET system that is quite different from that in Portugal, Olofsson and Östh (2007), using longitudinal data, have shown that the 2-year school-based vocational programmes in the 1980s increased the graduates' chances of obtaining employment. However, when these programmes were extended to 3 years' duration (with an assumed human capital-enhancing effect), no change in the probability of obtaining employment took place. If anything, that probability was slightly lower for graduates from three years' programme in the first few years after graduation (Hall, 2009). A related problem is nicely demonstrated by a study of subsidized apprenticeship in the former East Germany. The subsidized programmes were previously open only to physically or mentally challenged persons as well as to disadvantaged young people. From the early 1990s, however, they were opened to other young people in East Germany, the so-called ‘market-disadvantaged adolescents’. Participants in the subsidized programmes had a worse outcome than their counterparts in ‘normal’ apprenticeship programmes (adjusting for personal characteristics), even though there was no discernable difference between the programmes (Dettmann & Günther, 2009).

However, the participants in the subsidized German programmes did get employed and the main problem in youth unemployment is that technological change, in combination with labour market rigidities has made post-16 education more or less a necessity for entry into the labour market (Cedefop, 2009b; Olofsson, 2005, ch. 2). Thus, the relative attractiveness of vocational training programmes for young people with a weak academic background should have helped many to a position in the labour market. It is highly probable that the selection mechanisms for VET are different from those for secondary schooling of a more general or theoretical nature. These mechanisms may to some extent reflect individual differences in ‘ability’, but to a larger degree reflect differences in social, economic and cultural backgrounds (Colley et al., 2003). VET has for a long time been considered to be a suitable means to provide young people from groups with weak academic traditions with education over and above compulsory school. This leads to a different perspective on the relation between VET and unemployment: the indirect route which implies that in most countries, vocational training has been supplemented with theoretical studies to facilitate a transfer to higher education.

It is well known that the transfer to higher education is socially and economically skewed and that young people from non-academic background are under-represented at this level. It is equally well documented that the risk of unemployment generally decreases with the level of education. Several methods have been employed, over the years, to induce young people in vocational training to proceed to higher education but the results have been mixed. In a summary paper, Tsakarissianos (2008) stresses that earlier reforms in comprehensive schools and a contemporary trend towards convergence between general and vocational education at the secondary level have increased vocational students' enrolment in higher education, but not by very much. For Sweden, Hall (2009) found that the extension of vocational programmes from 2 to 3 years' duration, a reform that was partly motivated with its potential to increase enrolment in higher education from vocational programmes, did not have that effect at all. Tieben and Wolbers (2010) demonstrate that several changes in the Dutch secondary school system have implied that the effect of parents' education on young peoples' choice of educational path at the tertiary level has diminished. The Netherlands have a multi-track structured educational system and the most crucial decision lies in the choice of the track at the secondary level. At this stage, the socio-economic background is still an important determinant (if slightly less so than in the past), but that effect is much smaller in the transition to tertiary education. Hodgson and Spours (2010) investigate a new instrument in the English system – the 14–19 Diplomas that are intended to prepare young people for both employment and higher education. These diplomas are being gradually introduced from 2008. Earlier attempts (General National Vocational Qualification and Advanced Vocational Certificate of Education) to achieve similar goals have failed and the analysis by Hodgson and Spours suggests that there is a great risk that the new Diplomas will fail, too. The uncertainty of the qualifications and the conservatism of higher education providers point in that direction. Aynsley and Crossouard (2010) depart from previous research which shows that participation in higher education from lower socio-economic groups is resilient to change in the United Kingdom, and try to establish the reasons for this. They find that students from vocational programmes have high aspirations but these seldom include the decision to proceed to higher education.

A straightforward conclusion from this part of the review is that VET does improve the chances of young people to gain access to the labour market. There are, however, a number of caveats and qualifications that are highlighted in recent research. First, the employment opportunities improve in relation to young people who did not enter, or more commonly, did not complete a secondary school/vocational programme. Inability or unwillingness to complete such a programme is equated with ‘low ability’ in the labour market, with difficulties in gaining employment as a result. In consequence, part of the employment effects of VET is because of selection effects. A second qualification is that employers' perception of VET is extremely important. ‘Modern’ apprenticeship in England, subsidized apprenticeship training in Germany, and the prolongation of school-based vocational programmes in Sweden probably implied improvements but they were not perceived as such by the employers, and as a result there were no discernible employment effects. The examples cited are from countries with quite different VET systems, which make the persistence in results even more interesting. Third, a finding and certainly a political conviction in many countries is that apprenticeship constitutes a superior form of VET in terms of future employment opportunities. The comparatively low relative youth unemployment in countries where apprenticeship is a dominant or at least an important part of the VET system certainly supports that notion. But as numerous researchers have pointed out, that result is obtained under specific institutional arrangements in the labour market as well as in the education system. Active involvement by employers and by trade unions is obviously a minimum requirement, but in addition it seems as if regulation of trades and professions is needed. And this is precisely where an ‘apprenticeship-ization’ of VET meets a real obstacle. The institutional changes that such regulations imply are difficult to implement. The success of the Irish case is to a large extent because of the fact that it was restricted to ‘regulated trades’, that is, to sectors where only minor institutional change had to take place. The other requirement, active involvement of employers and trade unions, seems in comparison a much easier task to accomplish.

Policy implications and suggestions for future research

Two facts seem clear concerning relations between VET and economic and social development. First, from various quarters, positive effects from VET are expected and, indeed, presumed. Second, it is difficult to verify causality in empirical studies. Thus, the question mark in the title of this review is certainly appropriate. There are several reasons for the discrepancy between expectations and verifiable results. Section two demonstrated that the complexity of the concept of VET implies that it often eludes attempts at quantification and hence assessment of its impact. Furthermore, in empirical studies, only parts of the system are identified, making an overall assessment of effects all but impossible. Empirical studies referring to effects on productivity and economic growth at micro and macro levels, respectively, give results that are not easy to reconcile with each other. Studies at the micro (firm) level generally indicate that VET raises productivity and that both employers and workers benefit from it. At the same time, many employers fail to assess the results of the training they offer as the benefits are often difficult and/or costly to estimate. In some cases, the uncertainty of effects makes employers hesitant to offer training to their employees. One line of future research is to further improve the basis for decision-making at the firm level. This is already an expanding field of research, but the accessibility of micro-level data has improved considerably in recent years and there is still room for more research on them, not least in combination with improving econometric techniques. There is a risk that the strict modelling required by these techniques results in a number of ‘special cases’ where the results are not easy to generalize, but this seems like a risk well worth taking. In addition, large bases of register data are emerging, which should be exploited more. Here, researchers have access to large populations rather than small samples, which allows for more definitive conclusions.

An even more interesting area for future research – at present largely unexplored – is the long-term effects of initial VET. This training consists of programmes (school-based, company-based, or ‘dual’) of 2–4 years' duration that constitutes a massive investment which is expected to pay off over a long time period. Knowledge of the extent to which that pay-off takes place is, however, very limited. With access to register data and other sources, it is possible to carry out longitudinal studies and/or panel data studies of a high quality that would increase our understanding of the long-term economic effects of VET.

Turning to policy implications, the present situation in in-company training suggests a situation of underinvestment which could imply that a policy to subsidize such training should be advocated. That, however, would amount to subsidies to specific companies, and in the absence of a better understanding of when and where such subsidies would have detectable effects for society as a whole, such a strategy is difficult to justify. The causality chain from productivity changes at firm level to macroeconomic effects is complicated. Increasing productivity in some workplaces need not have effects on the rest of the economy if: (1) the number of workplaces with increased productivity constitute a small share of the total number, (2) workplaces with increased productivity are separated from the rest of the economy by institutional or legal barriers, or (3) the labour market exhibits rigidities that hinder mobility between workplaces. Thus, policies to counteract rigidity in labour markets seem to be more important, as that would increase the probability of diffusing productivity effects.

Over the last 15 or 20 years, the assumption that VET promotes social inclusion has been emphasized more strongly than the possible growth effects. Policy implications addressing this assumption cannot avoid affecting the institutional setting for VET. Because the said implications vary substantially between countries, the relevance of the implications varies, too. With the gradual erosion and, in many European countries, virtual extinction of a youth labour market, post-16 education is a necessity for young people in the present-day situation. As not all young people are suited for – and indeed not the least interested in – theoretical schooling, programmes with a more or less pronounced vocational orientation are the obvious, and in most cases, the only, solution to the problem. This has created new challenges to VET. A large number of the students in vocational programmes are looking for training that contains a lot of practical elements. If programmes are constructed to satisfy their demand, it would probably enhance the socialization effects because it motivates young people to enter a programme and to follow their training through. On the other hand, such training may not lead very far in the labour market as the number of jobs such training qualifies for is small and diminishing. Thus, although beneficial in terms of socialization, such programmes would do little to facilitate the school-to-work transition. Vocational programmes with a greater potential to qualify for skilled jobs, on the other hand, tend to contain substantial theoretical components and there is a risk that they will not attract students who are tired of school. There is a potential conflict of interest in the assumptions concerning social inclusion.

In their attempts to resolve the said conflict, various countries have differed in their methods, following different models of VET, but above all they have tried to revitalize the apprenticeship system in various ways. The different degrees of success that these reforms have enjoyed lead to one specific policy recommendation, namely, systems should be created which encourage trade unions as well as employers to take an active part. This seems to be a minimum requirement for an apprenticeship-based approach to work but is not done easily (Taylor, 2006; Winterton, 2000). As Billet and Seddon (2004) have shown, building what amounts to a social partnership is time-consuming.

A further and much more controversial policy proposal is to accept the conflict of interest and construct what is in reality a two-tier system. This may seem like, and to some extent also is, a step backwards. It is striking how the different VET systems changed in a similar fashion around 1990 to respond to the challenges posed by the ‘Knowledge Economy’. Regardless of whether the system was school-based, firm-based or ‘dual’, the theoretical components of the system increased. That was presumably unavoidable: not only did these changes respond to labour market demands but they also improved the opportunities for the students to proceed to higher education (another prioritized ambition at the time). However, an unforeseen side effect was an increasing dropout rate, with immense difficulties for dropouts and non-starters to gain access to the labour market. A conclusion which is difficult to avoid is that the attempts to construct a system that attracts ‘low-achievers’, as well as providing young people with the skills that are demanded in the more competitive parts of today's labour market, have failed. In this situation, it seems wise to accept realities and put some effort in constructing a viable system rather than simply proceed in the present manner.

Looking closely, an often unplanned development towards a two-tier system is already taking place in many countries. In Germany, the traditional apprenticeship system is shrinking and increasingly leaving room only for ‘high-achievers’ whereas the ‘low-achievers’ are absorbed in courses that have little connection with the regular system for vocational training (Kupfer, 2009; Schmidt, 2010). The English apprenticeship system with a ‘foundation’ and an ‘advanced’ part is a more developed two-tier system. The Dutch system with four ‘pillars’ at the secondary level (theoretical and/or vocational) is from this point of view almost fully developed. It is admitted that the results of existing two-tier systems are not overwhelmingly positive. The subsidized former East German system, for example, led to fewer jobs and lower pay than the non-subsidized parts, but the crucial point is that it led to employment. In a similar fashion, it could be claimed that the ‘Foundation’ part of the English system is fairly successful in enrolment and that it leads to employment. Facilitating access to the labour market is a key aspect of vocational education: it is undoubtedly a social good and it also affects the ‘softer’ aspects of growth. To the extent that VET increases the chances of young people entering the labour market, it enhances social equality, and that, in line with an important research tradition, may also facilitate economic growth (World Bank, 2006).

The lower part of a two-tier apprenticeship-based system could possibly be made to work if social partnerships could be constructed and if it was (possibly quite heavily) subsidized by the State. It seems highly unlikely that employers would be willing to carry the full financial responsibility for training programmes directed at ‘low-achievers’. It would be a bigger challenge to make the upper part successful, as the English experience, with stagnation in the enrolment rate in the ‘advanced’ part of the system, suggests. Institutional changes, including regulation and/or certification in the labour market, are presumably necessary to make that part of the system work (Spielhofer & Sims, 2004). As the Swiss experience shows, such regulation need not be very comprehensive but it would nevertheless imply institutional change that would require a strong political commitment. Even with such support, the inherent difficulties could be hard to overcome, as numerous examples show (Turbin, 2001). This leads to a final suggestion for future research. Our knowledge of institutional change is limited. One of the things we do know is that it is a time-consuming process. Therefore, historically based research on institutions directed at examples of successful as well as unsuccessful attempts to achieve institutional change is urgently needed.

In conclusion, it can be observed that the renewed interest in VET, partly reflected in the large body of papers published in recent years that forms the basis for this review, is a promising development. There are still large areas where our knowledge is limited, including when, how and where company-based training has an effect on productivity, the long-term economic and social effects of initial VET, and circumstances under which institutional change is likely to take place. It is urgent that these and related areas are subject to further research, because VET is crucial to social inclusion and possibly also to economic growth. This is particularly the case in developing countries (Blunch & Castro, 2007). VET, as we know it, is a response to challenges posed during the Second Industrial Revolution. Many developing countries today are in a similar position, and from this perspective the suggested research agenda has a particular relevance to these parts of the world.


  • 1

    The ‘Horndal effect’ denotes incremental and undocumented improvements in productivity and was described (in Swedish) by Professor Erik Lundberg. It refers to a sustained productivity increase in the Horndal steel mill in the 1930s and 1940s. The concept is discussed by, among others, Lazonick and Brush (1985).

  • 2

    Economic rent is an often used but not always well-defined concept. It denotes that an agent receives an excess return to an input, the excess return being derived from external factors. In this context, a compressed wage structure implies that enterprises pay skilled workers below their (marginal) productivity and as a consequence enterprises receive an excess income (or rent). For a readable account of the concept of economic rent, see Tollison (1982).

  • 3

    In neoclassical standard growth models, economic growth is ‘explained’ by the input of production factors (capital, natural resources and labour). The early models had a low explanatory power, but by disaggregating ‘labour’ into various components with different amounts of human capital, the models exhibit a better fit (that is, they are better at ‘explaining’ economic growth).

  • 4

    In a Mincerian schooling model earnings are ‘explained’ by the number of years in school and by the number of years of experience. This implies that earnings are determined by the amount of human capital an individual acquires in school and by learning-by-doing in the workplace.

  • 5

    Dummies are used to separate, for instance, separate parts of schooling. Individuals with the relevant schooling are given the value ‘1’, all others are given ‘0’. By calculating separate equations, the effect of the dummy is estimated.

  • 6

    Logit models are used to predict the probability that an event will occur by fitting data to a logistic curve (or logit function). Several explanatory variables are used in the prediction: in this case age, educational background and vocational training are used to predict the probability of employment. Gender is also an explanatory variable but due to the small number of women there is not a separate estimate.